WO2025059192A1 - Methods of using nanovials to associate secretions and transcriptomes of single cells - Google Patents

Abstract

A Secretion EnCoded (SEC) single-cell sequencing (seq) (SEC-seq) method and platform is disclosed which exploits hydrogel particles with sub-nanoliter cavities (nanovials) to capture individual cells and their secretions. This enables one to simultaneously measure the secretions and the transcriptome for a population of cells. In one example, the method was used measure secretion of vascular endothelial growth factor A (VEGF-A) and the transcriptome for thousands of individual mesenchymal stromal cells (MSCs). The highest VEGF-A secretion occurs in a subpopulation of MSCs characterized by a unique gene expression signature comprising a surface marker that defines this subpopulation to be enriched. SEC-seq enables the identification of gene-signatures linked to specific secretory states, enabling mechanistic studies, the isolation of secretory subpopulations, and the development of means to modulate cellular secretion. Specific improved therapeutic compositions of MSCs yielding enhanced secretions of therapeutic molecules are also disclosed.

Description

METHODS OF USING NANOVIALS TO ASSOCIATE SECRETIONS AND TRANSCRIPTOMES OF SINGLE CELLS

Related Application

[0001] This Application claims priority to U.S. Provisional Patent Application No. 63/582.215 filed on September 12, 2023 which is hereby incorporated by reference. Priority is claimed pursuant to 35 U.S.C. § 119 and any other applicable statute.

Technical Field

[0002] The technical field generally relates to methods and workflows that that link individual cell secretions to the cell transcriptional state. The particle-based platform enables the enrichment or identification of biomarkers used to enrich desirous sub-populations of cells. The method, which uses secretion encoded single-cell sequencing (referred herein as SEC-seq), enables the identification of gene-signatures linked to specific secretory states, enabling mechanistic studies, the isolation of secretory subpopulations, and the development of means to modulate cellular secretion. The technical field also relates to methods to improve the function or efficacy of a cellular therapeutic by selecting surface markers associated with secretory function using the SEC-seq technique, and specific improved selection and preparation processes for mesenchymal stromal or stem cell-based therapies.

Statement Regarding Federally Sponsored Research and Development

[0003] This invention was made with government support under GM099134, and DK128730 awarded by the National Institutes of Health. The government has certain rights in the invention.

Background

[0004] Cell function is defined by a myriad of biomolecules that they secrete. Over 3000 proteins are predicted to be secreted from human cells, and secreted proteins such as immunoglobulins, cytokines, chemokines, extracellular matrix proteins, proteases, morphogens and grow th factors span a diversity of critical functions. For example, mesenchymal stromal cells (MSCs) have been widely evaluated as therapeutics because they secrete bioactive factors including growth and neurotrophic factors (VEGF, HGF, GDNF), cytokines (IDO, PGE2, TGF-0, IL- 10), and extracellular vesicles, which promote immunomodulation and regeneration. Yet, successful translation of MSCs and other cell therapies have been hindered by' clinical outcome inconsistency attributed to functional differences that stem from the cell source and single-cell level heterogeneity⁷. The exploitation of the full therapeutic potential of cell therapies therefore necessitates an understanding of the cell sub-populations that secrete specific proteins, and how these change under distinct microenvironmental conditions associated with therapeutic production and delivery'. Given the heterogeneity' of cell cultures, such a method ideally would detect protein secretion at the single-cell level. Moreover, defining distinct secretory⁷ cell states through their gene expression state would enable a thorough characterization of the secreting cell subpopulations. In addition to enabling the identification of new regulatory⁷ mechanisms and required components in secretory pathways, the ability to link specific gene expression networks to secretion will also reveal targets for drug discovery to increase or decrease secretion of therapeutic or pathologic molecules, or genetic modifications to modulate secretion of biomolecules. Thus, a method that links the transcriptional state and secretions at scale would be highly valuable, both for a basic understanding of protein secretion and for creating and selecting sub-populations of cells for the next generation of cell therapies.

Summary

[0005] Here, a method was used determine secretory heterogeneity' in a population of cells and correlate it with the single-cell gene expression profile. Secretion encoded single-cell sequencing (SEC-seq) leverages microscale hydrogel particles (also referred to as nanovials herein) and single-cell (sc)RNA-seq in microfluidic droplet emulsions, to retain and analyze both transcriptome and secretion information from thousands of individual cells. Cells are loaded into gelatin-coated nanovials conjugated with capture antibodies for a secreted protein of interest, allowing the cells to adhere to the interior of the nanovial cavity and secreted protein to bind to the conjugated antibodies. After an appropriate incubation time, nanovials are incubated with oligonucleotide (oligo)-barcoded detection antibodies against the secreted product. Single-cell loaded nanovial samples, enriched by FACS, are then partitioned for downstream scRNA-seq, followed by library preparation for mRNA and oligo-barcode detection, sequencing, and data analysis employing established workflows used in other barcode-based multi-omics approaches. [0006] The utility of SEC-seq is demonstrated by exploring the relationship between the secretion of vascular endothelial growth factor A (VEGF-A) and the underlying transcriptome state in MSCs under normoxic and hypoxic culture conditions. VEGF-A secretion was the focus since this growth factor is an important secreted product that promotes angiogenesis and is involved in the mechanism of action of many MSC-based therapies. A relatively low correlation was found between VEGF-A secretion and VEGFA transcript levels across cells in each culture condition. Yet, the induction of the hypoxic response globally elevates VEGFA transcript levels and VEGF-A secretion across all cells, indicating that the modulation of VEGFA transcript levels can tune the secretion output under certain conditions. In addition, a subset of cells found in both normoxic and hypoxic culture conditions secrete highly elevated levels of VEGF-A. This subpopulation of cells is not defined by significantly increased VEGFA transcript levels; instead, it is characterized by a unique, secretory' gene-enriched transcription profile. This gene expression signature is referred to as the ‘Vascular Regenerative Signal’ (VRS) based on the presence of transcripts related to cell motility, blood vessel development, and wound response, and its link to the high VEGF-A secretory state. Inspection of the VRS gene signature revealed differential expression of several surface markers, one of which was exploited to enrich for the subpopulation of highly VEGF-A secreting MSCs and to show that this population stably maintains high VEGF-A secretion over time. Together, the findings suggest that multiple regulatory pathways control VEGF-A secretion, only revealed by simultaneous measurement of secretion and the transcriptome from thousands of single cells, demonstrating the need for a better understanding of secretory' sub-populations.

[0007] In one embodiment, a method of identifying one or more cell surface markers of cells associated with elevated secretion of a bioactive factor includes loading single cells from a population of cells into a plurality of ‘‘nanovial” particles, wherein the single cells are contained within a void or cavity in the plurality' of particles that opens to the exterior environment of the respective particles, wherein the particle contains a capture agent for the bioactive factor disposed on a surface of the particle within the void or cavity. Cells with elevated secretion levels of the bioactive factor are labelled by exposing the plurality of particles with single cells contained therein to oligo-barcoded labeling reagents (e g., detection antibodies) that bind to or associate with the bioactive factor. The single cells in the plurality of particles are exposed to one or more reagents of a single cell transcriptome assay. The transcriptome of the cells with elevated secretion levels is determined and one or more genes of the transcriptome that encode for the one or more cell surface markers associated with elevated secretion levels of the bioactive factor are identified.

[0008] In another embodiment, a method of enriching a sub-population of mesenchymal stromal cells or mesenchymal stem cells (MSCs) with elevated secretion of VEGF-A protein includes exposing a population of MSCs to anti-interleukin-13 receptor subunit alpha-2 (IL- 13RA2) antibody. The population of exposed MSCs are then sorted based on the presence of the anti-IL-13RA2 antibody to generate the enriched sub-population of MSCs with elevated secretion levels of VEGF-A protein. Sorting may be done using a fluorescent activated cell sorter (FACS) or a magnetic activated cell sorter (MACS). The enriched population may be expanded and used as a therapeutic (e.g., implantation, injection or infusion).

[0009] In another embodiment, a therapeutic composition is disclosed that includes an enriched sub-population of MSCs with elevated secretion levels of VEGF-A protein, the enriched sub-population of MSCs comprising IL13RA2-positive (+) MSCs. In another aspect, the therapeutic composition which includes the enriched sub-population of MSCs includes MSCs with elevated levels of gene expression comprising a vascular regenerative signal (VRS).

Brief Description of the Drawings

[0010] FIG. 1 illustrates a single particle or nanovial having a single cell loaded therein. Also illustrated is a capture antibody bound to a secretion from the cell.

[0011] FIG. 2 illustrates an illustrative workflow for identifying one or more cell surface markers of cells with elevated secretion of a bioactive factor.

[0012] FIG. 3 is an illustrative workflow for enriching a sub-population of cells with elevated secretion of the bioactive factor.

[0013] FIGS. 4A-4B: Overview of the SEC-seq workflow using nanovials. FIG. 4A illustrates that cells are loaded, adhered, and incubated in gelatin-coated nanovials functionalized with secretion capture antibodies. Secreted and captured proteins are labeled with oligonucleotide-barcoded detection antibodies. Nanovials loaded with single cells are then introduced into the 10X Chromium workflow for library preparation. Sequencing of resulting libraries results in linked secretion and transcriptomics data for downstream analyses. FIG 4B shows that gelatin-coated nanovials are fabricated using aqueous two-phase separation of gelatin and 4-arm polyethylene glycol (PEG) acrylate in a flow focusing device in which phase separation occurs followed by UV crosslinking. The resulting 35 pm nanovials are highly uniform with a cavity diameter coefficient of variation (CV) of 1.7% and nanovial diameter CV of 5.5%. MSCs loaded in nanovials settle in the center of the cavity, as seen in the confocal microscopy image where fluorescent streptavidin-labeled nanovials are loaded with calcein stained cells. Scale bars are 50 pm.

[0014] FIGS. 5A-5L illustrate an illustrative SEC-seq workflow using nanovials. FIG. 5A illustrates standard scRNA-seq equipment including Sony cell sorter, 10X Chromium scRNA-seq device, and Illumina sequencers are used in the different steps of the SEC-seq workflow (top to bottom). Checkmarks indicate the steps that were validated with corresponding equipment. FIG. 5B shows the standard curve of VEGF-A on nanovials using recombinant VEGF-A. immobilized via the VEGF-A capture antibody, and detected with fluorescently (AF647)-tagged anti- VEGF-A detection antibody shows a dynamic range across 2 orders of magnitude. Horizontal line represents the detection threshold (see methods). FIG. 5C (left) illustrates a schematic showing the steps of the VEGF-A secretion assay in single MSC-loaded nanovials. FIG. 5C (middle) shows flow cytometry' histograms of VEGF-A secretion (as detected by fluorescently (AF647)-tagged anti-VEGF-A) from single MSCs on nanovials after 2, 8, or 14 hours of incubation. FIG. 5C (right bottom) illustrates VEGF-A secretion assay on single MSC-loaded nanovials with and without VEGF-A capture antibody (Ab). 90% of cells in nanovials with capture antibody had fluorescence signal above the threshold (dotted line). FIG. 5C (right top) shows the fluorescence microscopy image shows single MSCs on nanovials with secreted VEGF-A detected with a fluorescently (AF647)-tagged VEGF-A detection antibody and cells stained with calcein AM. Scale bar is 50 pm. FIG. 5D illustrates a bar plot shows cell viability⁷ measured by image analysis of live/ dead stain (see methods) following flow sorting of cells in suspension or loaded in nanovials, with error bars for standard deviation. FIG. 5E (top) shows a schematic of nanovial loading into droplets with 10X training gel beads in the absence of detergent (to prevent cell lysis). FIG 5E (bottom) shows brightfield and fluorescence images of nanovials (red) with single MSCs (green) together with a gel bead in a droplet following emulsification. Scale bar is 50 pm. FIG. 5F illustrates a graph showing the proportion of all nanovial- containing droplets with the indicated number of nanovials. with error bars for standard deviation. FIG. 5G shows in an image that after droplet formation and in the presence of lysis buffer, lysis of calcein (green)-labeled cells on nanovials was observed by diffusion of the green fluorescent signal throughout the droplets containing single-MSC loaded nanovials. Overlaid fluorescence and brightfield images of droplets generated with a 10X Genomics NextGEM kit. Scale bar is 50 pm. FIG. 5H shows the distribution of species-specific reads from a scRNA-seq experiment with nanovials containing human MSCs or mouse fibroblasts pooled in a 1 : 1 ratio. Species identity was called by mapping to a joined genome contig and determining the ratio of reads from each species’ genome. FIG. 51 shows a comparison of transcripts per cell for either suspended (unsorted) MSCs, suspended and sorted (sorted) MSCs, or MSCs loaded on nanovials and sorted (nanovial). Number above plot indicates number of cells per sample. FIG. 5J illustrates an experiment where MSCs cultured under normoxic and hypoxic conditions, respectively, were loaded on nanovials labeled with different oligo-barcoded streptavidin molecules (‘normoxic’ and ‘hypoxic’ barcodes) and analyzed in a 1: 1 ratio in a single 10X channel. The scatter plot below depicts the assignment of cells based on the normoxic or hypoxic oligo-barcode attached to nanovials via streptavidin. Mixed cells have a signal for both barcodes. FIG. 5K shows UMAP plots of the scRNA-seq data derived from the experiments in (FIG. 5 J), where each cell is labeled by their oligo-barcode assignment. Mixed cells are excluded in UMAPs. FIG. 5L shows the UMAP from (FIG. 5K) labeled by the hypoxic gene expression signature to identify MSCs cultured in hypoxic conditions.

[0015] FIGS. 6A-6I illustrates SEC-seq measuring the transcriptome and VEGF-A secretion of normoxic or hypoxic MSCs. FIG. 6A is a schematic of the detection of secreted VEGF-A protein and corresponding global gene expression for individual MSCs using the SEC-seq method. FIG. 6B is a UMAP dimensionality reduction based on transcriptomes from SEC-seq experiments on normoxic and hypoxic MSCs in nanovials. Cells are labeled according to the culture condition. FIG. 6C is a UMAP displaying VEGF-A secretion level, shown as log transformation of the UMI collapsed anti-VEGF-A oligo-barcode reads per cell. FIG. 6D is a UMAP displaying VEGFA transcript levels, shown as normalized transcripts per cell. FIG. 6E is a distribution of VEGF-A secretion for cell-loaded nanovials in normoxic and hypoxic conditions, detected by FACS using a fluorescent anti-VEGF-A antibody (top) or by the SEC-seq experiment in (FIG. 6B) using the oligo-barcoded anti-VEGF-A antibody (middle). The last plot shows distribution of VEGFA transcript levels from the SEC-seq experiment cells in (FIG. 6B) (bottom). FIG. 6F is a UMAP displaying cluster assignment. FIG. 6G shows violin plots by cluster showing VEGFA transcripts and VEGF-A secretion levels for all cells in the normoxic clusters (N1-N5), hypoxic clusters (Hl-4), and mixed clusters (Ml-3) from (FIG. 6F). For mixed clusters, the levels are shown separately for normoxic and hypoxic cells. The dashed line represents the mean across all cells for each plot. Data below this threshold are lightened to highlight differences. FIG. 6H shows scatter plots showing the relationship between VEGFA transcript and VEGF-A secretion levels for individual normoxic (left) and hypoxic (right) cells from the experiment in (FIG. 6B). Best fit regression lines and Pearson correlation coefficients are shown. FIG. 61 is a plot showing the ranking of all detected genes based on the correlation of their transcript levels to the VEGF-A secretion level per cell for normoxic (top) and hypoxic (bottom) MSCs. The rank of the EEGFA gene is highlighted, and the top three genes per sample are also noted.

