# paper.pdf ## Metadata - PDFFormatVersion=1.4 - Language=en-US - IsLinearized=true - IsAcroFormPresent=false - IsXFAPresent=false - IsCollectionPresent=false - IsSignaturesPresent=false - CreationDate=D:20260708164120+05'30' - Creator=Adobe InDesign 15.1 (Windows) - ModDate=D:20260708164123+05'30' - Producer=Adobe PDF Library 15.0 - Trapped=False - Custom.jobID=7404a622-0328-4e1b-8f29-80d2b9b2e6fe - dc:identifier=doi:jitc-2026-014960 - dc:format=application/pdf - xmp:createdate=2026-07-08T16:41:20+05:30 - xmp:metadatadate=2026-07-08T16:41:23+05:30 - xmp:modifydate=2026-07-08T16:41:23+05:30 - xmp:creatortool=Adobe InDesign 15.1 (Windows) - xmpmm:instanceid=uuid:94c91f26-0cc8-4b02-8ba1-f0bcb07e199a - xmpmm:originaldocumentid=xmp.did:B2C59A163194E611BD009A7CB80B8EC7 - xmpmm:documentid=xmp.id:db4d9a57-55bb-ce40-b911-8df7af003494 - xmpmm:renditionclass=proof:pdf - xmpmm:derivedfrom=xmp.iid:e3ce06a0-cf4d-8440-b450-f9bfead3ae5cxmp.did:6277246c-bd4a-114e-88bf-c4dda312a33exmp.did:B2C59A163194E611BD009A7CB80B8EC7default - xmpmm:history=convertedfrom application/x-indesign to application/pdfAdobe InDesign 15.1 (Windows)/2026-07-08T16:41:20+05:30 - pdfx:jobid=7404a622-0328-4e1b-8f29-80d2b9b2e6fe - pdf:producer=Adobe PDF Library 15.0 - pdf:trapped=False ## Contents ### Page 1 1MontoyaflA, et al. J Immunother Cancer 2026;14:e014960. doi:10.1136/jitc-2026-014960Open accessUnbiased avidity-based isolation ofantigen-specificTcellsAmanda Montoya ,1 Meredith L Frank ,1 Peixin Jiang,1 Hui Nie,1 Minying Zhang,1 Emily Bontekoe,1 Jared K Slone,2 Ludovica L Posta,3 Sofia Rosy Caterina Sorice,3 Tina Cascone ,1 Maura Gillison,1 Don L Gibbons,1 Jianjun Zhang ,1,4 Eleonora Dondossola,3 Lydia Kavraki ,2 Pamela L Wenzel,5 John V Heymach,1 Alexandre Reuben 1To cite: Montoya A, Frank ML, Jiang P, et al. Unbiased avidity-based isolation of antigen- specific T cells. Journal for ImmunoTherapy of Cancer2026;14:e014960. doi:10.1136/ jitc-2026-014960► Additional supplemental material is published online only. To view, please visit the journal online (https:// doi. org/ 10. 1136/jitc- 2026- 014960). AM and MLF are joint first authors. Accepted 11 June 2026 1Department of Thoracic/Head and Neck Medical Oncology, The University of Texas MD Anderson Cancer Center, Houston, Texas, USA 2Deparrtment of Computer Science, Rice University, Houston, Texas, USA 3Department of Genitourinary Medical Oncology, The University of Texas MD Anderson Cancer Center, Houston, Texas, USA 4Department of Genomic Medicine, The University of Texas MD Anderson Cancer Center, Houston, Texas, USA 5The University of Texas Health Science Center, Houston, Texas, USACorrespondence toDr Alexandre Reuben;AReuben@ mdanderson. orgOriginal research© Author(s) (or their employer(s)) 2026. Re-use permitted under CC BY-NC. No commercial re-use. See rights and permissions. Published by BMJ Group.ABSTRACT Background Cancer immunotherapies have significantly improved treatment efficacy and patient survival by exploiting antigen-specific T cells to eliminate cancer cells. However, current approaches for identifying and isolating antigen-specific T cells typically require prior knowledge of target antigens, limiting discovery, and reducing the ability to consistently detect rare tumor-reactive T cells. We therefore sought to develop an unbiased platform for the identification and enrichment of antigen-specific T cells using naturally processed and presented tumor antigens.Methods We developed ATTACH (Assessment of T cells Tethered to Antigen Class I Histocompatibility), a microfluidic platform that applies controlled shear stress and leverages tumor cells as a natural source of endogenous major histocompatibility complex (MHC)-peptide complexes to isolate antigen-specific T cells based on MHC/peptide binding avidity. ATTACH was evaluated in both human and mouse systems for its ability to enrich rare tumor-reactive T-cell populations and deplete bystander virus-specific T cells.Results ATTACH resulted in up to a 10-fold enrichment of antigen-specific T cells across both human and mouse systems, enabling the isolation of clonotypes present at frequencies as low as 0.1%. In addition to enriching rare tumor-reactive T cells, ATTACH efficiently depleted virus-specific bystander T cells.Conclusions ATTACH provides a streamlined and unbiased approach for the rapid identification and isolation of antigen-specific T cells, and may facilitate the optimization of cellular therapies for the treatment of solid tumors.INTRODUCTIONIn the field of adoptive cell therapy (ACT), tumor infiltrating lymphocytes (TIL) have gained considerable attention for their ability to leverage the host’s immune system against cancer.1 Numerous studies have suggested that increased TIL density in solid tumors contributes to prolonged recurrence-free survival and lowers the risk of systemic recurrence.2–8 Addition- ally, in previous phase I and II clinical trials using TILs for the treatment of metastaticWHAT IS ALREADY KNOWN ON THIS TOPIC⇒ Tumor-infiltrating lymphocyte (TIL)-based therapies have demonstrated clinical benefit in solid tumors, but their efficacy remains limited by the high proportion of non-tumor-reactive T cells often present within tu- mors. Isolating these tumor-reactive T cells remains a major challenge. Existing strategies to identify antigen-specific T cells require prior knowledge of target antigens or rely on complex and resource-intensive workflows, limiting their scalability and discovery potential. Methods leveraging T-cell receptor avidity via microfluidic selection have been explored but re- main largely incapable of rapidly enriching functional antigen-specific T cells for downstream applications.WHAT THIS STUDY ADDS⇒ This study presents Assessment of T cells Tethered to Antigen Class I Histocompatibility (ATTACH), a microfluidic avidity-based platform that uses con- trolled shear stress and naturally presented tumor antigens to enrich antigen-specific CD8+ T cells. ATTACH enabled enrichment of rare tumor-reactive T- cell populations across both human and mu- rine systems, including clonotypes present at very low frequencies, while simultaneously reducing non-specific bystander T cells. The platform also preserved T-cell functionality after isolation and achieved enrichment without requiring predefined antigens or highly specialized instrumentation.HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY⇒ These findings support the use of ATTACH as a rapid and accessible approach for isolating tumor-reactive T cells for research and cellular immunotherapy ap- plications. By simplifying enrichment of rare antigen-specific T cells from heterogeneous populations, ATTACH could improve workflows for TIL-based thera- pies and facilitate development of personalized cancer immunotherapies. The platform also provides a useful framework for future avidity-based immune cell selec- tion technologies. ### Page 2 2MontoyaflA, et al. J Immunother Cancer 2026;14:e014960. doi:10.1136/jitc-2026-014960Open accessnon-small cell lung cancer (NSCLC), tumor regres- sion was achieved.2 9 However, only a small subset of patients experienced benefit, and durable responses were limited.2 9 One of the main challenges facing ACT using TILs is the limited frequency of antigen-specific T cells. Prior studies have demonstrated that within populations of TILs, only a minority of T cells exhibit tumor reactivity, with a significant presence of bystander T cells recog- nizing other non- tumor antigens, including those asso- ciated with viruses and/or bacteria.10–13 Notably, there is evidence supporting a correlation between clinical responses following TIL therapy and the reactivity of infused TILs toward mutated antigens.14 Typically, patients who achieve complete responses present higher amounts of neoantigen-specific CD8+ TILs in their infu- sion product.15 16 While TIL therapy has demonstrated efficacy in inducing tumor regression across various solid tumors, drawbacks such as diminished CD8+ T cell counts and the loss of tumor-specific clonotypes in the expanded TIL product have posed challenges in achieving success across broad patient populations. Thus, there has been a concerted effort to identify and isolate antigen-specific CD8+ TILs from solid tumors. The leading strategy for identifying and isolating antigen- specific