Defining Functional T Cell Receptor Repertoires With Nanovial‐Based Screening

ABSTRACT

T‐cell receptor repertoires are commonly characterized through sequence diversity and antigen binding, while the cellular functions associated with individual receptors are often measured separately. In this review, I consider how functional information can be incorporated more directly into repertoire analysis by linking receptor identity with measurable effector outputs at single‐cell resolution. I focus on Nanovial‐based technologies that combine antigen‐specific capture, stimulation, secretion measurement, viable‐cell recovery, and downstream sequencing in workflows compatible with flow cytometry and single‐cell sequencing platforms. Studies using conventional peptide–major histocompatibility complex ligands, the non‐classical antigen‐presenting molecules MR1 and CD1d, and secretion‐encoded single‐cell sequencing illustrate how these approaches can complement binding‐based enrichment by adding information about secretory effector responses. I also discuss the extension of this framework from ligand‐defined screening to dyad‐resolved assays involving target cells, which can reveal how receptor identity relates to transcriptional and functional programs arising during direct cell–cell interactions. Together, these developments provide tools for studying relationships among TCR sequence, antigen recognition, and cellular response, with potential applications in therapeutic receptor discovery and in generating datasets for predictive models of T‐cell function.

1. Introduction: From Sequence‐Based Clonal Maps to Functional TCR Repertoires

T cells are among the most potent sensing and response elements in biology. Through their T‐cell receptors (TCRs), they survey an immense landscape of antigenic information, distinguish subtle molecular differences on infected, transformed, or stressed cells, and convert recognition into actions that can reshape immunity across scales—from local cytokine release and direct target‐cell killing to durable memory formation. Understanding TCR repertoires is therefore important not only for basic immunology, where they provide a window into how the immune system recognizes and adapts to pathogens, tumors, and altered self, but also for biotechnology and medicine, where TCRs are increasingly used as components of engineered cell therapies and as biomarkers of immune exposure, vaccination, and disease [1, 2]. As cellular therapies emerge alongside molecular and genetic interventions as a major therapeutic modality, the ability to identify and prioritize useful receptors within complex T‐cell populations becomes correspondingly more important.

A central question in TCR biology is how receptor–ligand binding is translated into cellular function. Sequencing reveals which clonotypes are present and expanded, while peptide–MHC multimers identify cells whose TCRs bind defined ligands [3, 4, 5, 6]. These approaches provide essential information about receptor abundance, specificity, and binding, but the resulting cellular response also depends on receptor density, signaling thresholds, co‐receptor engagement, differentiation state, and antigen‐presentation context. Receptors with similar apparent binding properties may therefore produce different effector responses, while receptors with modest binding signals may remain functional [6, 7]. Defining when binding predicts function—and when it does not—is important for both fundamental immunology and receptor discovery.

These considerations motivate a broader view of the TCR repertoire that incorporates receptor sequence, ligand binding, and also the behaviors that follow antigen recognition. Relevant outputs may include cytokine secretion, cytotoxic granule release, target‐cell killing, serial engagement, proliferation, and the ability to sustain productive interactions with target cells. Here, I use the term functional TCR repertoire to describe the portion of receptor space associated with measurable cellular responses under defined antigenic and cellular conditions.

A central challenge is that function is harder to measure than sequence or binding. Secreted proteins diffuse, intracellular staining can compromise cell viability and recovery of nucleic acids representing the induced cellular state, and mRNA abundance does not always predict what a cell secretes or how effectively it will kill [8, 9]. These challenges have motivated screening strategies that preserve the linkage between receptor identity and cellular output at single‐cell resolution. In this review, I focus on Nanovial‐based approaches that couple antigen recognition with secretion‐based functional screening, viable‐cell recovery, and downstream sequencing [9, 10, 11]. Together, these methods provide a framework for examining how TCR identity and antigen recognition relate to measurable cellular responses, including secretion and, more recently, responses arising during direct cell–cell interactions.

2. Established Routes to TCR Repertoire Discovery

TCR repertoires have been studied through three broadly complementary approaches: sequencing‐based characterization of clonotype diversity, ligand‐binding‐based enrichment of antigen‐specific cells, and response‐based identification of cells following stimulation, including activation‐induced marker and cytokine assays [12, 13, 14]. Together, these methods have established the modern framework of repertoire biology. They have revealed clonal expansions associated with infection, vaccination, cancer, and therapy; enabled tracking of public and private clonotypes across individuals; and provided practical routes to candidate receptor discovery. Each approach, however, measures a different aspect of the repertoire. Sequencing identifies which receptors are present and how their abundance changes. Ligand‐binding assays identify receptors that engage a defined antigen under particular experimental conditions. Activation‐marker and cytokine assays indicate that a cell has responded to stimulation. Integrating these measurements remains important because receptor abundance, ligand binding, and induced cellular behavior are related but not interchangeable properties.

Bulk TCR sequencing can profile very large numbers of receptors, quantify clonal abundance, and track repertoire dynamics across tissues, disease states, and therapeutic interventions. Single‐cell methods add paired α‐ and β‐chain recovery and can connect receptor identity with transcriptional or surface‐phenotypic state [5, 15]. These approaches are particularly useful for describing repertoire architecture and nominating clonotypes associated with antigen exposure or clinical response. Sequence information alone, however, generally does not establish cognate specificity or reveal how a receptor will perform when reconstructed in another cellular context. Clonal expansion may indicate prior antigen‐driven selection, but the magnitude of expansion does not necessarily predict cytokine production, cytotoxicity, or other functional properties.