[0016] FIGS. 7A-7P illustrate the characterization of the high-VEGF-A secreting MSC subpopulation. FIG. 7A shows a scatter plot of the transcript to VEGF-A secretion correlation for all genes from SEC-seq experiments for normoxic and hypoxic MSCs from FIGS. 6A-6I. The 10 most highly correlating genes based on both experiments are labeled. FIG. 7B is a table giving the ranking (based on average correlation), gene name, correlation to secretion in normoxic and hypoxic cells, and the average of those two values for the ten top genes from FIG. 7A. FIG. 7C is a UMAPs showing VEGF-A secretion levels and expression of 5 select correlated genes from FIG. 7B in normoxic and hypoxic MSCs from FIGS. 6A-6I. The VEGF-A secretion UMAP is given from FIG. 6C for comparison. FIG. 7D as in (FIG. 7C), for a separate SEC-seq experiment performed on MSCs in the normoxic culture condition. FIG. 7E illustrates cell clusters projected onto the UMAP of the replicate SEC-seq experiment. FIG. 7F shows the heatmap of the top 10 differentially expressed genes from each cluster (indicated on top) of the SEC-seq experiment in (C, D) (rows=genes, columns=individual cells). Displayed at the top are the log transformed VEGF-A secretion and VEGFA transcript levels. Right: top 3 genes differentially expressed gene for each cluster. FIG. 7G shows the heatmap of the top GO terms found for all of the differentially expressed genes from the clusters in (FIG. 7E). The (-logP) value indicates if the term was enriched for a given cluster. FIG. 7H is a Venn diagram showing the overlap of differentially expressed genes from the highly secreting cluster in 3 SEC-seq experiments (top left: normoxic MSCs from (FIG. 6B), top right: hypoxic MSCs from (FIG. 6B), bottom: normoxic MSCs from (FIG. 7E). Overlapping genes form the Vascular Regenerative Signal (VRS). FIG. 71 is a gene ontology analysis for VRS genes from (FIG. 7H). Similar terms were collapsed. FIG. 7J is the average of the normalized transcripts level of VRS genes per cell, displayed for the SEC-seq experiment in (FIG. 7E) and (FIG. 6B). FIG. 7K as in (FIG. 7J), for MSCs loaded in oligo-barcoded nanovials (see FIG. 5J-5L). FIG. 7L, as in (FIG. 7 J), for a standard scRNA-seq experiment on unsorted, suspended MSCs. FIG. 7M is a comparison of gene type classification for VRS genes and genes differentially expressed in all clusters in (FIG. 7E) except for those from cluster c5. FIG. 7N illustrates enrichment of possible TF regulators of the VRS genes based on the TRRUST database. FIG. 70 is the consensus rank of VEGF-A secretion to gene correlation based the SEC-seq experiments used in (FIG. 7H), with lighter colored dots (left side) displaying all VRS genes. FIG. 7P is a schematic depicting the heterogeneity of VEGF-A secretion in MSCs under normoxic and hypoxic conditions, highlighting the importance of the VRS genes for marking high VEGF-A secretion.

[0017] FIGS. 8A-8I illustrate the enrichment of Vascular Regenerative Signal MSCs Using a Surface Marker. FIG. 8A is a UMAP from the SEC-seq experiment on normoxic MSCs from FIG. 8D, showing the expression level ofIL13RA2. FIG. 8B is a flow cytometry histograms of normoxic MSCs with anti-IL13RA2 labeling and without (negative control), along with gates used to sort IL13RA2+/- cells. FIG. 8C is a volcano plot showing the average log2 fold change per gene between IL13RA2-positive and negative cells from triplicate bulk RNA-seq experiments and the - log p-value for the significance of the difference between the two sample groups. Significant genes are highlighted in black (threshold >1.5 log2 fold change, <0.05 p-val), VRS genes are lighter colored data points, and VEGFA is identified with an arrow. FIG. 8D is a heatmap showing the expression of the top 20 VEGF-A secretion correlates, which overlaps with many VRS genes, in each of the three bulk RNA-seq replicates for the IL13RA2 + and -negative cells. Row values are normalized to show percent of row across samples. FIG. 8E is a UMAP from the SEC-seq experiment on normoxic MSCs from FIG. 7D showing the average expression of genes significantly upregulated in IL13RA2+ compared to IL13RA2- cells (>4-fold change, <0.05 p-val). FIG. 8F is an ELISpot assay spot count for VEGF-A secretion on IL13RA2+/- sorted MSCs immediately after isolation of the two cell populations. FIG. 8G shows VEGF-A secretion measured by ELISA from conditioned media taken from IL13RA2+/- cells expanded for 6 days after sort. FIG. 8H (top) is a flow cytometry⁷ histogram plot of normoxic MSCs stained for IL13RA2. FIG. 8H (bottom) shows the IL13RA2+/- populations were expanded for six days and each reanalyzed by FACS for IL13RA2 expression after expansion. FIG. 81 illustrates the workflow leveraging SEC-seq data for isolation of a subpopulation of interest.

[0018] FIG. 9A-9H illustrates the cell loading into nanovials, enrichment of single cell- loaded nanovials by FACS, and VEGF-A nanovial secretion assay validation. FIG. 9A is a schematic showing anchorage of single cells on nanovials coated with gelatin via integrin binding. Cell loading into nanovials is achieved by simple mixing. FIG. 9B is a flow cytometry histogram of cell-loaded nanovials stained with calcein AM viabi 1 ity dye. Cells are sorted via FACS (SONY SH800S) based on calcein signal into ‘Multiple Cells’ and ‘Singlecell’ gates. The distribution of the calcein signal has one peak with a tail at higher intensities, which represents nanovials containing more than one cell. FIG. 9C shows the results of three cell loading concentrations (0.4. 0.7. and 1 cell per nanovial) and the analyzed the fraction of nanovials carrying no, single or multiple cells using the gates described in (FIG. 9B). The graph quantifies cell loading into nanovials for these conditions. When loading cells at the 1: 1 cell-to-nanovial ratio, -23% single-cell loaded nanovials was achieved which could be separated by sorting for downstream approaches and analyses. FIG. 9D illustrates fluorescence microscopy images of nanovials sorted for the indicated gates as described in (FIG. 9B). By sorting nanovials in the ‘Single-cell’ gate, this enriched for nanovials carrying single cells as confirmed by Hoechst nuclei staining, whereas nanovials in the tail (‘Multiple Cells Gate’) represented mostly two or more loaded cells. Following sorting, it was estimated that 95% of the “Single-cell” gate sorted nanovials contained one cell based on image analysis. To isolate single cells on single nanovials, the following gating strategy (FIG. 9E) was used for cell-loaded nanovial samples: i) All nanovials were gated based on FSC/SSC, followed by ii) a single nanovial gate based on FSC-Width, and iii) the ‘Single-cell’ gate based on calcein signal intensity, iv) Flow cytometry fluorescence histogram of the fluorescent (AF647) anti-VEGF-A detection signal in single cell-loaded nanovials and empty nanovials isolated from the same nanovial cell-loading experiment after 12 hours of secretion incubation. Single cell-loaded nanovials have higher fluorescent (AF647) anti-VEGF-A signal than empty nanovials. showing low crosstalk. FIG. 9F illustrates the stability' of recombinant VEGF-A on nanovials over 24 hours. There is a 23% decrease in AF647 Anti- VEGF-A signal from 0 to 12 hours, and a 10% decrease in signal from 12 to 24 hours. FIG. 9G shows the level of autofluorescence and VEGF-A detection antibody signal for cell- loaded nanovials without VEGF-A capture/detection antibodies and cell-loaded nanovials without the VEGF-A capture antibodies, respectively. FIG. 9H shows the image of one well in the ELISpot assay measuring VEGF-A secretion from MSCs. An average of 99% of cells seeded formed spots across 3 wells. The range of integrated intensity of the spots across 3 wells, quantified in the histogram on the right of FIG. 9H, indicates that there is heterogeneity in secretion level. [0019] FIGS. 10A-10D illustrates how the nanovials protect viability of MSCs during flow sorting. FIG. 10A shows the effect of surfactant and sorting on viability of MSCs in nanovials or freely suspended in media, as measured by live/dead stain imaging. Typically, nanovial samples are kept in buffer with a surfactant (Pluronic) at low concentration for handling and sorting steps, as it prevents nanovials from aggregating. Here, the surfactant as a stressor was exploited to test the effect of sorting on the viability of suspendered cells and cells in nanovials. For suspended samples. MSCs were dissociated from flasks, resuspended in FACS buffer with and without Pluronic surfactant, and viability was measured for MSCs with and without sorting. For MSC-loaded nanovial samples, MSCs were loaded on nanovials, resuspended in wash buffer with and without Pluronic surfactant, and viability was measured for MSCs after sorting. Viability decreased significantly when MSCs suspended in FACS buffer with Pluronic are sorted, but all other conditions maintained high viability. Surfactant, even at low concentrations, can likely induce some membrane damage, which is enhanced during sorting; however, nanovials seem to protect cells from this damage. Error bars are for standard deviation. FIG. 10B shows finite element modeling using COMSOL results show that cells in nanovials are exposed to reduced levels of shear stress compared to cells in suspension when flowing through the nozzle of a flow sorter (see methods). Here, the shear stress is plotted on the cell and nozzle geometry', and shows how the suspended cell (right) experiences greater shear stress than the cell inside nanovial (left). FIG. 10C shows shear stress from the COMSOL model is plotted against position along the cell perimeter for suspended cells and cells adhered within a nanovial. The arrow in each schematic (based on the model geometry) indicates the direction of arc length and shear stress measurement. FIG. 10D illustrates the range of shear stress for suspended cells and cells adhered within a nanovial based on (FIG. 10C). with average shear stress plotted. The average shear stress is 400-fold higher for suspended cells than cells in nanovials.

[0020] FIGS. 11 A-l IF illustrates transcript changes related to cell loading and adhesion in nanovials. FIG. 11 A, for the scRNA-seq experiment with MSCs loaded into nanovials or freely suspended shown in FIG. 51, the graph shows the genes detected per cell for suspended and unsorted MSCs (unsorted), suspended FACS-sorted MSCs (sorted), and MSCs loaded on nanovials and sorted (Nanovial). FIG. 1 IB is a heatmap showing the average normalized transcript levels of known MSC markers (top) and markers from other cell ty pes (bottom) in each condition from the experiment in (FIG. 11 A). FIG. 11C shows UMAPs of the combined transcriptome data from the scRNA-seq experiment described in (FIG. 11 A). The cells from each condition are separately displayed and colored. FIG. 1 ID as in (FIG. 11C), showing the mean transcript level of genes significantly upregulated in cells adhered to nanovials relative to suspended MSCs (top) or upregulated in suspended MSCs (bottom). FIG. HE shows gene ontology for the two gene sets from (FIG. 1 ID). FIG. 1 IF shows the proportion of cells in cell-loaded and oligo-barcoded nanovials that could not be associated with the oligobarcoded nanovials and are therefore presumably detached, or of cells that could be associated with oligo-barcoded nanovials and therefore presumably are present within a nanovial, after the 10X emulsification step.

[0021] FIGS. 12A-12B show the effect of hypoxia inducers on VEGF-A secretion by MSCs. FIG. 12A shows the ELISA results for VEGF-A levels in conditioned media collected from MSCs grown on tissue culture plates under normoxic condition (normal growth media) or treated with indicated concentrations of cobalt chloride (C0CI2) and deferoxamine (DFX) hypoxia mimicking agents for 24 hours to induce hypoxic conditions, as indicated. Error bars are for standard deviation. FIG. 12B shows flow' cytometry histograms for two fluorescence channels indicating calcein positive MSC-loaded nanovials with anti-VEGF-A labeling on nanovials for MSCs treated with indicated concentrations of cobalt chloride (C0CI2) and deferoxamine (DFX) hypoxia mimicking agents for 14 hours total. Normoxia and no anti- VEGF-A capture antibody controls are also shown. 500 pM DFX yielded the largest increase in VEGF-A secretion (fluorescent (AF647) anti-VEGF signal) without compromising cell metabolic activity/viability (calcein).

[0022] FIGS. 13A-13C illustrate oligo-barcoded Anti-VEGF-A binding specificity on nanovials. FIG. 13A is an anti-VEGF-A antibody (used as detection antibody for VEGF-A secretion in the SEC-seq approach) was conjugated with a 10X compatible oligo-barcode along with additional sequences necessary for lOx library preparation. The schematic shows the sequence composition of the oligo attached to the VEGF-A detection antibody, along with the lOx primer sequence which hybridizes to the antibody-derived oligo and adds the unique molecular identifier (UMI) upon reverse transcription. FIG. 13B (Left) is a schematic showing the attachment of recombinant VEGF-A and the detection immunoassay via the oligo-barcoded antibody described in (FIG. 13 A) which was quantified with a fluorescently- labeled secondary antibody in nanovials by flow cytometry. (Right) Flow' cytometry histograms showing outcome of the experiment on the left with recombinant VEGF-A with and without the oligo-barcoded VEGF-A detection antibody, demonstrating antibody specificity of the detection assay and the validity of the oligo-barcoded VEGF-A detection antibody to the presence of VEGF-A protein. The numbers on the left indicate the mean of both histograms. FIG. 13C same as in (FIG. 13B). except that no biotinylated anti-VEGF-A capture antibody and recombinant VEGF-A was used in the assay.

[0023] FIGS. 14A-14C show the analysis of the SEC-seq experiments for normoxic and hypoxic MSCs. FIG. 14A show s the proportion of hypoxic and normoxic MSCs from the SEC-seq experiments described in FIGS. 6A-6I, per cluster depicted in FIG. 6F, as percent of each sample or as a percent of each cluster. FIG. 14B shows a scatter plot showing the correlation between VEGFA transcript and VEGF-A secretion for individual cells from comprising hypoxic and normoxic conditions, from FIGS. 6A-6I. FIG. 14C is a violin plot showing the loglO total transcript count per cell, for each cluster in FIGS. 6A-6I. An overlayed boxplot shows the median and first and third quartiles, in addition to the lo er and upper bounds of the data. Outliers are labeled as dots.

[0024] FIGS. 15A-15H illustrate the identification of a high-VEGF-A secreting MSC subpopulation in a replicate experiment. FIG. 15 A show s UMAPs showing the normalized transcript level of the indicated genes for the combined normoxic/hypoxic SEC-seq experiments from FIGS. 6A-6I. The five genes shown belong to the top 10 transcripts correlating best with VEGF-A secretion. FIG. 15B as in (FIG. 15A), for the replicate normoxic SEC-seq experiment from FIGS. 7D, 7E, and 7C, UMAP w ith cluster information (also shown in FIG. 7E) and violin plots showing VEGF-A secretion and VEGFA transcript levels for all cells in each cluster in the SEC-seq experiment for normoxic MSCs from FIG. 15B. FIG. 15C illustrates UMAPs and cluster information for VEGFA gene expression and VEGF-A Secretion. FIG. 15D shows UMAPs and cluster information in violin plots showing the average normalized transcript level of the 10 best correlating genes with VEGF-A secretion from FIG. 7B for the SEC-seq experiments with normoxic and hypoxic MSCs from FIGS. 6A-6I and the normoxic replicate from FIG. 7D, 7E. The UMAPs with cluster information are repeated here for ease of interpretation. FIG. 1 E illustrates for the SEC-seq experiment with normoxic MSCs in (FIG. 15B), all detected genes were ranked by the correlation of their transcript levels to the VEGF-A secretion level. Each gene is plotted by its rank and correlation. The ranks of the VEGFA and IGFBP6 genes are highlighted and the correlation values are given. FIGS. 15F-15G are scatter plots showing FIG. 15F, the correlation between VEGFA transcript and VEGF-A secretion for individual cells and FIG. 15G, the expression 0HGFBP6 normalized transcripts versus the log transformed VEGF-A secretion values in the replicate normoxic MSC experiment (from FIGS. 7D, 7E). The correlation value and linear regression line are shown in each graph. FIG. 15H shows the UMAP showing VEGFA transcript levels per cell for the replicate normoxic SEC-seq experiment (from FIGS. 7D, 7E).

[0025] FIGS. 16A-16E illustrate the high VEGF-A secretion cluster is not affected by cell cycle regression. FIG. 16A shows the UMAP showing the clusters in the normoxic SEC-seq replicate experiment from FIGS. 7A-7P (replicated from FIG. 7E) for easy comparison with the cell cycle-regressed data below. FIG. 16B shows VEGF-A secretion per cell for the normoxic SEC-seq replicate experiment from FIGS. 7A-7P (replicated from FIG. 7D) is shown for comparison with the cell cycle-regressed data below. FIG. 16C shows New UMAP coordinates and clustering of normoxic cells from FIG. 16A post cell cycle regression, with the original cluster information marked. Note that Cluster C5’s spatial separation from other clusters is preserved with low mixing. FIG. 16D shows VEGF-A secretion shown on the new UMAP coordinates post cell cycle regression. The cells in the newly arranged cluster C5 remain highly enriched for high VEGF-A secretion. FIG. 16E shows new' clustering of cells post cell cycle regression displayed on the UMAP from (FIG. 16C). While the borders between other clusters have shifted, the majority of cells that made up cluster C5 still distinctly form their own cluster (new cluster #6), demonstrating that the highly secretion cluster's special transcriptional profile is unaffected by cell cycle information.