T cells involves the use of MHC/peptide multimers, which exploit the antigen-binding avidity of T cell receptors (TCRs).10 17 18 However, the use of multi- mers demands prior knowledge of the epitope of interest and the MHC presenting it.10 17 18 Therefore, inherent biases exist toward pre-selected or “known” antigens when utilizing multimers for the identification and isola- tion of antigen-specific T cells.19 20 Moreover, custom multimer synthesis can be both time-consuming and costly.21 Furthermore, functional assays have emerged that leverage systematic peptide screens, T cell trogo- cytosis,22 bifunctional receptors on antigen-presenting cells,23 tandem minigenes,24 and yeast antigen display libraries,25 among others. These methods have over- come many hurdles associated with antigen discovery and T cell isolation but remain restricted by the need for prior knowledge of which antigens or TCRs to further investigate. Significant efforts have been invested to identify pheno- typic attributes associated with antigen specificity, such as increased clonotype frequency and expression of markers including 4–1BB, PD-1, CD39, CD103, OX40, and TOX,26–32 among others. However, many of these markers delineate activated, exhausted, or tissue-resident T cells, meaning they can be expressed by both anti-tumor and bystander T cells. In silico-based approaches such as netMHC predict potential antigens based on epitope length, sequence, and MHC-binding affinity. However, these algorithms rely heavily on MHC-binding affinity and are not yet capable of consistently distinguishing whether an epitope will be immunogenic, thereby requiring substantial subsequent in vitro screening.33–36 Databases cataloging TCR sequences and their cognate antigens have also been developed. However, these fail to predict novel TCR binding because of their polymorphisms and promiscuity in TCR/epitope binding interactions.13 37 38 All in all, current methods for identifying antigen-specific T cells exhibit clear limitations. Here, we describe a novel versatile platform called ATTACH (Assessment of T cells Tethered to Antigen Class I Histocompatibility), which enriches for antigen-specific T cells using tumor cells as a natural source of endogenously processed and presented MHC-peptide complexes. The process relies on the ability of antigen-specific T cells to recognize and bind to antigens presented on MHC molecules from target cells under a sustained, constant, low level of shear stress. Importantly, we show ATTACH can be used in both mouse and human systems and highlight its potential for future applications.MATERIALS AND METHODS Cell linesLewis Lung Carcinoma (LLC-JSP) cells were transduced to endogenously express the H-2Kb- restricted OVA257- 264 SIINFEKL antigen. Selection was maintained in trans- duced LLC-JSP-OVA cells by adding 0.05 mg/mL of Hygromycin B (Corning, #30-240-CR) to cell culture media. The NSCLC cell line H1975 was obtained from ATCC (RRID:CVCL_1511) and was transduced to express the HLA-A*03:01 (Vector Builder, VB241121-1833ydj), HLA- A*02:01 (Vector Builder, VB250619-2298uvq), and the FOXM1 genes (cloned with pENTER vector by GenScript USA, Inc). The K562 cell line was obtained from ATCC and was transduced to express HLA-A*02:01 and mutant HER2. All cell lines were cultured at 37°C, 5% CO2 in complete RPMI 1640 media (Corning, #10-040-CV) supplemented with 10% fetal bovine serum (FBS, GenDEPOT, F601-050) (referred to as “complete RPMI media” hereafter). Cell lines were routinely tested for mycoplasma contamination using the Lonza MycoAlert Detection Kit (#LT07-218).Murine CD8+ T cell isolation and expansionIsolated spleens were disrupted in complete RPMI media, and cell suspensions were passed through a 40 μm nylon mesh strainer (Fisher Scientific, #22363547) to remove debris and aggregates. Strained cell suspensions were pelleted by centrifugation at 600×g for 6 min. All centrif- ugation steps were performed at room temperature. Red blood cell lysis buffer (Santa Cruz Biotechnology, #sc-296258) was prepared at 1× concentration in sterile Milli- pore water and added to the pelleted cell suspension at a volume 10× that of the cell pellet. After incubating for 12 min at room temperature, complete RPMI media was added at 10× the volume of red cell lysis buffer to quench the lytic reaction. Cells were centrifuged at 600×g for 6 min and resuspended at ≤1×108 nucleated cells/mL in Ca2+ and Mg2+ free 1× phosphate-buffered saline (PBS) supplemented with 2% FBS and 1 mM EDTA (hereafter referred to as the “recommended media”). 50 μL/mL of ### Page 3 3MontoyaflA, et al. J Immunother Cancer 2026;14:e014960. doi:10.1136/jitc-2026-014960Open accessnormal rat serum (Sigma-Aldrich, #R9759) was added to samples before transferring them to 5 mL polystyrene round- bottom tubes. Negative CD8+ selection was used to isolate CD8+ splenocytes via magnetic sorting under sterile conditions using the STEMCELL Technologies EasySep mouse CD8+ isolation kit (#19853) per manufac- turer’s instructions. Isolated CD8+ T cells were counted using VitaStain AOPI staining solution (Nexcelom Biosci- ences, #CS2-0106) on a Nexcelom Cellometer K2 fluores- cent viability cell counter. An expansion protocol was adapted from Lewis et al.39 All cells were cultured at 37°C, 5% CO2. Freshly isolated CD8+ splenocytes from OT-I and C57BL/6 mice were resuspended in complete RPMI media supplemented with 50 uM 2-mercaptoethanol (Thermo Fisher Scien- tific, #21985023) and 1% Insulin-Transferrin- Selenium-Sodium Pyruvate (ITS, Thermo Fisher Scientific, #51300044) and plated at 5×106 cells/well in 6-well Nunc plates (Thermo Fisher Scientific, #140675) precoated with 0.5 μg/mL anti-mouse CD3 (clone 2C11, Bio X Cell, #BE0001-1) and 5 μg/mL anti-mouse CD28 (clone 37.51, Bio X Cell, #BE0015-1). Cells were cultured without cyto- kine stimulation for 24 hours. Recombinant mouse IL-7 (R & D Systems, 407 ML-005) and IL-2 (R&D Systems, #402 ML-020) were added to final concentrations of 0.5 ng/mL and 0.1–0.4 ng/mL, respectively. Cells were cultured for another 24 hours before subculturing cells at 1×106 cells/well in non- coated Nunc 6-well plates in freshly made complete RPMI media supplemented with 50 μM 2-mercaptoethanol, 1% ITS, 0.5 ng/mL recom- binant murine IL-7, and 0.1–0.4 ng/mL recombinant murine IL-2. Cells were subcultured every 24 hours for 8–10 days in freshly made complete RPMI media containing all supplements and cytokines. Cells were frozen at 80%–90% viability in FBS supplemented with 10% DMSO and stored in liquid nitrogen for later use.Generation of TCR-T cellsFresh leukapheresis products from healthy adult donors were purchased from Charles River and were obtained at U.S. Food and Drug Administration (FDA)-registered collection centers under written informed consent. Healthy donor PBMCs were processed within 48 hours of collection, preserved, and stored in liquid nitrogen. TCR- T cells targeting EGFR, FOXM1, E7, and HER2 were generated as follows. For retroviral production, 293GP packaging cells were seeded overnight on poly-D- lysine- coated 100 mm plates at 7×10⁶ cells per plate in complete DMEM. Cells were transfected with 6 μg of the pMSGV1-TCR construct and 4.5 μg of the RD114 enve- lope plasmid using Lipofectamine 3000 (Thermo Fisher Scientific, #L00015). Viral supernatants were collected at 48 and 72 hours post-transfection, pooled, and added to Retronectin-coated (20 μg/mL, Takara Bio, #T100B) non- tissue- culture- treated 24- well plates. Plates were centrifuged at 2000×g for 2 hours at 32°C to facilitate viral loading. CD8+ T cells were isolated from PBMCs by nega- tive selection using the Human CD8+ T Cell Isolation Kit (STEMCELL Technologies, #17953). Purified cells were activated with ImmunoCult Human CD3/CD28/CD2 T Cell Activator (STEMCELL Technologies, #10970) in the presence of 200 IU/mL recombinant human IL-2 and cultured for 48 hours in a 1:1 mixture of RPMI supple- mented with 10% FBS and AIM V medium. Activated T cells were transferred onto the virally loaded plates and spinoculated at 800×g for 20 min at 32°C. This trans- duction step was repeated on the next day to enhance efficiency. Transduced T cells were assessed 3–4 days later by flow cytometry for surface expression of cognate tetramer. Cells were subsequently expanded for 1 week in 24-well