Ready-to-load MHC-I nanoparticles for high-throughput T cell screening studies Authors: Hoang Anh T. Phan, Soomin Eom, Daniel Hwang, Michael C. Young, Shirley M. Sun, Molly C. Erdman, Yi Sun, Leena Mallik, Photis Rotsides, Mark M. Painter, E. John Wherry, Dimitri S. Monos, Nikolaos G. Sgourakis Published: 2026 ABSTRACT MHC-I proteins present epitopic peptides to CD8+ T cells to elicit multifaceted adaptive immune responses. The affinity and avidity of interactions between peptide-MHC molecules and their cognate T cell receptors (TCRs) are fundamental parameters that contribute to the induction of activated or anergic T cell states. Here, we present a multivalent, loadable platform, VLP-Open HLA, featuring a virus-like particle (VLP) which can display up to 60 peptide/HLA [human leukocyte antigen (HLA)] molecules. VLP-Open HLA can be used to stain antigen-specific CD8+ T cells, providing a screening tool for orphan TCRs. When combined with costimulation, VLP-Open HLA acts as an artificial antigen-presenting platform to induce antigen-specific T cell activation and expansion, providing a facile alternative to autologous antigen-presenting cells. Our platform can be adapted to encompass multiple HLA allotypes and costimulatory molecules as mosaic nanoparticles, enabling a range of applications in experimental immunology. INTRODUCTION Major histocompatibility complex class I (MHC-I) molecules play a central role in adaptive immunity of jawed vertebrates. MHC-I proteins comprise a polymorphic heavy chain, an invariable light chain β2 microglobulin (β2m), and an epitopic peptide of 8 to 15 amino acids in length which enables immunosurveillance of the intracellular proteome (1, 2). MHC-I molecules are assembled in the endoplasmic reticulum and loaded with a pool of up to 105 immunogenic peptides for trafficking to the cell surface (3). In the human population, the highly polymorphic human leukocyte antigen (HLA) proteins (the MHC in humans comprising >20,000 allomorphs) present distinct repertoires of endogenous peptides, thereby ensuring species adaptability to emerging intracellular pathogens (4). Specific engagement between an antigen-presenting cell and a CD8+ T cell is conferred by the intermediate affinity (micromolar range KD) interaction between a peptide-loaded MHC (pMHC) molecule and a clonotypic αβ T cell receptor (TCR), expressed on the T cell surface (5). Effective T cell activation and proliferation toward a sustained immune response are dependent on engagement of costimulatory receptors and adhesion molecules and necessitate an appropriate TCR:pMHC interaction to enact CD45 phosphatase exclusion and ultimately TCR-CD3 complex triggering via distinct mechanisms (6). Therefore, because of short half-life of the TCR:pMHC interaction, soluble monomeric pMHC-I molecules are insufficient to detect or to efficiently activate their antigen-specific T cells, and the immunology community has relied on multimerized MHC-I conjugates to identify cross-reactive T cells from polyclonal repertoires and to monitor responses in various settings (7). While several approaches have been developed to generate libraries of MHC-I molecules encompassing hundreds of different epitopic peptides (8–12), multimerized MHC-I molecules insufficiently stimulate T cells responses and are therefore limited to being used as probing reagents. There is thus a need for systems that can rapidly and accurately screen epitopic peptides that can elicit specific T cell responses for immunological studies and therapeutic development. In general, antigen-specific T cell stimulation relies on monocyte-derived dendritic cells (moDCs) generated from large numbers of peripheral blood mononuclear cells (PBMCs). The generation of moDCs typically requires at least 1 week of in vitro culture, and these cells are frequently irradiated to support T cell expansion. Because effective T cell stimulation and expansion often require several additional weeks, repeated rounds of moDC generation are commonly needed to obtain sufficient numbers of antigen-specific T cells (13, 14). In previous attempts to overcome this challenge, Schneck and colleagues (15–18) have developed nanoscale artificial antigen-presenting cells (aAPCs) for enrichment and expansion of antigen-specific T cells including iron-dextran paramagnetic nanoparticles (50 to 100 nm in diameter), avidin-coated quantum dot nanocrystals (30 nm in diameter), and larger size aAPCs (300 nm in diameter), as well as magnetic nanoparticle–based for detection of antigen-specific T cells. aAPCs can activate and rapidly expand antigen-specific T cells from patients with melanoma (19, 20). pMHC-based aAPCs can expand T cells that are specific for cancer and autoimmune epitopes (21–23). Other multimerization approaches for T cell activation include functionalization of Strep-Tactin (mutein of streptavidin) with anti-CD3 and anti-CD28 antibody fragments (Expamers) (24). Last, CombiCells, a synthetic biology system, enables combinatorial display of cell surface ligands, enabling studies of T cell antigen sensitivity as well as modulation by costimulation or coinhibition receptors (25). Recently, protein nanoparticles and cage-based protein assemblies have been gaining attention in the context of vaccine design (26, 27). Virus-like particles (VLPs) can offer multivalency and a symmetry that has been shown to be beneficial for antigen presentation in the context of vaccine development (28). Built on the 60-subunit nanocage based on 2-keto-3-deoxy-phosphogluconate aldolase developed by Baker et al., Howarth and colleagues (28–32) developed VLPs displaying the severe acute respiratory syndrome coronavirus 2 spike protein receptor–binding domain using SpyCatcher/SpyTag chemistry. The same group established the SpyCatcher003 scaffold that enabled efficient VLP conjugation and enhanced stability. To functionalize the nanoparticles, SpyCatcher/SpyTag chemistry offers irreversible isopeptide bond formation via site-specific ligation between a lysine residue on the SpyCatcher protein and an aspartic acid on its SpyTag protein partner (33, 34). In this study, we aimed to develop a universal, loadable VLP-based platform to screen for peptide/MHC-I–restricted T cell responses. Several methods have been proposed for generating loadable pMHC molecules, including ultraviolet (UV) irradiation of a photocleavable conditional ligand, disulfide bond direct stabilization of the MHC-I peptide–binding groove, elevated temperature treatment, or multivalent spheromer systems (8, 35–40). We have established a high-throughput method for chaperone-mediated exchange of a placeholder peptide for high-affinity peptides of choice (9). Recently, we used a structure-guided approach to developing conformationally stabilized, open, or ready-to-load MHC-I molecules (41). Here, we integrate our open MHC-I system with the enhanced avidity provided by VLPs and SpyCatcher/SpyTag-mediated multimerization to develop a loadable open MHC-I nanoparticle platform that is compatible with different HLA allomorphs (VLP-Open HLA). Nanoparticles displaying open MHC-I molecules allow efficient in vitro peptide