Page 11234..10..»

Category : Protein Folding

A novel system to map protein interactions reveals evolutionarily conserved immune evasion pathways on transmissible cancers – Science Advances

INTRODUCTION

Metastatic cancer affects most mammals, but the cancer incidence can vary widely across phylogenetic groups and species (Fig. 1 and table S1) (13). In humans, the lifetime risk of developing cancer is around 40% (4). This figure is in stark contrast to a general cancer incidence of 3% for mammals, 2% for birds, and 2% for reptiles reported by the San Diego Zoo (N = 10,317) (2, 5). A more recent study at the Taipei Zoo reported cancer incidence of 8, 4, and 1% for mammals, birds, and reptiles, respectively (N = 2657) (6). Cancer incidence in domestic animals is generally less than 10% (N = 202,277) (3). However, two studies performed 40 years apart reported that greater than 40% of Tasmanian devils develop spontaneous, often severe neoplasia in their lifetime (5, 7). Devils are also unique because they are affected by two of the three known naturally occurring transmissible cancers in vertebrate species (8, 9). Transmissible cancers are a distinct form of cancer in which the tumor cells function as an infectious pathogen and an allograft. Dogs (Canis lupus familiaris) are the only other vertebrate species affected by a transmissible cancer (10), and interestingly, some breeds of dogs also have high cancer incidence (3, 11).

Metastatic cancer has been reported in nearly all mammalian orders, and MHCs have been the most intensely studied molecules in most orders. In the past decade, studies of immune checkpoint molecules (PD1, PDL1, and CTLA4) have become a primary focus in humans and rodents. However, immune checkpoint studies in other species are limited, particularly at the protein level, because of the lack of species-specific reagents. This creates a vast gap in our understanding of the evolution of the mammalian immune system. The numbers in the columns represent the number studies matching Web of Science search results between 2009 and 2019. See table S1 for search terms.

The devil facial tumor (DFT) disease was first detected in Northwest Tasmania and has been a primary driver of an 80% decline in the wild Tasmanian devil population (8, 12). The clonal DFT (DFT1) cells have been continually transmitted among devils and are estimated to have killed at least 10,000 individuals since at least 1996. In 2014, a second independent transmissible Tasmanian DFT (DFT2) was found in wild devils (9), and 23 cases have been reported to date (13). Genetic mismatches, particularly in the major histocompatibility complex (MHC) genes, should lead to rejection of these transmissible tumors. Consequently, the role of devil MHC has been a focus of numerous studies (Fig. 1 and table S1) to understand the lack of rejection of the transmissible tumors. These studies have revealed that the DFT1 cells down-regulate MHC class I (MHC-I) expression (14), a phenomenon observed in many human cancers. In contrast to DFT1 cells, the DFT2 cells do express MHC-I (15). DFT1 and DFT2 cells have 2884 and 3591 single-nucleotide variants, respectively, that are not present in 46 normal devil genomes (16). The continual transmission of DFT1 and DFT2, despite MHC-I expression by DFT2 cells and genetic mismatches between host and tumor, suggests that additional pathways are likely involved in immune evasion.

Human cancer treatment has been transformed in the past decade by manipulating interactions among immune checkpoint molecules. These have proven broadly effective in part because they function across many different MHC types and tumor mutational patterns. However, these pathways have received little attention in transmissible cancers and other naturally occurring cancers in nonmodel species (Fig. 1 and table S1) (1719). We have previously shown that the inhibitory immune checkpoint molecule programmed death ligand 1 (PDL1) is expressed in the DFT microenvironment and is up-regulated by interferon- (IFN-) in vitro (17). This finding led us to question which other immune checkpoint molecules play a role in immune evasion by the transmissible cancers and the devils high spontaneous cancer incidence. Understanding immune evasion in a natural environment will support DFT vaccine development to help protect this endangered species (20) and has the potential to identify protein interactions that are conserved across divergent species to improve translational success of animal models (19). Unfortunately, a persistent limitation for immunology in nontraditional study species is a lack of species-specific reagents. Wildlife biologists and veterinarians are at the front lines of emerging infectious disease outbreaks, but they often lack species-specific reagents to fulfill the World Health Organizations call for cross-cutting R&D preparedness and perform mechanistic immunological investigations.

To solve the paucity of reagents available for Tasmanian devils and address ongoing limitations for nontraditional study species, we developed a Fluorescent Adaptable Simple Theranostic (FAST) protein system that builds on the diverse uses of fluorescent proteins previously reported (2123). This simple system can be used for rapid development of diagnostic and therapeutic (i.e., theranostic) immunological toolkits for any animal species (Fig. 2). We demonstrate the impact of the FAST system by using it to confirm seven receptor-ligand interactions among 12 checkpoint proteins in devils. We demonstrate the versatility of the system across species by fusing a fluorescent reporter to a well-characterized camelid-derived nanobody that binds human PDL1 (24).

(A) Schematic diagram of FAST protein therapeutic and diagnostic (i.e., theranostic) features. POI, protein of interest. (B) Graphic overview of FAST protein system including key steps: (i) characterize gene of interest (GOI) in silico; (ii) design expression vectors; (iii) digest FAST base vectors and insert alternative genes of interest or colors; (iv) transfect FAST vectors into mammalian cells and monitor using fluorescent microscopy or flow cytometry; (iv) purify the protein using 6xHis tag, visualize fluorescent color to show that protein is in frame and correctly folded. Image of microfuge tubes shows 100 l of mCitrine, mOrange, and mCherry FAST proteins (1 mg/ml) excited with blue light with an amber filter. Full protocols for vector construction and protein testing are available in the Supplementary Materials. (C) Results of flow cytometry binding assay with devil 41BB FAST proteins. The colored lines in the histograms show binding of devil 41BB fused to mTagBFP, mCerulean3, mAzurite, mCitrine, mOrange, mCherry, or mNeptune2 to CHO cells transfected with devil 41BBL, and the black lines show binding to untransfected CHO cells. FSC, forward scatter; SSC, side scatter.

In humans, checkpoint proteins have been targets of immunotherapy in clinical trials, but the functional role and binding patterns of these proteins are unknown for most other species. We have used the FAST system to show that the inhibitory checkpoint protein CD200 is highly expressed on DFT cells, opening the door to single-cell phenotyping of circulating tumor cells (CTCs) in devil blood. Furthermore, we are the first to report that coexpression of CD200R1 can block surface expression of CD200 in any species. Understanding how clonal tumor cells graft onto new hosts, evade immune defenses and metastasize within a host will identify evolutionarily conserved immunological mechanisms to help improve cancer, infectious disease, and transplant outcomes for human and veterinary medicine.

Initially, we developed FAST proteins to determine whether monomeric fluorescent proteins could be fused to devil proteins and secreted from mammalian cells (Fig. 2A and table S2). We used 41BB (TNFRSF9) for proof-of-concept studies by fusing the extracellular domain of devil 41BB checkpoint molecule to monomeric fluorescent proteins (Fig. 2, A and B, and fig. S1). We used wild-type Chinese hamster ovary (CHO) cells and CHO cells transfected with 41BBL (TNFSF9) to confirm specificity of the 41BB FAST proteins and demonstrate that the fluorescent proteins [mTag-blue fluorescent protein (BFP), mCerulean3, mAzurite, mCitrine, mOrange, mCherry, and mNeptune2] remained fluorescent when secreted from mammalian cells (Fig. 2C).

We chose mCherry, mCitrine, mOrange, and mBFP for ongoing FAST protein development. Initial batches of FAST proteins were purified using the 6xHis tag and eluted with imidazole. Following purification of FAST protein the color can be immediately observed with blue light and an amber filter unit, allowing confirmation that the fluorescent protein DNA coding sequences were in frame and the proteins were properly folded. After combining, concentrating, and sterile filtering the eluted fractions, 100 l at 1 mg/ml was aliquoted and visualized again using blue light to confirm fluorescent signal (Fig. 2B). A full step-by-step protocol and set of experimental templates for creating and testing FAST proteins for any species are available online in the Supplementary Materials.

We chose candidate immune checkpoint molecules for FAST protein development (Fig. 3A and table S2) based on targets of human clinical trials and then selected devil genes for which a reliable sequence was available either in the published devil genome or transcriptomes (19, 25, 26). We transfected the FAST protein expression vectors (table S3) into CHO cells and tested the supernatant against CHO cell lines expressing full-length receptors. 41BB FAST proteins in supernatant exhibited strong binding to 41BBL cell lines, but the fluorescent signals from most other FAST proteins were too weak to confirm binding to the expected receptors (fig. S2). As FAST proteins do not require secondary reagents, we next incubated target cells with purified FAST proteins and added chloroquine to block the lysosomal protein degradation pathway. This allowed us to take advantage of receptor-mediated endocytosis, which can allow accumulation of captured fluorescent signals inside the target cells (27). This protocol adjustment allowed confirmation that CD47-mCherry, CD200-mBFP, CD200-mOrange, CD200R1-mBFP, and CD200R1-mOrange, and PD1-mCitrine bound to their expected receptors (Fig. 3B). We also demonstrated the flexibility of the FAST proteins by showing that alternative fusion conformations (fig. S1, C and D), such as type II proteins (e.g., mCherry-41BBL) and a devil Fc tag (e.g., CD80-Fc-mCherry) bound to their expected ligands (Fig. 3B). The stability of the fusion proteins was demonstrated using supernatants that were stored at 4C for 2 months before use in a 1-hour live-culture assay with chloroquine (fig. S3).

(A) Diagram of soluble FAST proteins and full-length proteins used for testing of FAST proteins. 41BBL is a type II transmembrane protein; all other proteins are type I. CD80 and CTLA4 soluble FAST proteins included a devil immunoglobulin G (IgG) Fc tag. Arrows indicate interactions confirmed in this study. TNF, tumor necrosis factor. (B) Histograms showing binding of FAST proteins to CHO cells expressing full-length devil proteins. Target CHO cells were cultured with chloroquine to block lysosomal degradation of FAST proteins and maintain fluorescent signal during live-culture binding assays with purified FAST proteins (2 g per well) for 30 min or 18 hours to assess receptor-ligand binding (N = 1 per time point).

To further streamline the reagent development process, we next took advantage of the single-step nature of FAST proteins (i.e., no secondary antibodies or labels needed) in live-cell coculture assays (Fig. 4A). Cell lines secreting 41BB-mCherry, 41BBL-mCherry, or CD80-Fc-mCherry FAST proteins were mixed with cell lines expressing full-length 41BB, 41BBL, or CTLA4-mCitrine and cocultured at a 1:1 ratio overnight with chloroquine. Singlet cells were gated (Fig. 4B) and binding of mCherry FAST proteins to carboxyfluorescein diacetate succinimidyl ester (CFSE) or mCitrine-labeled target cells was analyzed (Fig. 4C). The strongest fluorescent signal from 41BB-mCherry, 41BBL-mCherr, and CD80-Fc-mCherry was detected when cocultured with their predicted receptors, 41BBL, 41BB, and CTLA4, respectively.

(A) Schematic of coculture assays to assess checkpoint molecule interactions (absent, weak, and strong). Cells were mixed and cultured overnight with chloroquine. Protein binding and/or transfer were assessed using flow cytometry. (B) Gating strategy for coculture assays. (C) CHO cells that secrete 41BBL-mCherry, 41BB-mCherry, or CD80-Fc-mCherry were cocultured overnight with target CHO cells that express full-length 41BB, 41BBL, or CTLA4. 41BB and 41BB-L were labeled with CFSE, whereas full-length CTLA4 was directly fused to mCitrine. Cells that secrete mCherry FAST proteins appear in the upper left quadrant. Cells expressing full-length proteins and labeled with CFSE or mCitrine appear in the lower right quadrant. Cells in the upper right quadrant represent binding of mCherry FAST proteins to full-length proteins on carboxyfluorescein diacetate succinimidyl ester (CFSE) or mCitrine-labeled cells. Results shown are representative of n = 3 per treatment. (D) CTLA4-Fc-mCherry FAST protein binding to DFT cells. DFT1 C5065 cells transfected with control vector (black), 41BB (gray), CD80 (red), or CD86 (blue) were stained with CTLA4-Fc-mCherry supernatant with chloroquine. Results are representative of N = 2 replicates per treatment.

The fluorescent binding signal of CD80-Fc-mCherry was lower than expected, so we next reexamined our Fc tag construct. In humans and all other mammals examined to date, the immunoglobulin G (IgG) heavy chain has glycine-lysine (Gly-Lys) residues at the C terminus; the initial devil IgG constant region sequence available to us had an incomplete C terminus, and thus, our initial CD80-Fc-mCherry vector did not have the C-terminal Gly-Lys. We subsequently made a new FAST-Fc construct with CTLA4-Fc-mCherry, which exhibited strong binding to both CD80 and CD86 transfected DFT cells (Fig. 4D).

Analysis of previously published devil and DFT cell transcriptomes suggested that CD200 mRNA is highly expressed in DFT2 cells and peripheral nerves, moderately expressed in DFT1 cells, and lower in other healthy devil tissues (Fig. 5A) (25, 26, 28). As CD200 is an inhibitory molecule expressed on most human neuroendocrine neoplasms (29), and both DFT1 and DFT2 originated from Schwann cells (26, 30), we sought to investigate CD200 expression on DFT cells at the protein level. Staining of wild-type DFT1 and DFT2 cells with CD200R1-mOrange FAST protein showed minimal fluorescent signal (Fig. 5B). However, overexpression of CD200 using a human EF1 promoter yielded a detectable signal with CD200R1-mOrange binding to CD200 on DFT1 cells. A weak signal from CD200-mOrange was detected on DFT1 cells overexpressing CD200R1 (Fig. 5B). To confirm naturally expressed CD200 on DFT cells, we digested CD200 and 41BB FAST proteins using tobacco etch virus (TEV) protease to remove the linker and fluorescent reporter. The digested proteins were then used to immunize mice for polyclonal serum production. We stained target CHO cell lines with preimmune or immune mouse sera collected after three-times immunizations. Only the immune sera showed strong binding to the respective CD200 and 41BB target cell lines (Fig. 5C). After the final immunization (four times), we collected another batch of sera and tested it on DFT1 and DFT2 cells (Fig. 5D). In agreement with the transcriptomic data for DFT cells (25), the polyclonal sera revealed high levels of CD200 on DFT cells, but low levels of 41BB.

(A) GOIs for this study are plotted as a log2-transformed transcripts per million (TPM) heat map with dark blue indicating the most highly expressed genes. Technical replicates (N = 2) from separate flasks were used for the cell lines (C5065, RV) and biological replicates (N = 2) were used for primary tissues, except peripheral nerve (PN) (N = 1). (B) Wild-type DFT1.C5065, DFT2.JV, DFT2.SN, and DFT1.C5065 transfected to overexpress CD200 or CD200R1 were stained with either CD200R1-mOrange or CD200-mOrange FAST protein. Histograms filled with blue or red highlight expected strong binding interactions. The percentage of events that falls within the marker is shown. Results are representative of N = 2 replicates per treatment. (C) Mice were immunized with 41BB or CD200 FAST proteins. Black, preimmune; gray, immune sera from a mouse immunized with 41BB; red, preimmune; blue, immune sera from a mouse immunized with CD200. CHO cells transfected with either full-length 41BB or CD200 were stained with sera and then anti-mouse AF647. Results are representative of N = 2 per treatment. (D) Sera were used to screen two strains each of DFT1 and DFT2 cells for 41BB and CD200 expression. Results are representative of N = 3 per treatment. (E) DFT1 C5065 transfected with either vector control, CD200, or CD200R1 was stained with purified polyclonal anti-CD200 and anti-mouse IgG AF647 (black, no antibodies; red, secondary antibody only; blue, primary and secondary antibody). Results are representative of N = 2 per treatment.

In humans, overexpression of some checkpoint proteins can block surface expression of heterophilic binding partners in cis (e.g., CD80 and PDL1) (31). As a potential route for disrupting the inhibitory effects of CD200 on antitumor immunity, we tested whether overexpression of CD200R1 on DFT cells could reduce CD200 surface expression. We stained a DFT1 strain, C5065, and DFT1 C5065 cells transfected to overexpress CD200 or CD200R1 with polyclonal anti-CD200 sera and secondary anti-mouse IgG Alexa Fluor 647 (AF647). We detected no surface protein expression of CD200 DFT1 cells overexpressing CD200R1 (Fig. 5E).

In addition to high expression of CD200 on neuroendocrine neoplasms (29), CD200 is used as a diagnostic marker for several human blood cancers (32). DFT cells metastasize in the majority of cases (33), and our transcriptome results (Fig. 5A) suggest that CD200 mRNA is more highly expressed in DFT cells than in peripheral blood mononuclear cells (PBMCs) (25, 26). As a result, we tested whether CD200 could be used to identify DFT cells in blood. We stained PBMCs and DFT2 cells separately with polyclonal anti-CD200 sera and anti-mouse AF647 and then analyzed CD200 expression by flow cytometry (fig. S4A). We then mixed the stained PBMCs and DFT2 cells at ratios of 1:10 (fig. S4A) and 1:5 (fig. S4B) and analyzed the mixed populations. PBMCs showed minimal CD200 expression and background staining (fig. S4), whereas CD200 was highly expressed on DFT2 cells. CD200+ DFT2 cells were readily distinguishable from PBMCs.

As our RNA sequencing (RNA-seq) results only included mononuclear cells, we next performed a pilot test to determine whether DFT cells could be spiked into whole devil blood and identified via flow cytometry using CD200 staining. DFT1 and DFT2 cells were labeled with CellTrace violet (CTV), and 10,000 cells were diluted directly into 100 l of whole blood from a healthy devil (N = 1 per treatment; n = 1 devil). The cells were then stained with purified polyclonal anti-CD200 with and without secondary anti-mouse IgG AF647 before red blood cell (RBC) lysis. Initial results showed that DFT2 cells expressed CD200 above the leukocyte background but that DFT1 cells could not be distinguished from leukocytes (fig. S5). To eliminate the secondary antibody step from the whole blood staining protocol, we next labeled the polyclonal anti-CD200 and normal mouse serum (NMS) with a no-wash Zenon mouse IgG AF647 labeling reagent (n = 1 per treatment; n = 2 devils). This system again showed that CD200 expression could be used to identify DFT2 cells in blood (Fig. 6, A to E), suggesting that CD200 is a candidate marker for identification of metastasizing DFT2 cells.

Color dot plots showing DFT cells in green (CFSE), PBMCs in black, DFT Alexa Fluor 647+ (AF647+) cells in red, and PBMC AF647+ in blue. Forward- and side-scatter plot of DFT2.JV cells alone (A) and DFT2.JV cells mixed with PBMCs (B). (C) Color dot plot showing dead cells stained with 4,6-diamidino-2-phenylindole (DAPI) (right quadrants) and CFSE-labeled DFT cells (upper quadrants). (D) The top row shows unmixed PBMCs. The middle row and bottom row show PBMCs mixed with DFT1.C5065 (middle) and DFT2.JV (bottom) cells. Cell mixtures were either untreated or incubated with Zenon AF647labeled NMS or Zenon AF647labeled -CD200 serum. AF647+ DFT (red) and PBMC (blue) are in the right quadrants. (E) Histogram overlays to highlight AF647+ (right quadrants) from DFT1-PBMC and DFT2-PBMC mixtures. Cells were analyzed on the Beckman Coulter MoFlo Astrios. (F) FAST nanobody proof of concept was accomplished using supernatant from untransfected ExpiCHO cells or ExpiCHO cells secreting human antiPDL1-mCitrine nanobody. Nanobody supernatant was used undiluted or at 1:10 or 1:100 dilutions in media and used to stain CHO cells that express either human PDL1 or human CD80. Results are representative of N = 2 per treatment.

Last, to test whether the FAST system could be applied to other species (e.g., camelid-derived nanobody) and applications (FAST nanobody), we reverse-translated the protein sequence for an anti-human PDL1 nanobody (24) and inserted the codon-optimized DNA sequence into a FAST mCitrine vector. The assembled plasmid was transfected into ExpiCHO cells, and the supernatant was tested for binding to CHO cells stably transfected with either full-length human PDL1 or human CD80; the human proteins were fused to miRFP670 (Addgene no. 79987) in a FAST vector. The nanobody supernatant was used undiluted or at 1:10 or 1:100 dilutions. The nanobody showed strong binding to PDL1-expressing cells, but not CD80-expressing cells (Fig. 6F).

Naturally occurring cancers provide a unique opportunity to study immune evasion and the metastatic process across diverse hosts and environments. The exceptionally high cancer rate in Tasmanian devils coupled with the two transmissible tumors currently circulating in the wild warrants a thorough investigation of the devil immune system. However, taking advantage of these natural disease models has been out of reach for most species because of a lack of reagents. The FAST protein system that we developed here is well suited to discovering additional DFT markers and, more generally, to filling the reagent gap for nontraditional species. For proteins like 41BB that have high affinity for 41BBL, FAST proteins can be used as detection reagents directly from supernatant. For other molecules with lower receptor-ligand affinity, the FAST proteins can be purified, digested with a protease to remove the nontarget proteins, and used for production of higher-affinity binding proteins (e.g., antibodies, aptamers, and nanobodies).

The versatility of the FAST system was demonstrated by fusing a validated human anti-PDL1 nanobody derived from a camel (Camelus bactrianus) heavy-chain variable region to mCitrine. The nanobody-reporter fusion allowed direct testing of the nanobody from supernatant without the need for purification or secondary labeling and provided a 1:1 ratio of nanobody and reporter to allow quantification of target proteins. In addition to fusing nanobodies to fluorescent proteins, fluorescently labeled target proteins could be used with nanobody display libraries to pull down or sort nanobodies that bind the target protein.

The simple cut-and-paste methods for vector assembly lend the FAST protein system to entry-level immunology and molecular biology skill sets. In addition, the ability of FAST proteins to be used in live coculture assays and with elimination of secondary reagents will increase efficiency and reduce experimental error for advanced human and mouse cancer immunology studies. For example, previous high-throughput studies have used a two-step staining process (i.e., recombinant protein and secondary antibody) to screen more than 2000 protein interactions (34); this type of assay can be streamlined using FAST proteins to eliminate the need for secondary antibodies. Fc tags or other homodimerization domains can be incorporated into FAST proteins to increase binding for low-affinity interactions and to assess potential Fc receptormediated functions.

Production of recombinant proteins in cell lines that closely resemble the physiological conditions of the native cell type (i.e., mammalian proteins produced in mammalian cell lines) is more likely to yield correct protein folding, glycosylation, and function than proteins produced using evolutionarily distant cell lines. The fluorescent fusion proteins developed here take advantage of natural receptor expression and cycling processes (e.g., CTLA4 transendocytosis) in eukaryotic target cells; bacterial protein production methods are not amenable to coculture with eukaryotic target cells in immunological assays. Our demonstration of the FAST protein system in CHO cells suggest that this method can be efficiently integrated into existing research and development pipelines for humans and other vertebrate species.

A primary question in transmissible tumor research is why genetically mismatched cells are not rejected by the host. Successful infection of devils with DFT cells relies on the ability of the tumor allograft to evade and manipulate host defenses. The missing-self hypothesis suggests that the lack of constitutive MHC-I expression on DFT1 cells should lead to natural killer (NK) cellmediated killing of the allograft tumor cells. Here, we used the FAST protein system to develop a tool set to address this question and show that DFT1 and DFT2 cells express CD200 at higher levels than most other devil tissues examined to date. CD200 has been shown to directly inhibit NK cells in other species (35), so overexpression of CD200 is a potential mechanism of immune evasion of NK responses by DFT cells.

We hypothesize that CD200 could be particularly important in DFT transmission as the CD200-CD200R pathway is critical to the initial stages of establishing transplant and allograft tolerance in other species (36). In line with this hypothesis, a recent study reported that overexpressing several checkpoint molecules, including CD200, PDL1, and CD47, in mouse embryonic stem cells could be used to generate teratomas that could establish long-term allograft tolerance in fully immunocompetent hosts (37). We have previously reported that PDL1 mRNA and protein are up-regulated on DFT2 cells in response to IFN- (17), and our transcriptome results show that CD47 is expressed at moderate to high levels in DFT cells. Here, we show that overexpression of CD200R1 on DFT1 eliminates binding of our polyclonal anti-CD200 antibodies, suggesting that DFT cells overexpressing CD200R1 could be used to test the role of CD200 in allograft tolerance. Alternatively, genetic ablation of CD200 in DFT cells could be used as a complementary approach to examine the role of immune checkpoint molecules in DFT allograft tolerance. Low MHC-I expression is a primary means of immune evasion by DFT1 cells, and disrupting the CD200-CD200R1 pathway could facilitate improved recognition of DFT1 cells by CD8 T cells by enhancing IFN-mediated MHC-I up-regulation. Recent work in mice has identified immunosuppressive natural regulatory plasma cells that express CD200, LAG3, PDL1, and PDL2; we have previously identified PDL1+ cells with plasma cell morphology near or within the DFT microenvironment (17).