plates and G-Rex culture vessels (Wilson Wolf, #80 660M) and cryopreserved in FBS containing 10% DMSO.Flow cytometryTo measure H-2Kb presentation of the SIINFEKL antigen on target cells, an APC anti-H-2Kb bound to SIINFEKL antibody (clone 25-D1.16, BioLegend, #141606) was added to target cells at 0.2 mg/mL and incubated on ice for 30 min. Non-transduced target cells served as a nega- tive control. For CD8 and CD4 surface staining, anti-mouse CD4-FITC (clone RM4-5, BioLegend, #100509), and anti-mouse CD8-FITC (clone 53–6.7, BD Pharmingen, #553030), or anti- mouse CD8- PE (clone 100708, BioLegend, #100708) antibody was used, and immediately prior to flow cytometric analysis, 7AAD viability staining solution (BioLegend, #420404) was added to cells. Tetramers were provided by the Baylor College of Medicine MHC Tetramer Production Core. Staining was performed for 30 min on ice prior to any other antibody staining. Samples were acquired by the University of Texas MD Anderson Flow Cytometry & Cellular Imaging Facility and collected on BD FACSCanto, Beckman Coulter Gallios and Gallios 561 cytometers. Acquired data were analyzed using FlowJo V.10.8.0 software (BD Biosciences), gating on live single cells.Immunofluorescence stainingAdherent LLC- JSP and LLC-JSP-OVA cells were detached and resuspended to ≤5×105 cells/mL in 1× PBS and stained with 6 μM of CellTracker blue CMAC dye (Invi- trogen, #C2110) following manufacturer’s instructions. Then, cells were resuspended in complete RPMI media and immediately used for ATTACH. OT-I CD8+ T cells were thawed and washed prior to staining. Cells were resuspended to ≤3 ×105 cells/mL in 1× PBS+0.1% FBS and stained with 0.7 μM of CFSE dye (Invitrogen, #65-0850-85) for 8 min at room temperature, obscured from light. An equal volume of 100% FBS was added to quench staining before transferring cells to a 37°C water bath and incubating for 10 min. After staining, cells were washed with 1× PBS+2% FBS and resuspended to 1×106 cells/mL in complete RPMI media for immediate use in ATTACH. C57BL/6 CD8+ T cells were resuspended to ≤1 × 106 cells/mL in 1× PBS followed by staining with 500 nM of ### Page 4 4MontoyaflA, et al. J Immunother Cancer 2026;14:e014960. doi:10.1136/jitc-2026-014960Open accessCellTracker Deep Red dye (Invitrogen, #C34565) or 250 nM MitoTracker Deep Red dye (Invitrogen, #M22426) for 30 min at 37°C, 5% CO2, washed, and resuspended in complete RPMI media for immediate use in ATTACH. For human experiments, T cells were resuspended to≤1×106 cells/mL in PBS followed by staining with 250 nM of CellTracker Deep Red dye (Invitrogen, #C34565) or 10 μM of CellTracker Blue CMAC dye (Invitrogen, #C2110) for 30 min at 37°C, 5% CO2, washed, and resuspended in complete RPMI media for immediate use in ATTACH.MIP-1β ELISAHuman MIP- 1β quantification was performed using Invitrogen’s Human MIP-1β (CCL4) Instant ELISA Kit (#BMS2030INST) following the manufacturer’s instruc- tions. Absorbance values were obtained using an Accuris SmartReader 96 microplate absorbance reader, and concentrations of unknown samples were interpolated using the standard curve’s equation of the line and absor- bance values.40ELISPOTFollowing recovery from ATTACH, T cells from each frac- tion were pulsed with either 1 μg/mL of HER2 peptide (GenScript) or CEF peptide pool (AnaSpec CEF Control Peptide Pool). Peptide-pulsed T cells were then subjected to IFN-γ ELISpot analysis using the Human IFN-γ ELISpot kit (CTL, catalog # hIFNg-2M/2), according to the manu- facturer’s instructions. Spot-forming units (SFUs) were quantified and compared between fractions and stimula- tion conditions using a CTL S6 Entry M2.Isolation of antigen-specific T cells in T-25LLC- JSP cells were seeded in tissue culture-treated vented T25 flasks and placed at 37°C, 5% CO2 overnight to adhere. Cells were seeded at 80% confluency so that they would be at 95% confluency for ATTACH the following day. The day of ATTACH, adhered cells were pulsed with 10 μg/ mL of SIINFEKL peptide by adding stock aliquots directly to T25 flasks containing adhered cells + 3 mL complete RPMI media. Cells were incubated for 1.5 hours at 37°C, 5% CO2. Meanwhile, expanded CD8+ T cells were thawed and resuspended to 1–2×106 cells/mL in complete RPMI media and placed in 6 well plates at 37°C, 5% CO2 for 1 hour. Following rest, T cell populations were stained as described above. Media containing SIINFEKL peptide was removed from flasks and flasks were washed with 1× PBS by gentle swirling. 1× PBS was removed and replaced with complete RPMI media containing the input popu- lation (either bulk OT-I splenocytes or a 1:1 mixture of OT-I: B6 CD8+ T cells). Part of this “input” subset was set aside at 37°C, 5% CO2 for downstream flow cytometric analysis. After a 90-minute incubation, media in flasks was carefully removed by pipette and placed on ice in a labeled tube (“aspirate” subset). Flasks were washed with 6 mL of 1× PBS, which was removed and placed on ice in labeled 15 mL tubes (“wash” subset). Finally, bound cells were detached with 0.25% Trypsin + 2.21 mM EDTA and transferred to labeled tubes (“detach” subset). Entire collections from the flask were stained and analyzed by flow cytometry as described above.Microfluidic isolation of antigen-specific T cellsSterilized ibiTreat: #1.5 polymer coverslip μ-Slide I0.4 luer tissue culture-treated channel slides (#80176), luer lock female connectors (#10825), elbow luer male connectors (#10802), and 1.6 mm silicone tubing (catalog #10842) were obtained directly from ibidi. Target cells were seeded in separate channel slides with complete RPMI and were incubated at 37°C, 5% CO2 overnight to allow target cells to adhere to the channel. The following day, slides were flushed by pipetting 125 μL of complete RPMI through channels to remove dead and detached cells prior to loading T cells. After fluorescently staining, CD8+ effector cells were loaded into channels at a 1:1 ratio of effector to target cells, followed by a 15-minute incubation. Shear stress was applied to channels by infusion via 140 mL syringes filled with prewarmed complete RPMI media connected to a Harvard Apparatus PHD ULTRA automated syringe pump. For ramped “enrichment” experiments, applied shear stress increased linearly with time y=1.56x+0.5, where y represents shear stress and xrepresents time, and ranged from 0 to 20 dyne/cm2 (0–21.1 mL/min) over the course of 12.5 min. Images were taken at 2.06, 8.3, 16.1, and 20 dyne/cm2. For de- enrichment experiments where a low shear stress was constantly applied to wash off non- specific CD8+ T cells, a shear stress of 1 dyne/cm2 (1.05 mL/min) was selected based on the point at which OT-I CD8+ became detached from negative control slides containing LLC-JSP cells (no antigen) during ramped shear experiments. Shear stress of 1, 2, and 3 dyne/cm2 was used for the EGFR-specific T cells and 3 dyne/cm2 for FOXM1-specific T cells and HER2-specific T cells. Slides were visualized under Nikon A1RSi (Center for Advanced Microscopy, Department of Integrative Biology & Pharmacology at McGovern Medical School, UTHealth) or Nikon ECLIPSE Ti confocal micro- scopes using Nikon NIS-Elements AR 5.21.03 and Zeiss Axio Observer 7. Fluorescent images were captured within 1 region of interest (ROI) in the middle of the channel. The ROI was consistent across all slides within an experiment. For experiments using 1:1000 ratios, images were taken at three randomly selected ROIs both before and after ATTACH. The average counts across these ROIs were calculated and used for subsequent analysis.Rapid immobilization of tumor cell linesIbidi μ-Slide VI0.4 channels were coated overnight at room temperature with 1 mg/mL lyophilized strepta- vidin (Thermo Fisher Scientific, 430431). The following day, channels were washed with PBS and blocked with 2.5% BSA (MilliporeSigma, #9048-46-8) for 30–60 min at room temperature, followed by a wash with RPMI 1640. Tumor cell biotinylation was performed.41 Briefly, tumor cell lines were washed with PBS and resuspended at 0.5×10⁶ cells/mL. Sulfo-NHS-LC-Biotin (Thermo ### Page 5 5MontoyaflA, et al. J Immunother Cancer 2026;14:e014960. doi:10.1136/jitc-2026-014960Open accessFisher Scientific, #A39257) was added at a final concen- tration of 1 mM, and cells were incubated at 4°C for 30 min. After incubation, cells