exchange, serving as stable antigen-presenting platforms. As a proof of concept, we demonstrate that our VLP-Open HLA (Open HLA-A*02:01/NY-ESO-1) can be fluorescently tagged to specifically stain primary T cells transduced with a TCR that recognizes the NY-ESO-1 peptide antigen (1G4) (42) and cytomegalovirus (CMV)+ PBMCs. We further demonstrate that our system can serve as an artificial antigen presentation platform to specifically activate T cells in vitro. Our platform provides a minimalistic protein-based system to systematically screen and evaluate T cell responses against a multitude of relevant peptide/HLA combinations, enabling a range of applications in research and clinical settings. RESULTS Design and characterization of multivalent nanoparticles displaying loadable HLA-I proteins To generate VLP-Open HLA nanoparticles, we displayed SpyTagged open MHC-I on VLP (SpyCatcher-mi3). First, the heavy chain ectodomain of Open HLA-A*02:01 was fused to SpyTag003 via a Gly-rich linker (see Supplementary Text of table S2). Our engineered Open HLA-A*02:01 features an interchain disulfide bond between the heavy (G120C) and light chain β2m (H31C) (Fig. 1, A and B), as previously described (41). Compared to native molecules, the engineered open MHC-I shows enhanced stability and rapid loading of synthetic peptides through the stabilization of a peptide-receptive conformation. To analyze interactions between the NY-ESO-1–specific TCR 1G4 (42) with either wild-type or Open HLA proteins presenting the heteroclitic NY-ESO-1C9V peptide (SLLMWITQV), biotinylated pMHC-I molecules were immobilized on separate flow channels of streptavidin-coated chips. After immobilization, various concentrations of NY-ESO-1–specific TCR were introduced to monitor their association and dissociation behaviors and to determine binding affinities. Surface plasmon resonance (SPR) results indicated that the NY-ESO-1–specific TCR exhibits nearly identical binding affinities and kinetic parameters for both wild-type and Open HLA-A*02:01/NY-ESO-1C9V molecules (Fig. 1C). Fitted KD values were 5.00 ± 0.03 μM for wild-type and 4.98 ± 0.02 μM for the Open HLA molecules. These data suggest that the engineered interchain disulfide bond in the Open HLA conformation does not introduce a marked effect on TCR recognition, and the Open HLA maintains effective binding to its cognate TCR (Fig. 1C). Nevertheless, it is conceivable that Cys mutations introduced in the Open HLA protein may affect T cell recognition in some peptide-MHC-I complexes, for example, by introducing changes at the heavy/light chain interface that could allosterically affect the peptide conformation (43). Design, production, and biophysical characterization of the VLP-Open HLA system. (A) Schematic of the design: A SpyTag Open HLA molecule is ligated to VLP-SpyCatcher to form VLP-Open HLA molecule. (B) Model of open HLA-A*02:01 with an engineered disulfide bond as previously described by our group (41). (C) Representative SPR sensorgrams for varying concentrations (0.5 to 50 μM) of NY-ESO-1 TCR flowed over a streptavidin chip coupled with biotinylated wild-type or Open HLA-A*02:01/SLLMWITQV (NY-ESO-1C9V) on the ligand channels from FC2 to FC4 and biotinylated wild-type or Open HLA-A*02:01/SLLMAITQV (NY-ESO-1W5A) on the reference channel FC1. The concentrations of analyte on the sensorgrams are noted. KD values were determined by saturation fit. Data are mean ± σ from n=3 technical replicates. (D) SEC traces of the SpyTagged G120C/H31C “Open HLA-A*02:01” molecules refolded with NY-ESO-1CV9 (pink), NY-ESO-1W5A (black), and NY-ESO-1gTAX (green). Representative SDS–polyacrylamide gel electrophoresis (SDS-PAGE) analysis of SpyTag fused Open HLA-A*02:01. (E) Ligation between SpyCatcher-mi3 (VLP) and SpyTag Open HLA-A*02:01 molecules refolded with different peptides as validated with SDS-PAGE gel. The gels showed pre- and postligation products. (F) Dynamic Light Scattering (DLS) scans of wild-type of VLP (SpyCatcher-mi3), VLP-Open A*02:01/NY-ESO-1C9V, and VLP-Open A*02:01/NY-ESO-1W5A. (G) Representative mass photometry data for VLP-Open A*02:01/NY-ESO-1C9V, showing successful ligation to most of the 60 sites of VLP. (H) Example from a cryo–electron micrograph of VLP-Open A*02:01/NY-ESO-1C9V. RU, resonance units; DTT, dithiothreitol. mAu, milli-Absorbance Units. These results establish a ready-to-load MHC-I system suitable for probing T cell activity and low-affinity ligand screening (41). We refolded SpyTag fused Open HLA-A*02:01 with the high-affinity peptides NY-ESO-1C9V (SLLMWITQV), the mutated decoy peptide NY-ESO-1W5A (SLLMAITQV), or a moderate-affinity placeholder peptide gTAX (LFGYPVYV) (Fig. 1D). We found that adding the SpyTag does not compromise the thermal stability of the protein, as evidenced by the high Tm values of the SpyTagged Open HLA-A*02:01 refolded with different peptides between 51.8° and 58.9°C using differential scanning fluorimetry (DSF) (Supplementary Text of fig. S2). We chose the protein-based nanoparticle platform VLP due to its versatility, ease of assembly, stability, and antigen display geometry (28, 44). Specifically, we focused on the SpyCatcher-mi3 scaffold (with 60 binding sites on VLP) that enabled efficient VLP conjugation and enhanced stability and can be expressed in Escherichia coli (see Supplementary Text of fig. S1 for expression and purification) (28, 32, 44). For the ligation reaction, we optimized the molar ratio of SpyCatcher-mi3 and Open HLA-A*02:01-SpyTag (see Supplementary Text of table S3 and fig. S3). Our results suggested that a ratio of 1:180, with final concentrations such as 50 nM SpyCatcher-mi3 particle and 9 μM Open HLA-SpyTag resulted in an optimal ligation efficiency. We then subsequently ligated SpyTag Open HLA-A*02:01 refolded with NY-ESO-1C9V, NY-ESO-1W5A, or gTAX peptide; a band shift on SDS–polyacrylamide gel electrophoresis (SDS-PAGE) gel confirmed the ligation via covalent bond (Fig. 1E). Unconjugated Open HLA molecules were removed using a high molecular weight cutoff (100 kDa MWCO) centrifugal spin column before characterization and subsequent experiments. To minimize the formation of unstable or aggregated open HLA molecules, we consistently prepared fresh open HLA proteins for each experiment. The spontaneous ligation via SpyCatcher/SpyTag chemistry resulted in the formation of VLP-Open HLA protein nanoparticle, as further confirmed by dynamic light scattering (DLS). The size of our VLP only particle (RH = 16.02 ± 0.12 nm) was consistent with previous reports [RH = 18.6 ± 4.8 nm from Tan et al. (32)] (Fig. 1F). The increase in particle size upon ligation with Open HLA-A*02:01/NY-ESO-1C9V or open HLA-A*02:01/NY-ESO-1W5A was consistent with the size of the Open HLA-I molecule (~47 kDa and 411 amino acids). We further characterized the particles using mass photometry, and the masses observed were consistent with the expected values for the unconjugated and conjugated particles (Fig. 1G and Supplementary Text of fig. S4). VLP-Open A*02:01/NY-ESO-1C9V showed a mass distribution for the conjugated particle at 4126 ± 402 kDa, indicating an average occupancy of approximately 46 of 60 