Previous DFT vaccine efforts have used killed DFT cells with adjuvants (38, 39). A similar approach to treat gliomas in dogs reported that tumor lysate with CD200 peptides nearly doubled progression-free survival compared to tumor lysate alone (40). Like devils, several breeds of dog are prone to cancer, and these genetically outbred large animal models provide a fertile ground for testing cancer therapies. The CD200 peptides are reported to provide agonistic function through CD200-like activation receptors (CD200R4) rather than by blocking CD200R1 (40). The functional role of CD200-CD200R pathway in devils remains to be elucidated, but the CD200R1NPLY inhibitory motif and key tyrosine residues are conserved in devil CD200R (19, 41, 42), demonstrating that this motif is conserved over 160 million years of evolutionary history (43). In addition to agonistic peptides, several other options for countering CD200-CD200R immune inhibition are possible. Human chronic lymphocytic leukemia cells often express high levels of CD200, which can be down-regulated in response to imiquimod (44). Likewise, we have previously shown that DFT1 cells down-regulate expression of CD200 mRNA in vitro in response to imiquimod treatment (25). In one of the longest running and most in-depth studies of host-pathogen coevolution, CD200R was shown to be under selection in rabbits in response to a myxoma virus biocontrol agent (45). As DFT1 and DFT2 have been circulating in devils for more than 20 and 5 years, respectively, it will be important to monitor CD200/R expression and the potential evolution of paired activating and inhibitory receptors in these natural disease models.

Immunophenotyping and single-cell RNA-seq of CTCs have a potential to identify key gene expression patterns associated with metastasis and tissue invasion. CD200 is a potential marker for the identification of CTCs from devil blood. As proof of concept, DFT2 cells could be identified in devil blood spiked with DFT2 cells. As CTCs are likely to be rare in the blood of most infected devils, CD200 alone would be insufficient for identifying DFT1 cells. Additional surface DFT markers would be required to purify CTCs for metastases and tissue invasion analyses. The FAST protein system provides a simple procedure to facilitate the production of a panel of DFT markers to help identify key proteins in the metastatic process.

In summary, the simple cut-and-paste production of the vectors and single-step testing pipeline of the FAST system provided multiple benefits. The FAST system allowed us to characterize receptor-ligand interactions and to identify evolutionarily conserved immune evasion pathways in naturally occurring transmissible cancers. Our initial implementation of the system confirmed numerous predicted protein interactions in a marsupial species and documented high expression of the inhibitory molecule CD200 on DFT cells. The high expression of CD200 in devil nervous tissues and neuroendocrine tumors, down-regulation of CD200 in response to imiquimod, and binding of CD200 to CD200R1 are consistent with results from human and mouse studies. Consequently, the CD200/R pathway provides a promising immunotherapy and vaccine target for DFTs (20). Beyond this study, FAST proteins meet the key attributes needed for reagent development, such as being straightforward to make, stable, versatile, renewable, cheap, and amenable to high-throughput testing. The direct fusion of the reporter protein to the protein of interest allows for immediate feedback during transfection, supernatant testing, and protein purification; proteins with frameshifts, introduced stop codons, or folded improperly will not fluoresce and can be discarded after a simple visualization, rather than only after extensive downstream testing. Efficient mapping of immune checkpoint interactions across species can identify evolutionarily conserved immune evasion pathways and appropriate large-animal models with naturally occurring cancer. This knowledge could inform veterinary and human medicine in the fields of immunological tolerance to tissue transplants, infectious disease, and cancer.

The objectives of this study were to fill a major gap in our understanding of the mammalian immune system and to understand how genetically mismatched transmissible tumors evade host immunity. To achieve this goal, we developed a recombinant protein system that directly fuses proteins of interest to a fluorescent reporter protein. The first phase was to determine whether the fluorescent protein remained fluorescent after secretion from mammalian cells and to confirm that proteins bound to their predicted receptors (i.e., ligands). Initial testing was performed in CHO cells and follow-up assays used devil cells. To reduce the risk of false positives in binding assays, we tested each FAST protein against the expected target protein and additional nontarget proteins. To further demonstrate the functionality of this system for antibody development, mice were immunized with either 41BB or CD200 proteins. Pre- and postimmunization polyclonal sera were used to confirm that the proteins used for immunization induced antibodies that specifically bound to surface-expressed recombinant proteins and native proteins on DFT cells. Last, to demonstrate the flexibility of the system, we replicated a known anti-human PDL1 nanobody that we fused to mCitrine. This shows that the FAST system can be used to target human proteins, to produce recombinant proteins derived from other species (e.g., camelid-derived nanobody), and for functions other than receptor-ligand interactions.

Target gene DNA sequences for vector construction were retrieved from Genbank, Ensembl, or de novo transcriptome assemblies (table S2). Target DNA was amplified from a complementary DNA template or existing plasmids using primers and polymerase chain reaction (PCR) conditions shown in tables S2 and S4 using Q5 High-Fidelity 2X Master Mix (New England Biolabs no. M0494L). Primers were ordered with 5 base extensions that overlapped expression vectors on either side of the restriction sites. The amplified products were identified by gel electrophoresis and purified using the NucleoSpin PCR and Gel Clean Up Kit (Macherey-Nagel no. 740609.5). Alternatively, DNA sequences were purchased as double-stranded DNA gBlocks (table S5) (Integrated DNA Technologies) for direct assembly into expression vectors.

All new plasmids were assembled using the NEBuilder kit (New England Biolabs; NEB no. E5520S) following the manufacturers recommendations unless otherwise noted. DNA inserts, digested plasmids, and NEBuilder master mix were incubated for 60 min at 50C and then transformed into DH5 included with the NEBuilder kit. Plasmid digestions were performed following manufacturer recommendations and generally subjected to Antarctic phosphatase (New England Biolabs no. M0289S) treatment to prevent potential reannealing. Sleeping Beauty transposon vectors pSBbi-Hyg (Addgene no. 60524), pSBbi-BH (Addgene no. 60515), pSBtet-Hyg (Addgene no. 60508), and pSBtet-RH (Addgene no. 60500) were gifts to Addgene from E. Kowarz (46). The pCMV(CAT)T7-SB100 containing the cytomegalovirus (CMV) promoter and SB100X transposase was a gift to Addgene from Z. Izsvak (Addgene no. 34879) (47). We first constructed an all-in-one Sleeping Beauty vector by inserting a CMV promoter and SB100X transposase from pCMV(CAT)T7-SB100 (47) into pSBi-BH (46) (tables S3 and S4). This was accomplished by using pAF111-vec.FOR and pAF111.1.REV primers to amplify an overlap region from pSBbi-BH (insert 1) and pAF111-2.FOR and pAF111-2.REV to amplify the CMV-SB100X region from pCMV(CAT)T7-SB100 (insert 2). The purified amplicons were then used for NEBuilder assembly of pAF111. The final all-in-one vectors pAF112 (hygromycin resistance and luciferase) and pAF123 (hygromycin resistance) were assembled from the pAF111 components. pAF112 was assembled by amplifying the Luc2 luciferase gene (insert 1) from pSBtet-Hyg and the P2A-hygromycin resistance gene (insert 2) from pSBbi-BH and inserting into the pAF111 Bsu36 I digest using NEBuilder. pSBbi-Hyg was Bsu36 Idigested to obtain the hygromycin resistance gene, and this fragment was inserted into Bsu36 Idigested pAF111 using T4 ligase cloning to replace the BFP-P2A-hygromycin segment in pAF111.

All full-length gene coding sequences except CTLA4 were cloned into the pAF112 Sfi I digest (table S2). All full-length vectors also contain luciferase with T2A peptide linked to the hygromycin resistance protein; luciferase was included for use in downstream functional testing that was not part of this study. Tasmanian devil CTLA4 was cloned into a NotI-HF and Xma I digest of pAF100 that was used in a different study but is derived from vectors pAF112 and pAF138. In addition, we also used devil PDL1 (CHO.pAF48) and 41BBL (CHO.pAF56) cell lines developed using a vector system described previously (17).

Plasmids containing fluorescent protein coding sequences mCerulean3-N1 (Addgene no. 54730), mAzurite-N1 (Addgene no. 54617), mOrange-N1 (Addgene no. 54499), and mNeptune2-N1 (Addgene no. 54837) were gifts to Addgene from M. Davidson. mTag-BFP was amplified from pSBbi-BH, mCitrine was amplified from pAF71, and mCherry was amplified from pTRE-Dual2 (Clontech no. PT5038-5). pAF137 was constructed by amplifying the devil 41BB extracellular domain with primers pAF137-1.FOR and pAF137-1.REV and amplifying mCherry with pAF137-2a.FOR and pAF137-2.REV (tables S3 and S4). 5 extensions on pAF137-1.FOR and pAF137-2.REV were used to create overlaps for NEBuilder assembly of pAF137 from a pAF123 Sfi Idigested base vector. 3 extensions on pAF137-1.REV and pAF137-2a.FOR were used to create the linker that included an Xma I/Sma I restriction site, TEV cleavage tag, GSAGSAAGSGEF linker peptide, and 6xHis tag between the gene of interest and fluorescent reporter. The GSAGSAAGSGEF was chosen because of the low number of large hydrophobic residues and less repeated nucleic acids than are needed with other flexible linkers such as (GGGS)4. The pAF137 primer extensions also created 5 Not I and 3 Nhe I sites in the FAST vector to facilitate downstream swapping of functional genes and to create a Kozak sequence (GCCGCCACC) upstream of the FAST protein open-reading frame. Following confirmation of correct assembly via DNA sequencing, the FAST 41BB-mCherry (pAF137) was digested and used as the base vector (Fig. 2B and fig. S1, A and B) for development of FAST vectors with alternative fluorescent proteins. This was accomplished by digestion of pAF137 with Sal I and Nhe I and then inserting PCR-amplified coding sequences for other fluorescent proteins using NEBuilder (tables S3 and S4).

Type I FAST (extracellular N terminus and cytoplasmic C terminus) protein vectors were constructed by digestion of 41BB FAST vectors with Not I and either Xma I or Sma I (Fig. 2B and fig. S1, A and B) and then inserting genes of interest (tables S2 to S4). To create an Fc-tagged FAST protein, we fused the extracellular domain of devil CD80 to the Fc region of the devil IgG (fig. S1C). The Fc region was amplified from a devil IgG plasmid provided by L. Corcoran (Walter and Eliza Hall Institute of Medical Research). All secreted FAST proteins in this study used their native signal peptides, except for 41BBL. 41BBL is a type II transmembrane protein in which the signal peptide directly precedes the cytoplasmic and transmembrane domains of the protein (cytoplasmic N terminus and extracellular C terminus). As type I FAST vectors cannot accommodate this domain architecture, we developed an alternative base vector for type II transmembrane FAST proteins (fig. S1D). To increase the probability of efficient secretion of type II FAST proteins from CHO cells, we used the hamster interleukin-2 (IL-2) signal peptide (accession no. NM_001281629.1) at the N terminus of the protein, followed by a Sal I restriction site, mCherry, an Nhe I restriction site, 6xHis tag, GSAGSAAGSGEF linker, TEV cleavage site, Xma I/Sma I restriction site, the gene of interest, and a Pme I restriction site following the stop codon.

Following transformation of assembled plasmids, colony PCR was performed as an initial test of the candidate plasmids. Single colonies were inoculated directly into a OneTaq Hot Start Quick-Load 2X Master Mix (NEB no. M0488) with primers pSB_EF1a_seq.FOR (atcttggttcattctcaagcctcag) and pSB_bGH_seq.REV (aggcacagtcgaggctgat). PCR was performed with 60C annealing temperature for 25 to 35 cycles. Colonies yielding appropriate band sizes were used to inoculate Luria broth with ampicillin (100 g/ml) for bacterial outgrowth overnight at 37C and 200 rpm. The plasmids were purified using standard plasmid kits and prepared for Sanger sequencing using the BigDye Terminator v3.1 Cycle Sequencing Kit (Thermo Fisher Scientific no. 4337455) with pSB_EF1a_seq.FOR and pSB_bGH_seq.REV primers. The BigDye Terminator was removed using Agencourt CleanSEQ (Beckman Coulter no. A29151) before loading samples to a 3500xL Genetic Analyzer (Applied Biosystems) for sequencing by fluorescence-based capillary electrophoresis.

DFT1 cell line C5065 and DFT2 cell line JV were cultured at 35C with 5% CO2 in cRF10 [10% complete RPMI (Gibco no. 11875-093) with 2 mM l-glutamine, supplemented with 10% heat-inactivated fetal bovine serum, and 1% antibiotic-antimycotic (Thermo Fisher Scientific no. 15240062)]. RPMI without phenol red (Sigma-Aldrich no. R7509) was used to culture FAST protein cell lines when supernatants were collected for downstream flow cytometry assays. Devil peripheral blood cells were cultured in cRF10 at 35C with 5% CO2. CHO cells were cultured at 37C in cRF10 during transfections and drug selection but were otherwise cultured at 35C in cRF5 (5% complete RPMI). For production of purified recombinant proteins, stably transfected CHO cells were cultured in suspension in spinner flasks in chemically defined, serum-free CHO EX-CELL (Sigma-Aldrich no. 14361C) media supplemented with 8 mM l-glutamine, 10 mM Hepes, 50 M 2-ME, 1% (v/v) antibiotic-antimycotic, and 1 mM sodium pyruvate and without hygromycin.

Stable transfections of CHO and DFT cells were accomplished by adding 3 105 cells to each well in six-well plates in cRF10 and allowing the cells to adhere overnight. The next day, 2 g of plasmid DNA was added to 100 l of phosphate-buffered saline (PBS) in microfuge tubes. Polyethylenimine (PEI) (linear, molecular weight, 25,000; Polysciences no. 23966-2) was diluted to 60 g/ml in PBS and incubated for at least 2 min. The PEI solution (100 l) was added to the 100 l of plasmid DNA in each tube to achieve a 3:1 ratio of PEI:DNA. The solution was mixed by gentle pipetting and incubated at room temperature for 15 min. While the solution was incubating, the media on the CHO cells were replaced with fresh cRF10. All 200 l from each DNA:PEI mix was then added dropwise to the CHO cells and gently rocked side to side and front to back to evenly spread the solution throughout the well. The plates were then incubated overnight at 37C with 5% CO2. The next day, the plates were inspected for fluorescence, and then the media were removed and replaced with cRF10 containing hygromycin (1 mg/ml) (Sigma-Aldrich no. H0654). The media were replaced with fresh cRF10 hygromycin (1 mg/ml) every 2 to 3 days for the next 7 days until selection was complete. The cells were then maintained in hygromycin (0.2 mg/ml) in cRF5 at 35C with 5% CO2. Supernatant was collected 2 to 3 weeks after transfection and stored at 4C for 2 months to assess stability of secreted FAST proteins.

Sixteen days after transfection, the first batch of FAST protein cell lines was adapted to a 1:1 mix of cRF5 and chemically defined, serum-free CHO EX-CELL media for 1 to 2 days to facilitate adaptation of the adherent CHO cells to suspension culture in serum-free media. At least 5 107 cells were then transferred to ProCulture spinner flasks (Sigma-Aldrich no. CLS45001L and no. CLS4500250) and stirred at 75 rpm at 35C in 5% CO2 on magnetic stirring platforms (Integra Bioscience no. 183001). Cells were maintained at a density ranging from 5 105 to 2 106 cells/ml for 8 to 14 days. Supernatant was collected every 2 to 3 days, centrifuged at 3200 relative centrifugal force (RCF) for 10 min, stored at 4C, and then purified using the KTA start protein purification system (GE Life Sciences no. 29022094). The supernatant was diluted 1:1 with 20 mM sodium phosphate (pH 7.4) and then purified using HisTrap Excel columns (GE Life Sciences no. 17-3712-05) according to the manufacturers instructions. Samples were passed through the columns using a flow rate of 2 ml/min at 4C; all wash and elution steps were done at 1 ml/min. Elution from HisTrap columns (GE Life Sciences no. 17-3712-05) was accomplished using 0.5 M imidazole and fractionated into 1-ml aliquots using the Frac30 fraction collector (GE Life Sciences no. 29023051). Fluorescence of FAST proteins was checked via brief excitation (Fig. 2) on a blue light transilluminator with an amber filter unit. In the case of mCherry, chromogenic color was visible without excitation. Fractions containing target proteins were combined and diluted to 15 ml with cold PBS, dialyzed (Sigma-Aldrich no. PURX60005) in PBS at 4C, 0.22-m sterile-filtered (Millipore no. SLGV033RS), and concentrated using Amicon Ultra centrifugal filter units (Sigma-Aldrich no. Z706345). The protein concentration was quantified using the 280-nm absorbance on a NanoDrop spectrophotometer. Extinction coefficients using for each protein were calculated using the ProtParam algorithm (48). The proteins were then aliquoted into microfuge tubes and frozen at 80C until further use. The CTLA4-Fc-mCherry protein was designed, assembled, and tested separately from the other FAST proteins and was tested directly in supernatant without purification.

CHO cells expressing full-length proteins were thawed in cRF10 and then maintained in cRF5 with hygromycin (0.2 mg/ml). The adherent CHO cells were washed with PBS and incubated with trypsin for 5 min at 37C to remove cells from the culture flask. Trypsin was diluted five times with cRF5 and centrifuged at 200 RCF for 5 min. Cells were resuspended in cRF5, counted (viability >95% in all cases), and resuspended and aliquoted for assays as described below.

Supernatants (cRF5) were collected from CHO cells expressing devil 41BB extracellular domain fused to either mCherry (pAF137), mCitrine (pAF138), mOrange (pAF164), mBFP (pAF139), mAzurite (pAF160), mCerulean3 (pAF161), or mNeptune2 (pAF163) (tables S2 to S4). The supernatant was spun for 10 min at 3200 RCF to remove cells and cellular debris and then stored at 4C until further use. CHO cells expressing devil 41BBL (CHO.pAF56) and untransfected CHO cells were prepared as described above. Flow cytometry tubes were loaded with 5 104 target CHO cells per well in cRF5, centrifuged 500 RCF for 3 min, and then resuspended in 200 l of supernatant from the 41BB FAST cell lines (N = 1 per treatment). The tubes were then incubated for 15 min at 4C, centrifuged at 500 RCF for 3 min, resuspended in 400 l of cold fluorescence-activated cell sorting (FACS) buffer, and stored on ice until the data were acquired on a Beckman Coulter Astrios flow cytometer (Fig. 2C). All flow cytometry data were analyzed using FCS Express 6 Flow Cytometry Software version 6 (De Novo Software).

U-bottom 96-well plates were loaded with 1 105 target CHO cells per well in cRF5, centrifuged 500 RCF for 3 min, and then resuspended in 175 l of cRF5 supernatant from FAST cell lines collected 11 days after transfection (N = 1 per treatment). The plates were then incubated for 30 min at room temperature, centrifuged at 500 RCF for 3 min, resuspended in 200 l of cold FACS buffer, centrifuged again, and fixed with FACS fix buffer [PBS, 0.02% NaN3, 0.4% formalin, glucose (10 g/liter)]. The cells were transferred to tubes, diluted with FACS buffer, and analyzed on a Beckman Coulter Astrios flow cytometer (fig. S2).

Purified FAST proteins were diluted to 20 g/ml in cRF5, aliquoted into V-bottom 96-well transfer plates, and then stored at 37C until target cells were ready for staining. Target cells were resuspended in cRF5 with 100 M chloroquine, and 100,000 cells per well were aliquoted into U-bottom 96-well plates. One hundred microliters of the diluted FAST proteins (N = 1 per treatment, two time points per treatment) was then transferred from the V-bottom plates into the U-bottom 96-well plates containing target cells. The final volumes and concentrations were 200 l per well in cRF5 with 50 M chloroquine and 2 g per well of FAST proteins. One set of plates was incubated at 37C for 30 min, and another set of plates was incubated at 37C overnight. The cells were then centrifuged 500 RCF for 3 min, the media decanted, and incubated for 5 min with 100 l of trypsin to dislodge adherent cells. The cells were then washed with 200 l of cold FACS buffer, fixed, resuspended in cold FACS buffer, and transferred to tubes for analysis on the Astrios flow cytometer (Fig. 3B).

The protocol for using FAST protein supernatants was the same above as the preceding experiment except for the modifications described here. Supernatants were collected 2 to 3 weeks after transfection, centrifuged at 3200 RCF for 10 min, and stored at 4C for 2 months. Before staining for flow cytometry, the supernatant was 0.22-m filtered. Supernatant was then loaded into V-bottom 96-well plates to facilitate rapid transfer to staining plates and stored at 37C until target cells were ready for staining. Target cells were prepared as described above except for being diluted in cRF5 with 100 M chloroquine. A total of 2 105 cells per well (100 l) were then loaded into U-bottom 96-well plates. One hundred microliters of FAST protein supernatant (N = 1 per treatment) was then transferred from the V-bottom plates to achieve 50 M chloroquine, and the cells were then incubated at 37C for 60 min. The plates were then washed, fixed, and analyzed on the Astrios flow cytometer (fig. S3). A similar procedure was used for staining stably transfected DFT cells with CTLA4-Fc-mCherry, except that the supernatant was used fresh (Fig. 4D).

CHO cells expressing full-length CTLA4 with a C-terminal mCitrine and CHO cells expressing full-length 41BB or 41BBL were labeled with 5 M CFSE; CFSE and mCitrine were analyzed using the same excitation laser (488 nm) and emission filters (513/26 nm). A total of 1 105 FAST proteinsecreting cells were mixed with 1 105 target cells in cRF5 with 50 M chloroquine and incubated overnight at 37C in 96-well U-bottom plates (Fig. 4A). The next day, the cells were rinsed with PBS, trypsinized, washed, fixed, and resuspended in FACS buffer before running flow cytometry. Cells were gated on forward and side scatter (FSC SSC) and for singlets (FSC-H FSC-A) (Fig. 4B). Data shown in Fig. 4C are representative of N = 3 technical replicates per treatment. Data were collected using a Beckman Coulter MoFlo Astrios and analyzed using FCS Express.

RNA-seq data were generated during previous experiments, aligned against the reference Tasmanian devil genome Devil_ref v7.0 (GCA_000189315.1), and summarized into normalized read counts as previously described (25, 26). Transcripts per millionnormalized read counts were calculated in R, and a heat map was produced from log2-converted values using the heatmap.2 function of gplots.

A total of 50,000 DFT cells per well were aliquoted into U-bottom 96-well plates, washed with 150 l of cRF10, and resuspended in 100 l of warm cRF10 containing 100 M chloroquine. Five micrograms of FAST protein per well was then added and mixed by pipetting. The plates were then incubated at 37C for 30 min. The cells were then transferred to microfuge tubes without washing, stored on ice, and analyzed on a Beckman Coulter MoFlo Astrios (N = 2 per treatment).

CD200 and 41BB FAST proteins were digested overnight with TEV protease (Sigma-Aldrich no. T4455) at 4C in PBS. The cleaved linker and 6xHis tag were then removed using a His SpinTrap kit (GE Healthcare no. 28-9321-71). Digested proteins in PBS were diluted 1:1 in Squalvax (OZ Biosciences no. SQ0010) to a final concentration of 0.1 g/l and were mixed using interlocked syringes to form an emulsion. Immunization of BALB/c mice for antibody production was approved by the University of Tasmania Animal Ethics Committee (no. A0014680). Preimmune sera were collected before subcutaneous immunization with at least 50 l of the emulsion. On day 14 after immunization, the mice were boosted using a similar procedure. On day 50, the mice received a booster with proteins in IFAVax (OZ Biosciences no. IFA0050); mice immunized with CD200 again received subcutaneous injections, whereas 41BB mice received subcutaneous and intraperitoneal injections. Preimmune and sera collected after three-times immunizations were then tested by flow cytometry against CHO cells expressing either 41BB or CD200. CHO cells were prepared as described above, and 2 105 cells were incubated with mouse serum diluted 1:200 in PBS for 30 min at 4C. The cells were then washed two times and stained with 50 l of anti-mouse IgG AF647 diluted 1:1000 in FACS buffer. The cells were then washed two times, stained with 4,6-diamidino-2-phenylindole (DAPI) to identify live cells, and analyzed on a CyAn ADP flow cytometer (Fig. 5C). CD200 and 41BB expression on DFT cells was tested using a procedure similar to the CHO cell staining, except that the sera used were collected after four-times immunizations and was diluted 1:500 and analyzed on the BD FACSCanto II (Fig. 5D).