Ligand‐binding approaches, especially peptide–major histocompatibility complex multimers, provide a direct route to enriching T cells that recognize a known antigen. Tetramers and related higher‐avidity reagents have been indispensable for studying antigen‐specific immunity and can efficiently recover rare cognate populations, particularly for well‐characterized viral epitopes. Their performance nevertheless depends on several experimental and biological factors, including TCR affinity and expression, multimer valency, staining conditions, and background binding. Moreover, multimer intensity does not consistently predict the magnitude or quality of a subsequent cellular response. Some receptors with measurable biological activity may stain weakly, whereas strongly staining receptors are not necessarily the most effective in secretion or cytotoxicity assays [10]. These considerations can be especially relevant for self‐like tumor antigens and other targets for which highly avid receptors may be uncommon or undesirable [16]. Binding‐based enrichment therefore provides valuable specificity information, but functional validation remains necessary when the goal is therapeutic receptor selection.

Response‐based assays provide a complementary route by identifying T cells that react following antigenic stimulation. Surface markers such as CD137, CD69, CD154, or OX40 can support viable‐cell enrichment and, depending on the stimulation format, can be applied without prior knowledge of the precise epitope. Cytokine assays offer an additional measure of induced effector activity. Activation‐marker expression can vary with cell state, stimulation duration, and marker choice, and may include indirectly activated or bystander cells in some settings [17]. Intracellular cytokine staining provides sensitive, multiparameter measurements of cytokine production, but requires secretion inhibitors, fixation, and permeabilization, which prevent recovery of viable cells and can complicate downstream RNA and paired‐TCR analysis [18]. Conversely, viable‐cell cytokine‐capture and secretion assays preserve cells but can be affected by diffusion and signal transfer between neighboring cells unless secretion is spatially localized. Thus, response‐based assays provide important evidence of reactivity, but linking the measured response unambiguously to receptor sequence or antigen identity remains technically challenging, highlighting the remaining opportunity for workflows that integrate receptor sequence, antigen recognition, and a directly measured cellular output within the same viable cell.

3. Adding Functional Measurements to Therapeutically Relevant Repertoire Analysis

Functional screening adds an important dimension to repertoire analysis by identifying receptors that produce desired biological outputs in living cells. This is particularly important when selecting TCRs for engineered cell therapies. A candidate receptor must retain sufficient potency and selectivity after transfer into a new cellular context and under the constraints of the disease microenvironment [16]. These properties are not fully determined by binding measurements alone, especially for self‐like tumor antigens, splice‐derived neoepitopes, or non‐classical antigens presented by MR1 and CD1d, where productive receptors may not exhibit strong binding signals [11, 16]. In these settings, binding‐ and function‐based assays may recover only partially overlapping receptor populations.

Single‐cell transcriptomics provides complementary information by linking paired TCR sequences to cellular state [19], but transcription is not a direct measure of effector output. Cytokine transcripts may correlate only weakly with the amount of protein ultimately secreted [10, 11], because secretion is also shaped by translation, intracellular storage, trafficking, and release kinetics. Granzyme B and perforin, for example, may be stored in secretory granules and rapidly released following stimulation. Direct measurement of secreted proteins can therefore provide information that is not captured by a transcriptional snapshot.

4. Nanovial Architecture and Principles of Single‐Cell Functional Screening

Directly linking a T cell to its secreted effector products presents a basic measurement challenge. Secreted proteins rapidly diffuse away from their source, yet the cell must remain viable and physically associated with the resulting signal if it is to be recovered for receptor sequencing, transcriptomic analysis, culture, or functional validation. A functional screening platform must therefore localize individual cells, concentrate their secretions sufficiently for detection, limit transfer of those molecules to other cells being simultaneously assayed, and preserve compatibility with high‐throughput analysis and sorting. Nanovials were developed as a lab‐on‐a‐particle approach to meet these requirements by combining microscale confinement, localized biochemical functionalization, and flow‐cytometric analysis within a suspendable structure [20, 21, 22, 23].

Nanovials are open‐cavity hydrogel microparticles, typically composed primarily of crosslinked polyethylene glycol, with a gelatin‐rich layer localized to the inner cavity (Figure 1). They are manufactured by generating aqueous droplets containing polyethylene glycol and gelatin, allowing the two polymers to phase‐separate and crosslinking the polyethylene glycol‐rich phase (Figure 1A). Removal of most of the gelatin‐rich phase leaves a cup‐shaped particle with an open cavity and a residual gelatin layer that can be selectively modified through its accessible amine groups (Figure 1B,C). In commonly used T‐cell formats, Nanovials have an outer diameter of approximately 35 μm and a cavity diameter of approximately 20–22 μm, dimensions that accommodate a single lymphocyte while remaining compatible with appropriately configured flow cytometers and droplet‐based single‐cell sequencing workflows [24].

FIGURE 1.

Nanovial fabrication, structure, and localized functionalization. (A) Nanovials are produced by microfluidic generation of aqueous droplets containing polyethylene glycol (PEG) and gelatin, which undergo phase separation before UV‐induced crosslinking of the PEG‐rich phase. Subsequent washing removes most of the gelatin‐rich phase, leaving an open‐cavity PEG hydrogel particle with a residual gelatin‐rich layer localized to the cavity surface. The process enables production of more than 10⁶ Nanovials in a single preparation. Representative brightfield and fluorescence images show the open‐cavity morphology and localized gelatin layer. (B) The residual gelatin within the cavity can be selectively biotinylated through NHS‐ester modification of accessible amine groups. Streptavidin binding to the resulting biotinylated gelatin provides a modular interface for subsequent attachment of biotinylated affinity reagents. (C) Representative fluorescence microscopy image of Alexa Fluor 488–streptavidin‐labeled Nanovials demonstrating localization of functional groups within the cavity. (D) Schematic illustrating a typical functionalization sequence in which biotin within the cavity is coupled to streptavidin and subsequently to a biotinylated reagent, shown here using anti‐CD45 as an example for cell capture. This localized chemistry allows the Nanovial cavity to be independently configured with cell‐capture, stimulatory, antigen‐presenting, and secretion‐capture reagents for different single‐cell assays. Adapted from Lee et al. [23], with permission.