sites on the VLP by pHLA-I molecules (Fig. 1G). Cryo–electron microscopy (cryo-EM) further confirmed the presence of a cage-like structural arrangement (Fig. 1H). Together, these studies support multivalent presentation of open MHC-I molecules on the surface of VLP nanoparticles using in vitro assembly from recombinant protein subunits expressed in E. coli. VLP-Open HLA nanoparticles can be loaded with exogenous peptides in vitro We previously established that our engineered open MHC-I design enables enhanced peptide exchange of placeholder peptides across polymorphic HLA molecules (41). We next examined whether open HLA (open HLA-A*02:01) ligated to VLPs retain their enhanced peptide exchange properties, to evaluate its potential as a screening system for immunogenic peptides. Open HLA-A*02:01 was refolded with the placeholder peptide, gTAX, before conjugation to the VLP scaffold. Via fluorescence polarization (FP), we observed rapid peptide exchange of gTAX for the labeled peptide TAMRATAX9 (TAMRA-KLFGYPVYV) on VLP-Open HLA (Fig. 2, A and B). The association of fluorescently labeled peptides was consistent across independent replicates of VLP sample preparation (Supplementary Text of fig. S5). Peptide loading was dependent on the nanoparticle concentration, showing an overall similar profile as observed in our previous study using soluble MHC-I molecules. We observed complete peptide exchange within 4 hours at room temperature, compared to minimal exchange using wild-type HLA-A*02:01 (41). In addition, an FP competition assay demonstrated that increasing concentrations of unlabeled high-affinity TAX9 peptide progressively displaced the TAMRA-TAX9 probe from the peptide binding groove of VLP-Open HLA-A*02:01 (Supplementary Text of fig. S6). Therefore, our VLP-Open HLA-A*02:01/gTAX design can enable efficient peptide exchange in vitro. VLP-Open HLA can capture and display exogenous peptides. (A) Principle of peptide exchange on VLP-Open HLA system. Peptide exchange can take place at room temperature for 1 hour and then overnight at 4°C (procedure used for subsequent immunological assays). (B) Loading and binding of TAMRATAX9 by VLP-Open A*02:01/gTAX monitored at room temperature. Peptide exchange was measured by FP (mP) of 40 nM TAMRATAX9 as a function of the VLP-Open A*02:01/gTAX concentrations. Individual traces were fit to an exponential association model. Data shown are mean values from n = 3 technical replicates. (C and D) Representative LC-MS analysis of peptide bound to VLP-Open A*02:01; the peptide is from the TCGA peptide library. VLP-Open A*02:01/gTAX can load and capture different antigenic peptides. (C) represents a total ion chromatogram of the sample, and (D) (red box) shows extracted m/z region of interest. RT and r.t., room temperature. Furthermore, we conducted bulk peptide exchange followed upon by mass spectrometry experiments to demonstrate that our platform can be used to identify potentially immunogenic peptides present in biological samples. As a proof of concept, VLP-Open HLA-A*02:01/gTAX was exchanged with either individual peptides or a mixture of 50 high-affinity peptides derived from the Cancer Genome Atlas (TCGA), previously validated by our group (41) (for the peptide information, see Supplementary Text table S4). Excess peptides were washed using spin columns to retain bound peptides; a negative control with peptide only (no VLP-Open HLA particle) and a negative control with VLP-Open HLA particle in the presence of an irrelevant peptide were included to ensure that our method efficiently washed out the excess peptide (Supplementary Text of figs. S7 and S8). Captured peptides were identified using liquid chromatography–mass spectrometry (LC-MS), whereas unbound placeholder peptides (gTAX) were not detected by mass spectrometry due to the washing step. As expected, we found that the predicted high-affinity peptides were retained on the VLP-Open HLA nanoparticle and readily identified in our data (Fig. 2, C and D). Additional mass spectrometry data from incubation of VLP-Open HLA-A*02:01/gTAX with individual or mixture of peptides from the TCGA epitope library are shown in the Supplementary Materials, using two complementary techniques—LC-MS (Supplementary Text of figs. S9 to S12) and matrix-assisted laser desorption/ionization time-of-flight mass spectrometer (MALDI-TOF MS) (fig. S14). Specific detection of CD8+ T cells depends on the density of peptide/HLA-I ligands on VLPs Since VLPs have spherical and hollow structures with two distinct phases (i.e., the interior and exterior surface) (45), we simultaneously functionalized both phases to enable antigen-specific T cell staining. To enable quantitative flow cytometry analysis, the interior of SpyCatcher-VLPs was labeled with Alexa Fluor 647 (AF647). Each SpyCatcher-VLP subunit contains cysteine residues on the interior surface, allowing site-specific conjugation of AF647 maleimide (46). After the removal of free dye molecules by dialysis, UV/Visible (Vis) spectrometry exhibited an average labeling efficiency of 0.41 AF647 molecules per SpyCatcher-VLP subunit (Fig. 3A). These cysteine residues may form a disulfide bond, which could influence labeling efficiency. In addition to conjugating fluorescent dyes to the VLP interior, Open HLA ligands were displayed on the exterior surface of SpyCatcher-VLPs using the SpyCatcher/SpyTag protein ligation system, allowing for antigen-specific T cell staining. When SpyCatcher-VLPs were mixed with Open HLA-SpyTag, a new band corresponding to the ligated product (~81 kDa) was observed by SDS-PAGE without any aggregation of VLPs (Fig. 3B). Because SpyCatcher/SpyTag ligation occurs in a stoichiometric manner, the number of displayed Open HLA ligands per VLP was precisely controlled by adjusting the input molar ratios. Densitometric analysis of SDS-PAGE indicated that each SpyCatcher-VLP particle (60 subunits) was ligated with approximately 13, 17, 18, 21, 26, 29, 37, 49, or 52 SpyTag fused Open HLA-I molecules on average (Fig. 3C). Optimization of Open HLA ligand density on VLPs enables antigen-specific T cell staining. (A) SDS-PAGE analyses of the AF647-conjugated SpyCatcher-VLP (SC-VLP). Protein bands are detected with Coomassie Blue (CS) and fluorescence emission from the conjugated dye, as indicated. (B) SDS-PAGE of ligation resultants between SpyCatcher-VLPs (SC-VLP) and SpyTag fused Open HLA-A*02:01 molecules (Open A02-ST) at various molar ratios. Ligated products (blue arrow), SpyTag fused Open HLA-A*02:01 ligands (red arrow), and SpyCatcher-VLP subunits (green arrow). (C) Band intensity analyses of the SDS-PAGE using ImageJ at various molar ratios. (D) Representative flow cytometry histograms of transduced 1G4 T cells stained with AF647-conjugated VLP and tetramer displaying different peptide/HLAs. A02/peptide indicates Open HLA-A*02:01 loaded with the peptide. Mean fluorescence intensity (MFI) of on-target (NY-ESO-1C9V) or off-target (CMVpp65) displaying (E) VLP Low, (F) VLP High, and (G) tetramer. (H) Signal-to-noise (S/N) ratios calculated as the ratio of on-target to off-target of MFI values for VLP/A02 Low (green square), VLP/A02 