[0026] FIGS. 17A-17D illustrate the purification and characterization of the IL13RA2+ MSC population. FIG. 17A is a dot plot showing the percent of cells expressing the indicating surface marker gene as well as average normalized transcripts, for each cluster from the SEC-seq experiment in FIGS. 7D, 7E. All surface markers shown are contained in the VRS gene list. FIG. 17B is a violin plot showing the normalized transcripts of IL13RA2 for cells per cluster as labeled in FIG. 7E. The dashed (vertical) line represents the mean across all cells for each plot. FIG. 17C illustrates FACS gating used for the isolation of IL13RA2+/- MSCs from three replicate experiments for downstream bulk RNA-seq. FIG. 17D shows how⁷ normoxic MSCs were sorted for IL13RA2+/- subpopulations as indicated in the FACS histogram on the top. Sorted cells were expanded for 4 (IL13RA2-) or 7 (IL13RA2+) days, to account for growth differences, and subsequently stained for IL13RA2 again and analyzed by flow cytometry (bottom).

[0027] FIGS. 18A-18C illustrate VEGFA splicing from bulk RNAseq of IL13RA2+/- sorted cells. FIG. 18A shows the IGV genome view of the RNA-seq reads across the VEGFA transcript from the triplicate bulk-RNA-seq experiments of IL13RA-positive and -negative cell populations described in (FIG. 18B). The exons of the EEGFA transcript are annotated below in FIG. 18B, alternative splicing output from computational evaluation of IL13RA2+ vs IL13RA2- RNAseq libraries, grading the splicing exclusion events using only reads that span exon-exon junctions for maximum accuracy. The same exon is shown multiple times if its size varied, or it had different donor/acceptor exon pairs. Given is exon location, exclusion p-value, exclusion false discovery rate, inclusion rates and inclusion difference between samples. FIG. 18C (top) shows the differential exon expression plot comparing the expression difference in VEGFA exon levels between the three IL13RA2+ and the three IL13RA2- samples, displayed as the loglO fold change (logFC) between the samples, where positive values indicate higher inclusion in IL13RA2+ samples. The exons are labeled. The bottom of FIG. 18C shows same as on the top, except for GATAD2A as an example of a gene with significant alternative splicing. Dots indicate the exons of this transcript.

[0028] FIG. 19 illustrates one example of a detailed SEC-seq protocol or workflow. Note that this protocol or workflow may apply to different cell types.

Detailed Description of Illustrated Embodiments

[0029] As seen in FIGS. 1 and 4A, the methods described herein employ three- dimensional shaped particles 10 (also referred to as nanovials 10) that are typically micrometer sized. The particles 10 generally have a longest dimensional length of between around 10 pm to 100 pm. In embodiments in which flow cytometers or fluorescence activated cell sorters are used to analyze or sort particles 10, the particles 10 are preferably between ~30 pm and ~60 pm in a maximum dimension (here particles 10 with an average outer diameter of 35 pm were predominantly formed for experiments described herein). The particles 10 may be formed from biocompatible materials or polymers. In one embodiment, the particles 10 are formed from a hydrogel material. For example, the particles 10 may be formed from polyethylene glycol (PEG) with a gelatin coating as seen in FIG. 1.

[0030] The particle 10 is shaped and includes a void or cavity 12 as seen in FIG. 1. The void or cavity 12 may have a single opening 14 that opens to the external environment of the shaped particle 10. The opening 14 of the void or cavity 12 is dimensioned to allow cells 100 (and in one preferred embodiment mesenchymal stromal cells or mesenchymal stem cells (MSCs)) to enter the void or cavity 12. Other cell types include, for example, T-cells or plasma cells (e g., B cells). The particles 10 may have a void or cavity 12 sized to fit a single cell 100 in some embodiments. For example, the particles 10 may have a void or cavity 12 with a longest dimension of 10 gm - 30 gm. In a preferred embodiment, the void or cavity 12 is dimensioned to hold a sub-nanoliter volume of fluid. The fluid may include an aqueousbased fluid and may include the cell 100 therein. As explained herein, in some preferred embodiments, the particles 10 preferably are designed to carry or hold cell(s) 100 and in particular mesenchymal stromal cells or mesenchymal stem cells (MSCs) within the void or caviW 12. The cell(s) 100 may be located within the volume of fluid located within the void or cavity 12 and preferably the cells 100 may adhere or become adherent to an inner surface of the shaped particle 10 within the void or cavity 12 by adhering to cell binding moiety (e.g., extracellular matrix proteins, cell adhesion peptide such as integrin or other peptide with cell binding functionality, or an antibody against a cell surface marker, biotin and/or streptavidin) disposed on a surface of the shaped particle within the void or cavity. As seen in FIG. 1, the inner surface of the void or cavity 12 includes a layer of gelatin 16.

[0031] The particles or nanovials 10 also include one or more secretion capture agents 18 located on the particle within the void or cavity 12. The one or more secretion capture agents may include, by way of example, capture moieties including antibodies (e.g.. capture antibody), antigens, aptamers, biotin, or biotin/streptavidin complexes. The particles 10 loaded with the cells 100 may, in some embodiments, be contained in an oil phase during certain operations or processes of the workflow (see FIGS. 4A, 5E, 19). For example, in one embodiment, the particles 10 loaded with cells 100 is contained in an aqueous phase droplet 20 surrounded by an immiscible oil phase during single cell transcriptome assay (e.g., FIG. 4A). As seen in FIG. 4A, the aqueous phase droplet 20 contains one or more reagents of a single cell transcriptome assay. For example, the droplet 20 may contain the 10X gel bead and 10X primers that are part of the 10X Chromium system (10X Genomics, Pleasanton.

CA). It should be appreciated that other single-cell transcriptome assays may be used with the particles 10. Examples include Parse Evercode, Seal eBio, Honeycomb HIVE, BD Rhapsody, or similar commercial assays, or other custom Drop-seq or related protocols. Regardless of the type of single cell transcriptome assay, the cell 100 loaded in the particle or nanovial 10 (which is typically lysed) is exposed to one or more reagents of a single cell transcriptome assay.

[0032] As explained herein a protocol is disclosed that enables linking the Secretion amount of a target protein EnCoded (SEC) by oligonucleotide barcodes w ith transcriptional sequencing (seq). for thousands of single cells 100. SEC-seq leverages the microscale particles or nanovials 10 to isolate cells 100 and capture their secretions in close proximity, oligonucleotide-labeled antibodies to tag secretions on the particles or nanovials 10, and flow cytometry and single-cell RNA-sequencing platforms for readout. Particle modification and cell loading takes, in some embodiments, under 4 hours, and once the desired incubation time is over, staining, cell sorting, and emulsion generation for scRNA-seq can also be completed in under 4 hours. Compared to related techniques which link secretions to a cell’s surface, SEC-seq provides a general solution across any secretion target due to the ease with which biotinylated particles or nanovials 10 can be modified. By linking gene expression and secretory strength, SEC-seq can expand the understanding of cell secretion, how it is regulated, and how it can be engineered to make better therapies.

[0033] FIG. 2 illustrates the exemplary workflow which is used to simultaneously measure the secretion levels of a bioactive factor of interest as well as the transcriptome of the cell 100 contained in the particle or nano vial 10 so that a cell surface marker of interest can be identified that can be used as a proxy to identify and enrich a population of cells that secrete the bioactive factor at a high level. In operation 200, single cells 100 from a larger population of cells 100 are loaded into the void or cavity 12 of the particle or nanovial 10 along with a capture agent for the bioactive factor of interest. To this end, the shaped particle 10 and loaded cells 100 or the droplets 20 containing the shaped particle 10 and loaded cells 100 may contain oligo-barcoded labeling reagents such as detection antibodies that bind to or associate with the bioactive factor. This allows for the identification of cells 100 that have elevated levels of the bioactive factor and linking of the transcriptome of these cells 100 to the level of the bioactive factor through the number of oligo-barcode reads. The labeling of cells with elevated secretion levels of the bioactive factor is seen in operation 210 of FIG. 2 wherein oligo-barcoded labeling reagents are used identify' cells 100 that have elevated levels of the bioactive factor of interest. With reference to operation 220, the single cells 100 contained in the particles or nanovials 100 are then exposed to one or more reagents of a single cell transcriptome assay.

[0034] Cells 100 with elevated secretion levels of the bioactive factor of interest (e.g., VEGF-A) are identified by labeling (operation 210) based on increased amounts of oligobarcode reads (using oligo-barcoded detection antibodies or other labeling reagents) and the transcriptome of these cells with elevated secretion levels is determined as seen in operation 230 of FIG. 2. Because the goal is to ultimately find cell surface markers of interest for facile sorting of live cells 100 using fluorescence activated cell sorting (FACS) or magnetic- activated cell sorting (MACS), the transcriptome analysis as seen in operation 230 is used to identify one or more genes of the transcriptome that encode for the one or more cell surface markers associated with elevated secretion levels of the bioactive factor. For example, as explained herein, the surface marker may include interleukin- 13 receptor subunit alpha-2 (IL- 13RA2) associated with elevated secretion levels of the bioactive factor, VEGF-A.

[0035] Once the surface marker of interest has been determined, labelled antibodies or other affinity agent (e.g.. anti- antibodies) to this surface marker of interest can then be used to enrich a sub-population of cells 100 (e.g.. mesenchymal stromal cells or mesenchymal stem cells (MSCs)). For example, a population of MSCs can be exposed to anti-interleukin-3 receptor subunit alpha-2 (IL-13RA2) antibody and this (exposed) population of MSCs can then be sorted with a cell sorter to generate the enriched sub-population of MSCs with elevated secretion levels of the particular bioactive factor of interest (e.g., VEGF-A) and MSCs with elevated levels of gene expression including a vascular regenerative signal (VRS). These cells 100 (which are alive) can then be expanded into a larger population of enriched cells 100. In one embodiment, a therapeutic composition is provided of these enriched cells 100 having elevated secretion levels of the bioactive factor of interest which can then be administered to a subject (e.g., implantation, implantation as part of a formulation with a biomaterial, infusion or injection).

[0036] FIG. 3 illustrates an illustrative workflow using the platform to generate a population of enriched sub-population of cells 100 which are used as a therapeutic agent. In operation 300, a population of cells 100 contained within particles 10 is labelled with an antibody or affinity agent specific to a surface marker associated with a biomarker of interest. The particles 10 and cells 100 contained therein are then sorted using a cell sorter (e.g., FACS or MACS device) to generate an enriched sub-population of cells with elevated secretion levels of the bioactive factor of interest as seen in operation 310. The cells 100 are liberated or removed from the particles 10 and the sub-population is expanded (enriched) to generate the therapeutic agent as seen in operation 320. The therapeutic agent, which in this embodiment is an enriched sub-population of cells 100 having elevated secretion levels, is then administered to the subject as seen in operation 330. This administration may include an implantation, injection, or infusion of the enriched sub-population of cells 100. [0037] Experimental

[0038] Results

[0039] Establishing the SEC-seq Workflow

[0040] To develop the SEC-seq assay, it was ascertained that (1) VEGF-A secretion from single MSCs can be captured on particles or nanovials 10, and cells 100 disposed on the nanovials 10 maintain viability after sorting; (2) cell-loaded nanovials 10 can be emulsified in a commercial microfluidic device; and (3) mRNAs can be captured and reverse transcribed from individual cells 100 in the presence of nanovials 10 and oligo-barcodes bound to nanovials 10 can be quantified (FIG. 5 A). An aim was to achieve these steps using standard lab equipment including a fluorescent activated cell sorter (FACS), a 10X Chromium controller for single-cell processing and library preparation, and an Illumina next-generation sequencer.

[0041] First a flow cytometry-compatible, single-cell VEGF-A secretion assay was developed using nanovials 10 (see methods and FIG. 4B for nanovial fabrication). Initially, it was confirmed that VEGF-A attached to nanovials 10 can be quantified using purified VEGF-A protein and a standard flow assay. Specifically, increasing amounts of recombinant VEGF-A were bound to nanovials 10 via VEGF-A capture antibodies and measured VEGF-A retained in nanovials 10 by flow cytometry upon incubation with fluorescently (AF647)- tagged anti-VEGF-A detection antibodies. It was found that VEGF-A can be detected across a dynamic range of at least two orders of magnitude (FIG. 5B).

[0042] Next, it was investigated whether VEGF-A secretion from single MSCs can be detected within nanovials 10. MSCs were loaded into 35 pm gelatin-coated nanovials 10 with a 20 pm cavity by simple pipet-mixing, which allows the cells to adhere to the gelatin coating via integrin binding (FIG. 9A). Up to 23% of nanovials 10 contained single MSCs. which could be reliably sorted by FACS (FIGS. 9B-9D). By using confocal microscopy to image the single-cell loaded nanovials 10, it was observed that cells 100 are centered in the nanovial cavity' 12 (Fig lb, bottom right). To measure VEGF-A secretion from single cells, MSCs were loaded into nanovials 10 conjugated with the VEGF-A capture antibody, and incubated the cells 100 on nanovials 10 to allow for VEGF-A secretion and binding to the capture antibody. Upon addition of the fluorescently (AF647)-labelled anti-VEGF-A detection antibody, FACS showed that nanovial fluorescence increased with the amount of time MSCs were allowed to secrete, from 2 to 14 hours (FIGS. 5C, 9E). An extended incubation length showed that recombinant VEGF-A captured in nanovials 12 remains bound over 1-2 days. potentially due to a large avidity effect of many clustered capture antibodies on the nanovial surface, which indicated that loss of protein would not affect results over the time frames of interest (FIG. 9F). A 14-hour incubation period was chosen for VEGF-A secretion accumulation for further experiments as it yielded a majority of the secretion signal in the observed dynamic range, with 90% of nano vials 10 displaying signal above the detection threshold based on control nanovials 10 with no VEGF-A capture antibody, indicating that most MSCs are secreting VEGF-A (FIG. 5C, dashed line, bottom right). The data from the nanovial secretion assay matched results obtained with an VEGF-A ELISpot assay, where 99% of seeded MSCs formed spots. Significant differences in the integrated intensity per spot were observed indicating heterogeneity in VEGF-A secretion levels for individual cells 100 (FIG. 9H), consistent with the broad distribution in the nanovial assay (FIG. 5C). Cell-loaded nanovials 10 had a 2.4-fold increase in mean fluorescent signal compared to empty nanovials 10, showing that secreted VEGF-A remained localized to nanovials 10 containing the secreting cells 100 with little crosstalk (FIG. 9E, plot iv). Note that these differences are observed even in the presence of a low baseline level of autofluorescence of nanovials 10 and non-specific binding of the detection antibody and/or VEGF-A protein (FIG. 9G). Together, these experiments demonstrate the ability to load MSC cells 100 into nanovials 10, robustly capture VEGF-A secretion from individual MSCs in nanovials 10, and isolate MSC-loaded nanovials 10 by FACS.

[0043] Viability is a key requirement for high quality scRNA-seq data, as it avoids cell composition bias and increases the number of cells 100 that pass quality filtering during postprocessing. Since FACS or other sorting is required to enrich for nanovials 10 containing single cells 100, it was explored if cell viability is preserved on nanovials 10 following FACS sorting. Suspended MSCs or nanovials 10 containing single MSCs were sorted and it was found that cells 100 in both conditions had a high viability post-sort (FIG. 5D). Additionally, nanovials 10 protected MSCs during sorting when exposed to surfactant, suggesting that nanovials 10 can shield cells 100 from external stressors (FIG. 10A). Consistent with this, modeling of fluid shear stresses in the flow cytometer nozzle yielded a 400-fold higher value for fluid dynamic shear stress acting on suspended cells 100 compared to cells 100 loaded in ananovial 10 (FIGS. 10B-10D), suggesting that nanovials 10 can shield cells from fluid shear stress.