The open‐cavity geometry provides both a physical niche for a cell and a localized solid phase for biochemical assays. The gelatin lining can be biotinylated and subsequently modified using streptavidin and biotinylated affinity reagents. Depending on the application, these reagents may include antibodies that capture cells through surface proteins, antigen‐presenting ligands that bind or stimulate cognate receptors, and antibodies that immobilize secreted cytokines or cytotoxic proteins (Figure 1D). A cell loaded into the cavity is therefore positioned immediately adjacent to a surface that can both provide an activating input and capture an induced output. Secreted molecules accumulate on this surface and are detected using fluorescent or oligonucleotide‐barcoded secondary antibodies in a sandwich immunoassay (Figure 2). Conceptually, the Nanovial acts as a suspendable, sortable microscale immunospot: the secreted product is recorded on the particle surrounding the viable cell rather than on a fixed plate.

FIGURE 2.

Nanovial‐based workflow for function‐first TCR discovery. Cavity‐containing hydrogel microparticles (Nanovials) are functionalized with activating ligands, including pMHC, MR1, CD1d, or other antigen‐presenting molecules, to capture and stimulate cognate T cells. Secreted effector molecules are captured by antibodies on the Nanovial surface and accumulate locally within the Nanovial cavity, where they are detected using fluorescent and/or oligonucleotide‐barcoded reagents. Functionally active cells are identified and sorted by flow cytometry, followed by single‐cell TCR sequencing to recover paired αβ receptor sequences. Recovered receptors can then be analyzed by effector function, re‐expressed and validated experimentally. Bottom panels show representative images of T cells loaded on Nanovials, secretion accumulation, gating of IFNγ‐positive events, sorting of loaded Nanovials, and ranking of recovered clones by secretion level, encoded through the oligobarcoded reagents, as well as mRNA levels. Images are adapted from references [10, 11] with permission.

The cavity does not form a sealed compartment. Instead, it reduces convective transport near the cell and provides a large nearby capture surface, allowing locally secreted molecules to be immobilized before they disperse into bulk solution. This distinction is important because Nanovials provide partial rather than complete compartmentalization. Secreted proteins can still diffuse out of the cavity and be captured by neighboring particles, particularly during long incubations, at high cell densities, or when abundantly secreted analytes are measured. Assay conditions must therefore balance the time required to generate a detectable signal against the accumulation of background through crosstalk from cells on neighboring Nanovials. In published T‐cell implementations, incubation periods have ranged from hours to approximately 12 h depending on the analyte and cellular state, with increasing signal but also increasing crosstalk at longer times.

A typical Nanovial assay consists of four stages (Figure 2). First, the cavity is functionalized with cell‐capture, stimulatory, and secretion‐capture reagents. Second, cells are incubated with the particles under conditions chosen to favor predominantly single‐cell occupancy, after which unbound cells are removed. Third, loaded Nanovials are incubated to allow stimulation and secretion, followed by staining of the cell and the captured extracellular products. Finally, Nanovials are analyzed and sorted by flow cytometry on the basis of cell occupancy, viability, surface marker staining, and secretion signal (Figure 2). Selected cells may remain attached to the particle for downstream sequencing, or may be released enzymatically or following cell division for subsequent culture and functional testing. Nanovials have been introduced directly into droplet‐based single‐cell sequencing systems, allowing receptor sequence, transcriptome, oligo‐barcoded ligand identity, and oligo‐barcoded cumulative secretion signal to be recovered from the same cell.

An additional strength of the platform is its modularity. The same basic Nanovial architecture can be adapted to different biological questions by changing the molecules displayed within the cavity. In early immune‐cell studies, Nanovials were functionalized with antibodies to capture T cells through broadly expressed surface markers and with cytokine capture reagents to profile secretion [10]. In later iterations, pMHC monomers were incorporated to selectively capture antigen‐reactive T cells and simultaneously stimulate them. This modularity also enabled stimulation and functional profiling of unconventional T cells with MHC‐like molecules like MR1 and CD1d, which present metabolite and lipid instead of peptide antigens [11]. Nanovials were also shown to be compatible with the loading of recombinant antigens, or antigens displayed on membrane‐coated vesicle or virus‐like‐particle‐inspired formats, enabling a more cell‐like presentation of membrane proteins or pMHC molecules [25, 26].

The platform is modular, but its performance depends on the assay configuration. Cell loading varies with particle dimensions, cell size, the capture ligand, and the cell‐to‐particle ratio. Because multiple reagents share a finite functionalized surface, increasing the density of one component can reduce the available sites for others, which should be considered when multiplexing to capture multiple secreted factors. Secretion is measured cumulatively over the incubation period and therefore represents an integrated output rather than an instantaneous or kinetic measurement. Longer incubations may increase sensitivity but also increase crosstalk between neighboring cells. Nanovial‐based assays therefore represent a set of experimentally controlled and quantitative views of T‐cell potency, although not comprehensive.

In the broader landscape of single‐cell functional screening, Nanovials occupy an intermediate position between static microwells and closed microfluidic droplets [20]. Microwells support imaging‐rich measurements and strong spatial registration, but scale is limited by the 2D footprint and recovery of selected cells may require specialized retrieval. Droplets provide strong confinement and rapid accumulation of secreted products, but reagent exchange and recovery can require dedicated microfluidic workflows. Nanovials provide less complete isolation than closed droplets but retain open access for washing and staining, preserve viable particle–cell complexes, and interface with flow cytometry and downstream sequencing. No single approach captures all dimensions that span from TCR identity to induced function. The methods differ in whether these properties are measured intrinsically within the assay or recovered through downstream sorting and sequencing, as well as in their compatibility with viable‐cell recovery, throughput, and multiplexing. Table 1 summarizes these complementary capabilities and the principal trade‐offs.

TABLE 1.

Comparison of approaches used for TCR repertoire discovery and functional prioritization.