High (purple circle), and Tetramers/A02 (triangle cyan). (I) Schematic view of ligand number–dependent staining efficiency and specificity. (J) Representative flow cytometry plots of CMVpp65-specific CD8+ T cells from primary human PBMCs stained with AF647-conjugated VLPs or tetramers displaying different peptides. (K) Frequencies of CMVpp65-specific CD8+ T cell from three CMV-seropositive donors stained with AF647-conjugated VLPs or tetramers displaying different peptides (NY-ESO-1C9V, gTAX, CMVpp65, and gTAX↔CMVpp65). All data shown here are presented as the mean ± SD (n = 3 per group) . Statistical significance was determined by one-way analysis of variance (ANOVA); ns, not significant; ****P < 0.0001. To evaluate ligand number–dependent staining efficiency and specificity, AF647-labeled VLPs displaying either a low copy number (22% occupancy, 13 of 60) or a high copy number (82% occupancy, 49 of 60) of Open HLA ligands were prepared for subsequent experiments. Since transduced T cells express the 1G4 TCR, NY-ESO-1C9V and CMVpp65 peptides were used as on- and off-target control, respectively (Fig. 3D). Flow cytometry analysis demonstrated that AF647 fluorescence signals increased for both low and high numbers of Open HLA-A*02:01/NY-ESO-1C9V displaying VLPs. In contrast, VLPs displaying a high number of Open HLA-A*02:01/CMVpp65 ligands exhibited increased AF647 signals, indicating nonspecific binding. Notably, the wild-type VLP group showed no detectable staining (Fig. 3D and Supplementary Text of fig. S15). For quantitative comparison, mean fluorescence intensities (MFIs) were analyzed across a wide range of VLP concentrations. While on-target MFI increased with VLP concentration for both low and high (Fig. 3, E and F), VLPs displaying a high number of Open HLA-A*02:01/CMVpp65 ligands showed increasing off-target staining at higher VLP concentrations (Fig. 3F), likely due to background binding to the HLA-I backbone under high avidity conditions and/or engagement through the CD8 coreceptor on 1G4 T cells. When signal-to-noise (S/N) ratios, defined as the ratio of on-target to off-target, were calculated, VLP/A02 Low (VLP displaying a low number of Open HLA-A*02:01) exhibited gradual increase of S/N compared to VLP/A02 High (VLP displaying a high number of Open HLA-A*02:01) (Fig. 3H), indicating that although increasing ligand amounts enhances avidity, it also promotes peptide-independent interactions that reduce staining specificity (Fig. 3I). We next compared the efficiency of staining 1G4 T cells with our optimized VLP/A02 Low versus tetramerized peptide/HLA. To perform a consistent comparison, we prepared refolded Open HLA-A*02:01 tetramers conjugated using the same fluorophore (AF647). As expected, Open HLA-A*02:01/NY-ESO-1C9V tetramers specifically stained 1G4 T cells, whereas Open HLA-A*02:01/CMVpp65 tetramers showed no detectable nonspecific binding. This is in agreement with our previous results which have shown that Open HLA-A*02:01/NY-ESO-1C9V tetramers perform very similarly, in terms of 1G4 T cell staining efficiency and background to tetramerized WT HLA-A*02:01/NY-ESO-1C9V (Fig. 3D) (41). MFI analysis across a wide concentration range of tetramers exhibited on-target tetramer signals increased with concentration, with limited nonspecific binding at higher concentrations (Fig. 3G). The calculated S/N ratios of Open HLA-A*02:01/tetramers (Tetramer/A02) were comparable to those obtained with VLP/A02 Low (Fig. 3H). These results suggest that displaying approximately 13 Open HLA molecules per VLP is sufficient to achieve selective target cell staining at comparable levels to tetramers (Fig. 3I). We further examined the utility of our VLP system to identify antigen-specific T cells present in PBMCs obtained from three CMV-seropositive donors (Fig. 3J). PBMCs were stained with AF647-labeled VLPs or tetramers for direct comparison. The VLPs or tetramers loaded with CMVpp65 peptide was used to identify antigen-specific CD8+ T cells, while the ones with NY-ESO-1C9V peptide served as a negative control. To further evaluate the performance of VLPs loaded with the on-target peptide, we used our established Open HLA-A*02:01/gTAX system and performed an exchange reaction. Across all three donors, CMVpp65 peptide containing VLPs or tetramers showed higher frequencies of CMVpp65+ CD8+ T cells compared to the control groups. Open HLA-A*02:01/CMVpp65 VLPs and Open HLA-A*02:01/gTAX↔CMVpp65 exchanged VLPs stained CMVpp65-specific T cell populations at comparable levels to Open HLA-A*02:01/CMVpp65 tetramers (Fig. 3K and Supplementary Text of fig. S16). These data collectively demonstrate the practical utility of our optimized, loadable VLP system for high-throughput T cell screening studies. On the basis of these results, we focused on VLPs displaying a low number of peptide/HLA-I ligands to evaluate their capacity for activating antigen-specific T cells in vitro. The VLP-Open HLA platform can induce potent and antigen-specific T cell activation To determine whether the antigen-specific binding enabled by the VLP-Open HLA platform translates into functional T cell responses, we evaluated T cell activation and proliferation using both transduced 1G4 T cells and primary human PBMCs (Fig. 4). For the transduced 1G4 T cell system, cells were stimulated with Open HLA-A*02:01 VLPs loaded either the cognate ligand, NY-ESO-1C9V or control peptides which aren't recognized by 1G4 such as gTAX or NY-ESO-1W5A. All experiments were done in the presence of anti-CD28 costimulation, offered in trans. T cell activation was quantified by flow cytometry based on the expression of multiple activation markers, including 4-1BB, CD25, and CD69, as well as interferon-γ (IFN-γ) production (Fig. 4, A and B and Supplementary Text of fig. S17). Stimulation with NY-ESO-1C9V displayed VLPs resulted in a significantly higher frequency of tetramer-positive CD8+ T cells than control peptide displaying VLPs, indicating selective enrichment of antigen-specific T cells. On the basis of these observations, tetramer-positive CD8+ T cell populations were further analyzed to assess antigen-specific activation. VLPs displaying Open HLA-A*02:01/NY-ESO-1C9V efficiently induced T cell activation, leading to significant up-regulation of activation markers and increased cytokine production compared with control VLPs displaying Open HLA-A*02:01/gTAX. VLPs displaying exchanged Open HLA-A*02:01/gTAX↔NY-ESO-1C9V peptides induced comparable levels of activation to refolded Open HLA-A*02:01/NY-ESO-1C9V VLPs, demonstrating that peptide exchange on the VLP surface preserves functional antigen presentation (Fig. 4B). In contrast, VLPs displaying Open HLA-A*02:01/gTAX↔NY-ESO-1W5A did not induce significant activation of transduced 1G4 T cells. These data suggest the potential of our VLP-Open HLA platform as a loadable artificial antigen-presenting system. VLP-Open HLA nanoparticles elicit antigen-specific T cell activation and expansion. (A) Representative flow cytometry histograms of transduced 1G4 T cells stimulated with VLP-Open HLA-A*02:01 loaded with different peptides (gTAX, gTAX↔NY-ESO-1W5A, NY-ESO-1C9V, and