[0044] After establishing the nanovial-based VEGF-A secretion assay and confirming cell viability on nanovials 10, it was explored whether cell-loaded nanovials 10 are compatible with the emulsion generation and cell lysis required for scRNA-seq. The 10X Chromium system was used as it is a commercially available scRNA-seq solution with microfluidics chip dimensions and resulting droplets that are compatible with the size of nanovials 10. 35 pm (diameter) nanovials 10 could be successfully loaded into microfluidic droplets with barcoded primer beads (FIG. 5E). Based on image analysis of the emulsion droplets, it was found that the majority of droplets with nanovials 10 contain one nanovial 10 across three repeats, and that the inclusion of multiple nanovials 10 is similar to the expected multiplet rate reported for cell loading by the manufacturer (9.7% vs 4%, FIG. 5F). Fluorescence microscopy of emulsions formed with nanovials 10 containing single calcein-stained MSCs in the presence of cell lysis buffer showed localized release of the dye, pervading only nanovial-containing droplets (FIG. 5G). This result indicated successful cell lysis post droplet formation and encapsulation. Based on these findings, it was concluded that cell-loaded nanovials 10 are compatible with the 10X Chromium-based emulsification workflow.

[0045] Next, it was explored whether scRNA-seq libraries could be successfully retrieved. Initially, a species mixing experiment was conducted, in which human MSCs and mouse fibroblasts were loaded separately into nanovials 10, sorted for single cells 100, and combined at a 1 : 1 ratio before loading into the 10X chip. This time, scRNA-seq I i bran construction and sequencing was pursued where transcripts from 6296 cells were retrieved. The fraction of human and mouse cells reflected the initial pooling ratio with 44.8% human cells. 47% mouse cells, and 8.2% mixed cells (2821 human cells. 2950 mouse cells, and 515 mixed cells; FIG. 5H). These results showed that it is possible to capture transcript data from individual cells 100 loaded into nanovials 10 without cell-to-cell mixing.

[0046] Next, it was studied how loading MSCs in nanovials 10 affects transcript recovery in comparison to scRNA-seq for free (suspended) MSCs. scRNA-seq was performed for suspended MSCs (unsorted), suspended MSCs sorted for positive calcein signal (sorted), and MSCs loaded in nanovials 10 sorted to enrich single cells 100 (gating strategy as seen in FIGS. 9B-9D). In this experiment, the nanovials 10 were not functionalized with the VEGF- A capture antibody. Although fewer cells 100 were detected in the nanovial sample (unsorted = 3559, sorted = 3297. nanovial = 2391; same sequencing depth per sample), transcript detection per cell was not adversely affected (FIG. 51). A similar trend was observed for genes detected per cell (FIG. 11 A). Both nanovial-loaded and suspended MSCs expressed standard MSC-specific surface markers (FIG. 1 IB); yet cells 100 on nanovials 10 clustered separately from suspended cells (FIG. 11C). Gene ontology analysis showed that genes significantly upregulated in nanovial -loaded cells 100 are related to cell division and DNA replication, which may reflect a healthier cell state for MSCs adhered to nanovials 10 compared to recently dissociated MSCs (FIGS. 1 ID, 1 IE). Together, these data suggest that adhered cells 100 maintain anchorage-dependent processes like cell division and show that nanovial loading does not adversely affect transcript capture or gene expression states. [0047] The last step in establishing the complete SEC-seq approach was to ensure accurate detection of oligo-barcodes within nanovials 10 together with the transcriptome of single cells 100. To validate this step, two populations of MSCs loaded on nanovials 10 were prepared, and functionalized each population of nanovials 10 with a specific oligo-barcoded streptavidin (FIG. 5 J). MSCs in these two nanovial samples differed as they were cultured in either standard normoxic conditions or treated with the hypoxia-mimetic agent deferoxamine (DFX), which is known to induce a hypoxic gene expression signature and to increase the secretion of angiogenic growth factors, including VEGF-A (FIG. 12A). Before progressing to scRNA-seq, the hypoxia-driven increase of VEGF-A secretion in hypoxic conditions was confirmed with flow cytometry for MSCs in nanovials 10 (FIG. 12B). It was expected that the hypoxia-induced gene signature would distinguish the two batches of MSCs and allow us to assess the accuracy of the oligo-barcode assignment in the pooling experiment. It was found that 85% of cells 100 could be assigned to one of the two oligo-barcodes and the tw o sets of cells 100 defined by the nanovial barcode separated when plotted together in a Uniform Manifold Approximation (UMAP) plot (1249 hypoxic cells and 865 normoxic cells; FIGS. 5J, 5K). The barcode-based classification matched the expected gene expression profile, as over 90% of cells 100 with high expression of a hypoxic gene signature (which includes VEGFA) were assigned to the respective (hypoxic) barcode (FIG. 5L). A similar strategy was used to infer how many cells 100 are released from barcoded nanovials 10 between sorting and emulsion formation and found that 93% of cells were linked to high levels of an oligo-barcode (i.e., associated with the nanovial 10), indicating low cell loss from nanovials 10 in this process (FIG. 1 IF). These results confirmed the ability to detect oligo- barcodes on nanovials 10 and link barcode information to single cell transcriptomes.

[0048] SEC-seq: Measuring VEGF-A Secretion from MSCs Reveals Uncoupled Secretion and Transcription

[0049] With all steps in the SEC-seq workflow validated (FIGS. 5A-5L), the SEC-seq approach was employed for simultaneously measuring secretion of VEGF-A and the global transcriptome for thousands of individual MSCs (FIG. 6A). To detect secreted VEGF-A through a sequencing readout, a custom VEGF-A detection antibody conjugated to a 1 OX Genomics compatible oligo-barcode was designed (FIG. 13 A). Capture primers from the 10X gel beads add unique molecular identifiers (UMIs) and cell barcodes to this antibody-derived barcode sequence upon capture and amplification. The oligo-barcoded antibody specifically detects VEGF-A on nanovials 10 as shown with a flow cytometry -based readout of nanovials 10 loaded with and without recombinant VEGF-A (FIGS. 13B-13C).

[0050] Upon validation of the oligo-barcoded VEGF-A antibody, the complete SEC-seq pipeline was executed, where single MSCs were loaded for 2 hours in nanovials 10 coated with VEGF-A capture antibodies and incubated for an additional 12 hours (14 hours total) to collect secreted VEGF-A. Nanovials 10 were then labeled with the oligo-barcoded anti- VEGF-A detection antibody, sorted to isolate the nanovials 10 containing single viable MSC cells 100, and processed for scRNA-seq and oligo-barcode library preparation using the 10X workflow. This experiment was performed in parallel for MSCs in normoxic and hypoxic culture conditions and the UMI counting of oligo-barcodes per cell was used as a measure for VEGF-A secretion in each condition.

[0051] MSCs cultured in normoxic and hypoxic conditions separated in transcriptional space as depicted upon dimensionality⁷ reduction in a UMAP plot (2943 hypoxic cells and 2566 normoxic cells; FIG. 6B), consistent with the standard scRNA-seq experiment in FIG. 5 J. As expected, cells with higher VEGF-A secretion and VEGFA transcripts were more abundant in the hypoxic culture condition (FIGS. 6C, 6D). On average, hypoxic cells secreted 1 .72 times more VEGF-A than normoxic cells (FIG. 6E, middle). The VEGF-A secretion measurement based on SEC-seq was confirmed when the VEGF-A secretion level was determined with a fluorescently labeled anti-VEGF-A detection antibody measured by flow cytometry (1.66 times higher on average in hypoxic conditions; FIG. 6E, top). This comparison revealed that the relative magnitude and sensitivity of SEC-seq and a fluorescence-based detection assay are similar.

[0052] Inspecting VEGF-A secretion and transcript levels in normoxic and hy poxic cells 100, two key observations can be made. First, VEGF-A secretion is highly heterogenous across MSCs cultured in normoxic and hypoxic conditions (FIG. 6C). Although the oligo- barcoded anti-VEGF-A detection antibody was detected in nearly all normoxic and hypoxic cells 100, some cells 100 in each culture conditions displayed higher VEGF-A secretion (FIG. 6C). Second, the heterogeneity⁷ in VEGF-A secretion was not matched by similar changes in VEGFA transcript levels in each culture condition (compare FIG. 6C and 6D), indicating that transcript levels are less dynamic within each culture condition than secretion levels. VEGFA transcripts were detected in 88% of cells 100, and 71% of cells 100 contained at least 0.5 normalized VEGFA transcripts, indicating that the majority' of MSCs had some level of both VEGFA transcript and secretion. VEGFA transcript levels were dramatically higher in MSCs cultured in hy poxic conditions compared to normoxic cells (FIG. 6D), with hypoxic MSCs, on average, having 13-fold higher VEGFA transcript levels than normoxic cells (FIG. 6E. bottom). Interestingly, VEGFA transcripts have a much greater dynamic range in hypoxic compared to normoxic conditions, in contrast to the milder VEGF-A secretion distributions (FIG. 6E, bottom). Thus, the increase in VEGF-A secretion between the two culture conditions does not match the magnitude of the change in transcript levels. These results suggested a surprising uncoupling between VEGF-A secretion and VEGFA transcript levels and an unexpected heterogeneity in VEGF-A secretion.

[0053] To take a closer look at the relationship between VEGF-A secretion and VEGFA transcript levels, the cells 100 were divided into clusters based on transcriptome (FIG. 6F). Clusters N1-N5 are predominantly populated by cells 100 from the normoxic condition, clusters H1-H4 by cells 100 from the hypoxic conditions, and clusters M1-M3 were formed by cells 100 from both culture conditions (FIG. 14A). The hypoxic cell clusters had higher VEGFA transcript and protein secretion levels than the normoxic cell clusters (FIG. 6G). However, the cluster with highest VEGF-A secretion contained cells 100 from the normoxic and hypoxic conditions (cluster Ml). Compared to all other clusters, this cluster had the highest minimum and median secretion levels. Interestingly, despite higher VEGF-A secretion in cluster Ml, VEGFA transcript levels in this cluster were similar to other clusters (FIG. 6G).

[0054] These data indicate that i) there is a subpopulation of normoxic cells (cluster Ml) with increased VEGF-A secretion in the absence of higher VEGFA transcript levels; ii) the hypoxia-induced global increase in VEGFA transcript level is associated with a broad increase in VEGF-A secretion; and iii) a subset of cells (cluster Ml) is secreting the highest level of VEGF-A regardless of culture condition. Together, these findings confirm the disconnect between VEGF-A secretion, measured over the 14 hours incubation period, and transcript levels assessed at the endpoint, which is most apparent in normoxic cells, and uncover a unique cell state associated with high VEGF-A secretion in both hypoxic and normoxic conditions. [0055] To further examine the relationship between transcript levels and protein secretion, the correlation between VEGFA transcripts and VEGF-A secretion was calculated for normoxic and hypoxic cells 100. It was found that these features were uncorrelated in normoxic cells 100 (r² = -0.01) and weakly correlated in hypoxic cells 100 (r² = 0.15; FIG. 6H). A repeat SEC-seq experiment for normoxic cells 100 corroborated this lack of association between VEGFA transcripts and VEGF-A secretion in normoxic cells 100 (see below, r²= 0.05). When ranking all genes according to the correlation of their transcript levels to VEGF-A secretion, the VEGFA transcript ranked lowly in the normoxic cells 100 (rank 13,323; FIG. 61). This analysis confirmed that there is no strong relationship between the basal levels of VEGFA transcripts and VEGF-A secretion in normoxic MSCs. In contrast, VEGFA was within the top 2% of correlated genes in hypoxic cells 100 (rank 215, FIG. 61), and when calculating transcript-to-secretion correlation across normoxic and hypoxic cells, a higher correlation between VEGFA transcripts and protein secretion was found (r² = 0.24, FIG. 14B). Thus, the induction of the hypoxia response triggers high levels of steady state VEGFA transcript which yields a modest correlation between transcript levels and secretion, either through a direct or secondary effect. However, overall, the VEGFA transcript level at the end of the secretion capture period has surprisingly minimal concordance with the VEGF- A secretion amount detected at a single cell level, and this effect is independent of transcript depth (FIG. 14C).

[0056] The MSC Subpopulation Secreting the Highest Levels of VEGF-A is Defined by a ‘Vascular Regenerative Signal’

[0057] Given that VEGFA transcript levels were not predictive at a single cell level of the highest secreting cells within the sample, in either normoxic or hypoxic culture conditions (Fig 3i), transcripts that are better predictors of VEGF-A secretion levels were pursued. An examination of the top-correlating genes (FIG. 61) demonstrated that they overlapped between normoxic and hypoxic cells 100 (FIG. 7A). These genes included insulin growth factor binding protein 5 (IGFBP5) and 6 (IGFBP6), tissue inhibitor of metalloproteinase 3 (TIMP3), heme oxygenase-1 (HMOXlj. platelet derived growth factor receptor beta (PDGFRB), and signal peptide-CUB-EGF-like-domain-containing protein 3 (SCUBE3) (FIG. 7B). IGFBP5 is a secreted signaling protein that can regulate cell growth and migration and IGFBP6 was previously linked to VEGF-A secretion in a growth factor array assay for an MSC product (Stempeucel), in which VEGF-A and IGFBP6 were the top two secreted proteins in conditioned media. As the transcript levels of these genes were all more highly correlated with VEGF-A secretion than VEGFA transcripts in both normoxic and hypoxic cells, it was reasoned that they could be used to identify the most highly VEGF-A secreting cells (FIGS. 6F, 6G3). Therefore, their expression distribution in normoxic and hypoxic cells 100 in the SEC-seq experiment described in FIGS. 6A-6I and it was found that cells in cluster Ml displayed the highest transcript levels for these genes (FIG. 7C, FIG. 15A).

[0058] This finding was validated by performing and analyzing data from an additional SEC-seq experiment for normoxic MSCs (2254 normoxic cells; FIG. 7D). As observed previously, a subpopulation of normoxic cells 100 marked by high VEGF-A secretion was found as well as high levels of IGFBP5, IGFBP6, TIMP3, and the rest of the top correlated genes (FIG. 7D, FIG. 15B). Clustering of the transcriptome data from this experiment revealed that the subpopulation with the highest VEGF-A secretion can be assigned to one cluster (cluster C5; FIG. 7E, 15C). As observed before for cluster Ml (FIGS. 6F, 6G), cluster C5 was not characterized by an increase in VEGFA transcript levels (FIG. 15C). Similar to the experiment described in FIGS. 6A-6I, IGFBP6 was the top ranked gene when correlating transcript levels of all genes to VEGF-A secretion (r²=0.48) and VEGFA transcripts lacked correlation to secretion (r²=0.05; FIGS. 15E-15H). Furthermore, a gene expression signature formed by the previously identified 10 top-correlating genes for VEGF-A secretion was most expressed in cluster C5 (FIGS. 15D. 15E), similar to the Ml cluster from FIGS. 6A-6I. Thus, the steady-state transcript levels defined a set of genes reproducibly correlated with the highest VEGF-A secretion in MSCs. Taken together, these findings suggest that a subpopulation of MSC cells 100 with a defined transcriptional state, identified by a specific expression cluster in each of the SEC-seq experiments, is reliably linked to highest VEGF-A secretion.

[0059] To further define this MSC subpopulation, differentially expressed genes for each cluster for the repeat SEC-seq experiment were determined (FIGS. 7D, 7E) and characterized them by gene ontology (GO) analyses (FIGS. 7F, 7G). Genes defining the highly VEGF-A secreting cluster C5 were associated with blood vessel morphogenesis and development, vasculature development, and extracellular matrix organization, indicative of an angiogenesis-related MSC activity and consistent with its high VEGF-A secretion state. Surprisingly, similar gene ontology terms to C5 were also found for the gene signatures defining clusters C6-C10, even though each had distinct differentially expressed genes (FIG. 7G). Due to the overlapping ontology enrichments between these clusters and lack of VEGFA transcript correlation, a traditional scRNA-seq analysis would have overlooked cluster C5’s identity as a super VEGF-A secretory subpopulation. The clusters with lowest VEGF-A secretion (C2, C3, and C4) were strongly enriched for mitotic signatures and likely represent actively dividing MSCs, and clusters Cl and C7 are associated with distinct but unremarkable cellular functions. Of note, the highly VEGF-A secreting cluster C5 could also be easily identified even after cell cycle regression (FIGS. 16A-16E).