Approximately, 200 l of NMS or anti-CD200 serum day 157 (after four-times immunizations) was purified using an Ab SpinTrap (GE Healthcare no. 28-4083-47) according to the manufacturers instructions. Serum was diluted 1:1 with 20 mM sodium phosphate and binding buffer (pH 7.0) and eluted with 0.1 M glycine-HCl (pH 2.7), and the pH was neutralized with 0.1 M glycine-HCl (pH 2.7). The eluted antibodies were then concentrated using an Amicon Ultra 0.5 centrifugal unit (Merck no. UFC500308) by centrifuging at 14,000 RCF for 30 min at 4C and then washing the antibodies with 400 l of PBS twice. The protein concentration was then quantified on a NanoDrop spectrophotometer at 280 nm using the extinction coefficients for IgG.

A total of 50,000 DFT cells per well were aliquoted into U-bottom 96-well plates and washed with 200 l of cold FACS buffer. Purified polyclonal anti-CD200 was diluted to 2.5 g/ml in cold FACS buffer, and the cells in appropriate wells were resuspend in 100 l per well (0.25 g per well) diluted antibody; wells that did not receive antibody were resuspended in 100 l of FACS buffer. The cells were incubated on ice for 20 min and then washed with 200 l of FACS buffer. While incubating, anti-mouse IgG AF647 was diluted to 1 g/ml in cold FACS buffer and then used to resuspend cells in the appropriate wells. The plates were incubated on ice for 20 min and then washed with 100 l of cold FACS buffer. The cells were then resuspended in 200 l of FACS fix and incubated on a rocking platform at room temperature for 15 min. The cells were then centrifuged 500 RCF for 3 min at 4C, resuspended in 200 l of FACS buffer, and stored at 4C until they were analyzed on a FACSCanto II (N = 2 per treatment) (Fig. 5E).

Blood collection from Tasmanian devils was approved by the University of Tasmania Animal Ethics Committee (permit no. A0014599) and the Tasmanian Department of Primary Industries, Parks, Water and Environment. Blood was collected from the jugular vein and stored in EDTA tubes for transport to the laboratory. Blood was processed within 3 hours by diluting 1:1 with serum-free RPMI and then layering onto Histopaque (Sigma-Aldrich no. 10771) before centrifuging at 400 RCF for 30 min. The interface containing the PBMCs was then collected using a transfer pipette, diluted with 50 ml of serum-free RPMI, and centrifuged for 5 min at 500 RCF. Cells were washed with again with cRF10 and then either used fresh or stored at 80C until further use.

Frozen devil PBMC was thawed and cultured in cRF10 at 35C with 5% CO2 for 2 hours; cells were then washed in FACS buffer and counted, and 3 105 PBMCs were used per sample. DFT2.JV cells were removed from culture flasks and counted, and 2 105 cells were used per sample. Samples were incubated with 50 l of normal goat serum (Thermo Fisher Scientific, catalog no. 01-6201) diluted 1:200 in FACS buffer for 15 min at 4C, and 50 l of anti-CD200 serum diluted 1:100 was added (1:200 final) for 30 min at 4C. Cells were then washed two times and stained with 50 l of anti-mouse IgG AF647 diluted to 1 g/ml in FACS buffer for 30 min at 4C. The cells were then washed two times, stained with DAPI (Sigma-Aldrich, catalog no. D9542) to identify live cells, and analyzed on the BD FACSCanto II. PBMC and DFT cells were run separately, and then PBMC and DFT2 were mixed at a ratio of 10:1 by volume for the combined samples (N = 1 per treatment) (fig. S4A). The experiment was repeated (N = 1 per treatment), except that PBMCs and DFT cells were mixed at a 5:1 ratio (fig. S4B).

DFT1.C5065 and DFT2.JV cells were labeled with 5 M CTV and cultured for 3 days at 37C. On the day of the assays, peripheral blood from one devil was collected and stored at ambient temperature for less than 3 hours. One hundred microliters of whole blood was aliquoted into 15-ml tubes and stored at ambient temperature while DFT cells were prepared. The media on CTV-labeled DFT cells were decanted, and the cells were detached from the flask by incubating in 2.5 ml of TrypLE Select for 5 min at 37C. The cells were washed with cRF10, resuspended in cRF10, and counted. DFT cells were then diluted to 1 104 cells/ml in cRF10, and 100 l was aliquoted into appropriate 15-ml tubes containing 100 l of whole blood. One microliter of purified anti-CD200 (0.5 g per tube) was diluted into the appropriate tubes and incubated for 15 min at ambient temperature. Next, anti-mouse IgG AF647 (0.5 g per tube) was added to each tube. Note: 0.5 l (0.5 g) of concentrated secondary antibody was accidentally added directly to the tube for the data shown in the top row, middle column of fig. S5A; for all other tubes, the secondary antibody was diluted 1:20 in PBS and 10 l was added to each tube. The cells were then incubated for 15 min at ambient temperature. The cells were then diluted in 1 ml of ammonium chloride RBC lysis buffer [150 mM NH4Cl, 10 mM KHCO3, and 0.1 mM EDTA disodium (Na2-2H2O)] and mixed immediately gently pipetting five times. The cells were incubated at ambient temperature for 10 min, then diluted with 5 ml of PBS, and centrifuged 500 RCF for 3 min. Some tubes contained residual RBCs, so the pellet was vigorously resuspended in 5 ml of RBC lysis buffer, incubated for 5 min, diluted with 5 ml of cold FACS buffer, and centrifuged 500 RCF for 3 min. The cells were then resuspended in 250 l of FACS buffer and stored on ice until analysis on a Beckman Coulter MoFlo Astrios (N = 1 per treatment). Data were analyzed in FCS Express version 6 (fig. S5).

The experiment above was repeated with the following modifications. DFT cells were labeled with 5 M CFSE and incubated for 2 days at 37C. On the day of the assays, fresh blood was collected from two devils. Purified anti-CD200 and NMS were labeled with Zenon mouse IgG AF647 (Thermo Fisher Scientific no. Z25008) and blocked with the Zenon blocking agent. A total of 1 104 CFSE-labeled DFT cells were diluted directly into 100 l of whole blood in 15-ml tubes, and 12 l (2-l antibody, 5-l labeling agent, and 5-l blocking agent) of Zenon AF647labeled purified NMS or anti-CD200 was added directly to the cells. The cells were incubated for 30 min at ambient temperature. The cells were then gently resuspended in 2.5 ml of RBC lysis buffer and incubated for 10 min at ambient temperature. The cells were diluted with 10 ml of PBS and centrifuged 500 RCF for 3 min. The cells were resuspended in 1.5 ml of RBC lysis buffer and incubated for another 10 min to lyse residual RBCs. The tubes were then resuspended in 9 ml of cRF10 and centrifuged 500 RCF for 3 min. The cells were resuspended in 350 l of cold FACS buffer containing DAPI (200 ng/ml) and stored on ice until analysis on a Beckman Coulter MoFlo Astrios (N = 1 per treatment for n = 2 devils) (Fig. 6, A to D).

The anti-human PD-L1 nanobody (KN035) (24) protein sequence was reverse-translated and as a double-stranded DNA gBlock (Integrated DNA Technologies) (table S5). The sequence was modified to include DNA extension that overlaps FAST vectors. The signal peptide from hamster IL-2 (also in pAF92) was added to the nanobody to increase secretion efficiency in CHO cells. The gBlock was inserted into a NotI-HF and Sma Idigested mCitrine FAST vector with NEB HiFi DNA Assembly Master Mix (NEB no. E2621). Transformation, purification of plasmid DNA, and sequencing were performed as described above.

ExpiCHO cells (Thermo Fisher Scientific no. A29127) for high-yield protein production were maintained at 37C with 8% CO2 with constant shaking at 200 rpm in ExpiCHO Stable Production Medium (SPM) (Thermo Fisher Scientific no. A3711001). ExpiCHO cells were added to a six-well plate at 3 105 cells per well in ExpiCHO SPM and cultured overnight. The next day, 2-g plasmid DNA was added to 100 l of PBS in a microcentrifuge tube. PEI was diluted to 60 g/ml in PBS and incubated at room temperature for 5 min. Diluted plasmid DNA was added to 100 l of PEI solution to achieve a 3:1 PEI:DNA ratio and incubated at room temperature for 15 min. During this time, ExpiCHO cells were transferred to 15-ml centrifuge tubes, washed with PBS at 300g for min, resuspended in 3-ml OptiPRO serum-free media (Thermo Fisher Scientific no. 12309019), and returned to the six-well plate. The PEI:DNA solution was then added directly to cells and incubated overnight. The next day, plates were inspected for fluorescence, and the media were removed and replaced with ExpiCHO SPM supplemented with hygromycin (1 mg/ml). Media were changed every second day until selection was complete. Once selection was complete, the cells were moved to 50-ml TPP TubeSpin bioreactor tubes (Sigma-Aldrich no. Z761028) and maintained at 4 106 to 6 106 cells/ml in ExpiCHO SPM with hygromycin (0.2 mg/ml). Supernatant was collected 2 weeks after transfection and stored at 4C.

CHO cells expressing either human PD-L1 or human CD80 fused to miRFP670 (table S3) were plated at 100,000 cells per well into a U-bottom 96-well plate and centrifuged at 300g for 5 min, and the supernatant was discarded. Two hundredmicroliter supernatant containing secreted PD-L1 nanobody was added to CHO cell lines either neat or diluted in 1:10 and 1:100 in FACS buffer. Cells were incubated at 4C for 30 min before being washed in FACS buffer for analysis on a Beckman Coulter FACSCanto II (Fig. 6F).

Read more:
A novel system to map protein interactions reveals evolutionarily conserved immune evasion pathways on transmissible cancers - Science Advances

Recommendation and review posted by Alexandra Lee Anderson

The structure of the RCAN1:CN complex explains the inhibition of and substrate recruitment by calcineurin – Science Advances

INTRODUCTION

Calcineurin (CN, PP2B, and PP3) is a Ca2+-dependent Ser/Thr phosphatase with critical functions in many physiological processes, including development, cardiac function, and the immune response (1, 2). Calcium-activated CN dephosphorylates many substrates, including the nuclear factor of activated T cell (NFAT) transcription factors (3). Once dephosphorylated, the NFATs translocate into the nucleus and initiate NFAT-regulated gene transcription. In 2006, Crabtree and co-workers (4) demonstrated that Nfatc2/ and Nfatc4/ knockout mice have facial characteristics similar to those observed in patients with Down syndrome (DS; also known as trisomy 21), suggesting that the disruption of NFAT signaling, potentially via CN, plays a central role in this disease. Consistent with this observation, mice with forebrain-specific deletions of CNB, a subunit of CN responsible for making CN activation sensitive to Ca2+, also exhibit defects in learning and memory, both established hallmarks of DS (5). Studies during the last two decades have revealed that deletions of CNB render CN incapable of forming the LxVP-binding pocket, which is essential for NFAT dephosphorylation and, in turn, prevents NFAT translocation to the nucleus.

Individuals with DS have three copies of chromosome 21, resulting in a 1.5 dosage of these genes. Studies of patients with incomplete trisomy 21 led to the identification of the Down syndrome critical region (DSCR), the chromosomal fragment of chromosome 21 that is hypothesized to contain the genes responsible for the major DS phenotypes. The protein product of one of these genes, DSCR1 (Adapt78, MCIP1, RCN1, and calcipressin; now known as RCAN1; Fig. 1A), was found to be an inhibitor of CN (68). Consistent with this, overexpression of both full-length and a C-terminal fragment (amino acids 115 to 197) of RCAN1 in calcium-stimulated human cells prevented NFAT translocation to the nucleus (6), suggesting that RCAN1 overexpression may contribute to DS by blocking NFAT signaling. Supporting this hypothesis, Northern blots showed that the levels of RCAN1 mRNA in DS brain tissue were significantly higher compared to tissue from non-DS patients (6). Together, these data suggest that RCAN1, via its ability to inhibit CN and in turn disrupt NFAT signaling, contributes to DS phenotypes.

(A) RCAN1 domain structure showing the N-terminal RRM domain (yellow, amino acids 6 to 82; PDB ID 1WEY), the LxVP motif (green, amino acids 96 to 99), the SPPASPP motif (cyan, amino acids 108 to 114), the PxIxIT motif (orange, amino acids 154 to 159), and the TxxP motif (purple, amino acids 186 to 189). Partially populated helices 1 and 2 [gray cylinders; see (D)] and the regions that remain flexible when bound to CN [dotted circles; flex1 and flex2; see (C)] are also indicated. Constructs used in this study are shown below, with mutations shown in red letters. Sequence alignment of multiple RCAN1 species (amino acids 89 to 197), with the residues corresponding to motifs highlighted; * indicates residues phosphorylated by active p38. RCAN1 orthologs used in sequence alignment are described in Materials and Methods. RCAN1 residue numbers are labeled according to human RCAN1 isoform 2. (B) Cartoon diagram of calcineurin (CN), with CNA shown as a white surface, CNB shown as a gray surface, the PxIxIT motifbinding pocket shown in orange, the LxVP motifbinding pocket shown in green, and the location of the active site indicated. (C) Overlay of the 2D [1H,15N] HSQC spectra of 15N-labeled RCAN1 in the absence (black) and presence (red) of CN. Peaks that do not interact directly with CN are labeled. (D) Secondary-structure propensity data plotted against RCAN1 residue numbers. These data indicate regions with transient secondary structure (SSP > 0, helix; SSP < 0, strand). (E) Overlay of the 2D [1H,15N] HSQC spectra of 15N-labeled RCAN1 (black) and 15N-labeled RCAN1core (blue).

RCAN1 is also overexpressed in patients with Alzheimers disease (AD), where RCAN1 levels are ~3-fold higher than in nondiseased individuals (9, 10). DS and AD are linked diseases, as it is well established that many patients with DS who reach middle age (40s) suffer from early-onset AD (1113). One of the hallmarks of AD is neurofibrillary tangles composed of hyperphosphorylated tau protein. Many Ser/Thr and Tyr phosphatases are critical for maintaining tau in a proper (healthy) phosphorylation state (14, 15). Because CN is highly enriched in the neuronal tissues, making up 1% of the total protein in brain, CN plays an especially critical role in the control of tau dephosphorylation (16). Consistent with this, overexpression of RCAN1 in mice increased tau phosphorylation (17). Thus, excess inhibition of CN due to increased levels of RCAN1 has profound consequences on the brain. However, despite extensive efforts (1821), very little is known about how RCAN1 binds and regulates CN at a molecular level.

Recently, it was shown that CN (composed of subunits A and B, CNA and CNB, respectively; Fig. 1B) recruits regulators, inhibitors, and substrates using two short linear motifs (SLiMs), the PxIxIT and the LxVP motif (22, 23). These SLiMs are typically found in intrinsically disordered proteins/regions (IDPs/IDRs) and bind to the corresponding PxIxIT- and LxVP-binding pockets in CN. The PxIxIT motif binds the catalytic domain of CNA (22), whereas the LxVP motif binds to a hydrophobic cleft at the interface of the CNA and B subunits (23). While it has been known for more than a decade that RCAN1 contains a PxIxIT motif, a number of studies have suggested that additional motifs (21, 24, 25), both N- and C-terminal to the noncanonical PxIxIT motif, play a key role in RCAN function and CN activity (inhibition), including a TxxP motif, which we have recently shown is critical for CN active site substrate recruitment (26). In Nhe1, the TxxP motif binds CN in an identical manner to the previous determined autoinhibitory domain (AID), which blocks the CN active site (27). Upon an increase in [Ca2+], the AID dissociates from the catalytic site, allowing for phosphatase activity.

Here, using a combination of structural and biochemical methods, we determined how, at a molecular level, RCAN1 binds and inhibits CN. We found that RCAN1 interacts extensively with CN, at both the canonical PxIxIT- and LxVP-binding pockets but also beyond these regions, including the catalytic site. Further, we found that CN binding leads to a folding-upon-binding event, which creates a novel extended PxIxIT-type interaction that defines the central core of the RCAN1:CN interaction. We also found that in addition to physically blocking substrate binding by binding the PxIxIT and LxVP binding pockets, RCAN1 also inhibits CN activity directly by binding and blocking its active site via two mechanisms. Last, both the activity and the binding of RCAN1 to CN are regulated by phosphorylation, providing an example of how SLiM interactions are actively regulated by phosphorylation. Together, these data reveal how RCAN1 controls CN activity and, by extension, how RCAN1-mediated inhibition disrupts CN signaling, ultimately leading to hyperphosphorylation of CN substrates.

RCAN1 is a two-domain protein with a structured N-terminal RNA-recognition motif (RRM) domain that binds mRNA (amino acids 1 to 88; Fig. 1A) and a C-terminal domain that is responsible for CN binding (amino acids 89 to 197, the RCAN1 CN-binding domain; hereafter referred to as RCAN1; Fig. 1A). The two-dimensional (2D) [1H,15N] heteronuclear single-quantum coherence (HSQC) spectrum of RCAN1 exhibits very little chemical-shift dispersion in the 1HN dimension, indicating that the CN-binding domain is an IDR (Fig. 1C). To gain structural insights, we used chemical shift analysis (CSI/SSP) to test for preferred secondary structure in RCAN1. Two regions with preferred -helical secondary structure (helix 1, amino acids 129 to 137, ~40% populated; helix 2, amino acids 168 to 178, ~55% populated; Fig 1D) were identified. Consistently, these regions also have reduced fast time scale dynamics as determined using a 15N-{1H} nuclear Overhauser effect (15N-{1H}NOE) experiment (fig. S1A). Together, these data show that the CN-binding domain of RCAN1 is intrinsically disordered with two partially populated -helices.

RCAN1 binds CN tightly, with a KD of ~10 nM [isothermal titration calorimetry (ITC); the thermogram is atypical, with two independent binding events that have differing enthalpy changes; fig. S1B]. Given the atypical ITC results, we used surface plasmon resonance (SPR) spectroscopy as a complementary approach, which reported a similar KD of ~1 nM (Table 1, table S1, and fig. S1B). As ITC does not report time-dependent events, the easiest interpretation for the atypical biphasic thermogram is two sequential processes that lead to CN binding, i.e., the first binding event is an RCAN1 rearrangement (i.e., possibly due to charge:charge interactions within RCAN1) and, upon successful rearrangement, the second binding event reports the interaction with CN, which we used to extract KD values. To identify the RCAN1 residues that bind directly to CN, we used nuclear magnetic resonance (NMR) spectroscopy. We formed the RCAN1:CN complex, in which only RCAN1 was 15N-labeled. Because free RCAN1 lacks any significant long-range intramolecular interactions, the CN-bound and CN-unbound residues of RCAN1 will have significantly different NMR relaxation properties. As a consequence, RCAN1 residues that bind directly to CN will be invisible in a 2D [1H,15N] HSQC spectrum, while unbound RCAN1 residues will retain their original positions in the 2D [1H,15N] HSQC spectrum, allowing all nonbinding residues to be identified (Fig. 1C). Using this well-established procedure, we found that most RCAN1 residues interact directly with CN (amino acids 89 to 110, 128 to 164, and 180 to 197), many more than expected on the basis of previously described SLiM-based interactions with CN. However, there are also RCAN1 regions that remain flexible and unbound, including amino acids 111 to 124 (flex1) and 168 to 178 (flex2) (Fig. 1, A and C); the latter flexible region, flex2, overlaps with helix 2, demonstrating that this helix does not bind CN.

The RCAN1 regions that bind directly to CN include the putative LxVP (96LAPP99), PxIxIT (154PSVVVH159), and TxxP (186TRRP189) motifs (Fig. 1A). To determine which motifs contribute to CN binding, we generated three RCAN1 variants in which the canonical residues of the putative motifs were mutated to alanines (RCAN1LxVPdead, HLAPPAAAAA; RCAN1PxlxlTdead, PSVVVHASAVAA; RCAN1TxxPdead TRRPAAAA) and measured the affinity of the variants for CN using either ITC or SPR. The data showed that the LxVP motif only minimally contributes to CN binding, as the KD of the RCAN1LxVPdead variant for CN is essentially unchanged. In contrast, both the TxxP and especially the PxIxIT motifs contribute significantly to binding, as the KD values of the corresponding dead variants increase (RCAN1TxxPdead variant, ~3-fold increase in KD; RCAN1PxlxlTdead, ~145-fold increase in KD; fig. S1B, Table 1, and table S1).

Together, the NMR and ITC data suggest that residues 128 to 164 constitute the core CN-binding domain of RCAN1, as this region is between the two flexible domains that do not bind to CN and also includes the tightly binding RCAN1 PxIxIT motif. ITC shows that RCAN128164 binds strongly to CN, albeit more weakly than RCAN189197 (Table 1 and fig. S1B). To facilitate our structural NMR studies (note that all other studies are performed with wt-RCAN1), we converted the RCAN1 PxIxIT motif to a strong PxIxIT sequence [~4-fold change; PSVVVH PSVVIT (22); hereafter referred to as RCAN1core; Table 1 and fig. S1B]. To confirm that RCAN1core behaves identically to the same domain within the context of the full RCAN1 CN-binding domain (RCAN1), we used NMR spectroscopy. First, an overlay of the 2D [1H,15N] HSQC spectra of free RCAN1 with free RCAN1core revealed that no chemical shift perturbations (CSPs) are observed for corresponding peaks, beyond those expected at the RCAN1core N and C termini (Fig. 1E and fig. S1C). This demonstrates that the free state of RCAN1core is identical to that present within the context of the full CN-binding domain. Second, a CSI/SSP analysis performed after completing the sequence-specific backbone assignment of RCAN1core showed that helix 1 is ~45% populated (Fig. 2A). This is identical to what was observed for the same helix in RCAN1 (Fig. 1D). These data demonstrate that the residues that define the RCAN1core behave identically in both constructs (RCAN1core and RCAN1).

(A) Secondary-structure propensity data plotted against RCAN1core residue numbers (SSP > 0, helix; SSP < 0, strand). RCAN1core, blue; CNA-bound RCAN1core, pink. (B) Overlay of the 2D [1H,15N] TROSY spectrum of free 15N-labeled RCAN1core (blue) with the 2D [1H,15N] TROSY spectrum of (2H,15N)-labeled RCAN1core bound to (2H)-labeled CNA (pink). Arrows indicate the peak shifts in RCAN1core upon binding CN. (C) 2mFo DFc electron density map (blue) contoured at 1 corresponding to the RCAN1core PxIxIT motif (orange sticks; PxIxIT motif residues underlined) bound to CN (gray). CN residues that form the PxIxIT motif hydrophobic docking pocket shown as sticks and labeled. (D) Structure of RCAN1core:CNA complex obtained using NMR and x-ray data corefinement. RCAN1core, orange (PxIxIT motif shown as sticks; strand 1, arrow, helix 1, cylinder); CNA, white surface with the CNA residues that experience CSPs upon RCAN1core binding shown in blue (>2 SD) or light blue (>1 SD); dark gray surface corresponds to unassigned residues. Residues that experience CSPs are also labeled. (E) Overlay of the five lowest-energy corefined structures. (F) Overlay of the 2D [1H,15N] TROSY spectra of (2H,15N)-labeled CNA in the absence (black) and presence (red) of the RCAN1core. Peaks that shift in the presence of the RCAN1core are shown with arrows and labels. Label colors correspond to chemical shift perturbation deviation magnitudes [see (D)].

The interaction of the RCAN1core domain with the CNA was then examined using NMR spectroscopy. In this experiment, both RCAN1core and CNA are isotopically labeled (2H,15N-labeled RCAN1core; 2H-labeled CNA), allowing all peaks to be detected. The data showed that RCAN1core folds upon CNA binding, as the dispersion of the peaks in the 1HN dimension widened significantly, owing to the formation of new hydrogen bonds in novel secondary-structure elements (Fig. 2B). Repeating this experiment using RCAN1 and CN (CNA/B) showed an identical folding pattern (fig. S2A), confirming that the RCAN1core binds identically to both CN and CNA and that the RCAN1core domain folds-upon-binding in a manner identical to that of RCAN1 (fig. S2A). To understand this conformational change in molecular detail, we completed the CNA-bound RCAN1core sequence-specific backbone assignment (2H,13C,15N-labeled RCAN1core, 2H-labeled CNA; 47-kDa complex; Fig. 2B). The CSI/SSP analysis showed that when RCAN1core is bound to CNA, most of the partially populated helix 1 becomes 100% populated, with the final turn being slightly lower populated. Further, two new strands are formed, 1 (amino acids 142 to 145) and 2 (amino acids 155 to 159; Fig. 2A).