gTAX↔ NY-ESO-1C9V) in the presence of anti-CD28 (αCD28). (B) Quantification of activation marker expression (4-1BB, CD25, and CD69) and cytokine production (IFN-γ) in tetramer-positive CD8+ 1G4 T cells. (C) Quantification of frequencies of CMVpp65-specific CD8+ T cell stimulated with A02/VLP displaying different peptides (NY-ESO-1C9V and CMVpp65) at various times. (D) Representative flow cytometry plots of PBMCs stimulated with A02/CMVpp65 VLPs in the presence of anti-CD28 at day 5, day 8, and day 11. (E) Quantification of activation marker expression (4-1BB and CD25) and cytokine production (IFN-γ) in CMVpp65-specific CD8+ T cells. All data shown here are presented as the mean ± SD (n = 3 per group). All statistical significance was determined by one-way ANOVA; **P < 0.01, ***P < 0.001, and ****P < 0.0001. On the basis of the antigen-specific staining and activation mediated by VLP-Open HLA platform, we next examined antigen-specific T cell expansion and activation in a more physiologically relevant system using primary human PBMCs obtained from a CMV-seropositive donor (Fig. 4, C to E). PBMCs were individually stimulated with VLPs displaying Open HLA-A*02:01/CMVpp65 and Open HLA-A*02:01/ NY-ESO-1C9V or with control conditions, in the presence of anti-CD28. After treatment of samples, tetramer-positive CD8+ T cell percentages were monitored over an 11-day period. CMVpp65-specific CD8+ T cells exhibited a time-dependent expansion only upon stimulation with VLPs displaying Open HLA-A*02:01/CMVpp65, whereas stimulation with VLPs displaying NY-ESO-1C9V or VLP-only controls did not result in significant expansion (Fig. 4, C and D and Supplementary Text of fig. S18A). These results demonstrate that VLP-Open HLA selectively induce antigen-dependent expansion of CD8+ T cells in primary PBMCs. To further evaluate the functional activation of antigen-specific T cells, tetramer-positive CD8+ T cell populations were analyzed at day 8 for activation markers and cytokine production (Fig. 4E and Supplementary Text of fig. S18B). Open HLA-A*02:01/CMVpp65 VLP-stimulated PBMCs exhibited significantly increased expression of 4-1BB and CD25, as well as enhanced IFN- γ production, compared with wild type of VLP and Open HLA-A*02:01/NY-ESO-1C9V VLP conditions. Therefore, these results demonstrate that our optimized VLP-Open HLA platform not only enables efficient staining but also induces antigen-specific T cell activation, and expansion. DISCUSSION In this study, we developed the VLP-Open HLA platform, a loadable antigen presentation system based on VLPs displaying our previously engineered Open HLA molecules (41). Using the SpyCatcher/SpyTag protein ligation system, we successfully produced and characterized VLPs displaying peptide-loaded Open HLA ligands. We demonstrated that these nanoparticles can be preloaded with a placeholder peptide and efficiently exchanged with candidate peptides, showing their utility as a screening tool for peptide mixtures derived directly from biological samples. For antigen-specific T cell staining, fluorescent dyes were conjugated to the interior of the VLP, while varying numbers of Open HLA molecules were displayed on the exterior surface. We found that a low number of approximately 13 Open HLA ligands per VLP was sufficient to selectively stain antigen-specific T cells, whereas higher ligand densities increased peptide-independent background interactions. VLPs displaying Open HLA-A*02:01/gTAX and subsequently loaded with exchanged peptides selectively stained antigen-specific CD8+ T cells across multiple human PBMC donors. We further applied our optimized VLP-Open HLA platform to induce antigen-specific T cell activation and expansion. Antigen-specific CD8+ T cells were efficiently activated and proliferated following stimulation with VLPs displaying Open HLA-A*02:01 loaded with on-target peptides, demonstrating the utility of this system as a loadable artificial antigen-presenting platform. This application highlights another advantage of our VLP platform for T cell stimulation and antigen recall studies which is its simplicity of use, particularly for TCR discovery applications, which often rely on time-consuming generation of autologous moDCs from a large number of PBMCs (13, 14). By completely bypassing moDC generation, our approach significantly reduces the number of starting PBMCs required and enables the use of freshly isolated CD8+ T cells for stimulation on the day of sample collection. Moreover, the VLP-Open HLA platform is a fully protein-based system that elicits appropriate stimulation/expansion of antigen-specific T cells without the need to remove particles or otherwise manipulate cultures before downstream applications. Among recently reported multivalent pMHC display platforms, the spheromer system (40) represents one of the closest published technologies to our VLP-based system. While the spheromer system relies on a biotinylated 24-subunit maxi-ferritin scaffold (expressed in mammalian cells) and requires refolded pMHC (prepared separately from E. coli expression), our HLA-VLP system offers greater modularity and ease of production using a single expression system. Furthermore, the SpyCatch-mi3 (VLP) scaffold consists of 60 identical subunits arranged in an icosahedral geometry, enabling controlled, ordered, and reproducible polyvalent display of pMHC and/or other costimulatory molecules via SpyCatcher/SpyTag ligation without altering the particle architecture. In contrast, in the spheromer system, pMHCs are multimerized in a less controlled manner using a biotinylation signal sequence on each of the 24 maxi-ferritin subunits to attach six streptavidin molecules that have been separately conjugated with pMHC “dimers.” While the resulting stoichiometry of 12 pMHCs per particle closely resembles our optimized VLP composition (~13 molecules per particle), assembling spheromers from their purified components using biotin/streptavidin interaction further complicates the preparation process and adds one more variable component which is the source of streptavidin (that may vary on the basis of the commercial provider). In contrast, the SpyCatcher/SpyTag-based VLP system uses genetically encoded components on the VLP to display purified pMHC molecules. Therefore, the use of our open VLP-Open HLA platform supports efficient peptide exchange post VLP subunit assembly, highlighting its ease of use and scalability for different antigens, which is not possible using the current inception of the spheromer system. Our platform can be extended beyond a research or screening tool to enable emerging cancer immunotherapy applications, including personalized T cell vaccines and T cell–based therapies (47). The VLP-Open HLA system can be adapted to individually present patient-derived HLA class I and class II allotypes loaded with candidate neoantigens as mosaic nanoparticles, enabling the generation of polyclonal T cell responses tailored to an individual patient’s tumor-specific HLA genotype and antigenic landscape. Our strategy can also support the development of personalized T cell vaccines, autologous T cell–based therapies like tumor