[0060] Since the GO analysis did not uniquely identify cluster C5 despite a unique set of highly expressed genes, efforts were made to create a gene expression signature most representative of the high VEGF-A secreting MSC subpopulation across all the SEC-seq experiments. By overlapping the differentially expressed genes in the high VEGF-A secretion cluster in each of the three SEC-seq experiments (for hypoxic and normoxic MSCs in FIGS. 6A-6I and normoxic MSCs in FIGS. 7D, 7E), a consensus set of 153 genes was determined (FIG. 7H). GO analysis linked this gene set to extracellular matrix organization, cell motility, blood vessel development, and wound healing (FIG. 71). On account of these GO terms and the association with highest VEGF-A secretion, this shared gene set was referred to as the Vascular Regenerative Signal (VRS). The VRS marked a subpopulation of cells 100 composed of 5-20% of each sample (at 75% threshold; see methods) in each of the MSC SEC-seq experiments (FIG. 7J), as well as in the scRNA-seq experiment testing MSCs on oligo-barcoded nanovials 10 described in FIGS. 5J-5L (FIG. 7K). To confirm that the VRS exists independently of nanovials 10 and possible autocrine and/or substrate effects caused by nanovial containment of individual MSCs, VRS-expressing cells in scRNA-seq experiments performed on free MSCs were investigated and it was found that 16% of cells expressed the VRS highly and that those cells 100 were gathered within one cluster (FIG. 7L).

[0061] Interestingly, a large proportion of VRS genes classified as secretory (29%), a much larger fraction compared to the percent of secretory genes marking all other clusters (FIG. 7M), and includes secretory genes implicated in promoting angiogenesis such as tropoelastin (ELN) and CXCL12. Other functional secretory' products include: HM0X1, a cryoprotective protein that maintains iron homeostasis; fibronectin 1, an extracellular matrix glycoprotein involved in tissue repair; TIMP3, an enzyme that stabilizes the extracellular matrix; and SCUBE3, a signaling protein involved in tumor angiogenesis and metastasis. The VRS also includes various cell surface proteins and some transcription factor (TF)-encoding genes know n for their role in mesenchymal fate control such as KL1- 6. PRRX2 and RBPJ (FIG. 7M). Using TRRUST database that infers regulator ' TFs for target genes based on comentions in publications (TRRUST v2: an expanded reference database of human and mouse transcriptional regulatory interactions. Nucleic Acids Research 26 Oct, 2017, which is incorporated herein by reference; https://www.gmpedia.org/trrust/), a link was found between NFKB1, RELA, TFAP2A, ERG, and the hypoxia regulator HIF1 A and VRS genes (FIG. 7N). As expected, VRS genes correlate highly to VEGF-A secretion, with 85% of them are within the top 500 correlating genes, as determined across all SEC-seq experiments (FIG. 70).

[0062] Taken together, the VRS is a unique signature identifying super secret ors of VEGF-A, the discovery of which was only made possible by combining the transcriptomic and secretory' data of individual cells 100 as achieved with the SEC-seq method. The SEC- seq data demonstrate that there are multimodal transcriptional states that control VEGF-A secretion from MSC cells 100: broadly induced cell states triggered by hypoxia as well as specialized subpopulations such as those marked by the VRS signal in both hypoxic and normoxic conditions (FIG. 7P).

[0063] Enrichment of the ‘Vascular Regenerative Signal’ Subpopulation

[0064] A subpopulation with a highly secretory phenotype may prove useful for therapeutic purposes and warranted further characterization. Having identified the VRS gene signature that distinguishes highly VEGF-A secreting cells, it was hypothesized that the gene expression information may be exploited to identify surface marker proteins that can be used to isolate this sub-population of MSC cells 100 by FACS. Out of the 153 VRS genes, 19 encoded surface proteins (FIG. 17A). Antibodies were obtained for the surface proteins translated by the most highly correlated genes to VEGF-A secretion and stained free MSCs for readout using flow cytometry'. It was found that an antibody targeting IL13RA2, a protein encoded by the third ranked surface marker gene in the VRS with high transcriptional enrichment (FIG. 17B), specifically marked a subpopulation of cells with similar frequency as the VRS-marked cells (FIGS. 8A, 8B; 4-12%). The IL13RA2-positive (+) cells are marked by high levels of CD248, another surface protein contained in the VRS gene list (FIG. 17A). Specifically, IL13RA2+ MSCs have 2.2x higher mean CD248 signal than IL 13RA2 -negative (-) cells (FIG. 17C).

[0065] To explore whether IL13RA2+ cells 100 are enriched for the VRS population. IL13RA2+/- cells were isolated by FACS in three replicates and performed bulk RNA sequencing and functional secretion assays. Virtually all VRS genes displayed higher expression in IL13RA2+ cells 100 compared to IL13RA2- cells 100 (FIG. 8C). On average, IL13RA2+ cells 100 had an approximately 2-fold increase in VRS gene expression across the three replicates; VRS genes with high VEGF-A secretion-to-transcript correlation were also highly enriched in IL13RA2+ cells 100 such as IGFBP5 (40-fold higher across replicates in IL13RA2+ vs IL13RA2-), ELN (11-fold), and NPR3 (15-fold) (FIG. 8D). Notably, VEGFA gene expression was not significantly different between both cell populations (FIG. 8C). In a complementary analysis approach, the genes significantly upregulated in IL13RA2+ cells 100 compared to IL13RA2- cells 100 were looked at if they mark VRS cells 100. Indeed, plotting the expression of these genes showed that they are specifically marking cluster C5 identified previously as the high VEGF-A secretion cluster in FIG. 7D and as highly IL13RA2 expressing (Fig 5e). This indicates that the IL13RA2+ cells 100 capture the gene expression state of the VRS subpopulation of MSCs. As a side note, bulk sequencing also allowed us to analyze the splicing pattern of the VEGFA transcript in IL13RA2+/- cells 100 (FIG. 18A), which revealed that splicing differences were not significant (p.val<0.05, FDR<0.10) between the IL13RA2+/- populations, ruling out alternative splicing as a potential source of the VEGF-A secretion enrichment in the VRS in the SEC-seq data (FIGS. 18B, 18C).

[0066] Once it was confirmed that IL13RA2+ cells 100 enriched for VRS cells 100 based on gene expression, it was questioned whether these cells 100 functionally recapitulate the VRS population, found to be high VEGF-A secreting in SEC-seq experiments. To this end, isolated IL13RA2+/- cells 100 were first isolated by FACS and determined the VEGF-A secretion level of both cell populations. It was found that IL13RA2+ cells produced 30% more spots than the IL13RA2- samples (p<0.05) using VEGF-A ELISpot assay for cells seeded right after sorting (FIG. 8F), indicating that the IL 13RA2+ isolated cells 100 possessed higher VEGF-A secretory' activity'. Finally, it was explored if high VEGF-A secreting IL13RA2+ cells 100 can be stably maintained in culture. Specifically, IL13RA2+/- cells 100 were isolated by FACS and expanded the cells 100 in culture for six days, then retested them for IL13RA2 staining by flow cytometry. It was found that 81.4% of the IL13RA2+ cells 100 retained high IL13RA2 levels above threshold after six days of culture and that expanded IL13RA2- cells 100 remained negative for this marker (FIG. 8H). In a repeat experiment, VEGF-A levels were also measured in conditioned media obtained from IL13RA2+/- expanded cells 100 by ELISA and found that VEGF-A secretion was 1.6-fold higher in the IL13RA2+ population following expansion (FIG. 8G, FIG. 17D). These data suggest that the enriched VRS population maintains higher VEGF-A secretion upon isolation by FACS and can be expanded for downstream assays. The fact that isolated highly VEGF-A secreting MSCs with the VRS gene signature display higher VEGF-A secretion (measured with two different methods - ELISpot and ELISA) corroborates the technical accuracy of the SEC-seq approach to measure secretion.

[0067] Together, these findings show that using SEC-seq one can characterize and identify cell subsets with unique secretory characteristics and that the transcriptomic data can be leveraged to identify' surface markers for separation and to isolate populations that can be expanded for applications where increased VEGF-A secretion is desirable (FIG. 81). Since IL13RA2+ cells 100 naturally occur as a minor subpopulation of MSC cells 100. an enriched formulation of IL13RA2+ cells 100 can be delivered by local injection, through the bloodstream (e.g., intravenous infusion), or incorporated into biomaterial scaffolds or hydrogel and used therapeutically to enhance the dose of VEGF-A secreted in vivo. This may provide therapeutic benefit for regenerating tissue, locally or systemically. Although this approach was applied to MSCs and VEGF-A secretion, the general approach is applicable to other cell types such as neural stem cells 100, regulatory T cells 100, beta cells 100, and other tissue-specific cells 100 as well as other secreted bioactive factors (or combinations of bioactive factors).

[0068] Discussion

[0069] Developing and using SEC-seq allows one to link the secretion function of -10,000 single MSCs with their transcriptomes. Although this work addressed secretion in adherent MSC cells 100, suspension cells 100, such as B cells 100, are also compatible with SEC-seq through antibody capture on nanovials. SEC-seq is easily adopted by anyone who has access to current single-cell sequencing instruments, like the 10X Chromium, and oligonucleotide-labeled antibodies, without the need for specialized microfluidics equipment and skills.

[0070] While simultaneous measurements of intracellular or cell surface proteins and transcriptome can be performed, methods to measure secreted proteins and the transcriptome from the same cell 100 have not been accessible. Cellular indexing of transcriptomes and epitopes by sequencing (CITE-Seq) uses oligonucleotide-barcoded antibodies to detect proteins on the cell surface together with the cell's trans criptome, which has facilitated important discoveries, including markers (CD80 and CD86) specific for activated regulatory T cells and a marker (CD201) for highly functional hematopoietic stem cells. The hope is that SEC-seq opens a similar dimension of inquiry' for secreted proteins. One recent method, TRAPS-seq, reported simultaneous measurement of transcriptome and secretions of leukocytes through incorporating affinity matrices on the cell's surface. This method should prove similarly powerful in appropriate systems but has many limitations, including the need for a strongly expressed cell surface receptor and an abundant quickly secreted target, the need for a complex microwell array to prevent crosstalk, and a lack of ECM substrate for adherent cells. The use of the dynamic and variable cell membrane for secretion capture limits the potential applications, scalability', and accuracy of measurement, while SEC-seq allows for more variety’ in secretion target, cell type, and assay time by nature of the modifiable nanovial. Extending the boundaries of multiomics, SEC-seq democratizes the ability to make discoveries by linking cell secretion to gene expression (and surface markers) for thousands of cells in parallel using nanovial technology.

[0071] The data highlights the significant heterogeneity in secretion of VEGF-A in MSC cells 100 grown under both normoxic and hypoxic conditions, and uncovers new transcriptionally-defined sub-populations of cells 100 (FIG. 7P), which may have improved therapeutic potential, overcoming previous translational failures. Only by using SEC-seq this sub-population of cells 100 was uncovered and a new vascular regenerative signal (VRS) gene netyvork was defined. The VRS contains a large fraction of secretory proteins, suggesting that SEC-seq will be valuable to further explore additional secreted products of VRS cells 100. Other works have also found functional sub-populations in MSCs, including studies where subpopulations/cl usters share multiple differentially expressed genes with the VRS gene signature, however, this previous work did not elucidate the connection with higher levels of VEGF-A secretion yvhich would result in a therapeutically useful subpopulation. In the future, SEC-seq yydll enable the exploration of the presence of VRS signature in primary MSCs from various tissue sources and donors.

[0072] As the primary' mechanism of action for MSC therapies is thought to be through the secretion of bioactive factors that promote immunomodulation and regeneration of resident cells 100, such as the groyvth factor VEGF-A, the VRS could also provide a foundation for developing critical quality attributes (CQAs) for MSC therapies. This neyv functional understanding of the link betw een the VRS and VEGF-A secretion could be exploited to enhance the frequency’ and efficacy of this potentially therapeutic population. Notably, the VRS provides genes and path-ways that could be genetically targeted or modified by pre-conditioning treatments and also contains candidate surface marker genes that could be more easily used for isolation or enrichment of high-secreting sub-populations. It has been shown that this is feasible for VRS cells 100 using IL13RA2 as a marker, and that sorted cells 100 maintain their gene expression and secretory characteristics for several population doublings (for at least 6 days of expansion). Thus. VRS cells 100 could be expanded and maintained for use in therapeutic approaches which would benefit from renewable sources of secretory VEGF-A. Streamlining the sorting process with magnetic based affinity methods to cell surface markers like IL13RA2 can be used to dramatically increase the scale of cells sorted and practicality of such an approach.

[0073] Often the transcriptional expression of a secretory protein is equated to secretion, however, the results suggest this assumption may not be correct across different cell types. It was found that VEGFA transcripts and VEGF-A secretion have very low correlation in normoxic and hypoxic MSCs. While surprising, the uncoupling of transcript level and protein secretion is in line with similar studies using CITE-seq where certain surface protein and transcripts feature low correlation. The finding that transcription and protein secretion are not tightly linked opens broad new questions and has important implications for engineering cell products. These questions include (1) how is secretion controlled and how dependent is it on cell state; (2) how are cells 100 facilitating the build-up and release of different amounts of secreted protein outside of gene regulation and (3) why would the uncoupling of secretion and transcript levels be beneficial for ultimate biological function compared to direct control of secretion via the regulation of steady state transcript levels. From a biotechnology perspective, the results suggest that approaches to drive secretion to modulate engineered cell therapy function should not rely alone on introducing copies to increase the transcript level of the target gene, but rather could benefit from a holistic engineering of the pathways involved in driving secretion, e g. modulating the VRS in MSCs.

[0074] There are some limitations in the SEC-seq assay. Due to the nature of the transcription measurement and duration of secretion, it is important to keep in mind that there is an inherent temporal mismatch between the cell’s mRNA detected at the instant of lysis and the gradual secreted protein accumulating over the length of the cell’s time in the nano vial 10. So far, no technique can measure real time secretion in single cells in a high throughput manner. Each secretion requires a unique cognate capture-detection antibody pair, meaning that the number of secreted proteins that can be assayed at one time is limited by the biotin sites on nanovials 10. Here, a single secreted protein was characterized, but the detection of multiple secreted proteins using nanovials 10 has also been demonstrated.

[0075] Using SEC-seq one can now answer critical questions about the secretome, including which cells secrete specific proteins (heterogeneity in protein secretion), whether there is coordinated secretion among proteins, and what mechanisms control secretion. The data obtained from SEC-seq adds another dimension to multi-omics approaches that can be used to inform genetic engineering to modulate secretion levels for bioproduction and cell therapy applications, determine markers to predict disease states, and decipher the biomolecular pathways of protein translation, post-translation modification, trafficking, and release. SEC-seq is opening the door to better cell-based therapeutics, new bioengineering approaches, and to advance the understanding of healthy and disease states throughout development and homeostasis.

[0076] Methods

Table 1 - List of reagents and resources

[0077] MSC Cell Culture

[0078] Immortalized human adipose-derived MSC cells 100 (ATCC SCRC-4000) were cultured in MSC Basal Medium (ATCC PCS-500-030) supplemented with Low Serum MSC Growth Kit for adipose MSCs (ATCC PCS-500-040) and Antibiotic-Antimycotic (A-A; Invitrogen) resulting in final complete MSC media concentrations of 2% Fetal Bovine Serum (FBS), 5 ng/mL FGF-1, 5 ng/mL FGF-2, 5 ng/mL EGF, 2.4 mM L-Alanyl-L-Glutamine and 1% A-A. MSCs were cultured in incubators at 37°C and 5% CO2 and passaged once 70-80% confluent, with MSCs up to passage 25 used in experiments.

[0079] MEF Cell Culture (for species mixing experiment)

[0080] Primary mouse embryonic fibroblasts (MEFs) were cultured in Dulbecco’s Modified Eagle Medium (Gibco) supplemented with 10% FBS and 1 % A-A. MEFs were cultured in incubators at 37°C and 5% CO2 and passaged once 70-80% confluent, with MEFs up to passage 7 used in experiments.

[0081] VEGF-A ELISA on MSC Conditioned Media Across Cell Culture Conditions [0082] To measure VEGF-A secretion from MSCs grown in tissue culture plates, 100,000 MSCs were seeded in 1.33 mL culture media in each w ell of 6-well plates. After 3 days, media was exchanged for normoxic or hypoxic media. Specifically, cell culture media was used for the normoxic condition and media supplemented with deferoxamine (500 pM and 1000 pM) or cobalt chloride (50 pM and 100 pM) for hypoxic conditions. After 24 hours, the conditioned media was collected and filtered, and cell number at the time of media collection was counted. The VEGF-A secretion amount was measured using a VEGF-A Human ELISA kit (Biolegend) by plate reader. The experiment was performed in triplicates for each tested condition. Secretion amount was calculated by first normalizing each sample by cell number, and then averaging the triplicate experiments.