To determine the RCAN1core:CNA 3D structure, we first used x-ray crystallography. The structure of the RCAN1core:CNA complex was determined by molecular replacement and refined to 1.85 resolution (table S2). Strong difference electron density was observed for the PxIxIT motif and adjacent residues, allowing RCAN1 residues 153TPSVVITVC161 to be readily modeled (Fig. 2C). The RCAN1 PxlxlT motif binds CN in an extended strand conformation, burying ~495 2 of solvent accessible surface area. As observed in other CN:PxIxIT complexes (fig. S2B), the RCAN1 PxIxIT motif strand hydrogen bonds with CNA strand 14 in a parallel arrangement, extending one of CNAs two central sheets (Fig. 2A and fig. S2C). In addition to the electron density corresponding to the PxIxIT sequence, weak difference density corresponding to a short strand was observed just above the PxIxIT motif strand, further extending the CNA central sheet by one more strand (fig. S2D). The presence of a second strand is consistent with the CSI/SSP NMR data that showed two strands form when RCAN1core binds CNA (Fig. 2A). Despite extensive efforts, no electron density was identified for RCAN1 residues 128 to 140, likely due to the fact that the PxIxIT binding pocket is located at a crystal contact and, as a consequence, displaced these residues in the crystal (fig. S2E).

To determine the structure of the RCAN1core:CNA complex, we used a hybrid structural biology approach, integrating the NMR and crystallographic data. First, we determined the solution structure of RCAN1core bound to CNA using chemical shiftbased dihedral angle restraints and NOE distance restraints, the latter of which were derived from 13C-ILV-methyl-methyl resolved [1H,1H] NOESY (NOE spectroscopy), 13C-methyl-ILV-15N resolved [1H,1H] NOESY, and 15N-resolved [1H,1H] NOESY spectra. A total of 84 NOE restraints were detected and used for structure determination (table S4). Next, we developed a corefinement protocol that used both the NMR-derived (NOE) and crystallographic-derived (H-bond) restraints with established protein stereochemical restraints to refine the structure of the CNA:RCAN1core complex (Fig. 2, D and E). Atoms allowed to change during corefinement included the atoms from all RCAN1core residues and CNA residues belonging to CNA strand 14. The ensemble of RCAN1core domains adopted a compact conformation defined by the secondary structures of a single helix (1, 129YDLLYAISKL138), two strands (1, 142EKYE145; 2, PxIxIT motif, 154PSVVIT159), and two turns. The strands bind one another in an antiparallel manner, with the helical axis of the helix aligning parallel and adjacent to both strands (because of the limited number of NOE restraints, the helix adopted a small range of orientations relative to the two-stranded sheet). The compact tertiary structure is stabilized by a central hydrophobic core defined by RCAN1 residues Leu131, Ile135, Leu138, Val156, and Ile158. The RCAN1core is anchored to CNA via its PxIxIT motif. This ordering of RCAN1core PxIxIT residues (including Val156 and Ile158), together with the CNA PxIxIT-binding pocket residues (especially Met290 and Ile331), provides a hydrophobic platform that enhances the stability of the folded conformation of the RCAN1core. The observation that the entire RCAN1core folds-upon-binding CNA has never been observed for any other CN-regulator.

To confirm the experimental accuracy of the NMR and crystallographically corefined structure, we performed an additional NMR experiment. Namely, we formed the RCAN1core:CNA complex using 2H,13C,15N-labeled CNA and unlabeled RCAN1core and completed the sequence-specific backbone assignment of CNA in its RCAN1core-bound conformation of most detectable NH cross peaks (Fig. 2F). Because of incomplete H/D back exchange (as commonly observed for proteins expressed in D2O), the backbone assignment of CNA is ~30% [this percentage is about the same as the published NMR backbone assignment of free CNA (28)]. Overlaying the 2D [1H,15N] TROSY (Transverse relaxation optimized spectroscopy) spectra of 2H,15N-labeled CNA in the presence and absence of unlabeled RCAN1core showed, as expected, that many CSPs belong to the residues corresponding to the CN PxIxIT-binding pocket (residues Val104, Asn192, Glu325, Asn327, Asn330-Arg332, and Phe334). However, CSPs were also observed beyond the expected PxIxIT CNbinding pocket including residues Ser294, Thr296, Gly298, and Ser301, in a pocket in which the N terminus of the RCAN1core engages CNA, confirming the structure of the RCAN1core:CN complex (Fig. 2, D to F).

Previous studies have shown that CN activity can be inhibited by distinct mechanisms. First, CN can be competitively inhibited by binding and blocking the active site (27); this is how the CN AID inhibits CN before Ca2+ activation. Second, CN can be inhibited by blocking substrate binding; this is how the viral protein inhibitor A238L inhibits CN, as it binds CN and blocks both the PxIxIT- and LxVP-binding pockets (23). To understand how RCAN1 regulates CN, we measured CN activity using para-nitrophenylphosphate (pNPP; a small substrate mimic that reports active site inhibition) in the presence and absence of RCAN1 and a series of RCAN1 motif variants. The data show that RCAN1 potently inhibits CN activity against pNPP (Fig. 3, A and B, and table S3). Mutating the primary SLiM motifs (LxVP; PxIxIT; Figs. 1A and 3B) in RCAN1 alters CN inhibition in a manner consistent with their role in CN binding. Namely, the loss of the LxVP motif (RCAN1LxVPdead), which contributes very little to CN binding, has no effect on the RCAN1-mediated inhibition of CN phosphatase activity against pNPP. In contrast, the loss of the PxIxIT motif (RCAN1PxIxITdead), which is essential for CN binding, significantly reduced the RCAN1-mediated inhibition of CN phosphatase activity (Fig. 3, A and B). Together, these data show that although RCAN1 binds to both CN-specific SLiM-binding pockets (LxVP and PxIxIT), it must also bind and block the CN active site, and this interaction requires binding via the PxIxIT motif.

(A) Enzymatic activity of CN in the absence (black) and presence of RCAN1 (red) and its SLiM-binding motif variants (PxIxITdead, orange; LxVPdead, green); pNPP assays; SE, n = 3. (B) RCAN1 domain diagram illustrating the RCAN1 mutants tested with the corresponding CN activity (relative to free CN). (C) Same as (A) but with RCAN1 TxxP variants. (D) Cartoon illustrating how the TxxP motif engages the CN active site. Inset: Model of the RCAN1 TxxP (TRRP; purple) motif bound to CN based on the structure of the CNA AID domain (27); CN is shown as an electrostatic potential energy surface. (E) Same as (A) but with RCAN1 SSPASSP and TxxP variants. (F) Same as (D), illustrating CN inhibition in the absence of a functional TxxP motif. Inset: Model of the RCAN1 108SPP (cyan) motif bound to CN based on the structure of the CNA AID domain.

Two additional motifs have previously been suggested to facilitate RCAN1-mediated inhibition of CN: the 108SPPASPP114 motif, which resembles NFAT substrate sequences (25), and the 186TxxP189 motif, which we recently showed is critical for active site substrate recruitment (26). To determine whether the RCAN1 TxxP motif contributes to the RCAN1-mediated inhibition of CN, we measured CN activity when bound to the RCAN1TxxPdead variant (Fig. 3C). Although mutating this motif has only a minor effect on CN binding (threefold), it strongly reduced the RCAN1-mediated inhibition of pNPP dephosphorylation (29), resulting in a 50% increase in CN activity (Fig. 3, B and C); this suggests that the TxxP motif interacts directly with the catalytic site. Using the crystal structure of CN bound to the AID as a model for TxxP binding (AID residues 481ERMP484; Glu481 carboxyl binds the CN active site metals), it became evident that the RCAN1 i+2 arginine (TRRP) is perfectly poised to bind a negatively charged pocket formed by the side chains of CN residues Cys153 and Glu220 and the peptide carboxyls from residues Asn150 and Pro221 (Fig. 3D, inset). To test whether Arg188 is important for RCAN1-mediated inhibition of CN, we generated a TxxP motif variant in which only this residue is mutated to alanine, RCAN1TRAP (Fig. 3, B and C). This single amino acid change results in a 50% reduction in RCAN1-mediated inhibition of CN. Inclusion of additional mutations in this motif (TAAP; TAAA) had only minor effects (Fig. 3, B and C). Together, the data show that the TxxP motif is critical for the RCAN1-mediated inhibition of CN at the active site, with TxxP residue Arg188 playing a key role in TxxP motif binding (Fig. 3D).

Although the TxxP motif is critical for the RCAN1-mediated inhibition of CN, the RCAN1TxxPdead variant does not allow CN activity to return to RCAN1-free levels, demonstrating that additional elements of RCAN1 also contribute to CN inhibition. Previous studies suggested that the RCAN1 108SPPASPP114 motif may function as an NFAT-like pseudosubstrate inhibitor of CN activity (25). To test this, we again used mutagenesis coupled with pNPP activity assays (Fig. 3, B to E). First, mutation of either the 108SPP or 112SPP motif to AAA did not alter RCAN1-mediated inhibition of CN. We hypothesized that this may be due to the presence of the 186TRRP189 motif, which may play a dominant role in CN inhibition. Thus, we generated the same mutants in the RCAN1TxxPdead variant, i.e., 108SPPdead/TxxPdead and 112SPPdead/TxxPdead. The data show that in the absence of a functional TxxP motif, mutating the 112SPP motif led to a further ~20% reduction in RCAN1-mediated CN inhibition. Mutating the 108SPP motif completely abolished the RCAN1-mediated inhibition of CN, rendering CN fully active. These data demonstrate that both the RCAN1 TxxP motif and the RCAN1 108SPP pseudosubstrate sequence are responsible for the RCAN1-mediated inhibition at the CN active site (Fig. 3, D to F). To confirm these results, we formed the 2H,15N-RCAN1TXXPdead:2H-CN complex and recorded a 2D [1H,15N] TROSY spectrum (fig. S3). The NMR data showed that upon deletion of the TxxP motif in RCAN1, residues that are part of or surrounding the 108SPPASPP114 motif are missing or shifted (e.g., Ser112, Gly116, Lys118, and Thr124), directly showing that these residues are now in a different chemical environment, i.e., binding the CN active site (as indicated by the activity assays).

RCAN1 has eight serine and six threonine residues (Fig. 1A). Of these, Ser93, Ser94, Ser108, Ser112, Thr124, Ser136, Thr153, Ser163, Thr186, and Thr192 have been experimentally identified to be phosphorylated in vivo (16, 30, 31), suggesting that they may be important for regulating RCAN1 binding to and/or inhibition of CN. Consistent with this, a subset of these residues are part of and/or adjacent to known RCAN1 motifs (Ser94, LxVP motif; Ser108/Ser112, SPPASPP pseudosubstrate motif; Thr153/Ser163, PxIxIT motif; Thr186/Thr192, TxxP motif; Fig. 1A). To determine how phosphorylation of RCAN1 alters its ability to bind and inhibit CN, we incubated 15N-labeled RCAN1 with MKK6-activated p38 (pp38) (32), a proline-directed kinase that has been shown to phosphorylate RCAN1 (31, 33). An overlay of the nonphosphorylated and phosphorylated 2D [1H,15N] HSQC spectra reveals large CSPs for several peaks, indicative of phosphorylation (fig. S4A). After completing the sequence-specific backbone assignment of p38-phosphorylated RCAN1 (p-RCAN1), we determined that Ser108, Ser112, Thr124, Thr153, and Thr192 are phosphorylated (pS108, pS112, pT124, pT153, and pT192; pT152 also becomes partially phosphorylated, although much more slowly, suggesting that it is nonspecific). CSI/SSP analysis shows that p-RCAN1 maintains the same secondary-structure preferences as nonphosphorylated RCAN1 (RCAN1), and thus, p-RCAN1 is identical to RCAN1 in solution (fig. S4B). Despite this, p-RCAN1 binds CN ~30-fold more weakly than RCAN1 (ITC; fig. S1B and Table 1), demonstrating that RCAN1 phosphorylation negatively affects CN binding.

We reasoned that the reduction in affinity was due to the phosphorylation of Thr153, which is in a loop connecting RCAN1core strands 1 and 2 and immediately N-terminal to the RCAN1 PxIxIT motif, which is essential for CN binding (Figs. 1A and 2D). To test this, we generated the RCAN1T153A variant and phosphorylated it using pp38. The same residues, with the exception of Thr153 (now T153A), were phosphorylated; further, no nonspecific phosphorylation of Thr152 was observed (Fig. 4A). However, in contrast to p-RCAN1, the p-RCAN1T153A variant binds CN with the same affinity as nonphosphorylated RCAN1T153A (fig. S1B and Table 1), both of which are nearly identical to RCAN1. Together, these data show that phosphorylation of Thr153 in RCAN1 is a key mechanism regulating RCAN1:CN complex formation and, in turn, the ability of RCAN1 to inhibit CN.

(A) Overlay of the 2D [1H,15N] HSQC spectra of free 15N-labeled RCAN1T153A with 15N-labeled p-RCAN1T153A. Peaks corresponding to phosphorylated residues in p-RCAN1T153A are labeled. (B) Progress curves monitoring the dephosphorylation of pNPP for the indicated p-RCAN1T153A:CN complexes. (C) Cartoon diagrams of the RCAN1 variants tested in (B), with the indicated phosphorylated residues and resulting CN activity at the 30-min time point relative to CN alone.

Next, we investigated how RCAN1 phosphorylation affects the ability of RCAN1 to inhibit CN. To exclude any effects from a change in RCAN1:CN binding affinity, RCAN1T153A was used throughout these studies. pNPP was used as a model substrate and its dephosphorylation was monitored over time. As expected, RCAN1T153A completely inhibited CN activity (Fig. 4, B and C). In contrast, p-RCAN1T153A showed a ~30% reduction in inhibition, demonstrating that phosphorylation of either Ser108, Ser112, Thr124, or Thr192 negatively affects the ability of RCAN1 to inhibit CN (Fig. 4, B and C). Since our results showed that the 108SPPASPP114 motif, especially 108SPP, plays a key role in the RCAN1-mediated inhibition of CN, we reasoned that the loss of inhibition was due to the phosphorylation of Ser108. To test this, we repeated the activity assay using a variant of RCAN1 in which both residues were mutated to alanine and subsequently phosphorylated by pp38 (p-RCAN1T153A/S108A). Preventing Ser108 phosphorylation restored the ability of RCAN1 to inhibit CN, demonstrating that phosphorylated pS108 relieves 108SPP-mediated inhibition (Fig. 4, B and C).

Last, we tested the importance of the TxxP motif in p-RCAN1-mediated inhibition. Thus, we measured the activity of CN bound to p-RCAN1T153A in the TxxPdead background, i.e., p-RCAN1T153A/TxxPdead. The data show that this variant completely lost its ability to inhibit CN. Moreover, the CN activity increased threefold in the presence of p-RCAN1T153A/TxxPdead (Fig. 4, B and C). This result is consistent with previous studies that have shown that the activity of CN increases in the presence PxIxIT- and LxVP-containing CN-specific regulators, due to stabilization of the enzyme (23). Together, these data not only confirmed the key role of the TxxP motif in the RCAN1-mediated inhibition of CN independent of the RCAN1 phosphorylation state but also revealed that phosphorylation of either Thr153 or Ser108 reduces RCAN1-mediated inhibition either by weakening the affinity of RCAN1 for CN (pT153) or by preventing the SPPASPP pseudosubstrate motif from engaging and blocking the catalytic site (pS108).

Next, we used NMR spectroscopy to determine whether p-RCAN1 is also a CN substrate. Because the affinity of CN for p-RCAN1, but not p-RCAN1T153A, is considerably weaker than the corresponding nonphosphorylated variant, we again used the RCAN1T153A variant for these experiments. The data show that all p-RCAN1 residues phosphorylated by pp38 are dephosphorylated by CN, with pS108 and pT192 being the residues most rapidly dephosphorylated (Fig. 5A; pS112 is also dephosphorylated, but the dephosphorylation does not go to completion).

(A) Dephosphorylation of p-RCAN1T153A residues pS108, pS112, pT124, and pT192. (B) Dephosphorylation of p-RCAN1T153A/LxVPdead residues pS108, pS112, pT124, and pT192. (C) Dephosphorylation of p-RCAN1T153A/S108A residues pS112, pT124, and pT192. (D) Dephosphorylation of p-RCAN1T153A/TxxPdead residues pS108, pS112, pT124, and pT192.

The 108SPPASPP114 motif is C-terminal to the RCAN1 96LxVP99 motif. Our NMR, ITC, and SPR data show that while the RCAN1 LxVP motif binds CN, it does not contribute significantly to its affinity. This weaker affinity of the LxVP versus PxIxIT motif in RCAN1 has also been observed in other CN substrates, which has led to the hypothesis that LxVP motifs facilitate CN-mediated dephosphorylation of specific substrate residues. To test this hypothesis for RCAN1, we repeated the NMR-based dephosphorylation experiments using an RCAN1 variant in which the LxVP motif was nonfunctional (RCAN1LxVPdead) (Fig. 5B). The data show that the dephosphorylation rates for the phosphorylated residues either slowed (pT124 and pT192) or went to zero (pS108 and pS112). Thus, the residue that is most rapidly dephosphorylated in wild-type (WT) p-RCAN1, pS108, is unable to be dephosphorylated in the absence of the LxVP motif (p-RCAN1LxVPdead). These data strongly support the hypothesis that, in at least a subset of CN substrates, the LxVP functions to optimally position phosphosites for rapid dephosphorylation by CN. Inhibiting Ser108 phosphorylation using p-RCAN1T153A/108SPPdead also completely prevented the partial dephosphorylation of pS112, demonstrating the inability of this residue to effectively engage the active site in the absence of the 108SPP110 motif (Fig. 5C). Last, we also measured RCAN1 dephosphorylation using a variant with an inactive TxxP motif: pRCAN1T153A/TxxPdead. As expected, the dephosphorylation rates increased markedly, demonstrating that the TxxP motif functions limit access to the active site, fully consistent with our molecular and inhibition data (Fig. 5D).

Over 20 years ago, RCAN1 was found to be a potent, endogenous inhibitor of CN; however, how RCAN1 interacted with and regulated the activity of CN has remained elusive. This lack of mechanistic knowledge has limited our ability to effectively combat RCAN1-mediated inhibition to enhance NFAT dephosphorylation in syndromes and diseases associated with RCAN1 up-regulation, including DS and AD. Here, we show that the RCAN1 CN-interaction domain, which is intrinsically disordered, forms a tight complex with CN via its multiple SLiMs, including the LxVP, PxIxIT, and TxxP motifs (Fig. 6A). Unexpectedly, we found that the RCAN1 PxlxlT interaction is distinct from canonical PxlxlT:CN interactions, as PxIxIT motif binding to CN causes its 30 N-terminal residues to undergo a folding-upon-binding event. This results in the formation of a stable, tertiary domain stabilized by an extensive network of hydrophobic residues from both CN and RCAN1. This observation is important for multiple reasons. First, this unique example highlights the emerging diversity of SLiM-based interactions, demonstrating that even established interactions like that of the PxIxIT with CN can be augmented and their interaction strengths can be modulated by additional protein stabilization interactions. It will be interesting to see whether similar extended interactions exist for other SLiMs. Second, detecting this folding-upon-binding event and determining the folded structure required a hybrid approach that integrated NMR and crystallographic data, highlighting the importance of using both methods, but especially NMR spectroscopy, for the study of IDP:protein interactions.

(A) RCAN1 is a potent CN inhibitor. RCAN1 (magenta, with key sequence and structural features indicated) binds and inhibits CN (gray/beige; active site in yellow) via two mechanisms. First, RCAN1 blocks CN-specific substrate LxVP and PxIxIT interaction grooves by binding these pockets using its LxVP and especially its PxIxIT motifs. This prevents canonical CN substrates, like the NFATs, from binding CN. Second, RCAN1 directly blocks the CN active site via its TxxP motif and, to a lesser extent, its SPPASSP motif. This further reduces CN activity against its endogenous substrates. The RCAN1core folds upon binding CN. (B) Phosphorylation of RCAN1 T153 (pT153) inhibits CN binding. Active p38 phosphorylates RCAN1 on pS108, pS112, pT124, pT153, and pT192. Phosphorylation of RCAN1 T153 (pT153), which is immediately N-terminal to the PxIxIT motif, lowers the affinity of RCAN1 for CN, leading to dissociation of the complex and increased CN activity. (C) Phosphorylated RCAN1 (pRCAN1) is weakly dephosphorylated by CN. In the absence of phosphorylation at T153 (pRCAN1T153A; T153A represented as an orange start), pRCAN1 is able to bind CN with strong affinity. This results in the slow dephosphorylation of pS108, pT124, and pT192, with pS112 becoming partially dephosphorylated (top). However, with the exception of pT192, these dephosphorylation events require the presence of the nearby LxVP motif. That is because, in the LxVPdead mutant (T153A and LxVPdead represented as orange stars), pS108, pS112, and pT124 remain phosphorylated, even after 20 hours (middle). Last, if the inhibitory TxxP sequence is inactivated (TxxPdead; T153A and TxxPdead represented as orange stars), the ability of CN to dephosphorylate pRCAN1 is substantially enhanced, with dephosphorylation rates increasing ~10-fold (bottom).

We also found that this novel RCAN1 interaction domain is the target of posttranslational modifications, namely, phosphorylation. Thr153 is a proline-directed kinase phosphorylation site that is readily phosphorylated by activated p38. As we show, phosphorylation of Thr153 reduces the binding affinity of RCAN1 for CN by more than 30-fold, which greatly attenuates its ability to bind, and, in turn, inhibit CN. Others have reported that phosphorylation of a nearby RCAN1 residue, Ser136, also transforms RCAN1 from an inhibitor to an activator, which enhances CN-NFAT signaling via NFAT translocation to the nucleus (34). Our new structural data show that Ser136 is located at the center of the newly formed PxIxIT domain, strongly suggesting that phosphorylation of Ser136 also destabilizes the RCAN1:CN interaction in a manner similar to that observed for Thr153. Together, these data show how phosphorylation functions as a molecular switch to convert RCAN1 from a potent to a weak inhibitor (sometimes called activator) of CN.

Our data reveal why RCAN1 is such a potent inhibitor of CN. Namely, it uses a combination of distinct yet complementary inhibitory mechanisms. First, RCAN1 binds to and blocks the PxIxIT and LxVP substrate-binding sites. This is the same mechanism used by the potent swine flu viral inhibitor A238L (23) and the blockbuster immunosuppressant drugs FK-506/cyclosporin-A (27, 35); that is, RCAN1 sterically occludes other substrates from binding the PxIxIT and LxVP substrate-binding pockets, which, in turn, prevents them from being dephosphorylated. The efficacy of FK-506/cyclosporin-A in preventing CN-mediated NFAT dephosphorylation demonstrates the effectiveness of this inhibitory strategy (36). However, RCAN1 also uses a second mechanism to inhibit CN activity. Namely, it binds and blocks the active site directly. Recently, we showed that S/TxxP motif substrate phosphosites are directly recruited to the CN active site (26). It is exactly such a motif, TRRP in RCAN1, which binds and blocks the CN active site, inhibiting its ability to dephosphorylate both small substrate mimetics (pNPP) and protein substrates (pRCAN1). In particular, we showed that the RP residues in TRRP are key for the local interaction at the active site and thus the potent inhibition of CN. Our data also showed that RCAN1 engages the active site using a second motif, one that was originally proposed to mimic known dephosphorylation sites in NFAT, SPPASPP (108SPPA112SPP). Thus, in addition to binding and sterically blocking known CN substrate interaction motifs, RCAN1 also exploits two distinct pseudosubstrate motifs to bind and inhibit the CN active site directly, demonstrating why RCAN1 potently inhibits CN activity.

What can we learn from these mechanisms of CN binding and inhibition by RCAN1 for CN inhibition and substrate selection? First, diffusion-controlled small-molecule dephosphorylation (e.g., for pNPP) and protein substrate dephosphorylation are different because the latter has the advantage of tethering via the CN-specific substrate interaction motifs (PxIxIT and LxVP). This tethering markedly increases the local substrate (phosphosite) concentration, which, in turn, enhances its dephosphorylation. This is why pNPP dephosphorylation by CN is completely inhibited by RCAN1, while substrate-like engagements, e.g., by p-RCAN1, leads to dephosphorylation, albeit exceedingly slowly (as we show, a functional TxxP motif greatly slows CN-mediated dephosphorylation of p-RCAN1). Furthermore, we show that for effective substrate dephosphorylation, the LxVP motif is critical, as, without it, the phosphosites immediately C-terminal to the LxVP motif are no longer dephosphorylated. Recently, we showed that the LxVP binding affinity for CN strictly depends on its on-rate (kon), i.e., how accessible the LxVP motif is in the ensemble of structures formed by RCAN1 in solution (37). Despite the fact that the RCAN1 LxVP site (LAPP) engages CN only very weakly, these data show that the LxVP is still essential for the specific and efficient dephosphorylation of 108SPPA112SPP, with a preference for 108SPP (the preference suggests that a PPA sequence N-terminal to a dephosphorylation site is unfavorable for dephosphorylation by CN). Together, these results reveal how RCAN1 inhibits CN and how this inhibition is regulated by phosphorylation. The mode of active site engagement is likely mirrored by CN substrates and thus this work also provides molecular insights into substrate engagement with the CN active site.