were washed with RPMI and seeded into coated channels at 1×10⁶ cells per channel. Cells were allowed to adhere for at least 2 hours at room temperature, after which channels were gently washed with 1 mL of RPMI.Live cell imagingTo assess cell death following ATTACH, 0.15 μM propidium iodide was introduced into the channels. Live time- lapse imaging was conducted using an EVOS microscope equipped with a heated environmental control chamber (37°C, 5% CO₂). Fluorescent and bright- field images were captured every 3 min for 24 hours using a 10× objective lens. The images were then stitched into a time-lapse sequence using ImageJ software (RRID:SCR_003070). For analysis, bounding boxes were drawn around cells using the OpenCV2 Python package (RRID:SCR_024202),42 and cell tracking across frames was performed using the OpenCV2 implementation of the CSRT tracking algorithm.43Statistical analysisAll statistical analyses were performed using GraphPad Prism V.10 (GraphPad Software; RRID:SCR_002798). Ordinary one-way and two-way analysis of variance tests (with multiple comparisons) and paired t-tests were performed when appropriate. Quantification of fluores- cent images captured during live imaging was performed using ImageJ Fiji software. Experiments were performed with independent biological replicates ranging from n=2 to n=4, as indicated in the corresponding figure legends.RESULTS Manual recovery of antigen-specific T cells with ATTACHIn order to leverage the avidity of TCRs for their anti- gens, we envisioned ATTACH to include the following steps. First, a population of target cells would be seeded and allowed to adhere overnight. Second, a population of T cells would be added and allowed to co-incubate and engage the target cells through their TCR based on their cognate MHC- presented epitopes. Third, culture media would be aspirated, resulting in the removal of any debris and dead cells. Fourth, cells would be washed gently, with T cells lacking the ability to bind to their target cells through MHC I being removed. And lastly, remaining cells (target cells and antigen-specific T cells) would be detached using trypsin for subsequent down- stream analysis and quantification (figures 1 and 2A). To investigate the feasibility of this process, in the first gener- ation of ATTACH, we first used OT-I T cells which recog- nize the MHC-I (H-2Kb)- restricted OVA257- 264 (SIINFEKL) epitope. Briefly, LLC-JSP cells transduced with an OVA257-264 minigene (LLC- JSP-OVA) to endogenously express and present the H-2Kb/OVA257- 264 epitope were seeded in T- 25 flasks overnight to be used as target cells (online supplemental figure S1A). The next day, bulk splenocytes from OT- I mice were added (Subset 1, S1) followed by a 90-minute co-incubation with tumor cells. Cells were then recovered in three subsets: aspirated culture media, debris and dead cells (S2), PBS used to wash cells and non- specific/unbound cells (S3), and antigen-specific T cells and their targets detached by trypsinization (S4), to assess the proportion of antigen-specific T cells within each. Interestingly and as expected, a gradual loss in CD4+OT I splenocytes and gradual increase in CD8+OT I splenocytes was observed with a fourfold enrichmentFigure 1 Overview of the ATTACH platform. Target cells and T cells are co-incubated on a surface followed by the application of fluidic shear stress. Higher avidity T cells (red) bind to their targets and are thus retained longer while lower avidity T cells (green) are discarded. ATTACH, Assessment of T Cells Tethered to Antigen Class I Histocompatibility. MHC, Major Histocompatibility Complex. TCR, T cell receptor. ### Page 6 6MontoyaflA, et al. J Immunother Cancer 2026;14:e014960. doi:10.1136/jitc-2026-014960Open accessfor CD8+ in presence versus absence of H-2Kb/OVA257- 264 (figure 2B–F). Furthermore, OT-I CD8+ enrichment was most significant in the final (S4) subset compared with earlier fractions, confirming differential binding avidities between antigen-specific CD8+ T cells and non-CD8+ T cells when exposed to OVA257- 264- presenting cells.Microfluidic recovery of antigen-specific T cells using ATTACHWe next sought to increase reproducibility of ATTACH by developing a microfluidic pipeline leveraging laminar flow. To do so, LLC-JSP- OVA cells were seeded overnight in chamber slides (online supplemental figure S1B). Subsequently, a 1:1 ratio of fluorescently labeled OT-I and C57BL/6 CD8+ T splenocytes was added to the channel followed by a 15-minute incubation. The reduced volume and height of the microfluidic system enabled rapid T cell settling and interaction with tumor cells, making this shorter incubation sufficient under these conditions. Next, a ramped pattern of shear stress, which increased linearly with time, was applied to channels via infusion by a programmable syringe pump to determine the level of shear stress needed to separate antigen-specific from non- specific CD8+ T cells. Shear stress values ranging from 0 to 20 dyne/cm2 were applied to channels over 12.5 min. Fluorescent images were taken at timepoints corresponding to shear stress of 2.06, 8.3, 16.1, and 20 dyne/cm2. Results showed a fourfold enrichment of OT- I CD8+ cells and a corresponding de-enrichment of C57BL/6 CD8+ splenocytes in the final recovered frac- tion, with the most pronounced enrichment occurring at low shear stress (<8.3 dyne/cm²) (figure 3A–D). However, a low viability of recovered T cells was observed at high levels of shear (online supplemental figure S1C,D).Low shear microfluidic de-enrichment of non-specific T cells using ATTACHTo increase the versatility of ATTACH and enhance viability of recovered cells, we hypothesized that applying a constant low level of shear stress sufficient to remove non- specific T cells but low enough to retain antigen-specific T cells regardless of their avidity would be pref- erable (i.e., de-enrichment, figure 4A). Fluorescent imaging was used to assess T cell populations in real-time, and a constant rate of 1 dyne/cm2 of shear stress was applied via infusion for 5 min. Images were taken within a single ROI in the center of the channel before and after the assay (figure 4B,C). Quantification of CD8+ T cell proportions pre-ATTACH and post-ATTACH confirmed a fivefold antigen-dependent enrichment of OT-I CD8+ TFigure 2 First generation of ATTACH. (A) Overview of ATTACH assay performed in flasks. (B) Relative proportions of CD4+ (red) and CD8+ (purple) T cell populations in fractions were analyzed by flow cytometry. (C) CD8 to CD4 ratio in different ATTACH fractions. (D) Percentage of CD8+ T cells recovered from different ATTACH fractions in LLC-JSP-OVA versus LLC-JSP flasks. (E) Percentage of CD4+ T cells recovered from different ATTACH fractions in LLC-JSP-OVA versus LLC-JSP flasks. (F) Percentage of CD8+ recovered in final fraction after ATTACH on LLC-JSP-OVA versus LLC-JSP flasks. Data are presented as mean±SEM; n=3 independent flasks. Statistical significance was assessed using two-tailed unpaired t-tests (**p<0.01). ATTACH, Assessment of T Cells Tethered to Antigen Class I Histocompatibility; LLC-JSP, Lewis Lung Carcinoma; PBS, phosphate-buffered saline. ### Page 7 7MontoyaflA, et al. J Immunother Cancer 2026;14:e014960. doi:10.1136/jitc-2026-014960Open accesscells in LLC-JSP-OVA slides compared with LLC-JSP slides (figure 4D, p=0.004). To further elucidate the specificity of T cell enrich- ment, a heterogeneous CD8+ T cell population was next used for de-enrichment experiments. A 1:1 ratio of OT-I to C57BL/6 CD8+ T splenocytes was added to chan- nels containing LLC-JSP- OVA cells, and 1 dyne/cm2 of shear stress was applied via infusion for 5 min. Quanti- fication of fluorescent images pre-ATTACH and post-ATTACH revealed a twofold enrichment of OT-I CD8+ T cells (p=0.0151, figure 4D- F) with 15% recovery of input OT- I CD8+ T cells. Conversely, the average retention of C57BL/6 CD8+ T splenocytes was 8% and no different between channels seeded with either LLC-JSP or LLC-JSP- OVA (p=0.7561).ATTACH allows recovery of human antigen-specific T cellsWe next sought to evaluate the feasibility of ATTACH using primary human T cells. To do so, we used previ- ously validated TCR- engineered T cells (TCR-T) recog- nizing an HLA-A*03:01-restricted epitope derived from EGFR.44 H1975 cells, which