Approach	Primary measurement	Antigen specificity retained	Direct extracellular functional readout	Viable cell recovery	Paired TCR αβ recovery	Transcriptomic integration	Throughput and accessibility	Principal strength	Principal limitation
Bulk TCR repertoire sequencing	Clonotype abundance and diversity	Not intrinsic	No	Not applicable	No	Not at single‐cell resolution	Very high; widely accessible	Broad and quantitative description of repertoire architecture	Does not directly establish antigen specificity or cellular function
Single‐cell TCR sequencing	Paired TCR sequence and cellular state	Can be inferred or added through separate assays	No	Cells are consumed during library preparation	Intrinsic	Intrinsic	High; commercial platforms available	Links paired receptor sequence to transcriptional or surface phenotype	Captures cellular state but not a directly measured effector output
pMHC multimer or tetramer staining	TCR binding to a defined antigen–MHC complex	Intrinsic	No	Yes	Demonstrated after sorting and sequencing	Demonstrated after sorting or with barcoded multimers	High; standard flow‐cytometry workflow	Efficient enrichment of antigen‐binding cells with known specificity	Binding intensity does not consistently predict downstream function
Activation‐induced marker assays	Surface‐marker induction after stimulation	Retained only if stimulation conditions identify the antigen	Indirect	Yes	Demonstrated after sorting	Demonstrated after sorting	High; standard flow‐cytometry workflow	Preserves responding cells and can be used with antigen pools	Markers are indirect measures of function
Intracellular cytokine staining	Intracellular cytokine accumulation after stimulation	Retained through the stimulation condition	Indirect; measures intracellular accumulation	No	Technically limited	Limited by fixation and RNA degradation	High; standard flow‐cytometry workflow	Sensitive, multiparameter assessment of cytokine‐producing cells	Requires secretion inhibitors, fixation, and permeabilization
ELISpot or FluoroSpot	Secreted cytokines accumulated around individual cells	Retained through the stimulation condition	Intrinsic	Limited given lengthy incubation and staining	Not typically retained	No	Medium; plate‐based and widely available	Detection of low‐frequency cytokine‐secreting cells	Limited recovery of the secreting cell and receptors
Microwell‐ or microchamber‐based functional screening	Secreted factors, imaging phenotypes, or co‐culture responses	Platform‐dependent	Intrinsic	Demonstrated in selected systems	Demonstrated with cell retrieval and sequencing	Demonstrated in selected workflows	Low to medium; often specialized	Rich imaging, temporal, and multiplexed functional measurements	Lower scalability and more labor‐intensive recovery of selected cells
Droplet microfluidic functional screening	Secreted factors or responses within closed droplets	Platform‐dependent with encoded ligands	Intrinsic	Demonstrated after droplet disruption or custom sorting	Demonstrated in selected workflows	Demonstrated in selected workflows	High, but often requires specialized microfluidics	Strong confinement, low crosstalk, and rapid secretion accumulation	Reagent exchange, cell recovery, and sorting may require custom workflows
Nanovial‐based screening	Antigen binding, induced secretion, and cell–cell interaction associated phenotype	Intrinsic when antigen‐presenting ligands or barcodes are displayed	Intrinsic	Yes	Demonstrated by index sorting or single‐cell sequencing	Demonstrated through SEC‐seq and related workflows	High; compatible with appropriately configured flow cytometry and single‐cell sequencing	Integrates antigen recognition, direct secretion measurement, and receptor recovery	Open cavities permit some diffusion and crosstalk

5. Nanovial‐Based Functional Screening of T Cells and TCR Repertoires

The application of Nanovials to TCR discovery developed through progressively integrated T‐cell assays. Initial studies established that viable T cells could be loaded onto Nanovials, stimulated through surface‐displayed ligands, analyzed according to induced secretion, and recovered by fluorescence‐activated cell sorting. These studies also defined practical assay parameters, including ligand density, cell‐loading conditions, secretion‐accumulation time, multiplexed cytokine capture, and the balance between signal accumulation and interparticle crosstalk [9, 10, 11].

A key step was demonstrating that the Nanovial surface could serve not only to retain cells and capture secretions but also to provide productive stimulation. Nanovials bearing anti‐CD3 and anti‐CD28 antibodies supported non‐antigen‐specific activation [27], whereas particles displaying peptide–MHC monomers stimulated antigen‐specific T cells and enabled measurement of induced cytokine secretion [10]. Because responding cells remained viable and physically linked to the accumulated secretion signal, they could subsequently be sorted, released, expanded, or analyzed by single‐cell sequencing.

These capabilities provided the basis for increasingly integrated TCR‐discovery workflows. The following sections trace this progression from model T‐cell systems to conventional pMHC‐restricted receptor discovery, secretion‐encoded single‐cell sequencing, and non‐classical MR1‐ and CD1d‐restricted repertoires and direct analysis of T‐cell–target‐cell interactions. Table 2 summarizes the major T‐cell‐focused Nanovial studies and the distinct capabilities demonstrated in each.

TABLE 2.

Progression of Nanovial‐based T‐cell assays from functional phenotyping to TCR and dyad‐resolved analysis.