Last, these data also provide the essential molecular details needed to develop therapeutics that disrupt RCAN1-mediated inhibition of CN in syndromes and disorders associated with RCAN1 overexpression. Namely, RCAN1 dissociation can be achieved by specifically limiting the formation of the hydrophobic network necessary to form the extended PxIxIT interaction, e.g., by small molecules. Furthermore, up-regulation of kinases that allow for RCAN1 phosphorylation on residues Ser136 and Thr153 are equally good routes to limit the effectiveness of this interaction and thus ultimately CN inhibition. Last, identification and inhibition of the necessary phosphatases for Ser136 and Thr153 will be another possibility for inhibition release. While none of these are short-term projects, the molecular insights presented here lay the foundation for these important efforts.

DNA coding the human RCAN1 CN-binding domain (residues 89 to 197) was subcloned into pTHMT (His6MBP-TEV-) as previously described (38). RCAN1 variants [RCAN1core (residues 128 to 164), 108SPPdead (S108A/P109A/P110A), 112SPPdead (S112A/P113A/P114A), TxxPdead (T186A/R187A/R188A/P189A), 108SPPdead/TxxPdead, 112SPPdead/TxxPdead, LxVPdead (H95A/L96A/P98A/P99A), PxlxlTdead (P154A/V156A/V158A/H159A), T153A, T153A/S108A, T153A/LxVPdead, and T153A/TxxPdead] were generated using site-directed mutagenesis following recommended protocols (QuikChange; Agilent). CNA (residues 27 to 348DD; for NMR; CN isoform was used throughout the study) and CNA (residues 27 to 339; for crystallography) were subcloned into pRP1B, containing an N-terminal His6-tag. CN391 (CNA1391/CNB1170) and CN370 (CNA1370/CNB1170) were cloned into the p11 bicistronic bacterial expression vector as a single cassette, which contains an N-terminal His6-tag followed by a TEV (Tobacco etch virus) protease cleavage site, as previously described (23). These CN constructs do not contain the AID. For protein expression, plasmid DNAs were transformed into Escherichia coli BL21 (DE3) RIL cells (Agilent). Cells were grown in Luria broth in the presence of selective antibiotics at 37C to an OD600 (optical density at 600 nm) of ~1.0, and expression was induced by the addition of 1 mM isopropyl--d-thiogalactopyranoside. Induction proceeded for ~4 hours at 37C or overnight at 18C before harvesting by centrifugation at 8000g. Cell pellets were stored at 80C until purification.

For NMR measurements, expression of uniformly (1H,15N)-, (2H,15N)-, (1H,15N,13C)-, or (2H,15N,13C)-labeled proteins was achieved by growing cells in H2O- or D2O-based M9 minimal media containing 15NH4Cl (1 g/liter) and/or [1H,13C]- or [2H,13C]-d-glucose [4 g/liter; CIL (Cambridge Isotope Laboratories) or Isotec] as the sole nitrogen and carbon sources, respectively, using established protocols (39). Selectively /-[13CH3, 12CD3]-Val/Leu, -[13CH3]-Ile, [U]-2H,15N-labeled RCAN1 were prepared in D2O-based M9 minimal media containing 15NH4Cl (1 g/liter) and/or [2H,12C]-d-glucose (4 g/liter) through addition, 1 hour before induction, of -ketoisovaleric acid (120 mg/liter; CDLM-7317; Cambridge Isotope Laboratories) and -ketobutyric acid (60 mg/liter; CDLM-7318; Cambridge Isotope Laboratories).

Cell pellets were lysed in lysis buffer [25 mM tris (pH 8.0), 500 mM NaCl, 5 mM imidazole, and 0.1% Triton X-100] containing EDTA-free protease inhibitor cocktail (Roche) using high-pressure homogenization (Avestin C3). The lysate was clarified by centrifugation at 42,000g and filtered through a 0.22-m PES (polyethersulfone) filter before loading onto a His-trap HP column (GE Healthcare). Bound proteins were washed with buffer A [50 mM tris (pH 8.0), 500 mM NaCl, and 5 mM imidazole] and eluted with increasing amounts of buffer B [50 mM tris (pH 8.0), 500 mM NaCl, and 500 mM imidazole] using a 5 to 500 mM imidazole gradient. Peak fractions were pooled and dialyzed overnight at 4C in high-salt dialysis buffer [50 mM tris (pH 8.0), 500 mM NaCl, and 0.5 mM TCEP] with 5:1 volume ratio of TEV protease overnight for RCAN1 or low-salt dialysis buffer [50 mM tris (pH 8.0), 50 mM NaCl, 0.5 mM TCEP, and 1 mM CaCl2] for CN. The next day, a subtraction His6-tag purification was performed to remove TEV and the cleaved His6-tag. RCAN1 was concentrated to ~6 ml; 5 mM (final concentration) DTT was added and further heat-purified (80C; 10 min). RCAN1 was further purified using size exclusion chromatography (Superdex 75 26/60) in either assay buffer [150 mM Hepes (pH 7.5), 150 mM NaCl, 0.5 mM TCEP, 1 mM CaCl2, and 0.5 mM MgCl2] or NMR buffer [20 mM Hepes (pH 6.8), 50 mM NaCl, 0.5 mM TCEP, and 1 mM CaCl2]. Cleaved CN was further purified using a HiTrap Q HP anion exchange column (GE Healthcare). Purified protein was either used immediately or flash-frozen in liquid nitrogen for storage at 80C.

The activities of freshly prepared CN in complex with RCAN1 and variants were measured in assay buffer containing varying concentrations of pNPP (0 to 120 mM). CN (0.1 M) and 0.5 M RCAN1s were incubated with the substrate at 30C for 30 min. The reaction was stopped using 1 M NaOH, and the absorbance was measured at 405 nm using a plate reader (BioTek). The rate of pNPP dephosphorylation was determined using the molar extinction coefficient for pNPP of 18,000 M1 cm1 and an optical path length of 0.3 cm (96-well plates). Km and Vmax were determined by fitting to the Michaelis-Menten equation, y = Vmax*x/(Km + x); kcat was extracted using y = Et*kcat*x/(Km + x). The catalytic efficiency was obtained as kcat/Km. SigmaPlot 12.5 was used for data analysis including the statistical analysis. The activities of freshly prepared CN in complex with p-RCAN1 and p-RCAN1 variants were measured in assay buffer containing 100 mM pNPP. CN (0.1 M) with 0.5 M p-RCAN1s were incubated with the substrate at 30C; the absorbance at 405 nm was measured every 30 s for 30 min. All experiments were carried out in triplicate.

NMR data were collected on either Bruker NEO 600- and 800-MHz spectrometers or a Bruker Avance III HD 850-MHz spectrometer equipped with TCI HCN z-gradient cryoprobes at 298 K. NMR measurements of RCAN1 were recorded using (1H,15N)-, (2H,15N)-, (1H,15N,13C)-, or selectively /-[13CH3, 12CD3]-Val/Leu, -[13CH3]-Ile, [U]-2H,15N-labeled protein at a final concentration of 0.6 mM in NMR buffer and 90% H2O/10% D2O. The sequence-specific backbone assignments of RCAN1 and variants, as well as CN-bound RCAN1 were achieved using 3D triple-resonance experiments including 2D [1H,15N] HSQC/TROSY, 3D HNCA, 3D HN(CO)CA, 3D HN(CO)CACB, and 3D HNCACB. All NMR data were processed using TopSpin 4.05 (Bruker BioSpin) and analyzed using CARA (http://cara.nmr.ch) and/or CcpNMR (40). 2D [1H,13C]-HMQC, 3D 13C-ILV-methyl-methyl resolved [1H,1H] NOESY, 13C-methyl-ILV-15N resolved [1H,1H] NOESY, and 15N-resolved [1H,1H] NOESY spectra were recorded with a mixing time (TM) of 120 ms using the selectively /-[13CH3, 12CD3]-Val/Leu, -[13CH3]-Ile, [U]-2H,15N-labeled RCAN1 and [U]-2H-labeled CNA complex. The NOE data were also used to assign the chemical shifts of /-CH3 of Val/Leu, -[CH3]-Ile.

Dephosphorylation was initiated by the addition of unlabeled active CN (CN activity was always tested using pNPP as a substrate before its use in NMR experiments) with 50 M 15N-labeled RCAN1 (and variants) with a molar excess of CN:RCAN1 of 1:10 or 1:20. A reference 2D [1H,15N] HSQC spectrum was recorded before the addition of CN. Dephosphorylation was monitored by extracting the intensities from a series of 2D [1H,15N] HSQC spectra. Apparent rates of dephosphorylation were extracted from global nonlinear least square fits of disappearing phosphorylated peaks and/or reporting neighbor peaks to single exponentials in SigmaPlot.

Purified CNA27339 and RCAN1core were incubated together on ice in a 1:1.2 molar ratio for 6 hours before complex purification using size exclusion chromatography in complex buffer [10 mM tris (pH 7.4), 50 mM NaCl, and 1 mM dithiothreitol (DTT)]. The peak corresponding to the complex was pooled and concentrated to 10.3 mg/ml. The CNA:RCAN1core complex crystallized in 0.1 M sodium cacodylate (pH 5.5), 12% polyethylene glycol 8000, and 0.1 M calcium acetate (sitting drop vapor diffusion). Crystals were cryoprotected using a 15-s soak in mother liquor supplemented with 30% (v/v) glycerol and immediately flash-frozen in liquid nitrogen. Data were collected at the APS GM/CAT 23ID-D and processed using HKL3000 (41). The structure was phased using the CNA subunit of Protein Data Bank (PDB) ID 4F0Z as a search model [PHASER as implemented in Phenix (42)]. A solution was obtained in space group P212121; clear electron density for the RCAN1 PxIxIT motif was observed in the initial maps. The initial models of the complex were built without RCAN1core using AutoBuild, followed by iterative rounds of refinement in Phenix and manual building using Coot (43). The RCAN1core PxIxIT sequence was then modeled, followed by additional rounds of refinement and model building. The final model includes RCAN1 residues 153 to 161. Additional electron density, corresponding to ~4 amino acids, was immediately adjacent to the RCAN1 residues at a crystal contact; however, no residues were built due to a lack of features that allowed the sequence or chain direction to be confidently determined. Data collection and refinement details are provided in table S2. Molecular figures were generated using PyMOL (44).

RCAN1128164 structures in complex with CNA were calculated using a corefinement protocol closely following that described in (45) and implemented using version 2.50 of Xplor-NIH (46). Interproton distances derived from NOEs were represented by restraints allowing a distance range of 1.8 to 6 . The dihedral angles and were calculated using TALOS (chemical shifts of HN, HA, HB, CA, and CB) (47). Xplor-NIHs PosDiffPot term was used to restrain the C coordinates to values determined in the crystal structure of CNA bound to the RCAN1128164 PxlxlT strand (amino acids 153 to 161) with 0.5 root mean square deviation allowed with zero energy penalty. In addition to the PosDiffPot term, the HBPot energy (48) term was included to form and improve any hydrogen bonding geometry. RCAN1 atoms, all CNA side-chain atoms, and all atoms of CNAs interfacial residues (amino acids 327 to 336, 318, 286, 288, 290, 293, 299, and 300) were allowed torsion angle degrees of freedom during simulated annealing and all degrees of freedom during a final Cartesian minimization. Backbone atoms of the noninterfacial CNA atoms were held rigid throughout. RCAN1128164 was folded from randomized extended coordinates in the initial step of refinement. An initial temperature of 3500 K and a final temperature of 25 K were used for simulated annealing. A total of 100 complex structures were calculated during the first cofold step and 10 structures with the lowest energies were used as inputs for a refinement step. A total of five refinement iterations were performed, and 10 conformers with the lowest energies were used to represent the structure of RCAN1128164 when bound to CN. For the final ensemble of RCAN1128164 structures, no distance violations more than 0.5 and no torsion angle violation more than 3 were identified.

Expression and purification of human p38 and MKK6 were carried out as previously described (32). Phosphorylated p38 was produced by p38 incubation with constitutively active MKK6S207E/S211E (1:40 molar ratio). The reaction components were added to a 50-ml conical tube in the following order to achieve the reported final concentrations in a 30-ml volume: 20 mM Hepes (pH 7.5), 0.5 mM EDTA, 2 mM DTT, 20 mM MgCl2, 0.05 M MKK6S207E/S211E, 2 M p38, and 4 mM adenosine 5-triphosphate (ATP). The mixture was mixed by pipetting up/down several times and was incubated at 27C for 5 min before adding ATP (Roche). After the addition of ATP, the reaction was incubated in a water bath at 27C for 5 hours. Following incubation, the mixture was exchanged into buffer A [20 mM tris (pH 7.6), 75 mM NaCl, and 0.5 mM TCEP (tris(2-carboxyethyl)phosphine)] to remove ATP using an Amicon Ultra-15 filter (Millipore). Upon ATP removal, the solution was filtered and loaded onto a Mono Q 5/50 GL column (GE Healthcare) pre-equilibrated in buffer A and eluted with a linear gradient of 0 to 100% buffer B [20 mM tris (pH 7.6), 0.4 M NaCl, and 0.5 mM TCEP]. RCAN1 variants were phosphorylated by pp38 in buffer [20 mM Hepes (pH 7.5), 20 mM MgCl2, 1 mM DTT, and 4.8 mM ATP]. pp38 (0.05 M) was added to 2 M RCAN1 and variants. The phosphorylation reaction was incubated at 27C for 24 hours and stopped by heating at 65C for 10 min.

Protein concentration of CN and RCAN1 variants was measured in triplicate using either the Pierce 660 (Thermo Fisher Scientific) or the AccuOrange Protein Quantitation assays (Biotium). CN and RCAN1 variants were equilibrated in 20 mM tris (pH 7.5), 150 mM NaCl, 1 mM CaCl2, and 0.5 mM TCEP. RCAN1 variants (80 to 100 M, syringe) were titrated into CN (5 to 10 M, cell) using a VP-ITC microcalorimeter (Malvern) or an Affinity-ITC microcalorimeter (TA Instruments) with a 250-s interval at 25C. Twenty-five injections were delivered during each experiment over a period of 20 s (VP-ITC microcalorimeter) or 10 s (Affinity-ITC microcalorimeter), and the solution in the sample cell was stirred at 307 rpm (VP-ITC microcalorimeter) or 125 to 200 rpm (Affinity-ITC microcalorimeter) to ensure rapid mixing. All ITC data were analyzed using NITPIC (49) and fitted to a single-site binding model using SEDPHAT (50); figures were generated using GUSSI (51). To distinguish between the different transitions, we defined a H 0.35 kcal/mol as baseline, which allows for a completely unbiased data analysis.

SPR measurements were performed at 25C and a sampling rate of 5 Hz using a four-channel Reichert 4SPR instrument fitted with an autosampler and a degassing pump (Reichert Technologies). SPR buffers containing 20 mM tris (pH 7.5), 500 mM NaCl, 1 mM CaCl2, 0.5 mM TCEP, and 0.05% Tween were sterile-filtered and degassed in autoclaved glassware. Running buffers were used to prime and run both the sample and syringe pump reservoirs before each experiment. Gold sensor chips modified with Ni-NTAfunctionalized dextran (NiD50L) were installed and equilibrated under flow conditions (100 l/min) for 60 min. Surface contaminants were cleared by a pair of 120-l injections of 10 mM NaOH during equilibration. Experiments were initiated by injecting 120 l of His6-CN370 (200 nM) diluted in 20 mM tris (pH 7.5), 500 mM NaCl, 1 mM CaCl2, 0.5 mM TCEP, and 0.05% Tween onto channels 1, 2, and 3 for 120 s at 50 l/min, which resulted in between 450 and 500 RIU of surface loading (channel 4 was used as reference surfaces). The sensor chip was allowed to equilibrate for 20 min at 50 l/min before beginning the experiments. Purified RCAN1 variants were diluted into running buffer to final concentrations of 1.25, 2.5, 5, 10, and 20 nM. A single 60-l RCAN1 injection was applied for 60 s at 50 l/min followed by a dissociation step of 180 s. For each concentration of RCAN1 injection, chip surface was prepared with stripping with 350 mM EDTA (pH 8), reconditioning the surface with 10 mM NaOH to remove nonspecifically bound CN aggregates, charging the surface with 40 mM NiSO4, and reloading fresh CN onto the surface. For all experiments, two buffer blank injections were included to achieve double-referencing. Technical replicates were obtained by using three channels per chip. Kinetic parameters were determined by curve-fitting using TraceDrawer software (Ridgeview Instruments AB) fit with a one-to-one model using local Bmax.

Multiple sequence alignment was performed using RCAN1 C-terminal CN-binding domains from human (Hs), Ovis aries (Oa), Bos taurus (Bt), Canis lupus dingo (Cl), Mus musculus (Mm), Rattus norvegicus (Rn), Xenopus tropicalis (Xt), Danio rerio (Dr), Drosophila novamexicana (Dn), Apis mellifera (Am), and Caenorhabditis elegans (Ce).

Acknowledgments: We are grateful to A. Oot for help at the beginning stages of this project, and we thank T. Moon for help with SPR data collection and evaluation. We thank M. Cyert for CN discussions. Funding: This work was supported by grant R01NS091336 from the National Institute of Neurological Disorders and Stroke to W.P. and grant R01GM098482 from the National Institute of General Medicine to R.P. Crystallographic data were collected on beamline 23-ID-D at APS, Argonne National Laboratory. This research used resources of the Advanced Photon Source, a U.S. Department of Energy (DOE) Office of Science User Facility operated for the DOE Office of Science by Argonne National Laboratory under contract no. DE-AC02-06CH11357. CDS was supported by the Intramural Research Program of the Center for Information Technology at the National Institutes of Health. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Author contributions: W.P., R.P., and Y.L. developed the concept. Y.L. designed, optimized, and performed in vitro phosphorylation and dephosphorylation experiments using NMR spectroscopy. S.G. performed RCAN189-197 backbone assignment, SSP, and hetNOE analysis. Y.L. performed CN and all other RCAN1 variant NMR backbone assignments and NMR studies. R.P. and S.R.S. designed, optimized, and performed crystallization and structure determination experiments. Y.L. and S.G. performed and analyzed ITC experiments. Y.L. performed SPR measurements and analysis and CN activity experiments and analysis. Y.L. and C.D.S. performed structure corefinement. W.P., R.P., and Y.L. wrote the manuscript with comments and inputs from all coauthors. Competing interests: The authors declare that they have no competing interests. Data and materials availability: All data needed to evaluate the conclusions in the paper are present in the paper and/or the Supplementary Materials. Additional data related to this paper may be requested from the authors. All NMR chemical shifts have been deposited in the BioMagResBank (BMRB: 27801, 27994, 27995, 27996, and 27997). Atomic coordinates and structure factors have been deposited in the Protein Data Bank (PDB: 6UUQ).

Read more from the original source:
The structure of the RCAN1:CN complex explains the inhibition of and substrate recruitment by calcineurin - Science Advances

Recommendation and review posted by Alexandra Lee Anderson

Rewriting the Rules of Vaccine Design With DNA Origami – Technology Networks

By folding DNA into a virus-like structure, MIT researchers have designed HIV-like particles that provoke a strong immune response from human immune cells grown in a lab dish. Such particles might eventually be used as an HIV vaccine.

The DNA particles, which closely mimic the size and shape of viruses, are coated with HIV proteins, or antigens, arranged in precise patterns designed to provoke a strong immune response. The researchers are now working on adapting this approach to develop a potential vaccine for SARS-CoV-2, and they anticipate it could work for a wide variety of viral diseases.

The rough design rules that are starting to come out of this work should be generically applicable across disease antigens and diseases, says Darrell Irvine, who is the Underwood-Prescott Professor with appointments in the departments of Biological Engineering and Materials Science and Engineering; an associate director of MITs Koch Institute for Integrative Cancer Research; and a member of the Ragon Institute of MGH, MIT, and Harvard.

Irvine and Mark Bathe, an MIT professor of biological engineering and an associate member of the Broad Institute of MIT and Harvard, are the senior authors of the study, which appears today inNature Nanotechnology. The papers lead authors are former MIT postdocs Rmi Veneziano and Tyson Moyer.

DNA design

Because DNA molecules are highly programmable, scientists have been working since the 1980s on methods to design DNA molecules that could be used for drug delivery and many other applications, most recently using a technique called DNA origami that was invented in 2006 by Paul Rothemund of Caltech.

In 2016, Bathes lab developed an algorithm that can automatically design and build arbitrary three-dimensionalvirus-like shapesusing DNA origami. This method offers precise control over the structure of synthetic DNA, allowing researchers to attach a variety of molecules, such as viral antigens, at specific locations.

The DNA structure is like a pegboard where the antigens can be attached at any position, Bathe says. These virus-like particles have now enabled us to reveal fundamental molecular principles of immune cell recognition for the first time.

Natural viruses are nanoparticles with antigens arrayed on the particle surface, and it is thought that the immune system (especially B cells) has evolved to efficiently recognize such particulate antigens. Vaccines are now being developed to mimic natural viral structures, and such nanoparticle vaccines are believed to be very effective at producing a B cell immune response because they are the right size to be carried to the lymphatic vessels, which send them directly to B cells waiting in the lymph nodes. The particles are also the right size to interact with B cells and can present a dense array of viral particles.

However, determining the right particle size, spacing between antigens, and number of antigens per particle to optimally stimulate B cells (which bind to target antigens through their B cell receptors) has been a challenge. Bathe and Irvine set out to use these DNA scaffolds to mimic such viral and vaccine particle structures, in hopes of discovering the best particle designs for B cell activation.

There is a lot of interest in the use of virus-like particle structures, where you take a vaccine antigen and array it on the surface of a particle, to drive optimal B-cell responses, Irvine says. However, the rules for how to design that display are really not well-understood.

Other researchers have tried to create subunit vaccines using other kinds of synthetic particles, such as polymers, liposomes, or self-assembling proteins, but with those materials, it is not possible to control the placement of viral proteins as precisely as with DNA origami.

For this study, the researchers designed icosahedral particles with a similar size and shape as a typical virus. They attached an engineered HIV antigen related to the gp120 protein to the scaffold at a variety of distances and densities. To their surprise, they found that the vaccines that produced the strongest response B cell responses were not necessarily those that packed the antigens as closely as possible on the scaffold surface.

It is often assumed that the higher the antigen density, the better, with the idea that bringing B cell receptors as close together as possible is what drives signaling. However, the experimental result, which was very clear, was that actually the closest possible spacing we could make was not the best. And, and as you widen the distance between two antigens, signaling increased, Irvine says.

The findings from this study have the potential to guide HIV vaccine development, as the HIV antigen used in these studies is currently being tested in a clinical trial in humans, using a protein nanoparticle scaffold.

Based on their data, the MIT researchers worked with Jayajit Das, a professor of immunology and microbiology at Ohio State University, to develop a model to explain why greater distances between antigens produce better results. When antigens bind to receptors on the surface of B cells, the activated receptors crosslink with each other inside the cell, enhancing their response. However, the model suggests that if the antigens are too close together, this response is diminished.

Beyond HIV

In recent months, Bathes lab has created a variant of this vaccine with the Aaron Schmidt and Daniel Lingwood labs at the Ragon Institute, in which they swapped out the HIV antigens for a protein found on the surface of the SARS-CoV-2 virus. They are now testing whether this vaccine will produce an effective response against the coronavirus SARS-CoV-2 in isolated B cells, and in mice.

Our platform technology allows you to easily swap out different subunit antigens and peptides from different types of viruses to test whether they may potentially be functional as vaccines, Bathe says.

Because this approach allows for antigens from different viruses to be carried on the same DNA scaffold, it could be possible to design variants that target multiple types of coronaviruses, including past and potentially future variants that may emerge, the researchers say.

Reference: Veneziano et al. (2020). Role of nanoscale antigen organization on B-cell activation probed using DNA origami. Nature Nanotechnology. DOI: 10.1038/s41565-020-0719-0.

This article has been republished from the following materials. Note: material may have been edited for length and content. For further information, please contact the cited source.