[0083] VEGF-A ELISpot

[0084] ELISpot Multiscreen plates were coated with endotoxin-free VEGF-A capture antibody overnight after ethanol wetting. The next day, plates were blocked with 1% BSA in PBS, and then cells were added to the plate in a total of 200 pl media. After 24 hour incubation, the plate was washed, with a special step of washing with ice-cold Milli-Q water and kept on ice to detach any remaining adherent cells. Then, biotinylated VEGF-A antibody was added as the detection antibody, followed by incubation with streptavidin-HRP. Finally, AEC substrate was added and incubated for 30 minutes at room temperature for spot development. Plates were washed thoroughly and stored in 4°C where they were allowed to dry for at least 4 days before read and analyzed on the ImmunoSpot S5UV analyzer. A negative control well was used with no capture antibody where cells were added, as well as a well where capture antibody was present but only plain media without cells added - both resulted in no spots. The positive control was a well with recombinant VEGF-A, resulting in the entire well appearing brightly colored. For the unsorted MSC ELISpot, 78-5,000 cells per well were tested, and between 156 and 625 cells per well led to a linear relationship between ELISpot counts and cells seeded per w^ell, so these wells were used for the total intensity per spot analysis. For IL13RA2+/- sorted cells ELISpot, 250 cells were seeded per well right after sorting.

[0085] Nanovial fabrication and modifications

[0086] FIG. 4B illustrates a schematic and images during fabrication of the nanovials 10. 35 pm nanovials 10 were fabricated using a three-inlet flow--focusing microfluidic device formed from polydimethylsiloxane (PDMS). PEG pre-polymer, gelatin and oil phases were infused at flow rates of 1.5 pl/min, 1.5 pl/min and 15 pl/min, respectively. The PEG prepolymer phase comprised 27.5% w/v 5 kDa 4-arm PEG acrylate (Advanced BioChemicals) with 4% w/v lithium phenyl-2,4,6-trimethylbenzoylphosphinate (LAP, Sigma) in phosphate buffered saline (PBS, pH 7.2). The gelatin phase comprised 20% w/v cold water fish gelatin (Sigma) in deionized water. The oil phase comprised Novec 7500 (3M) with 0.5% v/v Picosurf (Sphere Fluidics). Oil partitioned the aqueous phases into monodisperse water-in-oil droplets and the PEG and gelatin polymers phase separated into a PEG-rich and gelatin-rich phase. Phase-separated droplets were crosslinked with focused UV light through a DAPI filter set and 10X microscope objective (Nikon, Eclipse Ti-S) near the outlet region of the microfluidic device. Polymerized nano vials 10 were collected and any unreacted phases including oil were removed through a series of washing steps as previously described. Biotinylation of the gelatin-layer formed in the nanovial cavity was conducted by incubating nanovials 10 with 10 mM Sulfo-NHS-Biotin (APExBIO) overnight at room temperature while mixing. Nanovials 10 were then washed in pluronic buffer consisting of 0.05% Pluronic F-127 (Sigma), 1% antibiotic-antimycotic (Thermo Fisher) in PBS and sterilized in 70% ethanol overnight. Sterile nanovials 10 were stored 5x diluted (i.e., 100 pL of concentrated nanovial volume was resuspended in 400 pL pluronic buffer resulting in 6.5 million nanovials/mL) in this pluronic buffer at 4°C.

[0087] Conjugating VEGF-A capture antibodies to nanovials

[0088] 5x diluted nanovial suspension was incubated with an equal volume of 260 pg/mL of streptavidin for 30 minutes at room temperature on a tube rotator. Excess streptavidin was washed out by pelleting nanovials 10 at 200g for 3-5 minutes, removing supernatant and adding 1 rnL of wash buffer (pluronic buffer + 0.5% bovine serum albumin [BSA]) three times. All subsequent wash steps in assay workflows were performed similarly unless otherwise noted. The 5x diluted streptavidin-coated nanovial suspension was then incubated with an equal volume of 71.5 pg/mL of biotinylated anti-human VEGF antibody (Biolegend) for 30 minutes at room temperature on a tube rotator and washed. Nanovials 10 were resuspended at a 5x dilution in w ash buffer or cell culture medium prior to the next workflow step.

[0089] Loading cells in nanovials

[0090] MSC cells 100 were loaded into nanovials 10 by pipet-mixing nanovials 10 and the cell solution in 5 ml round-bottom polypropylene tubes (Coming Falcon), which provide a cell-adhesion resistant material and vent cap for gas exchange during incubation. First, the 5x diluted nanovial suspension was reconstituted in complete MSC media. MSCs were detached from tissue culture flasks using TrypLE (Gibco) and resuspended in media at concentrations of 0.9, 1.5 and 2.2 million per ml. Cells 100 and nanovials 10 are then pipette-mixed 10 times in a 3: 1 volume ratio for final cell to nanovial ratios of 0.4: 1, 0.7:1, and 1 : 1, respectively. A total volume of less than 0.5 ml per tube w as maintained throughout to reduce cell clumping during incubation. The nanovial suspensions were incubated at 37°C and 5% CO2 for 2 hours to allow MSCs to bind to the gelatin coating in nanovial cavities 12. Then, nanovial suspensions were first strained through a 20 pm strainer (CellTrics) to remove unbound cells. Subsequently, recovered nanovials 10 were strained through a 37 pm strainer (STEMCELL) to remove any large cell/nanovial aggregates. The 1: 1 cell to nanovial ratio was used for remaining experiments.

[0091] Confocal Microscopy of Cell-Loaded Nanovials

[0092] Nanovials 10 were labelled with AF647-conjugated streptavidin at the same concentration as streptavidin used for VEGF-A secretion assays (260 pg/mL) before cells 100 were loaded and incubated overnight. Samples were stained with calcein AM before FACS to sort out single-cell loaded nanovials 10. These nanovial samples were transferred to a well chamber slide with #1.5 glass bottom thickness and imaged on a Leica SP8 confocal microscope using a 40x oil immersion objective. The 488 nm laser was used to image the calcein stained cells and the 638 nm laser was used to image the AF647 streptavidin conjugated nanovials. Z stacks with a step size of 0.375 pm were taken. Leica Application Suite X software was used to process the acquired images. Due to the refractive index mismatch of sample media and objective media, the ratio of refractive indices was used to rescale the z-axis for final image and video renders.

[0093] Single-cell VEGF-A secretion on nanovials

[0094] VEGF-A capture antibody-conjugated nanovials 10 were loaded with MSC cells 100 as described above. After straining, nanovials 10 were incubated in 6-well plates for 12 hours, in addition to 2-hour loading, to allow for secretion, with up to 170,000 nanovials 10 per well in 2 mL media (for time sweep secretion experiment, total times of 2-, 8-, and 14- hours incubations were performed). Plates w ere shaken in horizontal cross-movements before placing in the incubator to space out nanovials 10. After incubation, nanovials 10 were collected in wash buffer, centrifuged, and resuspended as 5x diluted nanovial suspension. This suspension was incubated with an equal volume of 71.5 pg/mL AF647 anti-human VEGF antibody (Sigma) diluted in a buffer containing calcein AM for 30 minutes at 37°C with gentle vortexing every 10 minutes before washing excess antibody with one high- dilution wash (>100x nanovial volume). Samples were then analyzed by imaging using fluorescence microscopy and analyzed/sorted by FACS as described below'. For the hypoxia inducer concentration sw eep, MSCs in nanovials 10 were incubated in cell culture media for normoxic condition and media was supplemented with deferoxamine (100, 250, 500, 1000 pM) or cobalt chloride (50 pM and 100 pM) for hypoxic conditions. At least 10,000 single cell-loaded nano vials 10 were analyzed per condition.

[0095] Flow cytometry and sorting of nanovials using SONY SH800S

[0096] Nanovial samples were diluted 25x in washing buffer before sorting on the SONY SH800S cell sorter using a 130 pm sorting chip. The sorter featured violet (405 nm), blue (488 nm), yellow (561 nm) and red (640 nm) lasers and 450/50 nm (FL1), 525/50 nm (FL2), 600/60 nm (FL3) and 665/30 nm (FL4) filters were used. Typical sensor gain settings used for nanovial samples are given in Table 2, along with stains or antibodies measured using the filters. Compensation was performed for samples with spectral overlap (e.g., calcein spillover into FL4 for samples labeled with calcein and AF647 anti-VEGF) using an unstained control (plain nanovials 10) and single-stained controls (calcein-stained cells on nanovials 10; AF647 anti-VEGF-A-labeled recombinant VEGF in empty nanovials 10). The typical gating strategy for nanovials 10 is shown in FIG. 9E. Flow cytometry data was analyzed using FlowJo software version 10.8 (BD).

Table 2. Typical sensor gain settings used with SONY SH800S for nanovial sorts

[0097] Viability measurements for suspended cells and cells on nanovials after sorting

[0098] MSCs stained with CellTracker Orange CMTMR were either prepared in suspension or loaded in 35 pm nanovials 10. This stain was chosen to avoid spectral overlap with later live/dead Incucyte imaging. For MSCs in suspension, FACS buffer (PBS without calcium and magnesium + 1% A-A + 0.5% BSA) or FACS buffer with 0.05% Pluronic was used. For MSCs in nanovials 10. wash buffer (PBS with calcium and magnesium + 1% A-A + 0.5% BSA + 0.05% Pluronic) or wash buffer without 0.05% Pluronic was used.

Immediately after straining the sample as described previously, samples were sorted. The 405 nm, 488 nm, and 5 1 nm lasers were turned on for this experiment. Samples were sorted per well in a 96 black well plate in triplicate, and a similar amount of unsorted sample was prepared. Then calcein AM and propidium iodide were added to the existing media plus sorted sample resulting in a final 1 : 1000 dilution of each stain, and incubated for 30 minutes before imaging on the IncuCyte Live-Cell Analysis System using phase, green and red channels. Fluorescent images were thresholded manually, and the number of live/dead cells were quantified using the IncuCyte S3 software.

[0099] Finite element analysis for shear stress calculations on cells in nanovials versus suspended cells

[00100] Shear stress experienced by cells 100 in nanovials 10 and suspended cells were modeled using a 2D axisymmetric geometry' with COMSOL Multiphysics Single-Phase Laminar Flow physics. The full Navier-Stokes equations are solved by treating either the cell in nanovial or suspended cell as stationary in the center of the channel. This models the fluid flow at the sheath flow junction where the slower sample stream meets the much more rapidly flowing sheath fluid. The channel is set to a diameter of 780 pm, as measured on the junction of the SONY SH800 130 pm chip. The sample inlet boundary condition is set as fully-developed laminar flow with flow rate of 10 pL/min (sample pressure level 4), and the sheath inlet boundary condition has a flow rate of 5.5 mL/min (as measured experimentally). The outlet boundary condition is set to 0 atm pressure. The shear stress is then calculated along the surface of the suspended cell or cell 100 within a nanovial 10.

[00101] Dynamic range of VEGF-A immunoassay on nanovials using flow cytometry [00102] Nanovials 10 were conjugated with biotinylated anti-VEGF-A capture antibody as mentioned earlier. 5x diluted nanovials 10 were incubated with equal volumes of 0, 0.1 , 1 ,10, 100, 1000, 10000 ng/rnL of recombinant human VEGF-A (R&D Systems) for 12 hours at 4°C on a tube rotator. Excess recombinant VEGF-A was washed before reconstituting as 5x dilution of nanovials 10 and incubating with equal volume of 71.5 pg/mL AF647 anti- VEGF-A (Biolegend) at room temperature for 30 minutes in the dark on a tube rotator. After washing, nanovials 10 were reconstituted as 25x diluted suspension in wash buffer and transferred to a flow tube. Additionally, a small fraction of the sample was transferred to a 96-well plate to be imaged on a fluorescence microscope. Fluorescent signal on nanovials 10 was analyzed using the Sony sorter by gating for single particles. To generate the standard curve, the mean signal from each concentration of VEGF-A was calculated on FlowJo and plotted against concentration. This and all other thresholds in this paper are calculated as mean (0 ng/mL sample) + 2.5 * standard deviation (0 ng/ml sample).

[00103] Stability of VEGF-A captured on Nanovials [00104] VEGF-A was first incubated with VEGF-A capture antibody-conjugated nanovials 10 for 12 hours at 4°C on a tube rotator as mentioned above. VEGF-A was at a concentration (40 ng/ml) resulting in fluorescence signal in the normal cell secretion range. After washing the unbound VEGF-A, the 0-hour sample was labeled with fluorescent AF647 Anti-VEGF-A detection antibody similar to above, while 12- and 24-hour samples were placed in the incubator in a well plate to mimic the secretion assay, washed, and then labeled with detection antibody after the corresponding incubation time. Each sample was then analyzed on the Sony sorter and the resulting mean intensity values were measured after gating for single particles.

[00105] Oligo-barcoded anti- VEGF-A immunoassay validation

[00106] To validate the oligo-barcoded anti-VEGF-A specificity on nanovials 10, nanovials 10 were prepared either with or without recombinant VEGF-A bound (as discussed above, using 1000 ng/ml VEGF-A concentration). Then, 5x diluted nanovials 10 were incubated with an equal volume of either 0 or 71.5 ng/ml oligo-barcoded anti-VEGF-A (Biolegend) for 30 minutes at room temperature on a tube rotator. The four resulting samples were as follows: 1) VEGF-A⁺, anti-VEGF-A⁺, 2) VEGF-A; anti-VEGF-A⁺, 3) VEGF-A⁺, anti-VEGF-A; 4) VEGF-A; anti-VEGF-A". To measure the binding of the oligo-barcoded anti-VEGF-A detection antibodies, the four samples were then incubated with an equal volume of 71.5 ng/ml of AF647 goat anti-mouse IgG (Jackson ImmunoResearch), which binds to the oligo- barcoded Anti-VEGF-A antibody which is mouse species, for 30 minutes at room temperature on a tube rotator. Samples were then analyzed on the Sony sorter by gating for single particles.

[00107] Single-cell gene expression and feature barcode detection library generation for cell-loaded nanovials

[00108] The standard protocol for the Chromium Next GEM Single Cell 3' Kit v3. 1 was followed unless otherwise noted in the methods (10X Genomics, https://www.10xgenomics.com/products/single-cell/). Single cell containing nanovials 10 were isolated using FACS, as noted above. Nanovials 10 were prepared in PBS + 0.04% BSA at a typical concentration of 500 nanovials 10 per pL (experiments in FIGS. 5H, 5J were done at higher concentrations of 2k per pL). 16.5 pL of this solution containing approximately 8,250 nanovials 10 (target cell recovery ~ 5,000 cells) was loaded into Chip G microfluidic emulsion devices. As nanovials 10 settle more quickly than cells, the loading procedure was adjusted to reduce the time in between sample loading and emulsion generation. During the sample loading step, the nanovial + master mix suspension was first allowed to settle for 2 minutes after mixing, then the first half (35 pl) was taken from the supernatant and loaded in the lOx chip sample well (Row 1). The gel bead well and partitioning oil wells were then loaded. The second half of the nanovial sample (35 pl) was pipet-mixed and added to row 1 immediately before loading into a Chromium X controller to generate emulsions. After sequencing, approximately 40-50% of the target cell recovery number was recovered with this method, or -2500 nano vials 10 containing cells per experiment.

[00109] Libraries were assembled from emulsions according to the protocol (Chromium Single Cell 3' Reagent Kits User Guide (v3.1 Chemistry Dual Index) with Feature Barcoding technology for Cell Surface Protein and Cell Multiplexing), using the Chromium Single-Cell 3’ Library Kit (10X Genomics) for purification, amplification, fragmentation, end repair, A- tailing, adapter ligation, and final library indexing and amplification. Library cleanup was done with SPRI select reagent beads (Beckman Coulter, B23317). Separate libraries were made to detect the nanovial-associated streptavidin oligonucleotide barcode DNA and SEC- seq antibody-conjugated oligonucleotide DNA per cell using the protocol for Cell Surface Protein libraries in the above user guide. All libraries were quantified on a Tapestation 4200 (Agilent, G2991BA) according to the manufacturer's protocols using the D5000 screentape and reagents for cDNA QC (Agilent, 5067- 5588, 5067- 5589) and D1000 screentape and reagents for final library QC (Agilent. 5067- 5582, 5067- 5583). Libraries were pooled and paired-end sequenced at lOObp per end with an additional l Obp of index reads on aNovaSeq S4 flow cell (Illumina, Novaseq 6000).