endogenously express the EGFR mutation, were transduced to express HLA-A*03:01 and seeded into microfluidic channels over- night. The next day, antigen-specific CD8+ T cells were mixed at a 1:1 ratio with bulk CD8+ T cells isolated from healthy donor PBMCs and added to the channels at a 1:1 ratio of effector to target cell (figure 5A). To de-enrich for non-specific T cells, we tested three different constant shear levels for 5 min to determine the amount of shear stress needed to enrich for EGFR-specific CD8+ T cells. We observed a progressive enrichment of EGFR-specific T cells as shear stress increased from 1 to 3 dyne/cm2. At 3 dyne/cm2, the proportion of antigen- specific T cells surpassed that of non-specific T cells, suggesting that this level of shear stress may be sufficient for the enrichment of EGFR- specific T cells (figure 5A–C). To further evaluate the versatility of ATTACH, we tested these conditions using a different TCR-T popu- lation targeting an HLA-A*02:01 epitope derived from FOXM1 (figure 5D).45 Using CD8+ T cells isolated from HLA- A*02:01 healthy donor PBMCs (non-specific), we tested two different cell ratios (5:1 and 1:1). As shown in figure 5E, FOXM1-specific TCR-T cells were 2.6-fold (p=0.019) enriched compared with non-specific T cells in channels with a 1:1 (non-specific: FOXM1) ratio and by 4.4- fold (p=0.0084) in channels with a 5:1 ratio (figure 5E). Moreover, FOXM1- specific TCR-T cells were enriched by 22.4% (p=0.034) following ATTACH with an initial 1:1 input ratio, and by 35.2% (p=0.027) with an initial 5:1 input ratio (figure 5F,G). To confirm recovered T cells maintained their functional potential, channels were incubated overnight, and MIP-1β secretion and cyto- toxicity were quantified, confirming functional potential following ATTACH (figure 5H, online supplemental figure S2A,B).Selective depletion of bystander T cells using ATTACHLastly, we evaluated ATTACH’s ability to selectively enrich rare, antigen-specific T cells while simultaneouslyFigure 3 ATTACH enriches for antigen-specific T cells using ramping shear stress. B6 and OT-I splenocytes were combined at a ratio of 1:1 followed by a ramped shear stress pattern. (A) Immunofluorescence staining of B6 (green) or OT-I (red) splenocytes bound to LLC- JSP- OVA tumors (blue) before (left) and after (right) ATTACH. (B) Number of B6 and OT-I splenocytes per HPF based on shear stress being applied. (C) Proportion of B6 and OT-I remaining per HPF based on shear stress being applied. (D) Number of B6 and OT-I remaining after ramping shear stress was applied. Data are presented as mean±SEM; n=3 independent microfluidic chips. Statistical significance was assessed using two-tailed unpaired t-tests (*p<0.05 and ***p<0.001). ATTACH, Assessment of T Cells Tethered to Antigen Class I Histocompatibility; LLC-JSP, Lewis Lung Carcinoma; HPF, High-Power Field. ### Page 8 8MontoyaflA, et al. J Immunother Cancer 2026;14:e014960. doi:10.1136/jitc-2026-014960Open accessFigure 4 ATTACH allows recovery of antigen-specific T cells via low shear de-enrichment. (A) Overview of de-enrichment strategy. (B) OT-I CD8+ T cells were added to channels seeded with LLC-JSP-OVA or LLC-JSP cells, and ATTACH de-enrichment was applied. Immunofluorescence staining of OT-I (green) CD8+ T cells bound to LLC-JSP-OVA tumors (blue) before (left) and after (right) ATTACH using de-enrichment. B6 and OT-I CD8+ T cells were combined at a ratio of 1:1 followed by de-enrichment using ATTACH. (C) Immunofluorescence staining of B6 (red) and OT-I (green) CD8+ T cells bound to LLC-JSP (top) and LLC-JSP- OVA (bottom) tumors (blue) before (left) and after (right) ATTACH. (D) Proportion of OT-I remaining after ATTACH in LLC- JSP- OVA versus LLC-JSP. (E) Proportion of OT-I and B6 retained after ATTACH de-enrichment. (F) Proportion of OT-I and B6 retained after ATTACH de-enrichment in LLC-JSP-OVA versus LLC-JSP. Data are presented as mean±SEM. Each data point represents an independent microfluidic chip. For (D), n=7 independent microfluidic chips; for (E, F), n=2 independent microfluidic chips. Statistical significance was assessed using two-tailed unpaired t-tests (ns, not significant, *p<0.05 and ***p<0.001). ATTACH, Assessment of T Cells Tethered to Antigen Class I Histocompatibility; LLC-JSP, Lewis Lung Carcinoma. ### Page 9 9MontoyaflA, et al. J Immunother Cancer 2026;14:e014960. doi:10.1136/jitc-2026-014960Open accessdepleting bystander populations. Using a 1:1 mixture of HER2-specific TCR-T cells and non-specific CD8+ T cells, we collected the washed and retained fractions and stimulated them with either the HER2 peptide or a CEF peptide pool, a well-characterized mix of immuno- dominant viral epitopes derived from cytomegalovirus,Figure 5 Human antigen- specific T cell enrichment using ATTACH. (A) Immunofluorescence staining showing EGFR-specific CD8+ T cells (blue) and non- specific CD8+ T cells (red) bound to H1975-A*03:01 tumor cells (green) before (left) and after (right) ATTACH. (B) Percentage of EGFR-specific and non-specific CD8+ T cells remaining after ATTACH under different shear stresses (1, 2, and 3 dyne/cm²). (C) Proportion of EGFR-specific and non-specific CD8+ T cells under varying shear stresses applied during ATTACH. (D) Immunofluorescence staining of FOXM1-specific CD8+ T cells (blue) and non-specific CD8+ T cells (red) bound to H1975- A*02:01 tumor cells (green) before (left) and after (right) ATTACH. The T cell populations were mixed at two initial ratios of FOXM1- specific to non- specific cells (1:5 and 1:1). (E) Percentage of FOXM1-specific and non-specific CD8+ T cells remaining after ATTACH. (F) Proportion of FOXM1-specific and non-specific CD8+ T cells retained after ATTACH, starting with an input ratio of 1:1 (NS:FOXM1). (G) Proportion of FOXM1-specific and non-specific CD8+ T cells retained after ATTACH, starting with an input ratio of 5:1 (NS:FOXM1). (H) Concentration of MIP-1β measured after 24 hours of incubation on slides post-ATTACH. A separate chip with an equivalent amount of E7-specific T cells was used as a control. For (B, C), n=4 independent microfluidic chips; for (E), n=3; for (F), n=3; for (G), n=3; and for (H), n=3 independent microfluidic chips. Statistical significance was assessed using two-tailed unpaired t-tests (*p<0.05,**p<0.01 and ***p<0.001). ATTACH, Assessment of T Cells Tethered to Antigen Class I Histocompatibility; NS, Non-Specific. ### Page 10 10MontoyaflA, et al. J Immunother Cancer 2026;14:e014960. doi:10.1136/jitc-2026-014960Open accessEpstein-Barr virus, and influenza virus, commonly used as a control for detecting memory T cell responses.46 ATTACH preferentially retained HER2-reactive (anti-tumor) T cells while depleting CEF-reactive (anti-viral) bystanders, as assessed by ELISpot (figure 6A,B). While anti-tumor responses were largely preserved (~100%), anti-viral responses were largely lost (~4%–22%). To assess sensitivity, we next diluted antigen- specific populations to a 1:1000 ratio (0.1%). Even at this low frequency, ATTACH achieved a significant 10-fold enrich- ment of HER2-specific T cells along with a corresponding depletion of non- specific CD8+ T cells (figure 6C–E). These results highlight ATTACH’s robust ability to isolate rare, antigen-specific T cells from complex mixtures by enriching for desired clonotypes while depleting non-specific and bystander T cells.DISCUSSIONACT using TILs has shown promise for the treatment of solid tumors.1–9 However, a key limitation is the consid- erable proportion of T cells present in the tumor micro- environment lacking specificity toward tumor antigens, limiting its effectiveness in targeting and eliminating cancer cells.10–13 Several methods have been developed to identify tumor antigen-specific T cells; however, these methods have proven time- consuming and costly. While automated approaches to assess TCR avidity have been previously described,47–49 these methods are primarily limited to measurement, whereas ATTACH functions as a platform for the isolation of T cells based on their avidity, with the added advantage of recovering viable, functional cells for downstream applications. Herein, we present a method which leverages antigen-specific T cell avidity, using tumor cells as a natural source of endogenously-processed and -presented peptide-MHC complexes to