Original post:
Rewriting the Rules of Vaccine Design With DNA Origami - Technology Networks

Recommendation and review posted by Alexandra Lee Anderson

After this COVID winter comes an AI spring – VentureBeat

During boom times, companies focus on growth. In tough times, they seek to improve efficiency. History shows us that after every major economic downturn since the 1980s, businesses relied on digital technology and, specifically, innovations in software technology to return to full productivity with fewer repetitive jobs and less bloat.

The years Ive spent as a VC have convinced me that this is the best time to start an AI-first enterprise, not despite the recession, but because of it. The next economic recovery will both be driven by artificial intelligence and accelerate its adoption.

While the Great Recession is often thought of as a jobless recovery, economists at the National Bureau of Economic Research (NBER) found that the downturn accelerated the shift from repetitive to non-routine jobs at both the high and low ends of the spectrum. So, yes, existing tasks were automated, but companies empowered their employees with data and analytics to augment their judgment to improve productivity and quality, in a virtuous cycle of data and judgment that both increased profitability and created more rewarding work.

Indeed, the highest levels of unemployment during the Great Recession were followed by a surge in enrollment in post-secondary education in analytics and data science as people sought out opportunities to upskill. And the period was followed by a recovery in which despite increased automation unemployment fell to historic lows.

Through no fault of our own, were again thrust into the cycle of recession and recovery. Industries already expect to benefit from improved AI and machine learning in the next recovery. That expectation will create new opportunities for AI entrepreneurs.

Every economic recovery is defined by an emerging software technology and set of applications.

The companies that grew in the lackluster economy of the early 1980s staged the first software IPOs when the economy rebounded in the middle of that decade: Lotus, Microsoft, Oracle, Adobe, Autodesk and Borland.

Packaged software signified a unique turning point in the history of commercial enterprise; the category required little in the way of either CAPEX or personnel costs. Software companies had gross margins of 80% or more, which gave them amazing resilience to grow or shrink without endangering their existence. If entrepreneurs were willing to work for lower wages, software companies could be started quickly with minimal to no outside investment, and if they could find early product-market fit, they could often bootstrap and grow organically.

Those new software companies were perfectly adapted to foster innovation when recessions hit, because high-quality people were available and less expensive, and office space was abundant. At the same time, established companies put new product development on hold while they tried to service and keep existing customers.

I started working as a VC in 1990 for the first venture firm that focused purely on investing in software, Hummer Winblad. While it took hard work and tenacity for John Hummer and Ann Winblad to raise that first fund, their timing as investors turned out to be perfect. A recession began in the second quarter of that year and lasted through Q1 1991.

The software companies coming out of that recession pioneered cost-effective client-server computing. Sybase, which established this trend with its Open Client-Server Interfaces went public in 1991, after growing 54% in the previous year.

By then, universities had graduated many programmers, creating a talent pool for startups. New software developer platforms made those programmers more productive. The 1990s became the first golden era for enterprise computing. One Hummer Winblad company, Arbor Software, invented the category of Online Analytical Processing (OLAP). Another, Powersoft, became the dominant no-code client server development platform. It was the industrys first billion-dollar software acquisition.

The first CRM companies, spawned in that recession, held successful IPOs from 1993 to 1999. This class included Remedy a company that BusinessWeek breathlessly called Americas Number One Top Hot Growth Company in 1996. Scopus, Vantive, and Clarify all grew rapidly and went public or were acquired in this period or shortly thereafter.

That exuberance ended with the dot-com bust in March 2000.

At that time, Salesforce had existed for only a year. Concur was a relatively new company, forced to reinvent itself when its packaged software business collapsed. Many people would have thought their timing was terrible, but they were unhindered by the obligation to service an installed base during the 2001 recession that followed the bust. That left them free to innovate, and they became two of the very first SaaS businesses.

Salesforce went public in 2004, and now has a market cap of about $135 billion. In 2013, Concur sold to SAP for $8.3 billion. Amazon Web Services was also conceived during that recession and launched in July 2002. SaaS and cloud computing leveraged each other for the rest of the decade.

When the sub-prime mortgage crisis brought the entire economy down, companies had to retain customers and improve efficiency goals that are often at odds with each other. The idea of a big data future had already taken root, and forward-thinking executives suspected that the solution was already in their data, if they could only find it. But at the same time, established software companies also cut R&D spending. That opened up fertile ground for newer and more agile analytics companies.

Most software companies saw no growth in 2009, but Omniture, a leader in web analytics, grew more than 80% that year, prompting its acquisition by Adobe for $1.9 billion. Tableau had been founded back in 2003, but it grew slowly until the recession. From 2008-2010, it grew from $13 million to $34 million in sales. Over the same period Splunk went from $9 million to $35 million. Ayasdi, Cloudera, Mapr and Datameer were all launched in the depths of the Great Recession.

Of course, none of those companies could have flourished without data scientists. Just as universities accelerated the creation of software developers in the early 1990s, they again accelerated the creation of analytics experts and data scientists during the Great Recession, which again helped to spur the recovery and drive a decade of economic expansion, job growth, and the longest bull market in American history.

Even before the pandemic, many economists and corporate CFOs felt there was at least a 50% chance of recession in 2020.

Over a year ago, The Parliament the policy magazine published by the EU Parliament predicted that the next recession would usher in a wave of AI. The magazine quoted Mirko Draca, of the London School of Economics as saying, We expect to see another technology surge in the next 10 to 15 years, based on AI and robotics technology.

Those who predicted a mere recession were, to say the least, insufficiently pessimistic. Companies have reduced their labor costs more aggressively than ever to match the suddenness and seriousness of the situation. Once again, theyll rely on automation to boost production when the recovery begins.

The Atlantic Council surveyed over 100 technology experts on the impact that COVID-19 would have on global innovation. Even in the midst of the pandemic, those experts felt that over the next two to five years, data and AI would have more impact than medical bioengineering. The two are not mutually exclusive; Googles Deepmind Technologies recently used its AlphaFold tool to predict complex protein folding patterns, useful in the search for a vaccine.

Companies emerging from this recession will adapt processes to vaccinate their systems against the next pandemic. In response to supply-chain disruptions, Volkswagen is considering increasing its 3D printing capabilities in Germany, which would give the automaker a redundant parts source. The government-run Development Bank of Japan will subsidize the costs of companies that move production back to Japan.

Bringing production back onshore while controlling costs will require significant investment in robotics and AI. Even companies that dont have their own production capacity, such as online retailers, plan to use AI to improve the reliability of complex global supply chains. So a surge in demand for AI talent is inevitable.

In 2018, several major universities announced initiatives to develop that talent. MIT announced the largest-ever commitment to AI from a university: a $1 billion initiative to create a College of Computing. Carnegie-Mellon created the first bachelor of science in artificial intelligence degree program. UCBerkeleyannounced a new division of data science. And Stanford announced ahuman-centered AI initiative.

Dozens more schools have followed suit. Machine learning has moved from obscurity to ubiquity, just as software development did 30 years ago and data science did 10 years ago.

Back in 2017, a couple of my colleagues wrote about the AI risk curve, arguing that the adoption of AI is held back not by technology but by managers perception of the risks involved in replacing a worker (whose performance is known) with an unfamiliar software process.

Recessions increase the pressure on managers to reduce labor costs, and thus increase their tolerance for the risks associated with adopting new technology. Over the next year or two, companies will be more willing to take risks and integrate new technologies into their infrastructure. But the challenges of surviving in the recession will mean that AI-first companies must deliver measurable improvements in quality and productivity.

One relatively new risk that managers must tolerate pertains to data. Even companies that are not yet exploiting their data effectively now recognize it as a valuable resource. As startups deploy AI software systems that prove more accurate and cost-effective than human beings, their early-adopter customers must be more willing to trust them with proprietary data. That will allow AI companies to train new products and make them even smarter. And in return for taking this risk, companies must make their models more transparent, more easily reproducible, and more explainable to their customers, auditors, and regulators.

In the area of food and agriculture, AI will help us to understand and adapt to a changing climate. In infrastructure and security, machine learning models will improve the efficiency, reliability and performance of cloud infrastructure. Better and more dynamic risk models will help companies and the entire financial market handle the next crisis.

A host of new applied-AI companies will be needed in order accomplish all this and, especially, AI-enabling companies creating better developer tools and infrastructure, continuous optimization systems, and products that help disciplines improve data quality, security, and privacy.

Boom times favor established companies. They have the cash flow to fund skunkworks and conduct pure research. But its a truism that R&D spending is one of the first things big companies cut in a recession. As an entrepreneur, the idea of starting a company now of all times might be scary, but that retrenchment by established competitors leaves fresh ground open for you to seed with new ideas.

The first sign of AI spring will come when companies again forecast increased demand and seek to improve productivity. The only way to be there when that opportunity presents itself is to start now.

The best part is you wont just profit from the recovery, youll help to create it.

[VentureBeats Transform 2020 event in July will feature a host of disruptive new AI technologies and companies.]

Mark Gorenberg is founder and managing director at Zetta Venture Partners.

Continued here:
After this COVID winter comes an AI spring - VentureBeat

Recommendation and review posted by Alexandra Lee Anderson

Coupling chromatin structure and dynamics by live super-resolution imaging – Science Advances

INTRODUCTION

The three-dimensional organization of the eukaryotic genome plays a central role in gene regulation (1). Its spatial organization has been prominently characterized by molecular and cellular approaches including high-throughput chromosome conformation capture (Hi-C) (2) and fluorescent in situ hybridization (3). Topologically associated domains (TADs), genomic regions that display a high degree of interaction, were revealed and found to be a key architectural feature (4). Direct three-dimensional localization microscopy of the chromatin fiber at the nanoscale (5) confirmed the presence of TADs in single cells but also, among others, revealed great structural variation of chromatin architecture (3). To comprehensively resolve the spatial heterogeneity of chromatin, super-resolution microscopy must be used. Previous work showed that nucleosomes are distributed as segregated, nanometer-sized accumulations throughout the nucleus (68) and that the epigenetic state of a locus has a large impact on its folding (9, 10). However, to resolve the fine structure of chromatin, high labeling densities, long acquisition times, and, often, cell fixation are required. This precludes capturing dynamic processes of chromatin in single live cells, yet chromatin moves at different spatial and temporal scales.

The first efforts to relate chromatin organization and its dynamics were made using a combination of photoactivated localization microscopy (PALM) and tracking of single nucleosomes (11). It could be shown that nucleosomes mostly move coherently with their underlying domains, in accordance with conventional microscopy data (12); however, a quantitative link between the observed dynamics and the surrounding chromatin structure could not yet be established in real time. Although it is becoming increasingly clear that chromatin motion and long-range interactions are key to genome organization and gene regulation (13), tools to detect and to define bulk chromatin motion simultaneously at divergent spatiotemporal scales and high resolution are still missing.

Here, we apply deep learningbased PALM (Deep-PALM) for temporally resolved super-resolution imaging of chromatin in vivo. Deep-PALM acquires a single resolved image in a few hundred milliseconds with a spatial resolution of ~60 nm. We observed elongated ~45- to 90-nm-wide chromatin domain blobs. Using a computational chromosome model, we inferred that blobs are highly dynamic entities, which dynamically assemble and disassemble. Consisting of chromatin in close physical and genomic proximity, our chromosome model indicates that blobs, nevertheless, adopt TAD-like interaction patterns when chromatin configurations are averaged over time. Using a combination of Deep-PALM and high-resolution dense motion reconstruction (14), we simultaneously analyzed both structural and dynamic properties of chromatin. Our analysis emphasizes the presence of spatiotemporal cross-correlations between chromatin structure and dynamics, extending several micrometers in space and tens of seconds in time. Furthermore, extraction and statistical mapping of multiple parameters from the dynamic behavior of chromatin blobs show that chromatin density regulates local chromatin dynamics.

Super-resolution imaging of complex and compact macromolecules such as chromatin requires dense labeling of the chromatin fiber to resolve fine features. We use Deep-STORM, a method that uses a deep convolutional neural network (CNN) to predict super-resolution images from stochastically blinking emitters (Fig. 1A; see Materials and Methods) (15). The CNN was trained to specific labeling densities for live-cell chromatin imaging using a photoactivated fluorophore (PATagRFP); we therefore refer to the method as Deep-PALM. We chose three labeling densities 4, 6, and 9 emitters/m2 per frame in the ON-state to test on the basis of the comparison of simulated and experimental wide-field images (fig. S1A). The CNN trained with 9 emitters/m2 performed significantly worse than the other CNNs and was thus excluded from further analysis (fig. S1B; see Materials and Methods). We applied Deep-PALM to reconstruct an image set of labeled histone protein (H2B-PATagRFP) in human bone osteosarcoma (U2OS) cells using the networks trained on 4 and 6 emitters/m2 per frame (see Materials and Methods). A varying number of predictions by the CNN of each frame of the input series were summed to reconstruct a temporal series of super-resolved images (fig. S1C). The predictions made by the CNN trained with 4 emitters/m2 show large spaces devoid of signal intensity, especially at the nuclear periphery, making this CNN inadequate for live-cell super-resolution imaging of chromatin. While collecting photons from long acquisitions for super-resolution imaging is desirable in fixed cells, Deep-PALM is a live imaging approach. Summing over many individual predictions leads to considerable motion blur and thus loss in resolution. Quantitatively, the Nyquist criterion states that the image resolution R=2/ depends on , the localization density per second, and the time resolution (16). In contrast, motion blur strictly depends on the diffusion constant D of the underlying structure R=4D. There is thus an optimum resolution due to the trade-off between increased emitter sampling and the avoidance of motion blur, which was at a time resolution of 360 ms for our experiments (Fig. 1B and fig. S1D).

(A) Wide-field images of U2OS nuclei expressing H2B-PATagRFP are input to a trained CNN, and predictions from multiple input frames are summed to construct a super-resolved image of chromatin in vivo. (B) The resolution trade-off between the prolonged acquisition of emitter localizations (green line) and motion blur due to diffusion of the underlying diffusion processes (purple line). For our experimental data, the localization density per second is = (2.4 0.1) m2s1, the diffusion constant is D = (3.4 0.8) 103 m2s1 (see fig. S8B), and the acquisition time per frame is = 30 ms. The spatial resolution assumes a minimum (69 5 nm) at a time resolution of 360 ms. (C) Super-resolution images of a single nucleus at time intervals of about 10 s. Scale bars, 2 m. (D) Magnification of segregated accumulations of H2B within a chromatin-rich region. Scale bar, 200 nm. (E) Magnification of a stable but dynamic structure (arrows) over three consecutive images. Scale bars, 500 nm. (F) Fourier ring correlation (FRC) for super-resolved images resulting in a spatial resolution of 63 2 nm. FRC was conducted on the basis of 332 consecutive super-resolved images from two cells. a.u. arbitrary units.

Super-resolution imaging of H2B-PATagRFP in live cells at this temporal resolution shows a pronounced nuclear periphery, while fluorescent signals in the interior vary in intensity (Fig. 1C). This likely corresponds to chromatin-rich and chromatin-poor regions (8). These regions rearrange over time, reflecting the dynamic behavior of bulk chromatin. Chromatin-rich and chromatin-poor regions were visible not only at the scale of the whole nucleus but also at the resolution of a few hundred nanometers (Fig. 1D). Within chromatin-rich regions, the intensity distribution was not uniform but exhibited spatially segregated accumulations of labeled histones of variable shape and size, reminiscent of nucleosome clutches (6), nanodomains (9, 11), or TADs (17). At the nuclear periphery, prominent structures arise. Certain chromatin structures could be observed for ~1 s, which underwent conformational changes during this period (Fig. 1E). The spatial resolution at which structural elements can be observed (see Materials and Methods) in time-resolved super-resolution data of chromatin was 63 2 nm (Fig. 1E), slightly more optimistic than the theoretical prediction (Fig. 1B) (18).

We compared images of H2B reconstructed from 12 frames (super-resolved images) by Deep-PALM in living cells to super-resolution images reconstructed by 8000 frames of H2B in fixed cells (fig. S2, A and B). Overall, the contrast in the fixed sample appears higher, and the nuclear periphery appears more prominent than in images from living cells. However, in accordance with the previous super-resolution images of chromatin in fixed cells (6, 8, 9, 11, 17) and Deep-PALM images, we observe segregated accumulations of signal throughout the nucleus. Thus, Deep-PALM identifies spatially heterogeneous coverage of chromatin, as previously reported (6, 8, 9, 11, 17). We further monitor chromatin temporally at the nanometer scale in living cells.

To quantitatively assess the spatial distribution of H2B, we developed an image segmentation scheme (see Materials and Methods; fig. S3), which allowed us to segment spatially separated accumulations of H2B signal with high fidelity (note S1 and figs. S4 and S5). Applying our segmentation scheme, ~10,000 separable elements, blob-like structures were observed for each super-resolved image (166 resolved images per movie; Fig. 2A). The experimental resolution does not enable us to elucidate their origin and formation because tracking of blobs in three dimensions would be necessary to do so (see Discussion). We therefore turned to a transferable computational model introduced by Qi and Zhang (19), which is based on one-dimensional genomics and epigenomics data, including histone modification profiles and binding sites of CTCF (CCCTC-binding factor). To compare our data to the simulations, super-resolution images were generated from the modeled chromosomes. Within these images, we could identify and characterize chromatin blobs analogously to those derived from experimental data (see Materials and Methods; Fig. 2B).

(A) Super-resolved images show blobs of chromatin (left). These blobs are segmented (see Materials and Methods and note S1) and individually labeled by random color (right). Magnifications of the boxed regions are shown. Scale bars, 2 m (whole nucleus); magnifications, 200 nm. (B) Generation of super-resolution images and blob identification and characterization for a 25million base pair (Mbp) segment of chromosome 1 from GM12878 cells, as simulated in Qi and Zhang (19). Beads (5-kb genomic length) of a simulated polymer configuration within a 200-nm-thick slab are projected to the imaging plane, resembling experimental super-resolved images of live chromatin. Blobs are identified as on experimental data. (C) From the centroid positions, the NND distributions are computed for up to 40 nearest neighbors (blue to red). The envelope of the k-NND distributions (black line) shows peaks at approximately 95, 235, 335, and 450 nm (red dots). (D) k-NND distributions as in (B) for simulated data. (E) Area distribution of experimental and simulated blobs. The distribution is, in both cases, well described by a lognormal distribution with parameters (3.3 2.8) 103 m2 for experimental blobs and (3.1 3.2) 103 m2 for simulated blobs (means SD). PDF, probability density function. (F) Eccentricity distribution for experimental and simulated chromatin blobs. Selected eccentricity values are illustrated by ellipses with the corresponding eccentricity. Eccentricity values range from 0, describing a circle, to 1, describing a line. Prominent peaks arise because of the discretization of chromatin blobs in pixels. The data are based on 332 consecutive super-resolved images from two cells, in each of with ~10,000 blobs were identified.

For imaged (in living and fixed cells) and modeled chromatin, we first computed the kth nearest-neighbor distance (NND; centroid-to-centroid) distributions, taking into account the nearest 1st to 40th neighbors (Fig. 2C and fig. S2, C and D, blue to red). Centroids of the nearest neighbors are (95 30) nm (means SD) apart, consistent with previous and our own super-resolution images of chromatin in fixed cells (9) and slightly further than what was found for clutches of nucleosomes (6). The envelope of all NND distributions (Fig. 2C, black line) shows several weak maxima at ~95, 235, 335, and 450 nm, which roughly coincide with the peaks of the 1st, 7th, 14th, and 25th nearest neighbors, respectively (Fig. 2C, red dots). In contrast, simulated data exhibit a prominent first nearest-neighbor peak at a slightly smaller distance, and higher-order NND distribution decay quickly and appear washed out (Fig. 2D). This hints toward greater levels of spatial organization of chromatin in vivo, which is not readily recapitulated in the used state-of-the-art chromosome model.

Next, we were interested in the typical size of chromatin blobs. Their area distribution (Fig. 2E) fit a log-normal distribution with parameters (3.3 2.8) 103 m2 (means SD), which is in line with the area distribution derived from fixed samples (fig. S2E) and modeled chromosomes. Notably, blob areas vary considerably, as indicated by the high SD and the prominent tail of the area distribution toward large values. Following this, we calculated the eccentricity of each blob to resolve their shape (Fig. 2F and fig. S2F). The eccentricity is a measure of the elongation of a region reflecting the ratio of the longest chord of the shape and the shortest chord perpendicular to it (Fig. 2F; illustrated shapes at selected eccentricity values). The distribution of eccentricity values shows an accumulation of values close to 1, with a peak value of ~0.9, which shows that most blobs have an elongated, fiber-like shape and are not circular. In particular, the eccentricity value of 0.9 corresponds to a ratio between the short and long axes of the ellipse of 1:2 (see Materials and Methods), which results, considering the typical area of blobs in experimental and simulated data, in roughly 92-nm-long and 46-nm-wide blobs on average. A highly similar value was found in fixed cells (fig. S2F). The length coincides with the value found for the typical NND [Fig. 2C; (95 30) nm]. However, because of the segregation of chromatin into blobs, their elongated shape, and their random orientation (Fig. 2A), the blobs cannot be closely packed throughout the nucleus. We find that chromatin has a spatially heterogeneous density, occupying 5 to 60% of the nuclear area (fig. S6, A and B), which is supported by a previous electron microscopy study (20).

Blob dimensions derived from live-cell super-resolution imaging using Deep-PALM are consistent with those found in fixed cells, thereby further validating our method, and in agreement with previously determined size ranges (6, 9). A previously published chromosome model based on Hi-C data (and thus not tuned to display blob-like structures per se) also displays blobs with dimensions comparable to those found here, in living cells. Together, these data strongly suggest the existence of spatially segregated chromatin structures in the sub100-nm range.

The simulations offer to track each monomer (chromatin locus) unambiguously, which is currently not possible to do from experimental data. Since the simulations show blobs comparable to those found in experiment (Fig. 2), simulations help to indicate possible mechanisms leading to the observation of chromatin blobs. For instance, because of the projection of the nuclear volume onto the imaging plane, the observed blobs could simply be overlays of distant, along the one-dimensional genome, noninteracting genomic loci. To examine this possibility, we analyzed the gap length between beads belonging to the same blob along the simulated chromosome. Beads constitute the monomers of the simulated chromosome, and each bead represents roughly 5 kb (19).

The analysis showed that the blobs are mostly made of consecutive beads along the genome, thus implying an underlying domain-like structure, similar to TADs (Fig. 3A). Using the affiliation of each bead to an intrinsic chromatin state of the model (Fig. 3B), it became apparent that blobs along the simulated chromosome consisting mostly of active chromatin are significantly larger than those formed by inactive and repressive chromatin (Fig. 3C). These findings are in line with experimental results (10) and results from the simulations directly (19), thereby validating the projection and segmentation process.

(A) Gap length between beads belonging to the same blob. An exemplary blob with small gap length is shown. The blob is mostly made of consecutive beads being in close spatial proximity. (B) A representative polymer configuration is colored according to chromatin states (red, active; green, inactive; and blue, repressive). (C) The cumulative distribution function (CDF) of clusters within active, inactive, and repressive chromatin. Inset: Mean area of clusters within the three types of chromatin. The distributions are all significantly different from each other, as determined by a two-sample Kolmogorov-Smirnov test (P < 1050). (D) Distribution of the continuous residence time of any monomer within a cluster (0.5 0.3 s; means SD). Inset: Continuous residence time of any monomer within a slab of 200-nm thickness (1.5 1.6 s; means SD). (E) The blob association strength between any two beads is measured as the frequency at which any two beads are found in one blob. The association map is averaged over all simulated configurations (upper triangular matrix; from simulations), and experimental Hi-C counts are shown for the same chromosome segment [lower triangular matrix; from Rao et al. (40)]. The association and Hi-C maps are strongly correlated [Pearsons correlation coefficient (PCC) = 0.76]. (F) Close-up views around the diagonal of Hi-Clike matrices. The association strength is shown together with the inverse distance between beads (top; PCC = 0.85) and with experimental Hi-C counts [bottom; as in (E)]. The data are based on 20,000 polymer configurations.

Since chromatin is dynamic in vivo and in computer simulations, each bead can diffuse in and out of the imaging volume from frame to frame. We estimated that, on average, each bead spent approximately 1.5 s continuously within a slab of 200-nm thickness (Fig. 3D). Furthermore, a bead is, on average, found only 0.55 0.33 s continuously within a blob, which corresponds to one to two experimental super-resolved images (Fig. 3D). These results suggest that chromatin blobs are highly dynamic entities, which usually form and dissemble within less than 1 s. We thus constructed a time-averaged association map for the modeled chromosomes, quantifying the frequency at which each locus is found with any other locus within one blob. The association map is comparable to interaction maps derived from Hi-C (Fig. 3E). Notably, interlocus association and Hi-C maps are strongly correlated, and the association map shows similar patterns as those identified as TADs in Hi-C maps, even for relatively distant genomic loci [>1 million base pairs (Mbp)]. A similar TAD-like organization is also apparent when the average inverse distance between loci is considered (Fig. 3F, top), suggesting that blobs could be identified in super-resolved images because of the proximity of loci within blobs in physical space. The computational chromosome model indicates that chromatin blobs identified by Deep-PALM are mostly made of continuous regions along the genome and cannot be attributed to artifacts originating from the projection of the three-dimensional genome structure to the imaging plane. The simulations further indicate that the blobs associate and dissociate within less than 1 s, but loci within blobs are likely to belong to the same TAD. Their average genomic content is 75 kb, only a fraction of typical TAD lengths in mammalian cells (average size, 880 kb) (4), suggesting that blobs likely correspond to sub-TADs or TAD nanocompartments (17).