infiltrating lymphocytes, and screening tools for patient tumor-specific synthetic immunoreceptors such as in single-chain fragment variable antibodies, chimeric antigen receptors, or bispecific T cell engagers. Moreover, the intrinsic multivalency of the VLP scaffold enables the development of mosaic particles codisplaying antigen-loaded HLA molecules together with stimulatory or inhibitory molecules for diverse in vitro and in vivo applications. The size of VLPs displaying Open HLA-I ligands (~30 nm) would be well suited for in vivo dissemination, as they are sufficiently small to drain efficiently from the site of immunization directly to the lymphoid organs (48). Therefore, appropriately dosed VLPs displaying Open HLA-I may achieve broad access to T cell populations throughout the body. By displaying either positive or negative costimulatory molecules on the exterior surface of the VLP, this platform could be leveraged to promote effective T cell activation, survival, and expansion in cancer therapies (49) or, alternatively, to induce tolerance, anergy, or exhaustion in autoimmune diseases (50). Such approaches may also enable reduced dosing while maintaining antigen specificity, potentially minimizing systemic side effects. From both oncological and autoimmune perspectives, the ability to load VLP-Open HLA-I particles with defined mixtures of peptide antigens is of significant interest, as this enables simultaneous stimulation or suppression of multiple antigen-specific T cell populations and may increase the potential of sustained therapeutic efficacy. In conclusion, our VLP-Open HLA platform provides a versatile, modular, and loadable system for precise control of antigen-specific T cell responses in translational and therapeutic applications. MATERIALS AND METHODS Expression and purification of SpyTag Open HLA-A*02:01 The heavy chain of Open A*02:01 (G120C) ectodomain was C-terminally tagged with SpyTag003 (RGVPHIVMVDAYKRYK). This heavy chain was refolded with β2m (H31C) in the presence of excess peptide NY-ESO-1C9V (SLLMWITQV), NY-ESO-1W5A (SLLMAITQV; used negative control), or gTAX (LFGYPVYV; placeholder peptide). A Gly-rich linker (GGSGGGSGG) was included between the isopeptide fusion tag (SpyTag003) and the α3 domain of the heavy chain, to prevent interference between the tag function and the heavy chain and other undesired interactions. Plasmids encoding the luminal domain of the C-terminally linked SpyTag003 Open A*02:01 (G120C) and open human β2m (H31C) were transformed into E. coli BL21(DE3) cells using the pET22b(+). The specific protein sequences are described in Supplementary Text of table S2. The cells were grown and harvested in the LB medium, and inclusion bodies were pelleted and purified as previously described (41). The ternary pMHC-I complexes were generated by an in vitro refolding, by slowly diluting a 100 mg of Open HLA-A*02:01-SpyTag003 and open β2m at a 1:3 molar ratio over 4 hours in 500 ml of refolding buffer [0.4 M l-arginine HCl, 100 mM tris, 2 mM EDTA, 5 mM reduced l-glutathione, and 0.5 mM oxidized l-glutathione (pH 8.0)] supplemented with 5 mg of the peptide. The refolding reaction proceeded for 4 days, and the proteins were then dialyzed into a buffer containing 25 mM tris and 150 mM NaCl (pH 8.0) (tris-buffered saline). The protein complex was then purified by size exclusion chromatography (SEC), using a HiLoad 16/600 Superdex 75-pg column at 1 ml/min. The identity of the purified proteins was further confirmed in reduced and nonreduced conditions using SDS-PAGE gel electrophoresis. Expression and purification of His6-SpyCatcher-mi3 (VLP scaffold) The VLP with an N-terminal 6x-His Tag, His6-SpyCatcher-mi3, was produced as previously described (27, 32, 44). The plasmid encoding the SpyCatcher-mi3 particle (Addgene 113043) was transformed into E. coli BL21(DE3) cells. A starter culture of 50 ml in LB broth supplemented with kanamycin (50 μg/ml) was done overnight (16 hours). The starter culture was then diluted 1:100 into 1 liter of LB supplemented with kanamycin (50 μg/ml) and 0.8% (w/v) glucose; this was cultured at 37°C at 200 rpm until an optical density at 600 nm of 0.8 was reached. Protein expression was induced with isopropyl-β-d-thiogalactopyranoside (420 μM) and incubated at 18°C at 200 rpm for an additional 16 to 18 hours. The culture was centrifuged at 6000 rpm for 20 min at 4°C. The resulting pellet was resuspended in 30 ml of buffer containing 250 mM tris-HCl, 150 mM NaCl, 50 mM imidazole, 0.02% sodium azide (NaN3), supplemented with lysozyme (0.1 mg/ml), 2 mM phenylmethylsulfonyl fluoride, and 1000× benzonase nuclease. Cell suspension was incubated at room temperature for 1 hour on a platform shaker before being sonicated on ice five times of 30-s intervals at 40% amplitude (30 s on followed by 30 s off). Cell lysate was then clarified at 21,000g for 30 min at 4°C. The supernatant was filtered through 0.45-μm and then 0.22-μm syringe filters before being isolated by Ni-NTA chromatography using a prepacked HisPrep FF 16/10 column. For this immobilized metal affinity chromatography purification, the binding buffer was 20 mM tris-HCl (pH 8.0), 150 mM NaCl, 50 mM imidazole, and 0.02% NaN3; SpyCatcher-mi3 particle was eluted with the elution buffer composed of 20 mM tris-HCl (pH 8.0), 150 mM NaCl, 2 M imidazole, and 0.02% NaN3. Eluted particles were concentrated using a 30-kDa concentrator and further purified by SEC using a HiLoad 16/600 Superdex 200 column preequilibrated with 25 mM tris-HCl (pH 8.0), 150 mM NaCl, and 0.02% NaN3. SpyCatcher-mi3 particles were stored at 4°C and used for conjugations for up to 1 month after filtering with a 0.2-μm filter or centrifuging at 14,000 rpm for 15 min. The concentration of SpyCatcher-mi3 particles was measured by Pierce bicinchonic acid (BCA) Protein Assay (Thermo Fisher Scientific, 23225). Ligation of SpyTag Open HLA to SpyCatcher-mi3 Before the ligation, the proteins were centrifuged at 15,000 rpm for 10 min to remove any potential aggregates. VLP (SpyCatcher-mi3) was quantified using Pierce BCA Protein Assay. For a 500-μl reaction or 1000-μl reaction, 50 nM purified VLP (SpyCatcher-mi3) (molarity of 60-mer particle) was incubated with 9 μM SpyTag fused Open HLA-A*02:01/NY-ESO-1C9V, SpyTag fused Open HLA-A*02:01/NY-ESO-1W5A, or SpyTag fused Open HLA-A*02:01/gTAX in 25 mM tris-HCl (pH 8.0), 150 mM NaCl, and 0.02% NaN3 at room temperature for 16 to 18 hours. The SpyCatcher-VLP particle:Open HLA-SpyTag molar ratio explored was 1:180 as optimized. Ligation efficiency was visualized with SDS-PAGE with Coomassie staining. Ligated VLP-Open HLA complex was separated from unreacted HLA and monomers by washing the reactions three times with 25 mM tris-HCl (pH 8.0), 150 mM NaCl, and 0.02% NaN3 using a 100-kDa MWCO Amicon ultra centrifugal filter spin column. The concentration of VLP-Open HLA complex was measured by Pierce BCA Protein Assay. Dynamic light scattering For each sample, VLP-Open HLA protein complex in buffer [25 mM tris-HCl (pH 8.0), 150 mM NaCl, and 0.02% NaN3] was centrifuged at 15,000 rpm for 10 min to remove any potential aggregates right before DLS experiment. In a