rapidly enrich antigen-specific T cells within a matter of minutes. In the first generation of ATTACH, we show the feasibility to enrich for antigen-specific T cells based on binding avidity to cognate antigens without any need for specialized microfluidics slides or equipment. Given that our assay focuses on MHC I antigens, the first generation of ATTACH demonstrated specific MHC I/CD8 interac- tions by enriching for CD8+ T cells in the final fraction with a fourfold greater retention of OT-I CD8+ T cells in flasks where OVA257- 264 was presented, therefore, confirming antigen- specific CD8+ enrichment. Furthermore, micro- fluidic experiments demonstrated that shear stress can be effectively utilized to isolate antigen-specific T cell popu- lations across different HLAs with low, constant shear stress proving sufficient to selectively deplete non-specificFigure 6 ATTACH allows rapid depletion of bystander T cells and enrichment of rare anti-tumor T cell clonotypes. (A) ELISpot well images from input fraction, washed fraction, and retained fraction pulsed with CEF (anti-viral) or HER2 (anti-tumor) peptide after ATTACH. (B) Total IFN-γ spots in retained fraction normalized to the input value. HER2 and non-specific (NS) CD8+ T cells were mixed at a 1:1000 ratio and added to channels followed by ATTACH. (C) Percentage change of HER2-specific and NS T cells after ATTACH. (D) Fold change of HER2-specific T cells versus non-specific T cells after ATTACH. Fold change was calculated as the percentage of each population within the total cell population after ATTACH divided by its corresponding percentage before ATTACH. (E) Percentage of HER2-specific T cells before and after ATTACH based on total T cells. Data are presented as mean±SEM. Each data point represents an independent microfluidic chip. For (B), n=3 independent microfluidic chips; for (C–E), n=4 independent microfluidic chips. Statistical significance was assessed using two-tailed unpaired t-tests (*p<0.05, **p<0.01, ***p<0.001). ATTACH, Assessment of T Cells Tethered to Antigen Class I Histocompatibility; NS, non specific. ### Page 11 11MontoyaflA, et al. J Immunother Cancer 2026;14:e014960. doi:10.1136/jitc-2026-014960Open accessT cell clones. To our knowledge, these studies are the first to demonstrate the ability to isolate neoantigen-specific CD8+ T cells via microfluidic shear stress. Ashby and colleagues recently described an approach leveraging custom microfluidic chips for the isolation of tumor-associated antigen- specific CD8+ T cells,50 although the studies do exhibit numerous key tech- nical and conceptual distinctions. First, ATTACH can be performed manually and without any specialized equip- ment. Next, through ATTACH, we demonstrate the effi- cacy of applying 10–100 fold lower levels of shear stress, while still achieving almost 10-fold enrichment, which may present key advantages. Increasing shear stress during microfluidic selection has been shown to enhance the purity of enriched high-avidity T cells; however, we have observed that higher shear stress levels were associ- ated with an increase in cell debris in our hands, poten- tially due to mechanical damage or increased cell lysis caused by excessive force.51–53 Furthermore, the antigens and reactive clonotypes present in a tumor are seldom known beforehand, and antigen-specific T cells span a wide range of affinities, making it difficult to fine-tune shear stress to achieve pure enrichment without case-by-case optimization and limiting translational potential. Moreover, recovering a mix of low, intermediate, and high affinity T cells, rather than only high affinity cells, may be more favorable, as high affinity T cells are prone to overstimulation and activation-induced cell death,54–56 while low and intermediate affinity T cells can maintain functionality over extended periods and support robust immune responses.57 58 By focusing on constant low-level shear stress de-enrichment, we believe ATTACH is highly versatile and could increase the potential of this approach for downstream applications. To broaden the applicability of ATTACH, we demon- strate its capacity to selectively deplete bystander T cells while enriching for rare, antigen-specific populations. This capability addresses a critical challenge in ACT, particularly in the context of TILs, where a substantial proportion of recovered T cells are often reactive to irrel- evant antigens. For instance, T cells specific for common viral epitopes such as those found in the CEF peptide pool are frequently present among TILs and can dominate the expansion process during ex vivo culture, ultimately diluting the therapeutic impact of truly tumor-specific clonotypes.10 Our results show that ATTACH can effec- tively eliminate these non- specific bystanders from mixed populations, achieving robust depletion of CEF-reactive T cells while preserving antigen-specific reactivity toward neoantigens. This is notable given studies suggesting that only ~10% of TILs are truly tumor-specific.10 Accordingly, we tested ATTACH under increasingly stringent condi- tions, testing its ability to recover antigen-specific T cells representing just 0.1% of the total population. This is especially important for improving the precision and effi- cacy of TIL therapy, where conventional bulk expansion often fails to discriminate between functional and irrele- vant subsets. Notably, the T cells recovered post-ATTACH remain functionally competent, retaining their ability to recognize and respond to cognate antigen stimulation. Although the recovered fraction contains both T cells and tumor cells, ATTACH effectively increases the frequency of antigen- specific T cells within the recovered popula- tion. Together, these findings underscore the potential of ATTACH for numerous downstream and therapeutic applications. Furthermore, the antigen- specific nature of ATTACH is demonstrated by the low retention of C57BL/6 CD8+ T cells after the assay. The comparable retention of C57BL/6 CD8+ T cells on both LLC-JSP and LLC- JSP- OVA slides suggests that their retention was not driven by antigen- specific interactions. This observation aligns with previous studies in tumor microenvironments, where T cell interactions can be influenced by factors beyond antigen recognition.55–57 Moreover, a major advantage of ATTACH is the ability to purify antigen-specific T cells with a low amount of shear stress. In addition, it can enrich for antigen- specific T cells when these are at a lower frequency than non-specific T cells, closer in line with real-world circumstances. These findings suggest that ATTACH may have significant translational potential. ATTACH does exhibit certain limitations. First, T cell attachment is governed by stochastic cell–cell encounters and limited interaction time, such that not all antigen-specific T cells establish stable contacts. Consequently, shear stress may also remove a fraction of antigen-specific cells, reducing overall recovery. This effect may be further exacerbated when targeting endogenous neoantigen-specific T cells, which occur at very low precursor frequen- cies and may require additional expansion or stimulation to enable reliable detection after the assay. In addition, assay performance depends on target cell quality and antigen presentation, as reduced tumor cell viability or insufficient HLA/antigen expression can impair stable T cell–tumor interactions and limit enrichment efficiency, particularly in patient-derived samples. These challenges are compounded in clinical settings by the limited avail- ability of autologous tumor cell lines; while patient-derived xenograft models may provide an alternative source of tumor material, their use introduces additional logistical and translational constraints. In conclusion, we demonstrate the feasibility of isolating tumor antigen-specific T cells based on binding affinity using a microfluidic platform via ATTACH. By using tumor cells as a natural source of endogenously-processed and -presented MHC- peptide complexes, ATTACH can rapidly enrich for human and murine antigen-specific CD8+ T cells based on TCR avidity and circumvent many of the challenges associated with current antigen iden- tification methodologies in a rapid and cost-effective manner. All in all, ATTACH allows simple yet effective