To quantify the experimentally observed chromatin dynamics at the nanoscale, down to the size of one pixel (13.5 nm), we used a dense reconstruction of flow fields, optical flow (Fig. 4A; see Materials and Methods), which was previously used to analyze images taken on confocal (12, 14), and structured illumination microscopes (8). We examined the suitability of optical flow for super-resolution on the basis of single-molecule localization images using simulations. We find that the accuracy of optical flow is slightly enhanced on super-resolved images compared to conventional fluorescence microscopy images (note S2 and fig. S7, A to C). Experimental super-resolution flow fields are illustrated on the basis of two subsequent images, between which the dynamics of structural features are apparent to the eye (fig. S7, D and E). On the nuclear periphery, connected regions spanning up to ~500 nm can be observed [fig. S7D (i and ii), marked by arrows]. These structures are stable for at least 360 ms but move from frame to frame. The flow field is shown on top of an overlay of the two super-resolved images and color-coded [fig. S7D (iii); the intensity in frame 1 is shown in green, the intensity in frame 2 is shown in purple, and colocalization of both is white]. Displacement vectors closely follow the redistribution of intensity from frame to frame (roughly from green to purple). Similarly, structures within the nuclear interior (fig. S7E) can be followed by eye, thus further validating and justifying the use of a dense motion reconstruction as a quantification tool of super-resolved chromatin motion.

(A) A time series of super-resolution images (left) is subject to optical flow (right). (B) Blobs of a representative nucleus (see movie S1) are labeled by their NND (left), area (middle), and flow magnitude (right). Colors denote the corresponding parameter magnitude. (C) The average blob area, (D) NND, (E) density, and (F) flow magnitude are shown versus the normalized distance from the nuclear periphery (lower x axis; 0 is on the periphery and 1 is at the center of the nucleus) and versus the absolute distance (upper x axis). Line and shaded area denote the means SE from 322 super-resolved images of two cells. Scale bar, (A) and (B): 3 m.

Using optical flow fields, we linked the spatial appearance of chromatin to their dynamics. Effectively, the blobs were characterized with two structural parameters (NND and area) and their flow magnitude (Fig. 4B). Movie S1 shows the time evolution of those parameters for an exemplary nucleus. Blobs at the nuclear periphery showed a distinct behavior from those in the nuclear interior. In particular, the periphery exhibits a lower density of blobs, but those appear slightly larger and are less mobile than in the nuclear interior (Fig. 4, C to F), in line with previous findings using conventional microscopy (14). The peripheral blobs are reminiscent of dense and relatively immobile heterochromatin and lamina-associated domains (21), which extend only up to 0.5 m inside the nuclear interior. In contrast, blob dynamics increase gradually within 1 to 2 m from the nuclear rim.

To further elucidate the relationship between chromatin structure and dynamics, we analyzed the correlation between each pair of parameters in space and time. Therefore, we computed the auto- and cross-correlation of parameter maps with a given time lag across the entire nucleus (in space) (Fig. 5A). In general, a positive correlation denotes a low-low or a high-high relationship (a variable de-/increases when another variable de-/increases), while, analogously, a negative correlation denotes a high-low relationship. The autocorrelation of NND maps [Fig. 5A (i)] shows a positive correlation; thus, regions exist spanning 2 to 4 m, in which chromatin is either closely packed (low-low) or widely dispersed (high-high). Likewise, blobs of similar size tend to be in spatial proximity [Fig. 5A (iii)]. These regions are not stable over time but rearrange continuously, an observation bolstered by the fact that the autocorrelation diminishes with increasing time lag. The cross-correlation between NND and area [Fig. 5A (ii)] shows a negative correlation for short time lags, suggesting that large blobs appear with a high local density while small ones are more isolated. The correlation becomes slightly positive for time lags 20 s, indicating that big blobs are present in regions that were sparsely populated before and small blobs tend to accumulate in previously densely populated regions. This is in line with dynamic reorganization and reshaping of chromatin domains on a global scale, as observed in snapshots of the Deep-PALM image series (Fig. 1A).

(A) The spatial auto- and cross-correlation between parameters were computed for different time lags. The graphs depict the correlation over space lag for each parameter pair, and different colors denote the time lag (increasing from blue to red). (B) Illustration of the instantaneous relationship between local chromatin density and dynamics. The blob density is shown in blue; the magnitude of chromatin dynamics is shown by red arrows. The consistent negative correlation between NND and flow magnitude is expressed by increased dynamics in regions of high local blob density. Data represent the average over two cells. The cells behave similarly such that error bars are omitted for the sake of clarity.

The flow magnitude is positively correlated for all time lags, while the correlation displays a slight increase for time lags 20 s [Fig. 5A (vi)], which has also been observed previously (8, 12, 22). The spatial autocorrelation of dynamic and structural properties of chromatin are in stark contrast. While structural parameters are highly correlated at short but not at long time scales, chromatin motion is still correlated at a time scale exceeding 30 s. At very short time scales (<100 ms), stochastic fluctuations determine the local motion of the chromatin fiber, while coherent motion becomes apparent at longer times (22). However, there exists a strong cross-correlation between structural and dynamic parameters: The cross-correlation between the NND and flow magnitude shows notable negative correlation at all time lags [Fig. 5A (iv)], strongly suggesting that sparsely distributed blobs appear less mobile than densely packed ones. The area seems to play a negligible role for short time lags, but there is a modest tendency that regions with large blobs tend to exhibit increased dynamics at later time points [10 s; Fig. 5A (v)], likely due to the strong relationship between area and NND.

In general, parameter pairs involving chromatin dynamics exhibit an extended spatial auto- or cross-correlation (up to ~6 m; the lower row of Fig. 5A) compared to correlation curves including solely structural parameters (up to 3 to 4 m). Furthermore, the cross-correlation of flow magnitude and NND does not considerably change for increasing time lag, suggesting that the coupling between those parameters is characterized by an unexpectedly resilient memory, lasting for at least tens of seconds (23). Concomitantly, the spatial correlation of time-averaged NND maps and maps of the local diffusion constant of chromatin for the entire acquisition time enforces their negative correlation at the time scale of ~1 min (fig. S8). Such resilient memory was also proposed by a computational study that observed that interphase nuclei behave similar to concentrated solutions of unentangled ring polymers (24). Our data support the view that chromatin is mostly unentangled since entanglement would influence the anomalous exponent of genomic loci in regions of varying chromatin density (24). However, our data do not reveal a correlation between the anomalous exponent and the time-averaged chromatin density (fig. S8), in line with our previous results using conventional microscopy (14).

Overall, the spatial cross-correlation between chromatin structure and dynamics indicates that the NND between blobs and their mobility stand in a strong mutual, negative relationship. This relationship, however, concerns chromatin density variations at the nanoscale, but not global spatial density variations such as in euchromatin or heterochromatin (14). These results support a model in which regions with high local chromatin density, i.e., larger blobs are more prevalent and are mobile, while small blobs are sparsely distributed and less mobile (Fig. 5B). Blob density and dynamics in the long-time limit are, to an unexpectedly large extent, influenced by preceding chromatin conformations.

The spatial correlations above were only evaluated pairwise, while the behavior of every blob is likely determined by a multitude of factors in the complex energy landscape of chromatin (19, 22). Here, we aim to take a wider range of available information into account to reveal the principle parameters, driving the observed chromatin structure and dynamics. Using a microscopy-based approach, we have access to a total of six relevant structural, dynamic, and global parameters, which potentially shape the chromatin landscape in space and time (Fig. 6A). In addition to the parameters used above, we included the confinement level as a relative measure, allowing the quantification of transient confinement (see Materials and Methods). We further included the bare signal intensity of super-resolved images and, as the only static parameter, the distance from the periphery since it was shown that dynamic and structural parameters show some dependence on this parameter (Fig. 4). We then used t-distributed stochastic neighbor embedding (t-SNE) (25), a state-of-the-art dimensionality reduction technique, to map the six-dimensional chromatin features (the six input parameters) into two dimensions (Fig. 6A and see note S3). The t-SNE algorithm projects data points such that neighbors in high-dimensional space likely stay neighbors in two-dimensional space (25). Visually apparent grouping of points (Fig. 6B) implies that grouped points exhibit great similarity with respect to all input features, and it is of interest to reveal which subset of the input features can explain the similarity among chromatin blobs best. It is likely that points appear grouped because their value of a certain input feature is considerably higher or lower than the corresponding value of other data points. We hence labeled points in t-SNE maps which are smaller than the first quartile point or larger than the third quartile point. Data points falling in either of the low/high partition of one input feature are colored accordingly for visualization (Fig. 6D; blue/red points, respectively). We then assigned a rank to each of the input features according to their nearest-neighbor fraction (n-n fraction): Since the t-SNE algorithm conserves nearest neighbors, we described the extent of grouping in t-SNE maps by the fraction of nearest neighbors, which fall in either one of the subpopulations of low or high points (illustrated in fig. S9). A high n-n fraction (Fig. 6C) therefore indicates that many points marked as low/high are indeed grouped by t-SNE and are therefore similar. The ranking (from low to high n-n fraction) reflects the potency of a given parameter to induce similar behavior between chromatin blobs with respect to all input features.

(A) The six-dimensional parameter space is input to the t-SNE algorithm and projected to two dimensions. (B) The two-dimensional embedding of an exemplary dataset is shown and colored according to the magnitude of each input feature (blue to red; the parameter average is shown in beige). (C) Points below the first (blue) and above the third (red) quartile points of the corresponding parameter are marked, and the parameters are ranked according to the fraction of nearest neighbors that fall in one of the marked regions. (D) Data points marked below the first or above the third quartile points are labeled according to the feature in which they were marked. Priority is given to the feature with the higher n-n fraction if necessary. (E) t-SNE analysis is carried out for each nucleus over the whole time series, and it is counted how often a parameter ranked first. The results are visualized as a pie chart. The NND predominantly ranks first in about two-thirds of all cases. (F) Marked points in (C) and (D) are mapped back onto the corresponding nuclei, and the CDF over space is shown (means SE). Pie chart and CDF computations are based on 322 super-resolved images from two cells.

The relative frequency at which each parameter ranked first provides an intuitive feeling for the most influential parameters in the dataset (Fig. 6E). The signal intensity plays a negligible role, suggesting that our data are free of potential artifacts related to the bare signal intensity. Furthermore, the blob area and the distance from the periphery likewise do not considerably shape chromatin blobs. In contrast, the NND between blobs was found to be the main factor inducing the observed characteristics in 67% of all-time frames across all nuclei. The flow magnitude and confinement level together rank first in 26% of all cases (11 and 17%, respectively). These numbers suggest that the local chromatin density is a universal key regulator of instantaneous chromatin dynamics. Note that no temporal dependency is included in the t-SNE analysis and, thus, the feature extraction concerns only short-term (360 ms) relationships. The characteristics of roughly one-fourth of all blobs at each time point are mainly determined by similar dynamical features. Mapping chromatin blobs as marked in Fig. 6 (C and D) back to their respective positions inside the nucleus (Fig. 6F) shows that blobs with low/high flow magnitude or confinement level markedly also grouped in physical space, which is highly reminiscent of coherent motion of chromatin (12). In contrast, blobs with extraordinary low or high NND were found interspersed throughout the nucleus, in line with spatial correlation analysis between structural and dynamic features (Fig. 5). Our results point toward a large influence of the local chromatin density on the dynamics of chromatin at the scale of a few hundred nanometers and within a few hundred milliseconds. At longer time and length scales, however, previous results suggest that this relationship is lost (14).

With Deep-PALM, we present temporally resolved super-resolution images of chromatin in living cells. Our technique identified chromatin nanodomains, named blobs, which mostly have an elongated shape, consistent with the curvilinear arrangement of chromatin, as revealed by structured illumination microscopy (8) with typical axes lengths of 45 to 90 nm. A previous study reported ~30-nm-wide clutches of nucleosomes in fixed mammalian cells using STORM nanoscopy (6), while the larger value obtained using Deep-PALM may be attributed to the motion blurring effect in live-cell imaging. However, histone acetylation and methylation marks were shown to form nanodomains of diameter 60 to 140 nm, respectively (9), which includes the computed dimensions for histone H2B using Deep-PALM.

To elucidate the origin of chromatin blobs, we turned to a simulated chromosome model, which displays chromatin blobs similar to our experimental data when seen in a super-resolution reconstruction. The simulations suggest that chromatin blobs consist of continuous genomic regions with an average length of 75 kb, assembling and disassembling dynamically within less than 1 s. Monomers within blobs display a distinct TAD-like association pattern in the long-time limit, suggesting that the identified blobs represent sub-TADs. Transient formation is consistent with recent findings that TADs are not stable structural elements but exhibit extensive heterogeneity and dynamics (3, 5). To experimentally probe the transient assembly of chromatin blobs, it would be interesting to track individual blobs over time. However, this is a nontrivial task. While the size (area/volume) or shape of blobs could be used to establish correspondences between blobs in subsequent frames, the framework needs to be flexible enough to allow for blob deformations since blobs likely arise stochastically and are not rigid bodies. Achieving an even shorter acquisition time per frame in the future could help minimize the influence of blob deformations and make tracking feasible. The second challenge is to distinguish between disassembly and out-of-focus diffusion of a blob. The three-dimensional imaging at sufficient spatial and temporal resolution will be helpful in the future to overcome this hurdle.

Using an optical flow approach to determine the blob dynamics instead, we found that structural and dynamic parameters exhibit extended spatial and temporal (cross-) correlations. Structural parameters such as the local chromatin density (expressed as the NND between blobs) and area lose their correlation after 3 to 4 m and roughly 40 s in the spatial and temporal dimension, respectively. In contrast, chromatin mobility correlations extend over ~6 m and persist during the whole acquisition period (40 s). Extensive spatiotemporal correlation of chromatin dynamics has been presented previously, both experimentally (12) and in simulations (22), but was not linked to the spatiotemporal behavior of the underlying chromatin structure until now. We found that the chromatin dynamics are closely linked to the instantaneous but also to past local structural characterization of chromatin. In other words, the instantaneous local chromatin density influences chromatin dynamics in the future and vice versa. On the basis of these findings, we suggest that chromatin dynamics exhibit an extraordinary long memory. This strong temporal relationship might be established by the fact that stress propagation is affected by the folded chromosome organization (26). Fiber displacements cause structural reconfiguration, ultimately leading to a local amplification of chromatin motion in local high-density environments. This observation is also supported by the fact that increased nucleosome mobility grants chromatin accessibility even within regions of high nucleosome density (27).

Given the persistence at which correlations of chromatin structure and, foremost, dynamics occur in a spatiotemporal manner, we speculate that the interplay of chromatin structure and dynamics could involve a functional relationship (28): Transcriptional activity is closely linked to chromatin accessibility and the epigenomic state (29). Because chromatin structure and dynamics are related, dynamics could also correlate with transcriptional activity (14, 30, 31). However, it is currently unknown whether the structure-dynamics relationship revealed here is strictly mutual or whether it may be causal. Simulations hint that chromatin dynamics follows from structure (22, 23); this question will be exciting to answer experimentally and in the light of active chromatin remodelers to elucidate a potential functional relationship to transcription. Chromatin regions that are switched from inactive to actively transcribing, for instance, undergo a structural reorganization accompanied by epigenetic modifications (32). The mechanisms driving recruitment of enzymes inducing histone modifications such as histone acetyltransferases, deacetylases, or methyltransferases are largely unknown but often involve the association to proteins (33). Their accessibility to the chromatin fiber is inter alia determined by local dynamics (27). Such a structure-dynamics feedback loop would constitute a quick and flexible way to transiently alter gene expression patterns upon reaction to external stimuli or to coregulate distant genes (1). Future work will study how structure-dynamics correlations differ in regions of different transcriptional activity and/or epigenomic states. Furthermore, probing the interactions between key transcriptional machines such as RNA polymerases with the local chromatin structure and recording their (possibly collective) dynamics could shed light into the target search and binding mechanisms of RNA polymerases with respect to the local chromatin structure. Deep-PALM in combination with optical flow paves the way to answer these questions by enabling the analysis of time-resolved super-resolution images of chromatin in living cells.

Human osteosarcoma U2OS expressing H2B-PATagRFP cells were a gift from S. Huet (CNRS, UMR 6290, Institut Gntique et Dveloppement de Rennes, Rennes, France); the histone H2B was cloned, as described previously (34). U2OS cells were cultured in Dulbeccos modified Eagles medium [with glucose (4.5 g/liter)] supplemented with 10% fetal bovine serum (FBS), 2 mM glutamine, penicillin (100 g/ml), and streptomycin (100 U/ml) in 5% CO2 at 37C. Cells were plated 24 hours before imaging on 35-mm petri dishes with a no. 1.5 coverslip-like bottom (ibidi, Biovalley) with a density of 2 105 cells per dish. Just before imaging, the growth medium was replaced by Leibovitzs L-15 medium (Life Technologies) supplemented with 20% FBS, 2 mM glutamine, penicillin (100 g/ml), and streptomycin (100 U/ml).

Imaging of H2B-PAtagRFP in living U2OS cells was carried out on a fully automated Nikon Ti-E/B PALM (Nikon Instruments) microscope. The microscope is equipped with a full incubator enclosure with gas regulation to maintain a temperature of ~37C for normal cell growth during live-cell imaging. Image sequences of 2000 frames were recorded with an exposure time of 30 ms per frame (33.3 frames/s). For Deep-PALM imaging, a relatively low power (~50 W/cm2 at the sample) was applied for H2B-PATagRFP excitation at 561 nm and then combined with the 405 nm (~2 W/cm2 at the sample) to photoactivate the molecules between the states. Note that for Deep-PALM imaging, switched fluorophores are not required to stay as long in the dark state as for conventional PALM imaging. We used oblique illumination microscopy (11) combined with total internal reflection fluorescence (TIRF) mode to illuminate a thin layer of 200 nm (axial resolution) across the nucleus. The reconstruction of super-resolved images improves the axial resolution only marginally (fig. S1, E and F). Laser beam powers were controlled by acoustic optic-modulators (AA Opto-Electronic). Both wavelengths were united into an oil immersion 1.49-NA (numerical aperture) TIRF objective (100; Nikon). An oblique illumination was applied to acquire image series with a high signal-to-noise ratio. The fluorescence emission signal was collected by using the same objective and spectrally filtered by a Quad-Band beam splitter (ZT405/488/561/647rpc-UF2, Chroma Technology) with a Quad-Band emission filter (ZET405/488/561/647m-TRF, Chroma Technology). The signal was recorded on an electron-multiplying charge-coupled device camera (Andor iXon X3 DU-897, Andor Technology) with a pixel size of 108 nm. For axial correction, Perfect Focus System was applied to correct for defocusing. NIS-Elements software was used for acquiring the images.

The same cell line (U2OS expressing H2B-PAtagRFP), as in live-cell imaging, was used for conventional PALM imaging. Before fixation, cells were washed with phosphate-buffered saline (PBS) (three times for 5 min each) and then fixed with 4% paraformaldehyde (Sigma-Aldrich) diluted in PBS for 15 min at room temperature. A movie of 8000 frames was acquired with an exposure time of 30 ms per frame (33.3 frames/s). In comparison to Deep-PALM imaging, a relatively higher excitation laser of 561 nm (~60 W/cm2 at the sample) was applied to photobleach H2B-PATagRFP and then combined with the 405 nm (~2.5 W/cm2 at the sample) for photoactivating the molecules. We used the same oblique illumination microscopy combined with TIRF system, as applied in live-cell imaging.

PALM images from fixed cells were analyzed using ThunderSTORM (35). Super-resolution images were constructed by binning emitter localizations into 13.5 13.5 nm pixels and blurred by a Gaussian to match Deep-PALM images. The image segmentation was carried out as on images from living cells (see below).

The CNN was trained using simulated data following Nehme et al. (15) for three labeling densities (4, 6, and 9 emitters/m2 per frame). Raw imaging data were checked for drift, as previously described (12). The detected drift in raw images is in the range of <10 nm and therefore negligible. The accuracy of the trained net was evaluated by constructing ground truth images from the simulated emitter positions. The structural similarity index is computed to assess the similarity between reconstructed and ground truth images (36)SSIM=x,y(2xx+C1)(2xy+C2)(x2+y2+C1)(x2+y2+C2)(1)where x and y are windows of the predicted and ground truth images, respectively, and denote their local means and SD, respectively, and xy denotes their cross-variance. C1 = (0.01L)2 and C2 = (0.03L)2 are regularization constants, where L is the dynamic range of the input images. The second quantity to assess CNN accuracy is the root mean square error between the ground truth G and reconstructed image RRMSE=1NN(RG)2(2)where N is the number of pixels in the images. After training, sequences of all experimental images were compared to the trained network, and predictions of single Deep-PALM images were summed to obtain a final super-resolved image. An up-sampling factor of 8 was used, resulting in an effective pixel size of 108 nm/8 = 13.5 nm. A blind/referenceless image spatial quality evaluator (37) was used to determine the optimal number of predictions to be summed. For visualization, super-resolved images were convolved with a Gaussian kernel ( = 1 pixel) and represented using a false red, green, and blue colormap. The parameters of the three trained networks are available at https://github.com/romanbarth/DeepPALM-trained-models.

Fourier ring correlation (FRC) is an unbiased method to estimate the spatial resolution in microscopy images. We follow an approach similar to the one described by Nieuwenhuizen et al. (38). For localization-based super-resolution techniques, the set of localizations is divided into two statistically independent subsets, and two images from these subsets are generated. The FRC is computed as the statistical correlation of the Fourier transforms of both subimages over the perimeter of circles of constant frequency in the frequency domain. Deep-PALM, however, does not result in a list of localizations, but in predicted images directly. The set of 12 predictions from Deep-PALM were thus split into two statistically independent subsets, and the method described by Nieuwenhuizen et al. (38) was applied.

The super-resolved images displayed isolated regions of accumulated emitter density. To quantitatively assess the structural information implied by this accumulation of emitters in the focal plane, we developed a segmentation scheme that aims to identify individual blobs (fig. S3). A marker-assisted watershed segmentation was adapted to accurately determine blob boundaries. For this purpose, we use the raw predictions from the deep CNN without convolution (fig. S3A). The foreground in this image is marked by regional maxima and pixels with very high density (i.e., those with I > 0.99 Imax; fig. S3B). Since blobs are characterized by surrounding pixels of considerably less density, the Euclidian distance transform is computed on the binary foreground markers. Background pixels (i.e., those pixels not belonging to any blobs) are expected to lie far away from any blob center, and thus, a good estimate for background markers are those pixels being furthest from any foreground pixel. We hence compute the watershed transform on the distance transform of foreground markers, and the resulting watershed lines depict background pixels (fig. S3C). Equipped with fore- and background markers (fig. S3D), we apply a marker-controlled watershed transform on the gradient of the input image (fig. S3E). The marker-controlled watershed imposes minima on marker pixels, preventing the formation of watershed lines across marker pixels. Therefore, the marker-controlled watershed accurately detects boundaries and blobs that might not have been previously marked as foreground (fig. S3F). Last, spurious blobs whose median- or mean intensity is below 10% of the maximum intensity are discarded, and each blob is assigned a unique label for further correspondence (fig. S3G). The area and centroid position are computed for each identified blob for further analysis. This automated segmentation scheme performs considerably better than other state-of-the-art algorithms for image segmentation because of the reliable identification of fore- and background markers accompanied by the watershed transform (note S1).

Centroid position, area, and eccentricity were computed. The eccentricity is computed by describing the blobs as an ellipseE=1a2/b2(3)where a and b are the short and long axes of the ellipse, respectively.