cuvette, 60 μl of each sample was loaded; the total protein concentration should be between 0.2 and 0.9 mg/ml. Each sample was measured at 20°C on a Nanobrook 90Plus Submicron Particle Sizer (Brookhaven Instruments, Nashua, NH, USA) for two to four runs. The intensity distribution was plotted, and the particle size/radius average along with its SD was calculated in Excel. Mass photometry Before characterization with mass photometry, either wild type of VLP particles or VLPs ligated with Open HLA molecules were centrifuged at 15,000 rpm for 15 min at 4°C. The stock particles were stored in a buffer of 25 mM tris-HCl (pH 8.0), 150 mM NaCl, and 0.02% NaN3, and they could be further diluted in 1× phosphate-buffered saline (PBS) to the final concentrations right before data acquisition. The mass photometry instrument was tested with sterile 1× PBS buffer and calibrated with a mass standard of β-amylase (112 and 224 kDa) and thyroglobulin (670 kDa). The particles of concentration between 200 and 1000 μg/ml were examined by localizing 1 to 2 μl per sample on the mass photometry grid. The data were collected on a TwoMP mass photometer (Refeyn, Oxford, United Kingdom) and analyzed with the Refeyn Discover MP analysis software in histogram mode and mass plot using default settings. For each sample, an automated Gaussian curve was fitted to the particle distribution to estimate the molecular counts, median masses, and their SD (sigma values). Mass photometry data for all particles are shown in Supplementary Text of fig. S4. Cryo–electron microscopy VLP (SpyCatcher-mi3) was ligated to SpyTag Open HLA-A*02:01/NY-ESO-1C9V using the method in the previous section. The particle VLP-Open A*02:01/NY-ESO-1C9V at a total protein concentration of 0.919 mg/ml was used for grid preparation. The sample was prepared on Quantifoil Cu 300 2/2 (Quantifoil) grids and was frozen in liquid ethane using the Vitroblot Mark IV (Thermo Fisher Scientific) at 4°C and 100% humidity. The micrographs were collected on a Titan Krios G3i 300-kV cryo electron microscope (Thermo Fisher Scientific) with a K3 Summit camera (Gatan). Surface plasmon resonance SPR experiments were conducted in triplicate using a BiaCore T200 instrument (Cytiva) in SPR buffer [150 mM NaCl, 20 mM sodium phosphate (pH 7.4), and 0.1% Tween 20]. Biotinylated pMHC-I was injected into the experimental flow cells (FCs) to achieve the desired level of approximately 1000 resonance units, matched as closely as feasible in each FC and chip. FC1 was a reference for FC2 to FC4. Biotinylated wild-type or Open HLA-A*02:01/SLLMAITQV (NY-ESO-1W5A) used as a negative control was immobilized at 10 μl/min on FC1 of a streptavidin-coated chip (Cytiva). Biotinylated wild-type or Open HLA-A*02:01/SLLMWITQV (NY-ESO-1C9V) was immobilized at 10 μl/min from FC2 to FC4 of a streptavidin-coated chip (Cytiva). Various concentrations (0, 0.5, 1, 2, 5, 10, 20, and 50 μM) of TCR NY-ESO-1 (51, 52) were selected and injected over the chip at 25°C at a flow rate of 30 μl/min for 60 s followed by a buffer wash with 180-s dissociation time. The SPR sensorgrams were analyzed using the surface-bound analysis settings in BiaCore T200 evaluation software (Cytiva) and fitted using one site-specific binding by GraphPad Prism v9. SPR sensorgrams and saturation curves were prepared in GraphPad Prism v9. Fluorescence polarization The ligation of SpyCatcher-VLP and SpyTag Open HLA-A*02:01/gTAX was conducted in the same way as described in the previous section. Open A*02:01 was refolded with the placeholder peptide, gTAX, was exchanged, and loaded with labeled peptide TAMRATAX9 (TAMRA-KLFGYPVYV). Before the assays, the VLPs and all components of the assays were centrifuged for 15,000 rpm; VLP concentration was determined via Pierce BCA Protein Assay. In a 96-well plate, for each 100-μl reaction, VLP-Open HLA-A*02:01/gTAX of different concentrations (2.5 to 100 nM) were incubated with 40 nM TAMRATAX9 in FP buffer [150 mM NaCl, 20 mM sodium phosphate, and 0.05% Tween 20 (pH 7.4)]. The concentration of the TAMRA-labeled peptide was previously optimized so that the polarization baseline was between 0 and 50 mP. The kinetic association of fluorescently labeled peptides loaded on MHC-I was monitored via FP for 4 to 6 hours. The fluorescence of the TAMRA-labeled peptides was monitored at the excitation and emission wavelengths of 531 and 595 nm on a SpectraMax iD5 plate reader. All experiments were performed in triplicates at room temperature. Raw parallel (III) and perpendicular emission intensities (I⊥) were measured and then converted to polarization (mP) values using the equation 1000*[(III − (G*I⊥))/(III + (G*I⊥))], in which the G-factor was 0.33 for TAMRA-labeled peptides. The data were fitted in GraphPad Prism 10. Mass spectrometry To prepare for the particles for the mass spectrometry study, in a total reaction volume of 1000 μl, 50 nM purified His6-SpyCatcher-mi3 (SpyCatcher-VLP) was incubated with 9 μM SpyTag Open HLA-A*02:01/gTAX in 25 mM tris-HCl (pH 8.0), 150 mM NaCl, and 0.02% NaN3 at room temperature overnight (16 to 18 hours). Using the same protocol for ligation and wash steps as outlined here in the main text, SDS-PAGE was used to confirm, and the VLP-Open HLA-A*02:01/gTAX particles were washed three times using a 100-kDa spin column to remove excess unreacted SpyTag Open HLA molecules before being quantified by Pierce BCA Protein Assay (Thermo Fisher Scientific, catalog no. 23225). For the TCGA mass spectrometry peptide study, the peptides were selected on the basis of our previous study (41), and the peptides were purchased from Mimotopes (Australia); peptide information for the TCGA peptide is shown in Supplementary Text of table S4. For the peptide exchange experiment, 4 nM (or 20 μg/ml) VLP-Open HLA-A*02:01/gTAX particle was incubated without and with 2.4 μM peptide in 25 mM tris-HCl (pH 8.0), 150 mM NaCl, and 0.02% NaN3 at room temperature for 1 hour and then overnight at 4°C. Each reaction was prepared in a total volume of 500 μl; the different control and experimental conditions tested are summarized in Supplementary Text of table S5. The peptide exchanged samples were then washed twice by centrifuging at 10,000 rpm using a 100-kDa spin column to remove any excess peptide. The resulting sample (about 50 μl) was diluted two times for the mass spectrometry experiment with LC-MS or with MALDI-TOF MS. For the LC-MS experiment, the sample was analyzed on a Waters Acquity UPLC SQD instrument, using a Waters Acquity C8 column followed by electronspray ionization–MS with a mass detection range of 0 to 2000 mass/charge ratio. Analysis of LC-MS data was done with Xcalibur and UniDec. LC-MS spectra are shown in Supplementary Text of figs. S7 to S12. For the MALDI-TOF MS experiment, the samples were diluted two times in 0.05% trifluoroacetic acid/Milli-Q water and then spot on the MALDI plate with α-Cyano-4-hydroxycinnamic acid matrix. The samples were analyzed on the Bruker RapifleX (MALDI-TOF/TOF) (Billerica, MA, USA). MALDI-MS spectra are shown in Supplementary Text of z S13 and S14. Fluorescent labeling of VLP and displaying Open HLA molecules After