Study	T‐cell system	Nanovial configuration	Functional readout	Selection and analysis	Downstream linkage	Key contribution
Kizerwetter et al., 2025 [26].	Primary mouse OT‐I CD8 T cells	Anti‐CD45 for cell capture; anti‐CD3/anti‐CD28 or OVA‐loaded H‐2Kᵇ monomers for stimulation; cytokine‐capture antibodies	IFNγ and TNFα secretion; pre‐stimulation surface phenotype	Flow‐cytometric analysis and sorting according to secretion level	Live‐cell release after collagenase treatment; regrowth and phenotypic analysis	Established non‐antigen‐specific and antigen‐specific on‐particle stimulation, multiplexed cytokine capture, viable sorting, and linkage of pre‐existing phenotype to subsequent secretion
Koo et al., 2024 [10].	Human primary CD8 T cells and PBMC‐derived antigen‐specific T cells	pMHC monomers displaying viral or tumor‐associated epitopes; cytokine‐capture antibodies; antigen and secretion barcodes	IFNγ and granzyme B secretion	Binding‐ and secretion‐based FACS using a standard sorter	Paired αβ TCR sequencing, barcode‐based epitope assignment, secretion encoding, and cell regrowth	Integrated antigen recognition, induced secretion, and paired receptor recovery; expanded recovery of functional conventional TCRs, including functional receptors not selected by tetramer staining
van den Dijssel et al., 2025 [9].	Human influenza‐specific CD8 T cells	HLA‐A*02:01–M1₅₈–₆₆ monomer and anti‐IFNγ capture antibody	IFNγ secretion	Nanovial binding and secretion compared directly with tetramer plus intracellular cytokine staining	Index sorting followed by paired TCR sequencing	Independently validated linkage of antigen‐specific binding, secretion, and paired TCR sequence while showing lower total cell recovery than conventional tetramer/ICS analysis under the tested conditions
Soemardy et al., 2026 [25].	Human MAIT and iNKT cells directly from PBMCs	Antigen‐loaded MR1 or CD1d monomers; anti‐IFNγ capture antibody; fluorescent or oligonucleotide particle barcodes	IFNγ secretion and transcriptomic state	Multiplexed MR1/CD1d loading, secretion‐based enrichment using standard FACS	Paired αβ TCR sequencing, SEC‐seq, receptor reconstruction, and in vitro and in vivo validation	Extended functional TCR discovery to non‐classical MR1‐ and CD1d‐restricted repertoires and recovered receptors with validated antitumor activity
Baghdasarian et al., 2026 — Cell–Cell‐seq [28].	Engineered human tumor–T‐cell dyads	Physical confinement of defined cell pairs; cell‐capture chemistry and granzyme B capture reagents	Granzyme B secretion and interaction‐induced transcriptional programs	Standard FACS enrichment of dyads followed by microscopy and sequencing	Dyad‐resolved transcriptomics and computational partner attribution	Extended the platform from ligand‐displayed stimulation to direct target‐cell interactions, enabling analysis of receptor function within defined cell–cell dyads

5.1. Establishing On‐Particle T‐Cell Stimulation and Secretion Analysis

A key technical advance was the integration of peptide–MHC (pMHC) ligands and cytokine‐capture antibodies within the same Nanovial cavity [10]. In a representative implementation, 35 μm PEG‐based Nanovials with biotinylated gelatin localized to the inner cavity were functionalized with pMHC monomers and cytokine‐capture reagents. This configuration allowed the particle to serve simultaneously as an antigen‐binding surface, a stimulatory interface, and a local capture site for secreted proteins. Cells were loaded in well plates, unbound cells were removed by straining, and secretions were accumulated for three hours before fluorescent staining and flow‐cytometric analysis. Under these conditions, recombinant cytokines were detected across an approximately two‐log dynamic range, while interparticle crosstalk remained low, with fewer than 0.04% of empty control Nanovials entering the positive secretion gate.

Antigen‐specific capture and activation were first demonstrated using PBMCs engineered to express the NY‐ESO‐1‐reactive 1G4 TCR. Nanovials displaying HLA‐A*02:01‐restricted NY‐ESO‐1 pMHC selectively retained the engineered cells: approximately 17% of Nanovials contained viable cells, and 93.9% of the bound population expressed the NGFR transduction marker. Cell loading increased with pMHC density, whereas untransduced PBMCs showed little binding across the same conditions. The immobilized ligand also provided productive stimulation. Compared with control Nanovials bearing anti‐CD45, NY‐ESO‐1 pMHC‐functionalized Nanovials induced detectable IFNγ secretion within three hours, with signals one to two orders of magnitude above background.

The assay was then evaluated using five NY‐ESO‐1‐specific TCRs spanning a range of affinities. Antigen‐specific cells were identified by Nanovial binding alone, Nanovial binding combined with IFNγ secretion, or dual‐color tetramer staining. Adding secretion to the Nanovial gate increased the purity of the recovered engineered population, while pMHC‐functionalized Nanovials recovered substantially more cells expressing several receptors with weaker tetramer staining. For example, 1996 cells expressing the 4A2 TCR were recovered using Nanovials compared with 15 by dual‐color tetramer staining, with similarly large differences observed for the 5G6 and 9D2 receptors. These findings were consistent with the high local density of immobilized pMHC increasing effective avidity and enabling retention of receptors that were less efficiently detected by tetramers. Once retained, the cells could be further distinguished according to induced secretion, separating antigen engagement from the functional response that followed.

This combined binding‐and‐secretion logic is particularly relevant for tumor‐associated antigens, for which useful receptors may occupy an intermediate affinity range because the targets are often structurally related to self. In such settings, antigen‐binding and functional assays may identify overlapping but non‐identical receptor populations, making their integration valuable for receptor prioritization.

5.2. Linking Antigen Identity, Secretion, and TCR Sequence

The next advance was to encode antigen identity directly into the discovery workflow. Distinct Nanovial populations were functionalized with HLA‐A*02:01‐restricted pMHCs displaying CMV pp65, CMV IE‐1, or EBV BMLF1 epitopes and labeled with corresponding oligonucleotide barcodes [10]. These populations were pooled and used to screen PBMCs that had been pre‐expanded with the same viral peptides. After cell loading and a three‐hour secretion period, viable CD3⁺CD8⁺ cells associated with IFNγ signal were sorted and introduced into the 10× Genomics Chromium workflow while still attached to their Nanovials. This preserved the linkage among paired TCR sequence, antigen barcode, and secretion phenotype within the same single‐cell assay. Productive paired αβ TCR sequences were recovered at similar frequencies across the Nanovial, tetramer, and CD137‐selected samples, while the Nanovial barcode additionally assigned a dominant target epitope to most productive TCR calls. In contrast to activation‐marker screening with pooled peptides, this allowed receptor sequence and antigen identity to be recovered without a separate deconvolution step.