[00110] Emulsion imaging

[00111] Reservoir devices with a height of 130 pm made using standard soft lithography techniques (PDMS devices plasma bonded to glass slides) were used to image nanovial- containing emulsions produced from the 10X Chromium system in a single layer. 5 pl of 10X partitioning oil was added first to displace air, followed by 3-5 pl of 10X emulsion added gently without disrupting the droplets. To quantify nanovial loading in droplets, the emulsion was imaged using a fluorescence microscope. Then a custom MATLAB script was used to first detect all droplets, and then determine how many nano vials 10 were in each detected droplet. [00112] Species mixing experiment

[00113] Nanovials 10 were separately loaded with either human MSCs (1 cell : 1 nanovial) or MEFs (0.4 cell : 1 nanovial). Single-cell loaded nanovials 10 for each species were separately sorted and then combined in a 1 : 1 ratio. The sample was reconstituted as 2000 nanovials/pl before loading into a 10X chip for single-cell sequencing library preparation.

[00114] Comparing transcriptomes of suspended cells and cells on nanovials

[00115] MSC cells 100 were prepared either suspended or loaded in nanovials 10. The nanovial sample was incubated for 12 hours, similar to secretion assay experiments, and labeled with oligo-barcoded streptavidin (Biolegend) after loading. Suspended cells and nanovial samples were sorted for single cells or single cells 100 on nanovials 10, respectively. An additional sample of suspended cells was left unsorted. Each of the three samples (suspended and sorted, suspended and unsorted, and cells 100 on nanovials 10) was reconstituted as 500 cells/pl or nanovials/pl and loaded in separate 10X chip channels for single-cell sequencing library preparation.

[00116] Barcoded normoxic and hypoxic conditioned MSCs on nanovials pooled for scRNA-seq

[00117] MSCs were loaded into nanovials 10 and were incubated for 12 hours in either normoxic or hypoxic (500 pM deferoxamine) media. After incubation, each nanovial sample was conjugated with a different oligo-barcoded streptavidin (Biolegend). Nanovial samples were then sorted using the sorting gate for single cells on nanovials, and then normoxic and hypoxic cell-loaded nanovials were combined in a 1: 1 ratio. The pooled sample was reconstituted as 1000 nanovials/pl before loading into a 10X chip for single-cell sequencing library preparation.

[00118] SEC-seq for MSC transcriptome and VEGF-A secretion

[00119] VEGF-A capture antibody-conjugated nanovials 10 were loaded with MSCs as described above. After straining, nanovials 10 were incubated in 6-well plates for 12 hours to accumulate VEGF-A secretions, with up to 170,000 nanovials per well in 2 mL media (cell culture media for the normoxic condition and media supplemented with 500 pM deferoxamine for the hypoxia-inducing condition). Plates were shaken in horizontal crossmovements before placing in the incubator to space out nanovials. After incubation, nanovials were collected in wash buffer, centrifuged, and resuspended as 5x diluted nanovial suspension. This suspension was incubated with an equal volume of 71.5 pg/mL of oligo- barcoded anti-VEGF-A detection antibody (Biolegend) diluted in a buffer containing calcein AM for 30 minutes at 37°C with gentle vortexing every 10 minutes before washing excess antibody with one high-dilution wash (>100x nanovial volume). Samples were then sorted using the SONY SH800S using the single cell on nanovial gate. The normoxic and hypoxic MSC on nanovial samples were then reconstituted separately as 500 nanovials/pl before loading into a 10X chip for single-cell sequencing library preparation.

[00120] Surface marker staining and enrichment of VRS cells

[00121] For these experiments, MSCs were dissociated from culture in flasks, and directly used without nanovial loading. Cells were stained with APC/Vio 770 anti-IL13RA2, diluted 1 : 10 in staining buffer (1% BSA in PBS) containing calcein AM, for 15 minutes at room temperature. Cells were washed twice before flow cytometry and sorting. During flow cytometry, samples were first gated for single cells (FSC-A vs FSC-H) and then gated for calcein positive signal.

[00122] For ELISpot, IL13RA2+/- cells 100 were sorted into 1.5 ml tubes, spun down and resuspended in media at an appropriate concentration for seeding into ELISpot plates (see above for ELISpot methods). Three wells were seeded per sample.

[00123] For bulk RNA-seq experiments, IL13RA2+/- cells 100 were sorted into 15 ml tubes (at least 100,000 cells per sample). Slightly different gates were used for each of three replicates w ith respect to IL13RA2 cutoff for positive and negative cells, see FIG. 17C.

[00124] For expansion experiments, IL13RA2+/- cells 100 were sorted into 1.5 ml tubes, spun down and resuspended in media for plating into 6 well plates. For VEGF-A ELISA, once cells in two replicate wells per condition were -60% confluent, their media was changed after PBS w ash and then collected after 22 hours of secretion in culture. Media w as centrifuged to remove cell debris, and then used undiluted for ELISA. For IL13RA2 restain experiments, cells were collected when -80% confluent and stained as described above before flow' analysis.

[00125] Single cell sequencing data analysis

[00126] Sequencing data w ere demultiplexed in Basespace and mapped, barcode collapsed, and filtered in the Cell Ranger software (lOx Genomics). Reads were mapped to the hg38 Refseq human reference transcriptome, or for the mixed species experiment, a fusion of the hg38 genome and the mouse mm38 genome build using the Cell Ranger mkref function. The output from Cell Ranger, a raw' sparse matrix with digital expression of cell barcodes by genes, was used for downstream analysis. [00127] Using Seurat 4. 1 in R (https://github.com/satijalab/seurat), normalization of transcripts and clustering of cells was performed, and obtained reduced dimensionality PCA and UMAP coordinates for each cell. Cells were regressed using depth as the variable. To identify and remove doublets, the cell matrix was processed by DoubletFinder (https://github.com/chris-mcginnis-ucsf/DoubletFinder).

[00128] Streptavidin barcodes linked to nanovials 10 were used to separate mixed hypoxic and normoxic MSCs in FIGS. 5A-5L. Barcode reads for this feature were matched to each cell 100 using Cell Ranger’s multi-config workflow. A pseudocount was added to each barcode read and, for each cell, calculated the ratio of barcodes. Cells with a ratio favoring one barcode at least 2.5-fold were called for that tagged sample (“Nonnoxic” or Hypoxic”), while cells with ratios between 2.5 and -2.5 fold were considered “Mixed” and removed from the analysis. Cells with less than 25k streptavidin reads for either sample (-15% of max) were considered unannotated and also discarded. In FIGS. 11 A-l IF, cell separation from nanovials 10 was identified during emulsion formation using detection of a single barcode; here escaped cells 100 were called which had fewer than 800 barcode reads (25% of average read number).

[00129] Analysis and plots were mostly created using R and the libraries ggplot2 and pheatmap. The hypoxic gene signature shown in FIG. 5L was derived de novo using the gene-gene correlation method GEND, as previously described. Cell clusters were derived using Seurat’s FindNeighbors and FindClusters function, and differential expression among clusters was determined by using the FindMarkers function. Differential expression between samples was determined by testing each gene for a distribution p- value of less than 0.05 and a fold change greater than 2 between the average gene value in the cells for each sample. Histograms for sequencing data was generated by grouping normalized reads into 100 equally sized bins. Gene ontology for various gene lists was determined using Metascape.

[00130] Cell cycle regression was performed using Seurat. Briefly, cell cycle genes marking S phase and G2/M phases were used to create cell cycle scores for each cell 100 using the CellCycleScores function. These scores were used to adjust the gene by cell matrix using the ScaleData function to regress out the effect of cell cycle, and then the output was processed and analyzed using normal parameters.

[00131] SEC-seq reads were recovered from the Cell Surface Protein library' workflow and matched to the transcriptome cell barcodes using Cell Ranger’s multi-config tag workflow. Secretion reads were log-transformed to more closely match the dynamic range of the normalized gene transcripts. Gene correlates were determined using Pearson's correlation of all gene transcripts against the log of secretion reads. Secretion correlate genes were ranked by order of correlation. For generation of a consensus ordering between multiple SEC-seq experiments, the Pearson’s correlation values were averaged for each gene and anew ordering rank was determined from that average.

[00132] The Vascular Regenerative Signal (VRS) was determined by identifying the cluster with high IGFBP6 expression in three MSC scRNA-seq experiments (two Normoxic replicates and a Hypoxic MSC run), running differential gene expression analysis for the respective cluster against all other cells in the given experiment, and then taking the overlap of genes betw een the three samples. The percent of cells 100 in each experiment called as “VRS expressing" was calculated by averaging the expression of all genes in the VRS per cell and using a threshold of 75% of the max averaged value. Potential regulators of the VRS genes were determined by testing the VRS genes for enrichments in the TRRUST database. VRS gene types were annotated using separate databases for the secretome (SPRomeDB), transcription factors (ATFDB), and surfactome (SURFY), after overlaps between the secretome and the other two databases were pruned.

[00133] Bulk RNA-sequencing analysis

[00134] RNA was isolated from cells 100 pelleted after sorting using TRIzol reagent (ThermoFisher, #15596026) to dissociate, then phenol chloroform extraction, and binding and purification with the RNA concentrator kit (Zymogen, cat R1080). RNA-sequencing libraries were prepared from this RNA using a strand specific poly-A capture kit (KAPA mRNA HyperPrep Kit, Roche KK8580). Libraries were paired end sequenced at lOObp per read on aNovaSeq S4 flow- cell (Illumina, Novaseq 6000).

[00135] RNA-seq reads were filtered and aligned using the STAR aligner v2.7. 10b and Samtools, using the hg38 human genome, and reads were assigned to genes using the R packages EdgeR and Rsubread. Alternative splicing comparisons were calculated using RMATs, using comparison testing for splice junction reads only to describe spliced exons. Isoform data was calculated per exon using EdgeR and the exon usage was tested using the Simes test method. Data graphs were created in R using the packages ggplot2 and pheatmap.

[00136] FIG. 19 illustrates an exemplary, detailed SEC-seq protocol. Particles or nanovials 10 are first functionalized with cell and secretion capture moieties. Cells 100 are suspended in media and mixed with that partici es/nanovials 10 in a tube or well plate followed by straining to remove unbound cells 100. Cell-loaded particles or nanovials 10 are incubated to collect secretions which are then stained with oligo-labeled detection antibodies. The sample is then sorted utilizing either live cell and/or secretion signals for sorting, followed by singlecell sequencing. The workflow can be broadly broken down in the four phases as seen in FIG. 19. This includes Nanovial Functionalization, Cell Loading and Secretion Capture, OligoSecretion Labeling, FACS and Library’ Preparation.

[00137] Nanovial coating (Volumes are based on 200k particles or nanovials 10 per vessel) [00138] Timing: Steps 1-13, 50 min; Steps 14-18, 50 min, or overnight if pausing;

[00139] 1. Confirm the particle/nanovial 10 and cell 100 numbers to use for loading. For one SEC-seq experiment, with a target of 10k-40k cell-loaded nanovials 10 use ~200k nanovials 10. Multiple vessels with this 200k increment can be processed in parallel to scale the experiment, to test multiple conditions or larger cell numbers. The stock concentration of nanovials 10 is -4000 nanovials/pl for 35 pm diameter nanovials 10 and -2000 nanovials/pl for 50 pm diameter nanovials 10. The cell-to-nanovial ratio should be 1: 1 to load -10-20% of nanovials 10 with cells 100, accounting for loss in the loading process.

[00140] CRITICAL STEP: The number of nanovials 10 used will depend on the loading rate for the cells 100 of interest. Lower cell-to-nanovial ratios (e g. 1:5 or 1 : 10) can be used for rare cell samples to isolate more of the inputted cells 100 into nanovials 10, and to reduce doublets for smaller cells 100. However, this will result in a smaller fraction of nanovials 10 loaded with cells 100.

[00141 ] 2. Precoat all pipette tips used to handle nanovials 10 with Wash Buffer by pipetting the maximum volume up and then expelling the volume.

[00142] 3. Precoat a 1.5 ml eppendorf tube with 500 pl of Wash Buffer by flushing the surface and removing the volume.

[00143] CRITICAL STEP: Precoating reduces the loss of nanovials 10 that may stick to the pipette tip and vessel surfaces.

[00144] 4. Pipette the nanovial stock to resuspend equally throughout the volume and make sure there is no visible pellet.

[00145] 5. Pipette the nanovial suspension from the stock and transfer it to a precoated 1 .5 ml eppendorf tube with a precoated tip. The volume should be determined based on the number of nanovials 10 (ultimately cells 100) that are desired to be analyzed (see Step 1). For example, pipette 50 pl from the 4000/pl stock to transfer 200k 35-pm-diameter nanovials 10 sufficient for 10k-40k loaded cells 100. [00146] 6. In a separate tube, prepare an equal volume of streptavidin solution at a concentration of 300 pg/ml in Wash Buffer.

[00147] 7. Transfer the streptavidin solution to the tube containing the nanovial suspension.

[00148] 8. Mix thoroughly by pipetting up and down (for at least 30 seconds) and incubate for 30 min, RT (20-25°C).

[00149] 9. Add a 10X volume of Wash Buffer to the reaction.

[00150] 10. Centrifuge at 200g for 5 mins, RT.

[00151] 11. Aspirate supernatant being careful not to aspirate the nanovial pellet.

[00152] 12. Repeat wash steps 9-11, three times to completely wash off any unbound streptavidin.

[00153] 13. Resuspend nanovials 10 back into 1 volume (ex. 50 pl) of Wash Buffer.

[00154] 14. Prepare a Capture Cocktail; this cocktail can include (i) a biotinylated surface marker-specific cell capture antibody (or other biotinylated cell binding agents) at a final concentration of 50 pg/ml and (ii) a biotinylated secretion capture antibody at a final concentration of 50 pg/ml in 1 volume of Wash Buffer (ex. 50 pl).

[00155] CRITICAL STEP: Creating the Capture Cocktail is a necessary step since both cell capture and secretion capture moieties bind to the same streptavidin functional groups on the nanovials 10. Sequential addition will lead to low binding of the subsequently added capture moiety.

[00156] 15. Add 1 volume of Capture Cocktail to the streptavidin-coated nanovial suspension from step 13.

[00157] 16. Incubate for 30 min at RT (18-22°C) or 4°C overnight if pausing the experiment.

[00158] PAUSE POINT: The nanovials 10 mixed with Capture Cocktail can be kept overnight at 4°C and the experiment can be resumed at the next step the following day. This is recommended for secretion durations of longer than 1 hour to reduce the overall work duration in a single day.

[00159] 17. Add 10X volume of Wash Buffer to the nanovial suspension from step 16.

[00160] 18. Spin at 200g for 5 min and aspirate the supernatant, being careful not to aspirate the nanovial pellet. Repeat the wash steps three times total.

[00161] Cell loading

[00162] Timing: Steps 19-24, 110 min; [00163] 19. Optional: For cells 100 with low viability' (below- 70%), it is recommended to remove dead cells 100 by live cell enrichment or Ficoll gradient separation prior to loading into nanovials 10.

[00164] 20. Resuspend cells 100 in a 20X volume of media to a final concentration such that there is 1 cell per 1 nano vial (ex. 200k cells in 1 ml per 200k nanovials 10 in 50 pl). [00165] 21. Precoat either a 24-well plate or a 5 ml eppendorf tube with 2 ml of Wash Buffer per well or tube.

[00166] 22. Add the entire volume (ex. 50 pl) of nanovial suspension into a single well or tube from step 21.

[00167] 23. Add cell suspension into the well or tube and pipette up and down gently for 30 seconds in circular motions to mix cells 100 and nanovials 10.

[00168] 24. Incubate the plate or tube at 37°C for 90 min, with agitation (e.g. 20 rpm) if possible. Pipette up and down gently to mix every' 30 min.

[00169] CRITICAL STEP: Secretions, such as immunoglobulins, that occur rapidly at high levels may lead to high background during the standard 90 min cell loading process. In this case, use this Cell loading process below for high-secretion-rate cells instead, which uses blocking and cooling to limit background.

[00170] Cell loading: High-secretion-rate cells (Alternative to Steps 19-24)

[00171] Timing: Steps 25-35, 90 min;

[00172] 25. Place the nanovial suspension from step 18 on the ice and place 20 ml cell culture medium on ice to equilibrate temperature.

[00173] 26. Transfer target number of cells 100 in culture medium into a 15 ml falcon tube.

[00174] 27. Centrifuge cells 100 at a speed suitable for the cell type for 5 min at 4°C.