enrichment of functionally intact antigen-specific T cells, which can be used for multiple downstream applications.Acknowledgements Some figures were created with BioRender.com. ### Page 12 12MontoyaflA, et al. J Immunother Cancer 2026;14:e014960. doi:10.1136/jitc-2026-014960Open accessContributors AMA: conceptualization, data curation, formal analysis, investigation, visualization, writing–original draft, writing–review, and editing. MLF: conceptualization, data curation, formal analysis, validation, investigation, methodology, writing–review, and editing. PJ: conceptualization, data curation, software, investigation, visualization, writing–review, and editing. HN: conceptualization, data curation, investigation, methodology, writing–review, and editing. MZ: data curation, investigation, writing–review, and editing. EB: data curation, investigation, writing–review, and editing. JKS: data curation, software, investigation, visualization, writing–review, and editing. LLP: data curation, investigation, writing–review, and editing. SRCS: data curation, investigation, writing–review, and editing. TC: resources, writing–review, and editing. MG: resources, writing–review, and editing. DLG: resources, writing–review, and editing. JZ: resources, data curation, writing–review, and editing. ED: resources, data curation, writing–review, and editing. LK: writing–review and editing. PLW: conceptualization, data curation, formal analysis, investigation, writing–review, and editing. JVH: supervision, writing–review, and editing. AR: conceptualization, resources, supervision, funding acquisition, writing–original draft, project administration, writing–review, and editing. AR is the guarantor of this work, responsible for the overall content and ensuring the integrity of the research.Funding AR is supported by an American Association for Cancer Research (AACR) Career Development Award in Lung Cancer Research (23-20- 01- REUB), Cancer Prevention and Research Institute of Texas awards including a High Impact High Reward Award (RP210137) and Individual Investigator Research Awards (RP230363 and RP260357), a Department of Defense Lung Cancer Research Program Idea Development Award (HT9425-23- 1- 1021), the Exon 20 Group, the Happy Lungs Project, LUNGevity Foundation awards including an EGFR Resisters/LUNGevity Foundation EGFR-positive award and a RETpositive/LUNGevity Translational Research Award, MD Anderson’s Lung Cancer Moon Shot, an NCI SBIR Innovative Concept Award (75N91025C00012), an NIH/NCI R21 (R21CA283852), the Petrin Fund, Rexanna’s Foundation (Bruce Campbell Research Grant and The Lance Bertrand Research Grant), the Salgado Family Charitable Fund, the Troper Wojcicki Foundation, the University Cancer Foundation Institutional Research Grant program at the University of Texas MD Anderson Cancer Center, and the Waun Ki Hong Lung Cancer Research Fund.Disclaimer The study sponsors had no role in the study design, collection, analysis, interpretation of the data; in the writing of the report; or in the decision to submit the paper for publication.Competing interests J.Z. reports research funding from Merck, Johnson and Johnson, Novartis, Summit, Hengenix and consultant fees from BMS, Johnson and Johnson, AstraZeneca, Geneplus, OrigMed, Innovent, Varian, Catalyst outside the submitted work. J.V.H. reports research funding from AstraZeneca, Boehringer-Ingelheim, Mirati, Bristol-MyerMyers Squibb, Takeda, Taiho, has served on advisory committees for AbbVie, Amgen, AnHeart Therapeutics, Arrivent, AstraZeneca, BioNTech AG, BI, BMS, Curio Science, DAVA Oncology, Eli LilyLilly & Co, EMD Serono, Janssen Pharmaceuticals, Jazz Pharmaceuticals, Mirati Therapeutics, Moffitt Cancer Center, ModeX, Novartis Pharmaceuticals, OncoCyte, Pfizer, Sanofi, Spectrum Pharmaceuticals, Takeda, Ottimo Pharma, ALK Positive, served at speakers bureau for MJH Events, Guidepoint, Ideology Health, and received licensing/royalties from Spectrum.T.C. has acted as a consultant and/ or advisor and/or speaker for, and/or has received travel support from ASCO Post, AstraZeneca, Bio Ascend, Bristol Myers Squibb, Clinical Care Options, Daiichi Sankyo, Genentech, IDEOlogy Health, Medical Educator Consortium, Medscape, Merck, OncLive, oNKo-innate, PEAK Medicals, PeerView, Pfizer, Physicians’ Education Resource, RAPT Therapeutics, Regeneron, and Targeted Oncology. D.L.G. reports honoraria for scientific advisory boards from AstraZeneca, Sanofi, Alethia Biotherapeutics, Menarini Richerche, Eli Lilly, 4D Pharma, and Onconova and research support from Janssen, Takeda, Astellas, Ribon Therapeutics, Boehringer Ingelheim, NGM Biopharmaceuticals, Mirati Therapeutics, and AstraZeneca. A.R., A.M.A., M.L.F., H.N., P.J., P.W. and J.V.H. are inventors on patent application PCT/ US2025/054052 (published as WO2026097098A1), entitled “Methods and Systems for Unbiased Microfluidic Isolation of Antigen-Specific T Cells,” which relates to the ATTACH platform used in this study.Patient consent for publication Not applicable.Ethics approval Ethical approval for all animal studies was obtained from the MD Anderson Institutional Animal Care and Use Committee (IACUC) under protocol #00002019-RN01. Female transgenic OT-I mice (RRID: IMSR_JAX:003831) and C57BL/6 mice (RRID: IMSR_JAX:000664) were obtained from Jackson Laboratories and used at 8–12 weeks of age at the time of sacrifice. All animal procedures were conducted in accordance with institutional guidelines for animal care and use. TCR transduction experiments were performed under approved Institutional Biosafety Committee (IBC) protocol #RM00006717-RN00_AR003. This study did not involve human participants; therefore, informed consent was not required.Provenance and peer review Not commissioned; externally peer reviewed.Data availability statement Data sharing not applicable as no datasets were generated and/or analyzed for this study. Data are available on reasonable request. All data relevant to the study are included in the article or uploaded as supplementary information. Data are available to all researchers on request.Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer- reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.Open access This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See https://creativecommons.org/licenses/by-nc/4.0/.ORCID iDsAmanda Montoya https://orcid.org/0009-0001-2055-5084 Meredith L Frank https://orcid.org/0000-0002-3658-0057 Tina Cascone https://orcid.org/0000-0003-3008-5407 Jianjun Zhang https://orcid.org/0000-0001-7872-3477 Lydia Kavraki https://orcid.org/0000-0003-0699-8038 Alexandre Reuben https://orcid.org/0000-0003-4510-0382REFERENCES1 Rosenberg SA, Restifo NP. Adoptive cell transfer as personalized immunotherapy for human cancer. Science 2015;348:62–8.2 Creelan BC, Wang C, Teer JK, et al. Tumor-infiltrating lymphocyte treatment for anti- PD- 1-resistant metastatic lung cancer: a phase 1 trial. Nat Med 2021;27:1410–8.3 Panahi M, Rezagholizadeh F, Mollazadehghomi S, et al. The association between CD3+ and CD8+tumor- infiltrating lymphocytes (TILs) and prognosis in patients with pancreatic adenocarcinoma.Cancer Treat Res Commun 2023;35:100699.4 Rodrigo JP, Sánchez- Canteli M, López F, et al. Tumor- Infiltrating Lymphocytes in the Tumor Microenvironment of Laryngeal Squamous Cell Carcinoma: Systematic Review and Meta-Analysis.Biomedicines 2021;9:486.5 Romagnoli G, Wiedermann M, Hübner F, et al. Morphological Evaluation of Tumor- Infiltrating Lymphocytes (TILs) to Investigate Invasive Breast Cancer Immunogenicity, Reveal Lymphocytic Networks and Help Relapse Prediction: A Retrospective Study. Int J Mol Sci 2017;18:1936.6 Köhn P, Lalos A, Posabella A, et al. High density of CXCL12-positive immune cell infiltration predicts chemosensitivity and recurrence-free survival in ovarian carcinoma. J Cancer Res Clin Oncol2023;149:17943–55.7 Sano S, Akiyoshi T, Yamamoto N, et al. Intratumoral Budding and CD8-Positive T-cell Density in Pretreatment Biopsies as a Predictor of Response to Neoadjuvant Chemoradiotherapy in Advanced Rectal Cancer. Clin Colorectal Cancer 2023;22:411–20.8 Horne ZD, Jack R, Gray ZT, et al. High Levels Of Tumor-infiltrating Lymphocytes And The Absence Of Angiolymphatic Invasion Increase Recurrence-free Survival In Stage 1A Non- Small Cell Lung Cancer.Journal of Surgical Research 