The reconstructed receptors showed that the three enrichment strategies sampled overlapping but non‐identical repertoires. At a clonotype‐frequency threshold of at least five cells, the Nanovial workflow recovered TCRs directed against all three viral epitopes, whereas the tetramer arm was restricted to the CMV pp65 specificity used for staining. Comparing within this arm, six CMV‐reactive TCRs identified by Nanovials were not recovered in the tetramer‐selected population. Across the Nanovial‐derived candidates tested, 17 TCR combinations induced antigen‐dependent NFAT reporter activity after reconstruction, including receptors reactive to both CMV and EBV epitopes. Importantly, the peptide specificity predicted by the Nanovial barcode was confirmed experimentally: receptors assigned to CMV responded to the corresponding CMV peptide, whereas those assigned to EBV responded to the EBV peptide. When analyzed at the cell‐clonotype level, most clonotypes contained at least one reactive αβ‐chain pairing. Re‐expression in primary T cells further confirmed peptide‐induced IFNγ secretion from the reactive candidates. These results showed that the combined assay could move beyond enrichment to recover experimentally validated receptor–epitope pairs prioritized by an induced secretory response.

The workflow was also evaluated without prior peptide expansion. Freshly thawed PBMCs were screened directly using CMV1 pMHC‐functionalized Nanovials or dual‐color CMV1 tetramers. Nanovials recovered a larger population of CD3⁺CD8⁺ events and identified a subset with detectable IFNγ secretion, whereas tetramer staining recovered a smaller antigen‐binding population. Because the two assays apply different gates—binding alone for tetramers and binding plus induced secretion for Nanovials—the comparison does not represent a direct measure of sensitivity. It nevertheless demonstrated that antigen‐specific, secretion‐based sorting could be performed directly from PBMCs in a single‐day workflow, without a seven‐day pre‐expansion step. This direct‐from‐PBMC strategy established the feasibility of screening low‐frequency antigen‐reactive populations without prolonged pre‐expansion and provided the basis for subsequent application to MR1‐ and CD1d‐restricted T cells [11].

5.3. Multiplexed Antigen and Secretion Encoding With SEC‐Seq

The workflow was next extended to a more challenging cancer‐relevant setting by integrating secretion‐encoded single‐cell sequencing, or SEC‐seq [29]. In this approach, captured secretions are labeled with oligonucleotide‐barcoded detection reagents, allowing secretion magnitude to be quantified alongside paired TCR sequence, antigen barcode, and transcriptomic state in a standard droplet‐based single‐cell sequencing workflow. This converts secretion from a fluorescence‐based sorting parameter into a molecular feature that can be assigned directly to each recovered clonotype.

To search for rare receptors recognizing prostate‐associated antigens and cancer‐enhanced splicing epitopes, ten HLA‐A*02:01‐restricted pMHC Nanovial populations were separately oligonucleotide‐barcoded and pooled. These included PAP14, PAP21, PAP22, PAP23, CTNND1, CLASP1, MEAF6, PXDN, SCAMP3, and TCF12. Granzyme B was used as the functional readout because of its closer relationship to cytotoxic effector activity than cytokine secretion alone. Captured granzyme B was detected both fluorescently for sorting and through an oligonucleotide‐tagged anti‐APC reagent for sequencing‐based quantification. Starting from approximately 20 million donor PBMCs, 698 granzyme B‐positive CD3⁺CD8⁺ Nanovial‐bound cells and 4764 granzyme B‐negative controls were sorted. Sequencing recovered 87 productive paired αβ TCRs from the positive fraction and 570 from the negative fraction. Granzyme B barcode values were substantially higher in the sorted positive population, supporting the use of the molecular barcode as a quantitative measure of secretion.

The secretion barcode then provided a basis for prioritizing clonotypes by associated granzyme B output. Clonotypes were stratified into high‐, medium‐, and low‐secretion groups, and selected candidates were reconstructed for validation. Among the tested receptors, functional TCRs were recovered from the granzyme B‐high and ‐medium groups, whereas none were identified from the low group. The validation rate was 33% among granzyme B‐high clonotypes, compared with 6.7% among medium and 0% among low clonotypes. Although the number of reconstructed candidates was limited, these results indicated that secretion magnitude could enrich for receptors more likely to retain function after transfer into new T cells.

This study therefore added a second level of prioritization to antigen‐specific enrichment. High‐density pMHC presentation enabled capture of candidate antigen‐reactive cells, while secretion encoding ranked the associated receptors according to a directly measured effector output. Rather than treating binding and function as interchangeable, the workflow combined them to recover and prioritize rare receptor–epitope pairs that would otherwise require extensive downstream screening.

5.4. Extending Screening to MR1‐ and CD1d‐Restricted T Cells

The Nanovial framework was subsequently extended from conventional peptide–MHC screening to unconventional T‐cell populations restricted by MR1 and CD1d [11]. MAIT and iNKT cells recognize non‐peptide antigens presented by these non‐classical antigen‐presenting molecules and are attractive therapeutic candidates because of their rapid effector responses and reduced dependence on classical HLA matching. Their relative scarcity in PBMCs, however, and the potential for expansion or pre‐enrichment to alter clonal representation make direct receptor discovery challenging.