[00175] 28. Wash cells 100 with 15 ml ice cold culture medium and centrifuge again at a speed suitable for the cell type for 5 min at 4°C.

[00176] 29. Resuspend cells 100 in cell culture medium at a concentration of 4000k cells/ml (50 pl for 200k cells 100).

[00177] 30. Optional: For high secretion rate cells 100. it may be beneficial to add blocking antibodies against the secreted protein in order to reduce or eliminate the capture of secreted proteins during cell loading steps. Therefore, dilute the blocking antibody (stock concentration 0.5 mg/ml) in culture medium at a final concentration of 50 pg/ml into the cell suspension. The blocking antibodies against the secreted protein should be the same clone as the secreted protein capture antibody.

[00178] 31. Pipette to mix the cell suspension (50 pl) with blocking antibody (7.7 pl) 3 times.

[00179] 32. Transfer the cell suspension (57.7 pl) into the nanovial suspension tube (~20 pl) from step 18 on ice. Blocking antibody is ~10-fold diluted in step 31-32.

[00180] 33. Pipette up and down gently for 30 seconds in circular motions to mix cells 100 and nano vials 10 evenly.

[00181] 34. Add 1 ml ice cold cell culture medium per 200k nanovials 10.

[00182] 35. Incubate tube on ice for 1 hr allowing cells 100 to adhere to nanovials 10.

[00183] Removing unloaded cells

[00184] Timing: Steps 36-42, 30 min;

[00185] 36. Set aside either two 15 ml conical tubes, or one 15 ml conical tube and one 6- well plate. One 15 ml conical tube is for collection of waste (unloaded cells 100). The second 15 ml conical tube or the 6-well plate is for collecting nanovials 10.

[00186] CRITICAL STEP: A 6-well plate should be used for longer secretion incubation times (> 90 minutes) to maintain normal culture conditions during incubation, or when a larger fraction of cells 100 are expected to secrete, as the larger area of the plate spaces cell- loaded nanovials 10 further apart.

[00187] 37. Precoat one of the 15 ml conical tubes or a 6-well plate with Wash Buffer and set aside (for collecting loaded nanovials 10).

[00188] 38. Precoat a 20pm cell strainer with Wash Buffer by pipetting 500 pl of Wash Buffer on top of the strainer and letting it flow through.

[00189] 39. Place the cell strainer on top of another 15 ml conical tube (for unloaded cells 100), with the narrow end facing up (inverted compared to normal usage).

[00190] 40. Transfer the nanovials 10 and cell mix onto the strainer and let the liquid drain through the strainer.

[00191] 41. After the sample flows through, wash at least two more times by pipetting 500 pl of Wash Buffer on top of the strainer and letting it flow through.

[00192] 42. Flip the strainer over and recover the nanovials 10 into either the precoated 15 ml conical tube or 6-well plate by pipetting 2 ml of media and letting it drain through the inverted strainer. [00193] Incubation to collect secretions on nanovials:

[00194] Timing: Steps 43-44, timing depends on the cell secretion rate (e.g., 30 min for plasma cells secreting IgG, 3-6 hours for T cells secreting cytokines, 12-24 hours for stromal cells secreting VEGF-A).

[00195] 43. For normal secretion studies, incubate at 37°C in the well plate or tube. Time varies by secretion, 1-24 hours.

[00196] 44. Optional: For high-secretion-rate cells 100, spin down nanovials 10 at 200g for 5 min after straining, resuspend strained nanovials 10 with 2 ml pre-warmed culture medium per 200k nanovials 100 and transfer into a pre-coated 5 ml eppendorf tube. Load the tube on a rotator/shaker at low speed at 37°C and incubate for 30 min to collect secretions.

[00197] Labeling secretions with detection antibodies:

[00198] Timing: Steps 45-52, 50 min; Step 53, 50 min; Step 54, 1 min;

[00199] 45. Precoat a 5 ml eppendorf tube with Wash Buffer and leave 2 ml of Wash

Buffer in the tube.

[00200] 46. Transfer the nanovial suspension from step 43 or 44 to the tube.

[00201] 47. Centrifuge samples at 200g for 5 min, RT. and aspirate the supernatant to wash.

[00202] 48. Prepare detection Staining Cocktail in Wash Buffer for labeling secreted proteins and cell surface markers. Typical dilutions are 1: 10 to 1 :2000 for 0.5 mg/mL antibody stocks in 100 pl of Wash Buffer. For IgG oligo-barcoded detection antibodies (TotalSeq™ B), this was 0.5 pl in 1 ml for a 1 :2000 dilution, for the VEGF-A oligo-barcoded detection antibody this was 1:700 to a final concentration of 71.5 pg/ml.

[00203] CRITICAL STEP: The Staining Cocktail should also contain a cell stain to enable selecting cell-loaded nanovials 10 by FACS. Option 1 : Add viability dye to the Staining Cocktail. For Calcein AM, add 0.2 pl of stock per 1 ml. Option 2: For heterogeneous cell populations, add fluorescent antibodies against cell surface markers of target cells. For anti-CD3 and anti-CD8, 1:30 of 100 pg/ml and 500 pg/ml stock, respectively.

[00204] 49. Add 100 pl of the Staining Cocktail solution to the 5 ml tube containing nanovials 10 from step 47, pipette to mix gently, and incubate at 37°C for 30 min.

[00205] Optional: For high-secretion-rate cells 100, stain on ice for 20 min.

[00206] 50. Repeat steps 51-52 two times to wash samples with 1 ml of Wash Buffer.

[00207] 51. Centrifuge samples at 200g for 5 min, 4°C. [00208] 52. Aspirate supernatant, being careful not to aspirate the nanovial pellet. If no secondary staining is desired, proceed to step 54 (route A).

[00209] 53. Optional (route B) - If using a secondary secretion detection antibody for either oligo-barcoding or fluorescence-based sorting, repeat staining steps as from step 49-52 for secondary reagents. This secondary' staining can also occur after FACS for addition of an oligo-barcoded secondary antibody targeting the primary fluorescent secretion detection antibody.

[00210] 54. Resuspend the pellet obtained from either step 52 or 53 in 300 pl Wash Buffer and transfer to a FACS tube for sorting.

[00211] Sorting:

[00212] Timing: Steps 55-58, variable, depending on the samples and sorting rate of the machine;

[00213] 55. Optional - For 35 pm nanovials 10, filter samples with a 40 pm strainer to remove clumps. For 55 pm nanovials 10, filter samples with a 70 pm strainer instead.

[00214] 56. Choose appropriate flow cell or microfluidic chip size compatible with nanovials 10 (100 pm nozzle for 35 pm nanovials 10 or 130 pm nozzle for 35 or 50 pm nanovials 10). The smaller nozzle size of 100 pm is preferred for the 35 pm nanovials 10 as it enables sorting at higher rates.

[00215] 57. Set up the gates for live single-cell-loaded nanovials 10, and/or distinguishing cells of interest (based on surface markers or secreted markers) loaded onto the nanovials 10. [00216] 58. Collect sorted nanovial samples for the scRNA-seq workflow into a 2ml eppendorf tube pre-filled with 500 pl of Wash Buffer.

[00217] lOx Genomics scRNA-seq workflow, library preparation and sequencing: [00218] Timing: Steps 59-66, 40 min; Steps 67-68, 4-7 days;

[00219] 59. Calculate nanovial number targeting loading of 2.5k-10k nanovials 10 per sample for each lOx Genomics chip well.

[00220] 60. Centrifuge nanovials 10 at 200g for 5 min at 4°C.

[00221] 61. Resuspend nanovials 10 in Cell Resuspension Buffer such that the final concentration is 500 Nanovials/pl.

[00222] 62. Prepare lOx Genomics workflow as usual. Make a mixture of the nanovial suspension and Reverse Transcriptase solution. Prepare this mixture just before loading of the Chip G/X. Refer to the “Cell Suspension Volume Calculator’" in the lOx Genomics user handbook to determine the optimum final volume for the nanovial suspension based on concentration.

[00223] 63. Allow nanovials 10 to sediment in Reverse Transcriptase mixture during reagent preparation (~3 min). Take approximately 70% of sample volume supernatant without disturbing nanovials 10 and load into the cell well. This is ~50 pl in the Chip G workflow.

[00224] 64. Load all other wells with appropriate reagents.

[00225] 65. Return to the remaining 30% of nanovial/Reverse Transcriptase mixture, resuspend and load over the top of the cell well volume and attach the gasket (approx. 20 pl in Chip G workflow)

[00226] 66. Proceed immediately to the controller and start flow.

[00227] 67. Prepare libraries using lOx Genomics Chromium Next GEM Single Cell v3. 1 kit following the lOx Genomics User Guide (CG000317) as usual. Be sure to also prepare the Feature Barcode library. Post step 65, there are no modifications to the normal library preparation workflow. The cDNA and SEC-seq detection oligo barcodes are successfully released from the nanovials 10, so nanovials 10 may be discarded with other remaining cellular debris after library preparation.

[00228] 68. Sequencing the libraries using the following minimum read lengths for a 138 cycle sequencing run: 28 bp Readl, 10 bp i7 Index, 10 bp i5 Index, and 90 bp Read2. Refer to lOx Genomics User Guide for library sequencing recommendations.

[00229] SEC-seq data analysis:

[00230] 69. Demultiplex any indexed data and convert to FastQ.

[00231] 70. Process Fastq files using the Cellranger multi-function to process both the transcriptome and secretion oligo-barcoded detection antibody reads. Providing the appropriate barcodes in the multi configuration file allows for correct assignment of detected capture antibody oligo barcodes to each cell at this step.

[00232] 71. Perform data analysis including normalization, batch correction, dimensional reduction, hierarchical clustering, clustering, and differential gene analysis using typical workflows. Secretion values can be extracted from the raw data.

[00233] Python: See https://github.com/Rene2718/SEC-seq_plasma-cell_nanovial

[00234] R: Using Seurat, read the Cellranger output with the Readl0x() function, secretion data per cell will be output in a slot of the created object, named as in your Cellranger configuration settings. [00235] Timing Summary for the Protocol

[00236] Nanovial coating - 2 hours, or overnight if pausing;

[00237] Cell loading - 2 hours

[00238] Removing unloaded cells 100 - 30 min;

[00239] Incubation to collect secretions on nanovials 10: timing depends on the cell secretion rate (e.g.. 30 min for plasma cells 100 secreting IgG, 3-6 hours for T cells 100 secreting cytokines, 12-24 hours for stromal cells 100 secreting VEGF-A).

[00240] Labeling secretions with detection antibodies: 2 hours

[00241] Sorting: variable, depending on the samples and sorting rate of the machine; expect 1 -2 hours

[00242] lOx Genomics scRNA-seq workflow day 1 : 3 hours

[00243] Library preparation and sequencing: 2+ days

[00244] Reagents

[00245] Wash Buffer:

[00246] DPBS (+ Ca^2+/+Mg²⁺) + 0.5% BSA + 0.05% Pluronic F-127 + 1% anti-anti or pen- strep. Mix these reagents and pass them through a Stericup vacuum filter unit. This buffer can be prepared in advance and stored at 4°C for up to a month.

[00247] Cell Resuspension Buffer:

[00248] DPBS (-Ca²7 -Mg²⁺) + 0.4% BSA. Mix these reagents and pass them through a Stericup vacuum filter unit. This buffer can be prepared in advance and stored at 4°C for up to a month.

[00249] Cell Culture:

[00250] Plasma cells: Utilizing the EasySep Human B cell isolation kit, B cells 100 were isolated from healthy donors’ peripheral blood mononuclear cells. The isolated cells 100 were cultured in Iscove’s modified Dulbecco’s medium supplemented with 2- mercaptoethanol (55 pM) and 10% FBS, followed by activation and differentiation. Cells 100 were incubated at 37°C and 5% CO2.

[00251] Immortalized human adipose-derived MSCs: Immortalized human adipose-derived MSCs were cultured in MSC basal medium which was supplemented with low-serum MSC growth kit (per manufacturer’s protocol) and antibiotic-antimycotic. The cells 100 were incubated at 37°C and 5% CO2 and passaged at 70-80% confluency, with cell passages up to 25 used in experiments. [00252] Human donor PBMCs: Naive PBMCs for were acquired from a commercial vendor and cultured in TCRPMI with 50 U/mL IL-2 and chemically synthesized peptides of interest. TCRPMI media includes RPMI 1640, 10% FBS, Glutamax, 10 mM Hepes, nonessential amino acids, sodium pyruvate, and 50 uM P-mercaptoethanol. The cells 100 were incubated at 37°C and 5% CO2.

[00253] CAUTION: It is advised to check your research cell lines regularly to ensure they are authentic and free of infection with mycoplasma.

[00254] While embodiments of the present invention have been shown and described, various modifications may be made without departing from the scope of the present invention. For example, while VEGF-A was identified as the bioactive factor of interest, it should be appreciated that the method is applicable to different bioactive factors. Further, the invention is not limited to MSCs 100 as the platform may be used with other cell types as noted herein. The invention, therefore, should not be limited, except to the following claims, and their equivalents.

Claims

What is claimed is:

1. A method of identifying one or more cell surface markers of cells associated with elevated secretion of a bioactive factor comprising: loading single cells from a population of cells into a plurality⁷ of particles, wherein the single cells are contained within a void or cavity in the plurality of particles that opens to the exterior environment of the respective particles, wherein the particle contains a capture agent for the bioactive factor disposed on a surface of the particle within the void or cavity: labeling cells with elevated secretion levels of the bioactive factor by exposing the plurality⁷ of particles with single cells contained therein to oligo-barcoded labeling reagents that bind to or associate with the bioactive factor; exposing the single cells in the plurality of particles to one or more reagents of a single cell transcriptome assay; and determining the transcriptome of the cells with elevated secretion levels and identifying one or more genes of the transcriptome that encode for the one or more cell surface markers associated with elevated secretion levels of the bioactive factor.

2. The method of claim 1, wherein the cells comprise mesenchymal stromal cells or mesenchymal stem cells (MSCs).

3. The method of claim 1 , further comprising sorting the plurality of particles in a cell sorter prior to determining the transcriptome of the cells with elevated secretion levels and identifying one or more genes of the transcriptome that encode for the one or more cell surface markers associated with elevated secretion levels of the bioactive factor.

4. The method of claim 1, wherein the oligo-barcoded labeling reagents comprise oligo-barcodes that correspond to cells incubated in different environments.

5. The method of claim 4. wherein the different environments comprise a normoxic environment or a hypoxic environment.

6. The method of claim 1, wherein the operations of labeling cells with elevated secretion levels of the bioactive factor and determining the transcriptome of the cells with elevated secretion levels are done simultaneously.

7. The method of claim 2, wherein the bioactive factor comprises VEGF-A protein.

8. The method of claim 2, wherein the one or more cell surface markers comprises interleukin- 13 receptor subunit alpha-2 (IL-13RA2).

9. The method of claim 1. wherein the oligo-barcoded labeling reagents bind directly to the bioactive factor.

10. The method of claim 1, wherein the oligo-barcoded labeling reagents bind indirectly to the bioactive factor via fluorescent reporter.

11. The method of claim 1, wherein the oligo-barcoded labeling reagents comprise oligo-barcoded antibodies.

12. A method of enriching a sub-population of mesenchymal stromal cells or mesenchymal stem cells (MSCs) with elevated secretion of VEGF-A protein comprising: exposing a population of MSCs to anti-interleukin-3 receptor subunit alpha-2 (IL-13RA2) antibody; and sorting the population of MSCs to generate the enriched sub-population of MSCs with elevated secretion levels of VEGF-A protein.

13. The method of claim 12, wherein the sorting comprises using a fluorescent activated cell sorter (FACS) or a magnetic activated cell sorter (MACS).

14. The method of claim 12, further comprising expanding the enriched subpopulation of MSCs with elevated secretion levels of VEGF-A protein.

15. The method of claim 14, further comprising administering the enriched subpopulation of MSCs with elevated secretion levels of VEGF-A protein to a subject.

16. A therapeutic composition comprising an enriched sub-population of MSCs with elevated secretion levels of VEGF-A protein, the enriched sub-population of MSCs comprising IL13RA2-positive (+) MSCs.

17. A therapeutic composition comprising an enriched sub-population of MSCs with elevated secretion levels of VEGF-A protein, the enriched sub-population of MSCs comprising MSCs with elevated levels of gene expression comprising a vascular regenerative signal.