2011;165:189–90.9 Schoenfeld AJ, Lee SM, Doger de Spéville B, et al. Lifileucel, an Autologous Tumor-Infiltrating Lymphocyte Monotherapy, in Patients with Advanced Non- Small Cell Lung Cancer Resistant to Immune Checkpoint Inhibitors. Cancer Discov 2024;14:1389–402.10 Simoni Y, Becht E, Fehlings M, et al. Bystander CD8+ T cells are abundant and phenotypically distinct in human tumour infiltrates.Nature 2018;557:575–9.11 Scheper W, Kelderman S, Fanchi LF, et al. Low and variable tumor reactivity of the intratumoral TCR repertoire in human cancers. Nat Med 2019;25:89–94. ### Page 13 13MontoyaflA, et al. J Immunother Cancer 2026;14:e014960. doi:10.1136/jitc-2026-014960Open access12 Reuben A, Zhang J, Chiou S-H, et al. Comprehensive T cell repertoire characterization of non-small cell lung cancer. Nat Commun 2020;11:603.13 Chiou S-H, Tseng D, Reuben A, et al. Global analysis of shared T cell specificities in human non- small cell lung cancer enables HLA inference and antigen discovery. Immunity 2021;54:586–602.14 Robbins PF, Lu Y-C, El-Gamil M, et al. Mining exomic sequencing data to identify mutated antigens recognized by adoptively transferred tumor-reactive T cells. Nat Med 2013;19:747–52.15 Kristensen NP, Heeke C, Tvingsholm SA, et al. Neoantigen- reactive CD8+ T cells affect clinical outcome of adoptive cell therapy with tumor- infiltrating lymphocytes in melanoma. J Clin Invest2022;132:e150535.16 Gros A, Robbins PF, Yao X. tumor-reactive repertoire infiltrating human tumors. J Clin Invest 2014;124:2246–59.17 Altman JD, Moss PA, Goulder PJ, et al. Phenotypic analysis of antigen-specific T lymphocytes. Science 1996;274:94–6.18 Schmidt J, Dojcinovic D, Guillaume P, et al. Analysis, Isolation, and Activation of Antigen-Specific CD4(+) and CD8(+) T Cells by Soluble MHC- Peptide Complexes. Front Immunol 2013;4:218.19 Dolton G, Zervoudi E, Rius C, et al. Optimized Peptide- MHC Multimer Protocols for Detection and Isolation of Autoimmune T-Cells. Front Immunol 2018;9.20 Rius C, Attaf M, Tungatt K, et al. Peptide-MHC Class I Tetramers Can Fail To Detect Relevant Functional T Cell Clonotypes and Underestimate Antigen- Reactive T Cell Populations. J Immunol2018;200:2263–79.21 Effenberger M, Stengl A, Schober K, et al. FLEXamers: A Double Tag for Universal Generation of Versatile Peptide-MHC Multimers.J Immunol 2019;202:2164–71.22 Li G, Bethune MT, Wong S, et al. T cell antigen discovery via trogocytosis. Nat Methods 2019;16:183–90.23 Joglekar AV, Leonard MT, Jeppson JD, et al. T cell antigen discovery via signaling and antigen-presenting bifunctional receptors. Nat Methods 2019;16:191–8.24 Tran E, Turcotte S, Gros A, et al. Cancer immunotherapy based on mutation-specific CD4+ T cells in a patient with epithelial cancer.Science 2014;344:641–5.25 Birnbaum ME, Mendoza JL, Sethi DK, et al. Deconstructing the peptide-MHC specificity of T cell recognition. Cell2014;157:1073–87.26 Pasetto A, Gros A, Robbins PF, et al. Tumor- and Neoantigen-Reactive T-cell Receptors Can Be Identified Based on Their Frequency in Fresh Tumor. Cancer Immunol Res 2016;4:734–43.27 Yossef R, Tran E, Deniger DC, et al. Enhanced detection of neoantigen-reactive T cells targeting unique and shared oncogenes for personalized cancer immunotherapy. JCI Insight 2018;3:e122467.28 Seliktar-Ofir S, Merhavi-Shoham E, Itzhaki O, et al. Selection of Shared and Neoantigen-Reactive T Cells for Adoptive Cell Therapy Based on CD137 Separation. Front Immunol 2017;8:1211.29 Thommen DS, Koelzer VH, Herzig P, et al. A transcriptionally and functionally distinct PD-1+ CD8+ T cell pool with predictive potential in non-small-cell lung cancer treated with PD-1 blockade. Nat Med2018;24:994–1004.30 Duhen T, Duhen R, Montler R, et al. Co-expression of CD39 and CD103 identifies tumor-reactive CD8 T cells in human solid tumors.Nat Commun 2018;9:2724.31 Gros A, Parkhurst MR, Tran E, et al. Prospective identification of neoantigen-specific lymphocytes in the peripheral blood of melanoma patients. Nat Med 2016;22:433–8.32 Scott AC, Dündar F, Zumbo P, et al. TOX is a critical regulator of tumour-specific T cell differentiation. Nature 2019;571:270–4.33 Reynisson B, Alvarez B, Paul S, et al. NetMHCpan-4.1 and NetMHCIIpan- 4.0: improved predictions of MHC antigen presentation by concurrent motif deconvolution and integration of MS MHC eluted ligand data. Nucleic Acids Res 2020;48:W449–54.34 Trolle T, Nielsen M. NetTepi: an integrated method for the prediction of T cell epitopes. Immunogenetics 2014;66:449–56.35 Pei B, Hsu Y-H. IConMHC: a deep learning convolutional neural network model to predict peptide and MHC-I binding affinity.Immunogenetics 2020;72:295–304.36 Fasoulis R, Rigo MM, Antunes DA, et al. Transfer learning improves pMHC kinetic stability and immunogenicity predictions.Immunoinformatics (Amst) 2024;13:100030.37 Tickotsky N, Sagiv T, Prilusky J, et al. McPAS-TCR: a manually curated catalogue of pathology-associated T cell receptor sequences. Bioinformatics 2017;33:2924–9.38 Shugay M, Bagaev DV, Zvyagin IV, et al. VDJdb: a curated database of T-cell receptor sequences with known antigen specificity. Nucleic Acids Res 2018;46:D419–27.39 Lewis MD, de Leenheer E, Fishman S, et al. A reproducible method for the expansion of mouse CD8+ T lymphocytes. J Immunol Methods 2015;417:134–8.40 Reuben A, Phénix M, Santos MM, et al. The WT hemochromatosis protein HFE inhibits CD8+ T-lymphocyte activation. Eur J Immunol2014;44:1604–14.41 Davis KA, Goh JZ, Sebastian AH, et al. Increased Retention of Cardiac Cells to a Glass Substrate through Streptavidin-Biotin Affinity. ACS Omega 2021;6:17523–30.42 Bradski G. The opencv library. Dr. Dobb’s Journal of Software Tools; 2000.43 Lukežič A, Vojíř T, Čehovin Zajc L, et al. Discriminative Correlation Filter Tracker with Channel and Spatial Reliability. Int J Comput Vis2018;126:671–88.44 Bontekoe E, Zhang M, Jiang P, et al. 358 T cell receptor engineering targeting egfr-mutant non-small cell lung cancer. J Immunother Cancer 2024;12:A414.45 Bontekoe E, Zhang M, Jiang P, et al. FOXM1-Specific TCR-Engineered T Cells Target Non-Small Cell Lung Cancer. Cancer Immunol Res 2026;14:958–73.46 Currier JR, Kuta EG, Turk E, et al. A panel of MHC class I restricted viral peptides for use as a quality control for vaccine trial ELISPOT assays. J Immunol Methods 2002;260:157–72.47 Kamsma D, Bochet P, Oswald F, et al. Single-Cell Acoustic Force Spectroscopy: Resolving Kinetics and Strength of T Cell Adhesion to Fibronectin. Cell Rep 2018;24:3008–16.48 Moser J, Bar–Ephraim Y, Eberlein J, et al. 180 The in-vitro measure of T cell avidity between tumor-effector pairs, not affinity, predicts optimal anti-tumor response in vivo. J Immunother Cancer2022;10.49 Carr A, Mateyka LM, Scheu SJC, et al. Advances in preclinical TCR characterization: leveraging cell avidity to identify functional TCRs.Biol Chem 2024;405:517–29.50 Ashby JF, Schmidt J, Kc N, et al. Microfluidic T Cell Selection by Cellular Avidity. Adv Healthc Mater 2022;11:e2200169.51 Chittur KK, Mcintire LV, Rich RR. Shear Stress Effects on Human T Cell Function. Biotechnol Prog 1988;4:89–96.52 Radley G, Laura Pieper I, Thomas BR, et al. Artificial shear stress effects on leukocytes at a biomaterial interface. Artif Organs2019;43:E139–51.53 Lien S-C, Chang S- F, Lee P-L, et al. Mechanical regulation of cancer cell apoptosis and autophagy: Roles of bone morphogenetic protein receptor, Smad1/5, and p38 MAPK. Biochim Biophys Acta2013;1833:3124–33.54 Wijewarnasuriya D, Bebernitz C, Lopez AV, et al. Excessive Costimulation Leads to Dysfunction of Adoptively Transferred T Cells. Cancer Immunol Res 2020;8:732–42.55 Spear TT, Evavold BD, Baker BM, et al. Understanding TCR affinity, antigen specificity, and cross-reactivity to improve TCR gene- modified T cells for cancer immunotherapy. Cancer Immunol Immunother 2019;68:1881–9.56 Singhaviranon S, Dempsey JP, Hagymasi AT, et al. Low-avidity T cells drive endogenous tumor immunity in mice and humans. Nat Immunol2025;26:240–51.57 Knudson KM, Goplen NP, Cunningham CA, et al. Low- affinity T cells are programmed to maintain normal primary responses but are impaired in their recall to low-affinity ligands. Cell Rep2013;4:554–65.58 Martinez RJ, Evavold BD. Lower Affinity T Cells are Critical Components and Active Participants of the Immune Response. Front Immunol 2015;6:468.