The unconventional T‐cell workflow retained the same basic design used for pMHC‐directed screening. Nanovials were functionalized with 5‐OP‐RU‐loaded MR1 or PBS57‐loaded CD1d together with anti‐IFNγ capture antibodies, allowing the displayed ligand to mediate selective binding and stimulation while the cavity accumulated induced secretion. Ligand titration established conditions that supported selective loading of MAIT and iNKT cells, with negligible binding in the absence of MR1 or CD1d or when control T cells lacking the corresponding receptors were used. Ligand‐bearing Nanovials also induced antigen‐dependent IFNγ secretion above the low background observed with anti‐CD45 control particles. These experiments established that the platform could be adapted from classical pMHC recognition to non‐peptide antigen systems without changing its underlying assay logic.

MR1‐ and CD1d‐functionalized Nanovials were then pooled and applied directly to donor PBMCs. Fluorescent or oligonucleotide barcodes distinguished the two ligand classes within the same screen, while secretion‐based sorting enriched responding cells. The relative abundance and IFNγ response of MR1‐ and CD1d‐reactive populations varied among donors, illustrating that unconventional repertoires differ not only in clonotype composition but also in the frequency and responsiveness of ligand‐reactive cells.

Integration with SEC‐seq linked four measurements from the same cell: paired αβ TCR sequence, transcriptome, Nanovial‐encoded ligand identity, and IFNγ secretion. The workflow progressively narrowed the input population from 10 million PBMCs to ligand‐bound secretors and finally to prioritized paired receptor sequences selected for reconstruction (Figure 3). Recovered cells showed the expected invariant α‐chain usage for MAIT and iNKT populations while retaining diversity in their paired β chains. Secretion barcodes were enriched in cells sorted through the IFNγ‐positive gate, and cells with the strongest secretion signals also expressed broader effector programs including GZMB and PRF1. The assay therefore provided both receptor recovery and a means of prioritizing clonotypes according to measured secretion and transcriptional state.

FIGURE 3.

Funnel view of function‐first discovery of unconventional TCRs using MR1‐ and CD1d‐functionalized Nanovials. Starting from approximately 10 million donor PBMCs, MR1‐ and CD1d‐functionalized Nanovials were used to selectively capture and activate unconventional T cells, enriching first for cell‐loaded Nanovials, then for IFNγ‐positive events, and finally for sorted cells suitable for sequencing. The workflow yielded roughly 180 recovered sequences, including paired receptors from 68 MAIT and 107 iNKT cells, followed by prioritization of a small number of top candidate hits for downstream validation. This schematic highlights the progressive narrowing of the candidate space while preserving functional and molecular information at each stage.

Five MAIT TCRs spanning a broad range of secretion signals were reconstructed for validation. All five conferred antigen‐dependent cytokine secretion and cytotoxicity after re‐expression, despite substantial diversity in their paired β chains and CDR3β sequences. Selected MAIT TCR‐engineered T cells mediated antigen‐dependent killing across several MR1‐positive tumor models and maintained activity during longer‐term coculture. The top candidates also supported tumor localization and measurable antitumor activity in vivo. The CD1d branch similarly yielded an iNKT receptor that induced IFNγ secretion and killing of CD1d‐expressing tumor cells. These studies showed that Nanovial‐based screening could begin with rare unconventional T‐cell populations in PBMCs and recover receptor candidates that retained functional activity after reconstruction.

Altogether, the studies discussed in this section established that ligand‐defined Nanovial screening could identify functionally active receptors across both classical and non‐classical antigen systems. They also highlighted the next limitation: stimulation by an immobilized ligand still does not reproduce the full cellular context in which a TCR encounters and alters a target cell.

6. From Ligand‐Defined Screening to Dyad‐Resolved TCR Discovery

T‐cell responses in vivo are shaped by a broad cellular context that includes antigen density and processing, adhesion, costimulation, inhibitory signaling, target‐cell state, and reciprocal responses between the interacting cells. A logical extension of functional repertoire analysis is therefore to evaluate TCRs during direct encounters with living target cells.

Cell–Cell‐seq provides an initial framework for this transition (Figure 4) [28]. Nanovials are used to confine defined cell pairs, coordinate the onset and duration of contact, protect fragile conjugates during handling and sorting, and introduce intact dyads into droplet‐based sequencing workflows. In tumor–T‐cell models, this approach has enabled profiling of interaction‐induced transcriptional programs that were obscured in conventional co‐culture or dissociated single‐cell measurements. The present implementation is primarily a method for analyzing heterogeneous cell–cell interactions and, although compatible, has not yet been applied to de novo receptor discovery.

FIGURE 4.

Nanovial‐based dyad workflow extends function‐first receptor profiling to direct cell–cell interaction assays. Nanovials can be used to confine defined cell pairs, here an antigen‐presenting cell (APC) or target cell together with a T cell, creating a microscale assay compartment in which receptor engagement, secretion, and transcriptional state can be measured in the context of direct cell–cell contact. In the workflow shown, Nanovials co‐load the two cells, allow local accumulation of secreted factors, and are then washed, stained, and sorted by flow cytometry before downstream single‐cell sequencing. This enables linked recovery of TCR sequence, dyad‐level transcriptomes, and functional readouts from interacting cell pairs. Representative panels show brightfield and fluorescence images of tumor (PC3‐A2‐PAP)–T cell dyads in Nanovials, flow‐cytometric gating of granzyme B‐positive events, UMAP visualization of dyad‐associated transcriptional states, and correlation analysis revealing coordinated gene‐expression programs across interacting cells. Green fluorescence corresponds to green fluorescent protein, magenta fluorescence to anti‐CD45 labeling of T cells, and blue fluorescence to granzyme B captured on Nanovials. Adapted from reference [28] with permission. Together, this workflow illustrates how Nanovials extend function‐first discovery from single‐cell ligand screens to dyad‐resolved assays that place TCR function into the context of direct target‐cell interaction.

A dyad‐based discovery workflow could begin with heterogeneous target cells paired with donor‐derived or library‐derived T cells. Dyads could then be enriched using readouts such as cytokine or granzyme secretion, target‐cell death, or transcriptional changes in either partner. This would allow receptor discovery to begin from a biological interaction—for example, recognition of a tumor subtype, infected cell, or stressed tissue state—without requiring that the relevant antigen be fully specified in advance. TCRs recovered from responding cells could subsequently be reconstructed and their antigen specificity resolved through targeted validation.