purification, SpyCatcher-VLPs were centrifuged at 16,000 rpm for 15 min at 4°C to remove any potential aggregated forms. The concentration of SpyCatcher-VLP subunit was determined by measuring the UV absorbance at 280 nm. To enable efficient thiol-maleimide conjugation, SpyCatcher-VLPs were dialyzed overnight against phosphate buffer (pH 7.4). After dialysis, the VLPs were concentrated using an Amicon ultra centrifugal filter, 100-kDa MWCO. The concentration of SpyCatcher-VLP subunit was then redetermined by UV absorbance at 280 nm to enable accurate calculation for the labeling reaction. AF647 C2-Maleimide (Thermo Fisher Scientific, A20347) was reconstituted in dimethyl sulfoxide according to the manufacturer’s instructions. SpyCatcher-VLP which contains accessible cysteine residues on each subunit was incubated with a fivefold molar excess of AF647 maleimide at 4°C for overnight (46). Unconjugated dye molecules were removed by dialysis against phosphate buffer (pH 7.4). Then, the concentrations of the conjugated dye and SpyCatcher-VLP subunits were determined separately by measuring their corresponding UV absorbance. The degree of labeling was calculated as the molar ratio of conjugated dye per SpyCatcher-VLP subunit. To display the open MHC ligands on the surface of VLPs, optimized amounts of SpyTag fused open MHC ligands were incubated with AF647-conjugated SpyCatcher-VLPs at 4°C for overnight to allow SpyCatcher/SpyTag-mediated ligation. T cell staining with AF647-labeled VLP-Open HLA T cells transduced with variants of a single TCR (1G4) specific to the cancer antigen NY-ESO-1 were used in these experiments. Details regarding the production of these cells are provided in the Supplementary Materials. Cells were cultured in advanced RPMI (Gibco), supplemented with 10% heat-inactivated fetal bovine serum (FBS; Gibco), 1× GlutaMAX (Gibco), 1× penicillin-streptomycin (Gibco), 10 mM Hepes (Gibco), and recombinant interleukin-2 (IL-2; 10 ng/ml). For staining experiments, 8 × 105 cells per well were seeded in 96-well TC-treated plates and cultured at 37°C and 5% CO2 overnight. On the next day, cells were harvested by centrifugation at 300g for 5 min and stained with LIVE/DEAD IR 876 Fixable Dead Cell Stains (Invitrogen) for 10 min at room temperature. After washing with fluorescence-activated cell sorting (FACS) buffer [1× Dulbecco’s Phosphate Buffered Saline (DPBS) supplemented with 2.5% heat-inactivated FBS], cells were individually incubated with AF647-labled VLPs or tetramers displaying different peptides in blocking buffer (6% bovine serum albumin in PBS) for 1 hour at 4°C in the dark. After staining, cells were further incubated with fluorescein isothiocyanate (FITC)–conjugated anti-human CD4 antibody (BioLegend, 317408) for 30 min at 4°C in the dark. Cells were subsequently washed with FACS buffer and fixed with 4% paraformaldehyde for 30 min at room temperature. After two additional washes with FACS buffer, fluorescence signals were detected using a CytoFLEX LX flow cytometer, and data were analyzed using FlowJo software. CMVpp65-seropositive primary human PBMCs were also stained using the same staining procedure described above. For CD8 T cell–specific staining, phycoerythrin (PE)–conjugated anti-human CD4 antibody (BioLegend, 344606) and FITC-conjugated anti-human CD8 antibody (BioLegend, 980908) were used. T cell activation assay using the VLP-Open HLA system For T cell activation assays, CD8+ T cells expressing the 1G4 TCR were used. Cells were cultured in CTS Optimizer T Cell Expansion Medium (Gibco) supplemented with human IL-7 and IL-15 (5 ng/ml) (PeproTech) and 5% human serum (Valley Biomedical). On day 0, T cells were cultured at 1 × 106 cells/ml at 37°C and 5% CO2 overnight. On day 1, cells were plated at 1 × 105 per well in sterile U-bottom 96-well plates and stimulated either with VLP-Open HLA platforms or with Dynabeads Human T-Activator CD3/CD28 (Gibco) for 5 days. A negative control condition containing media (no VLP and no beads) was included. VLP displaying Open HLA-A*02:01 loaded with different peptides were supplemented with anti-CD28 antibody (20 μg/ml; BioLegend, 302934). On day 5, cells were harvested by centrifugation at 300g for 5 min. For flow cytometry analysis, cells were stained with LIVE/DEAD IR 876 Fixable Dead Cell Stains (Invitrogen) for 10 min at room temperature and washed with FACS buffer (1× DPBS supplemented with 2.5% heat-inactivated FBS). Cells were then stained with Open HLA-A*02:01/NY-ESO-1C9V tetramer (APC-conjugated) for 15 min, followed by incubation with an antibody cocktail containing Brilliant Ultra Violet 421–conjugated anti-human CD69 antibody (BioLegend, 310930), FITC-conjugated anti-human CD8 antibody (BioLegend, 980908), PE/cyanine7-conjugated anti-human 4-1BB antibody (BioLegend, 300808), and PE/Dazzle 594–conjugated anti-human CD25 antibody (BioLegend, 385224) for 30 min at 4°C in the dark. After two additional washes with FACS buffer, fluorescence signals were detected using a CytoFLEX LX flow cytometer, and data were analyzed using FlowJo software. For intracellular cytokine staining, cells were stimulated with 1× Cell Activation Cocktail (BioLegend; containing phorbol 12-myristate 13-acetate, ionomycin, and brefeldin A) for 4 hours before fixation and permeabilization. Cells were then stained as described above, fixed, and permeabilized. Subsequently, cells were incubated with Brilliant Violet 510–conjugated anti-human IFN-γ antibody (BioLegend, 506540) for 30 min at 4°C in the dark. Cells were washed twice with BD Perm wash buffer (BD Biosciences), followed by a final wash with FACS buffer before analysis. To assess antigen-specific T cell proliferation and activation in CMV-seropositive human PBMCs, CD8+ T cells were purified from PBMCs by magnetically activated cell sorting using a negative selection kit for CD8+ T cells (Miltenyi Biotec) and stimulated with VLP-Open HLA-A*02:01/CMVpp65 or wild-type VLP in the presence of anti-CD28 antibody. At the indicated time points (day 5, day 8, and day 11), cells were harvested and stained with Open HLA-A*02:01/CMVpp65 tetramers (APC-conjugated) to quantify the frequency of CMVpp65-speicifc CD8+ T cells by flow cytometry. Unstimulated cells (day 0) were also analyzed. Activation marker expression and cytokine production of tetramer-positive CD8+ T cells from PBMCs at day 8 were further analyzed by flow cytometry using the same experimental and staining procedures applied to 1G4 T cells. Ethical approval and informed consent The Human Immunology Core at the University of Pennsylvania provided all primary cells used in this study. The studies involving human participants were reviewed and approved by the University of Pennsylvania Institutional Review Board (IRB no. 850312). A written informed consent to participate in this study was provided by the participants. Supplementary materials The Supplementary Materials contain additional experimental details, including information regarding peptide and protein sequences, recombinant protein purification, DSF, mass photometry, FP, mass spectrometry, 1G4 TCR T cell production, and T cell staining and activation assays.