We chose to use a computational chromatin model, recently introduced by Qi and Zhang (19), to elucidate the origin of experimentally determined chromatin blobs. Each bead of the model covers a sequence length of 5 kb and is assigned 1 of 15 chromatin states to distinguish promoters, enhancers, quiescent chromatin, etc. Starting from the simulated polymer configurations, we consider monomers within a 200-nm-thick slab through the center of the simulated chromosome. To generate super-resolved images as those from Deep-PALM analysis, fluorescence intensity is ascribed to each monomer. Monomer positions are subsequently discretized on a grid with 13.5-nm spacing and convolved with a narrow point-spread function, which results in images closely resembling experimental Deep-PALM images of chromatin. Chromatin blobs were then be identified and characterized as on experimental data (Fig. 2, A and B). Mapping back the association of each bead to a blob (if any) allows us to analyze principles of blob formation and maintenance using the distance and the association strength between each pair of monomers, averaged over all 20,000 simulated polymer configurations.

The radial distribution function g(r) (also pair correlation function) is calculated (in two dimensions) by counting the number of blobs in an annulus of radius r and thickness dr. The result is normalized by the bulk density = n/A, with the total number of blobs n and, A, the area of the nucleus, and the area of the annulus, 2r drdn(r)=g(r)2rdr(4)

Super-resolved images of chromatin showed spatially distributed blobs of varying size, but the resolved structure is too dense for state-of-the-art single-particle tracking methods to track. Furthermore are highly dynamic structures, assembling and dissembling within one to two super-resolved frames (Fig. 3D), which makes a single-particle tracking approach unsuitable. Instead, we used a method for dynamics reconstruction of bulk macromolecules with dense labeling, optical flow. Optical flow builds on the computation of flow fields between two successive frames of an image series. The integration of these flow fields from super-resolution images results in trajectories displaying the local motion of bulk chromatin with temporal and high spatial resolution. Further, the trajectories are classified into various diffusion models, and parameters describing the underlying motion are computed (14). Here, we use the effective diffusion coefficient D (in units of m2/s), which reflects the magnitude of displacements between successive frames (the velocity of particles or monomers in the continuous limit) and the anomalous exponent (14). The anomalous exponent reflects whether the diffusion is free ( = 1, e.g., for noninteracting particles in solution), directed ( > 1, e.g., as the result from active processes), or hindered ( < 1, e.g., because of obstacles or an effective back-driving force). Furthermore, we compute the length of constraint Lc, which is defined as the SD of the trajectory positions with respect to its time-averaged position. Denoting R(t; R0), the trajectory at time t originating from R0, the expression reads Lc(R0) = var(R(t; R0))1/2, where var denotes the variance. The length of constraint is a measure of the length scale explored of the monomer during the observation period. A complementary measure is the confinement level (39), which computes the inverse of the variance of displacements within a sliding window of length : C / var(R(t; R0)), where the sliding window length is set to four frames (1.44 s). Larger values of C denote a more confined state than small ones.

The NND and the area, as well as the flow magnitude, were calculated and assigned to the blobs centroid position. To calculate the spatial correlation between parameters, the parameters were interpolated from the scattered centroid positions onto a regular grid spanning the entire nucleus. Because not every pixel in the original super-resolved images is assigned a parameter value, we chose an effective grid spacing of five pixels (67.5 nm) for the interpolated parameter maps. After interpolation, the spatial correlation was computed between parameter pairs: Let r = (x, y)T denote a position on a regular two-dimensional grid and f(r, t) and g(r, t) two scalar fields with mean zero and variance one, at time t on that grid. The time series of parameter fields consist of N time points. The spatial cross-correlation between the fields f and g, which lie a lag time apart, is then calculated asC(,)=1Ntx,yf(r,t)g(r+,t+)x,yf(r,t)g(r,t+)(5)where the space lag is a two-dimensional vector = (x, y)T. The sums in the numerator and denominator are taken over the spatial dimensions; the first sum is taken over time. The average is thus taken over all time points that are compliant with time lag . Subsequently, the radial average in space is taken over the correlation, thus effectively calculating the spatial correlation C(, ) over the space lag =x2+y2. If f = g, then the spatial autocorrelation is computed.

We denote as global parameters those that reflect the structural and dynamic behavior of chromatin spatially resolved in a time-averaged manner. Examples involve the diffusion constant, the anomalous exponent, the length of constraint, but also time-averaged NND maps, etc. (fig. S8). Those parameters are useful to determine time-universal characteristics. The spatial correlation between those parameters is equivalent to the expression given for temporally varying parameters when the temporal dimension is omitted, effectively resulting in a correlation curve C().

The distance from the periphery, intensity, their NND, area, flow magnitude, and confinement level of each identified blob form the six-dimensionalinput feature space for t-SNE analysis. The parameters for each blob (n = 3,260,232; divided into subsets of approximately 10,000) were z-transformed before the t-SNE analysis. The t-SNE analysis was performed using MATLAB and the Statistics and Machine Learning Toolbox (Release 2017b; The MathWorks Inc., Natick, MA, USA) with the Barnes-Hut approximation. The algorithm was tested using different distance metrics and perplexity values and showed robust results within the examined ranges (note S3 and fig. S10).

Acknowledgments: We acknowledge support from the Ple Scientifique de Modlisation Numrique, ENS de Lyon for providing computational resources. We thank B. Zhang (Massachusetts Institute of Technology) for providing data of simulated chromosomes and S. Kocanova (LBME, CBI-CNRS; University of Toulouse) for providing PALM videos for fixed cells. We thank H. Babcock (Harvard University), A. Seeber (Harvard University), and M. Tamm (Moscow State University) for valuable feedback on the manuscript. Funding: This publication is based upon work from COST Action CA18127, supported by COST (European Cooperation in Science and Technology). This work is supported by Agence Nationale de la Recherche (ANR) ANDY and Sinfonie grants. Author contributions: H.A.S. designed and supervised the project. R.B. designed the data analysis and wrote the code. H.A.S. carried out experimental work. R.B. carried out the data analysis. H.A.S. and R.B. interpreted results. H.A.S., R.B., and K.B. wrote the manuscript. Competing interests: The authors declare that they have no competing interests. Data and materials availability: All data needed to evaluate the conclusions in the paper are present in the paper and/or the Supplementary Materials. Additional data related to this paper may be requested from the authors.

View post:
Coupling chromatin structure and dynamics by live super-resolution imaging - Science Advances

Recommendation and review posted by Alexandra Lee Anderson

Disordered proteins follow diverse transition paths as they fold and bind to a partner – Science Magazine

Shedding light on disordered proteins

Disordered proteins often fold as they bind to a partner protein. There could be many different molecular trajectories between the unbound proteins and the bound complex. Most methods to measure transition paths rely on monitoring a single distance, making it difficult to resolve complex pathways. Kim and Chung used fast three-color single-molecule Foster resonance energy transfer (FRET) to simultaneously probe distance changes between the two ends of an unfolded protein and between each end and a probe on the partner protein. They show that binding can be initiated by diverse conformations and that the molecules are held together by non-native interactions as the disordered protein folds. This allows the association to be diffusion limited because most collisions lead to binding.

Science, this issue p. 1253

Transition paths of macromolecular conformational changes such as protein folding are predicted to be heterogeneous. However, experimental characterization of the diversity of transition paths is extremely challenging because it requires measuring more than one distance during individual transitions. In this work, we used fast three-color single-molecule Frster resonance energy transfer spectroscopy to obtain the distribution of binding transition paths of a disordered protein. About half of the transitions follow a path involving strong non-native electrostatic interactions, resulting in a transition time of 300 to 800 microseconds. The remaining half follow more diverse paths characterized by weaker electrostatic interactions and more than 10 times shorter transition path times. The chain flexibility and non-native interactions make diverse binding pathways possible, allowing disordered proteins to bind faster than folded proteins.

Read the original here:
Disordered proteins follow diverse transition paths as they fold and bind to a partner - Science Magazine

Recommendation and review posted by Alexandra Lee Anderson

Case Medical Awarded Patent for Multi Enzymatic Solution for Cleaning Medical Devices and Food Industry Utensils and Surfaces Exposed to Brain Wasting…

Patent is a significant step toward commercializing cleaning products to effectively inactivate and degrade prions

Case Medical today announced that it was awarded U.S. patent number 10,699,513 B2, by the U.S. Patent and Trademark office for "compositions and methods for handling potential prion contamination." The patent is a significant step for the company toward commercializing cleaning products that will enable prion contaminated devices and surfaces to be processed without resorting to the extraordinary methods required today.

Prions are a type of protein that can cause unfolding in normal prion proteins most commonly found in the brain, but also in the spine, eye, spleen, and lymphoid tissues. Prion diseases are described by the CDC as "a family of rare progressive neurodegenerative disorders that affect both humans and animals. They are distinguished by long incubation periods, characteristic spongiform changes associated with neuronal loss, and a failure to induce inflammatory response." The CDC also indicates that "the abnormal folding of the prion proteins leads to brain damage... Prion diseases are usually rapidly progressive and always fatal."

Prions are transmitted by eating of meat infected with prions, but also in healthcare settings from blood transfusions and from medical devices, especially from surgical instruments, even from apparently cleaned devices, having residual prion contamination.

"The challenge with prions is that they are almost impossible to detect before a fatal occurrence of the disease and they are also extremely hard to remove from contaminated devices and surfaces," said Marcia Frieze, CEO of Case Medical. "The logical solution would be to make prion decontamination a standard part of medical device processing but the current options are extremely time consuming and so harsh that they significantly reduce the useful life of the devices themselves."

Currently, prion contaminated materials are either incinerated or pre-treated with sodium hypochlorite, sterilization, oxidizing agents, peracetic acid, or pre-treatment at temperatures above 100C for extended periods of time. These methods and materials are environmentally unfriendly and excessively corrosive to the materials being cleaned. The cleaning solution patented by Case Medical uses a multi enzymatic formulation to achieve a safer, more thorough result and requires much less time and effort, suggesting a feasible process for healthcare settings and the food processing industry.

In brief, Case Medicals formulation uses specific enzymes combined with a surfactant. The enzymes effectively digest or inactivate prions rendering them ineffective and the surfactant lowers the level of friction to allow easy rinsing. The process is easy, biodegradable, and environmentally preferred.

"While prion diseases are currently rare and a much bigger issue in Europe than in the U.S., the coronavirus pandemic has hopefully taught us the value of being prepared," said Frieze. "We still have many regulatory steps before we can fully commercialize this product and process, but we are continuing to work as fast as we can."

Testing and validation were performed in conjunction with the U.S. Geological Service (USGS) through their National Wildlife Health Center at the Class III prion lab in Madison, Wisc.

About Case Medical

Case Medical is a FDA registered, ISO certified manufacturer of validated, sustainable, and cost effective products for instrument processing. Our reusable sterilization containers and instrument chemistries meet the highest standards for patient safety and environmental preference. Case Medical was an inaugural recipient of the U.S. EPA Safer Choice Partner of the Year award. Visit http://www.casemed.com for more information.

View source version on businesswire.com: https://www.businesswire.com/news/home/20200616005865/en/

Contacts

Lisa Forsell, Director of MarketingPhone: 201-313-1999 x302Email: lforsell@casemed.com Web: http://www.casemed.com

Read more:
Case Medical Awarded Patent for Multi Enzymatic Solution for Cleaning Medical Devices and Food Industry Utensils and Surfaces Exposed to Brain Wasting...

Recommendation and review posted by Alexandra Lee Anderson

What’s the Difference Between Prokaryotic and Eukaryotic Cells? – HowStuffWorks

Advertisement

You know when you hear somebody start a sentence with, "There are two kinds of people..." and you think to yourself "Oh boy, here it comes." Because reducing the whole of humanity down to "two kinds of people" seems like an odious activity at best.

But what if I were to tell you that there are just two kinds of organisms?

According to scientists, the world is split into two kinds of organisms prokaryotes and eukaryotes which have two different types of cells. An organism can be made up of either one type or the other. Some organisms consist of only one measly cell, but even so, that cell will either be either prokaryotic or eukaryotic. It's just the way things are.

The difference between eukaryotic and prokaryotic cells has to do with the little stuff-doing parts of the cell, called organelles. Prokaryotic cells are simpler and lack the eukaryote's membrane-bound organelles and nucleus, which encapsulate the cell's DNA. Though more primitive than eukaryotes, prokaryotic bacteria are the most diverse and abundant group of organisms on Earth we humans are literally covered in prokaryotes, inside and out. On the other hand, all humans, animals, plants, fungi and protists (organisms made up of a single cell) are eukaryotes. And though some eukaryotes are single celled think amoebas and paramecium there are no prokaryotes that have more than one cell.

"I think of a prokaryote as a one-room efficiency apartment and a eukaryote as a $6 million mansion," says Erin Shanle, a professor in the Department of Biological and Environmental Sciences at Longwood University, in an email interview. "The size and separation of functional 'rooms,' or organelles, in eukaryotes is similar to the many rooms and complex organization of a mansion. Prokaryotes have to get similar jobs done in a single room without the luxury of organelles."

One reason this analogy is helpful is because all cells, both prokaryotes and eukaryotes, are surrounded by a selectively permeable membrane which allows only certain molecules to get in and out much like the windows and doors of our home. You can lock your doors and windows to keep out stray cats and burglars (the cellular equivalent to viruses or foreign materials), but you unlock the doors to bring in groceries and to take out the trash. In this way, all cells maintain internal homeostasis, or stability.

"Prokaryotes are much simpler with respect to structure," says Shanle. "They have a single 'room' to perform all the necessary functions of life, namely producing proteins from the instructions stored in DNA, which is the complete set of instructions for building a cell. Prokaryotes don't have separate compartments for energy production, protein packaging, waste processing or other key functions."

In contrast, eukaryotes have membrane-bound organelles that are used to separate all these processes, which means the kitchen is separate from the master bathroom there are dozens of walled-off rooms, all of which serve a different function in the cell.

For example, DNA is stored, replicated, and processed in the eukaryotic cell's nucleus, which is itself surrounded by a selectively permeable membrane. This protects the DNA and allows the cell to fine-tune the production of proteins necessary to do its job and keep the cell alive. Other key organelles include the mitochondria, which processes sugars to generate energy, the lysosome, which processes waste and the endoplasmic reticulum, which helps organize proteins for distribution around the cell. Prokaryotic cells have to do a lot of this same stuff, but they just don't have separate rooms to do it in. They're more of a two-bit operation in this sense.

"Many eukaryotic organisms are made up of multiple cell types, each containing the same set of DNA blueprints, but which perform different functions," says Shanle. "By separating the large DNA blueprints in the nucleus, certain parts of the blueprint can be utilized to create different cell types from the same set of instructions."

You might be wondering how organisms got to be divided in this way. Well, according to endosymbiotic theory, it all started about 2 billion years ago, when some large prokaryote managed to create a nucleus by folding its cell membrane in on itself.

"Over time, a smaller prokaryotic cell was engulfed by this larger cell," says Shanle. "The smaller prokaryote could perform aerobic respiration, or process sugars into energy using oxygen, similar to the mitochondria we see in eukaryotes that are living today. This smaller cell was maintained within the larger host cell, where it replicated and was passed on to subsequent generations. This endosymbiotic relationship ultimately led to the smaller cell becoming a part of the larger cell, eventually losing its autonomy and much of its original DNA."

However, the mitochondria of today's eukaryotes have their own DNA blueprints that replicate independently from the DNA in the nucleus, and mitochondrial DNA has some similarity to prokaryotic DNA, which supports the endosymbiotic theory. A similar model is thought to have led to the evolution of chloroplasts in plants, but the story begins with a eukaryotic cell containing a mitochondria engulfing a photosynthetic prokaryote.

Eukaryotes and prokaryotes they're different! But even though it can be hard to see the similarities between humans and bacteria, we are all made of the same stuff: DNA, proteins, sugars and lipids.

Read more:
What's the Difference Between Prokaryotic and Eukaryotic Cells? - HowStuffWorks

Recommendation and review posted by Alexandra Lee Anderson

Origami Therapeutics, Inc. selected as a CONNECT 2020 Cool Company – thepress.net

SAN DIEGO, June 8, 2020 /PRNewswire/ -- Origami Therapeutics http://origamitherapeutics.com, an early stage biotech company taking a precision medicine approach to find disease-modifying treatments for neurodegenerative diseases caused by protein folding, announced today it has been selected as one of the 60 "Cool Companies" for 2020 by Connect with San Diego Venture Group. Origami was selected from a pool of over 300 tech and life science applicants.

Cool Companies is an annual capital program designed to match San Diego's best technology and life sciences startups ready to raise Series A with quality venture capital. The program selects top tier, local entrepreneurs raising institutional funding, and grants them opportunities for direct access to capital providers. The program regularly attracts over 200 VCs to the region annually. Since 2016, Cool Companies have raised over $400M, in just Series A institutional funding.

"We received a record number of applications from extraordinary companies for the 'Cool Companies' program this year," said Mike Krenn, CEO of Connect.

"We are very excited to be included in such a stellar group of new, innovative companies," said Beth Hoffman, Founder, President & CEO of Origami. "We are happy to be part of the vibrant San Diego biotech ecosystem and look forward to showcasing our novel therapeutics to investors."

Leveraging the Founder's experience in discovering transformational therapies for Cystic Fibrosis that modulate CFTR conformation, Origami's focus is to treat neurodegeneration by directly modulating the pathogenic proteins that cause disease. Their platform enables discovery of both protein degraders and conformation correctors, allowing them to match the best drug to treat each disease by using patient-derived disease models to ensure success in clinical trials.

About Connect

Connect is a community nonprofit organization passionate about helping tech and lifesci entrepreneurs build great companies. Connect serves entrepreneurs throughout their growth journey with a suite of curated programs aimed to help companies grow, gain access to capital, and scale. Connect helps innovative companies thrive so they can make a meaningful impact on the economic development of the region, and together create a world-class tech ecosystem.

About Origami Therapeutics

Origami is generating a pipeline of small molecule therapeutics that prevent or delay the onset and the progression of neurodegenerative diseases by targeting the underlying genetic cause of disease. Currently, they are selecting the optimal protein degrader molecule to advance into preclinical testing for Huntington's disease, a devastating fatal disease that strikes at the prime of life. Origami's core technology should be applicable to multiple neurological disorders where the proximate cause of the disease is a misfolded protein. These include Alzheimer's disease, Parkinson's disease, amyotrophic lateral sclerosis, frontotemporal dementia, Lewy body diseases, and other polyglutamine diseases. For more information, please visit http://www.origamitherapeutics.com

See the article here:
Origami Therapeutics, Inc. selected as a CONNECT 2020 Cool Company - thepress.net

Recommendation and review posted by Alexandra Lee Anderson

Why the buzz around DeepMind is dissipating as it transitions from games to science – CNBC

Google Deepmind head Demis Hassabis speaks during a press conference ahead of the Google DeepMind Challenge Match in Seoul on March 8, 2016.

Jung Yeon-Je | AFP |Getty Images | Getty Images

In 2016, DeepMind, an Alphabet-owned AI unit headquartered in London, was riding a wave of publicity thanks to AlphaGo, its computer program that took on the best player in the world at the ancient Asian board game Go and won.

Photos of DeepMind's leader, Demis Hassabis, were splashed across the front pages of newspapers and websites, and Netflix even went on to make a documentary about the five-game Go match between AlphaGo and world champion Lee SeDol. Fast-forward four years, and things have gone surprisingly quiet about DeepMind.

"DeepMind has done some of the most exciting things in AI in recent years. It would be virtually impossible for any company to sustain that level of excitement indefinitely," said William Tunstall-Pedoe, a British entrepreneur who sold his AI start-up Evi to Amazon for a reported $26 million. "I expect them to do further very exciting things."

AI pioneer Stuart Russell, a professor at the University of California, Berkeley, agreed it was inevitable that excitement around DeepMind would tail off after AlphaGo.

"Go was a recognized milestone in AI, something that some commentators said would take another 100 years," he said. "In Asia in particular, top-level Go is considered the pinnacle of human intellectual powers. It's hard to see what else DeepMind could do in the near term to match that."

DeepMind's army of 1,000 plus people, which includes hundreds of highly-paid PhD graduates, continues to pump out academic paper after academic paper, but only a smattering of the work gets picked up by the mainstream media. The research lab has churned out over 1,000 papers and 13 of them have been published by Nature or Science, which are widely seen as the world's most prestigious academic journals. Nick Bostrom, the author of Superintelligence and the director of the University of Oxford's Future of Humanity Institute described DeepMind's team as world-class, large, and diverse.

"Their protein folding work was super impressive," said Neil Lawrence, a professor of machine learning at the University of Cambridge, whose role is funded by DeepMind. He's referring to a competition-winning DeepMind algorithm that can predict the structure of a protein based on its genetic makeup. Understanding the structure of proteins is important as it could make it easier to understand diseases and create new drugs in the future.

The World's top human Go player, 19-year-old Ke Jie (L) competes against AI program AlphaGo, which was developed by DeepMind, the artificial intelligence arm of Google's parent Alphabet. Machine won the three-game match against man in 2017. The AI didn't lose a single game.

VCG | Visual China Group | Getty Images

DeepMind is keen to move away from developing relatively "narrow" so-called "AI agents," that can do one thing well, such as master a game. Instead, the company is trying to develop more general AI systems that can do multiple things well, and have real world impact.

It's particularly keen to use its AI to leverage breakthroughs in other areas of science including healthcare, physics and climate change.

But the company's scientific work seems to be of less interest to the media.In 2016, DeepMind was mentioned in 1,842 articles, according to media tracker LexisNexis. By 2019, that number had fallen to 1,363.

One ex-DeepMinder said the buzz around the company is now more in line with what it should be. "The whole AlphaGo period was nuts," they said. "I think they've probably got another few milestones ahead, but progress should be more low key. It's a marathon not a sprint, so to speak."

DeepMind denied that excitement surrounding the company has tailed off since AlphaGo, pointing to the fact that it has had more papers in Nature and Science in recent years.

"We have created a unique environment where ambitious AI research can flourish. Our unusually interdisciplinary approach has been core to our progress, with 13 major papers in Nature and Science including 3 so far this year," a DeepMind spokesperson said. "Our scientists and engineers have built agents that can learn to cooperate, devise new strategies to play world-class chess and Go, diagnose eye disease, generate realistic speech now used in Google products around the world, and much more."

"More recently, we've been excited to see early signs of how we could use our progress in fundamental AI research to understand the world around us in a much deeper way. Our protein folding work is our first significant milestone applying artificial intelligence to a core question in science, and this is just the start of the exciting advances we hope to see more of over the next decade, creating systems that could provide extraordinary benefits to society."

The company, which competes with Facebook AI Research and OpenAI, did a good job of building up hype around what it was doing in the early days.

Hassabis and Mustafa Suleyman, the intellectual co-founders who have been friends since school, gave inspiring speeches where they would explain how they were on a mission to "solve intelligence" and use that to solve everything else.

There was also plenty of talk of developing "artificial general intelligence" or AGI, which has been referred to as the holy grail in AI and is widely viewed as the point when machine intelligence passes human intelligence.

But the speeches have become less frequent (partly because Suleyman left Deepmind and works for Google now), and AGI doesn't get mentioned anywhere near as much as it used to.

Larry Page, left, and Sergey Brin, co-founders of Google Inc.

JB Reed | Bloomberg | Getty Images

Google co-founders Larry Page and Sergey Brin were huge proponents of DeepMind and its lofty ambitions, but they left the company last year and its less obvious how Google CEO Sundar Pichai feels about DeepMind and AGI.

It's also unclear how much free reign Pichai will give the company, which cost Alphabet $571 million in 2018. Just one year earlier, the company had losses of $368 million.

"As far as I know, DeepMind is still working on the AGI problem and believes it is making progress," Russell said. "I suspect the parent company (Google/Alphabet) got tired of the media turning every story about Google and AI into the Terminator scenario, complete with scary pictures."

One academic who is particularly skeptical about DeepMind's achievements is AI entrepreneur Gary Marcus, who sold a machine-learning start-up to Uber in 2016 for an undisclosed sum.

"I think they realize the gulf between what they're doing and what they aspire to do," he said. "In their early years they thought that the techniques they were using would carry us all the way to AGI. And some of us saw immediately that that wasn't going to work. It took them longer to realize but I think they've realized it now."

Marcus said he's heard that DeepMind employees refer to him as the "anti-Christ" because he has questioned how far the "deep learning" AI technique that DeepMind has focused on can go.

"There are major figures now that recognize that the current techniques are not enough," he said. "It's very different from two years ago. It's a radical shift."

He added that while DeepMind's work on games and biology had been impressive, it's had relatively little impact.

"They haven't used their stuff much in the real world," he said. "The work that they're doing requires an enormous amount of data and an enormous amount of compute, and a very stable world. The techniques that they're using are very, very data greedy and real-world problems often don't supply that level of data."

Read the original post:
Why the buzz around DeepMind is dissipating as it transitions from games to science - CNBC

Recommendation and review posted by Alexandra Lee Anderson


Page 11234..10..»