Target identity could also be encoded directly within living‐cell libraries. Target cells expressing barcoded peptide or minigene constructs could each present a defined candidate antigen. Following dyad formation and functional enrichment, sequencing could recover the target‐cell barcode together with the responding T cell's receptor and transcriptional state. This would extend the logic of barcoded pMHC Nanovials from purified ligand presentation to physiologically processed antigens on living cells. Related approaches could use genetic or metabolic perturbations to generate libraries of target‐cell states, enabling T cells to be screened against variation in antigen processing, metabolism, or non‐classical ligand presentation.

Dyad‐resolved assays could also broaden how receptor function is defined. Rather than ranking TCRs only by secretion from the T cell, receptors could be evaluated according to the coordinated state transition produced across both cells. Relevant outcomes might include induction or suppression (for Treg cells) of target‐cell stress or death programs, sustained T‐cell activation, favorable chemokine exchange, or escape from suppressive or exhaustion signaling. Such measurements would define receptor function through the consequences of the interaction rather than through a single output from one partner.

Substantial technical challenges remain. Dyad occupancy and interaction timing must be controlled, target‐cell libraries must preserve physiological antigen processing, and sequencing workflows must retain linkage among receptor sequence, both cellular transcriptomes, target identity, and functional readouts. Composite dyad transcriptomes also require computational methods to distinguish interaction‐induced programs from the additive expression of the two cells. Nanovials provide a useful physical foundation for addressing these challenges by stabilizing defined pairs while maintaining compatibility with flow‐based enrichment and single‐cell genomics [28]. Dyad‐resolved TCR discovery therefore remains an emerging direction, but one that could extend functional repertoire analysis from ligand recognition to the reciprocal biology of living cell–cell interactions.

7. Conclusions and Outlook

TCR repertoires are likely to be most informative for therapeutic discovery when receptor identity is linked to the functions it enables in living cells. Nanovial‐based screening has helped make this linkage experimentally accessible by combining antigen recognition, secretion measurement, viable‐cell recovery, paired receptor sequencing, and transcriptomic analysis within scalable workflows. In conventional pMHC‐restricted systems, these approaches enabled recovery and functional prioritization of viral‐ and tumor‐reactive receptors. Their extension to MR1‐ and CD1d‐restricted populations showed that the same framework could be applied to non‐classical antigen systems and rare unconventional T cells. SEC‐seq added a quantitative layer by encoding secretion magnitude alongside receptor sequence and cellular state, extending the output from a binary list of responsive clonotypes toward a multidimensional description of receptor‐associated function.

The next opportunity is to generate datasets that more directly capture how receptor sequence gives rise to cellular behavior in context. Dyad‐resolved assays offer one route by placing TCR recognition within interactions with living target cells, where activation is shaped by receptor–ligand binding but also by antigen processing, ligand density, adhesion, costimulation, inhibitory signaling, metabolism, and reciprocal target‐cell responses. With encoded target‐cell libraries, such experiments could link TCR sequence, target identity or perturbation, secretory output, transcriptional state, and the resulting dyadic response within the same measurement framework. These data would support a richer definition of receptor function based on the consequences of a cell–cell encounter rather than on binding or one‐sided activation alone.

This type of data generation is increasingly relevant as biological research becomes more closely coupled to machine learning and predictive modeling [30]. Some of the most successful models in biology have benefited from exceptionally large sequence, structure, and binding datasets. Comparable datasets that connect molecular sequence to induced cellular function remain much less developed, particularly for immune recognition, where behavior depends strongly on the state of both the responding and target cells. Functional TCR atlases generated through ligand‐defined and dyad‐resolved screening could therefore complement existing receptor catalogs by providing experimentally measured examples of what defined receptors cause cells to do under specified conditions. Such datasets may be useful for developing models that predict receptor specificity and context dependence of the resulting response.

More broadly, the opportunity may be not simply to measure known TCR–ligand relationships more efficiently, but to recover the biological context that reductionist binding assays necessarily leave behind. When a defined pMHC is presented to a TCR, the experiment asks a focused and important question, but one whose relevant molecular interaction has already been specified. Direct cell–cell measurements create a less constrained discovery space. The state of the target cell, the molecules it presents, its metabolic and stress programs, co‐stimulatory and inhibitory signals, and the reciprocal response induced by the T cell can all become part of the measured phenotype. Such experiments therefore have the potential to reveal determinants of T‐cell recognition and function that were not anticipated when the assay was designed. The growing appreciation of unconventional T cells provides one precedent: broadening attention beyond classical peptide–MHC recognition has exposed metabolite‐ and lipid‐reactive repertoires with distinct biology and therapeutic potential. Cell‐context‐resolved screening may uncover analogous surprises that do not fit neatly within our current definitions of antigen recognition.

From this perspective, function‐first TCR discovery is both a practical route to therapeutic receptors and a means of exploring what information actually governs productive immune interactions. Nanovial‐based approaches provide a way to progressively reconnect levels of biology that have often been separated experimentally—from receptor sequence and ligand recognition to secretion and cell state, and ultimately to the reciprocal behavior of interacting cells. The most valuable datasets may therefore be those that preserve enough of this context to reveal not only which receptors perform best against targets we already understand but also new rules of recognition and response that we did not know to look for.

Funding

This work was supported by Silicon Valley Community Foundation, 2023‐332384. Japan Science and Technology Agency, EXPERT‐J.

Conflicts of Interest

The author and the Regents of the University of California have financial interests in Partillion Bioscience, which is commercializing Nanovial technology.

Data Availability Statement

Data sharing not applicable to this article as no datasets were generated or analysed during the current study.