All methods performed in this study adhered to the relevant guidelines and regulations. Human specimens were collected in accordance with the specifications of the Amgen Human Tissue Science Center study, under the approval of site-specific Institutional Review Boards or Ethics Boards, with informed consent obtained in compliance with all applicable laws and regulations.

Tagap knockout mice were generated using CRISPR-based extreme genome editing (EGE™) technology by Biocytogen Pharmaceuticals. Two sgRNAs targeting the intron 4 and intron 7 were designed to mediate the deletion of 2.3kb chromosomal region (Exon5, Exon6, and Exon7) at the Tagap locus in the mouse genome.

Briefly, two sgRNAs were designed by the CRISPR design tool (http://cctop.cos.uni-heidelberg.de:8043/) and then were screened for on-target activity using a Universal CRISPR Activity Assay (UCA™, Biocytogen Pharmaceuticals). The T7 promoter sequence was added to the sgRNA template by PCR amplification in vitro.

The nuclease mRNA and two sgRNAs were co-microinjected into the cytoplasm of one-cell stage fertilized C57BL/6J eggs and then the injected embryos were transferred into oviducts of pseudopregnant females to generate F0 mice. Of 866 zygotes transferred, 131 pups were obtained, of which 6 F0 mice were confirmed as positive founders. F0-positive founders were bred with wild-type (WT) C57BL/6J mice to generate F1 pups. Germline transmission had been achieved from founder line E2S0192-0019. F1 heterozygous mice were confirmed by two rounds of tail genomic PCR and DNA sequencing.

Six- to eight-week-old male and female WT and TAGAP KO mice on C57BL/6 background and female Rag2^-/- mice purchased from Taconic Biosciences were used in this study. All the mice were maintained on a 12-hour light/dark cycle under specific pathogen-free conditions with access to food and water ad libitum. Protocols were approved by the Institutional Animal Care and Use Committee of Amgen.

All recombinant TAGAP proteins (WT, K147, and A123) were produced and purchased from WuXi Biologics. The sequence design is attached in Supplementary Table 1 . GAP activity was determined with the RhoGAP assay kit (Cytoskeleton, Inc.; #BK105). Briefly, GTPase activity was measured by detecting inorganic phosphate (Pi) released from GTP hydrolysis with or without GAP molecules in the reaction. The reactions were initiated with the addition of 200 µM GTP to a RhoGTPase in the presence or absence of TAGAP at 37 °C for 5 min, followed by adding CytoPhos reagent. The absorbance at 650 nM indicated the concentration of released phosphate and was recorded at 2-min intervals for 1 hr to monitor reaction kinetics.

Cells were lysed in RIPA buffer (ThermoFisher; #89900) in the presence of a protease inhibitor cocktail (ThermoFisher; #78429). Protein concentrations in cell lysates were quantified using the BCA assay (ThermoFisher; #23225). Thirty μg of total protein lysate were used for Western blotting. Western blotting was performed using 4-12% Bis-Tris gels, followed by transferred onto NC membranes. After blocking with 5% skim milk powder, the membranes were incubated overnight at 4°C with primary antibodies at 1:1000 dilution and secondary antibodies were used at 1:10000 dilution at room temperature for 2 h. All primary and secondary antibodies are listed in Supplementary Table 2 . Protein samples were visualized using a ChemiDoc XRS+ imaging system (BioRad).

Three single guide RNAs (sgRNAs) targeting the TAGAP gene as well as two non-targeting (NT) control sgRNAs were predesigned and purchased from IDT. The sequences of the sgRNAs are shown in Supplementary Table 3 . Transfection was performed using the Neon Transfection System (ThermoFisher; #MPK10025 and MPK1025) according to the manufacturer’s guide and protocols. Briefly, human primary monocytes/pan T cells/HL-60/THP-1/Jurkat cells were used in individual reactions mixed with sgRNA-nuclease ribonucleoprotein complexes. Knockout efficiency was confirmed by Western blotting. KO pool was used to minimize clonal bias in each experiment.

HL-60 cells were differentiated in 1.3% DMSO for 6 days as described before (14). Two hundred and fifty thousand differentiated cells in RPMI containing 0.5% BSA were then seeded in inserts of transwell plates (Corning; #3421) with the bottom wells loading different concentrations of fMLP (Sigma; #F3506) or LTB4 (TOCRIS; #F3506) depending on experimental settings. After the incubation at 37°C for 45 min, migrated cells (cells in the bottom wells) were quantified with CellTiter-Glo (Promega; #G7570) per manufacturer’s protocols.

The phagocytic activity of HL-60 cells was measured by the uptake of opsonized red fluorescent pHrodo Escherichia coli bioparticles (Invitrogen; #P35361) as described by Jun et al. with modification (15). Briefly, 5 × 10⁴ DMSO-induced differentiated HL-60 cells were mixed with pHrodo E. coli bioparticles at a 1-to-30 ratio in HBSS containing 10 mM HEPES and 5% FBS. The mixture was incubated at 37°C and the kinetic uptake activity was recorded and analyzed using the Incucyte live cell imaging and analysis system (Sartorius).

Human PBMCs were purified from human whole blood using LeucoSep tubes according to the manufacturer’s recommendations. Briefly, 2X diluted whole blood with PBS was overlaid on the separation medium Ficoll (Cytiva, #17144003) in LeucoSep tubes (Greiner, #227290). Samples were spun at 800g with low acceleration and low brake for 20 min. After centrifugation, the buffy coat was collected for further purification using EasySep Human Monocyte or T Cell Isolation Kits (STEMCELL; #19359 or #17951, respectively).

Purified human monocytes were transfected as described above to generate NT and TAGAP KO cells. After the electroporation, cells were rested in RPMI containing 10% FBS and 50 ng/ml M-CSF (Gibco; # 300-25) for overnight. Cells were then cultured in fresh M-CSF-supplemented media for 7 days with replenishment of fresh media on day 3 and 5. On day 7, cells were harvested and stimulated with TLR or Dectin-1 ligands for 24 hr. All TLR and Dectin-1 ligands are listed in Supplementary Table 4 . After the stimulation, cells were harvested for Western blotting and supernatant was collected for MSD cytokine/chemokine analysis.

Jurkat and Jurkat-NFAT-GFP cells were transfected as described above to generate NT and TAGAP KO cells. For TCR-dependent stimulation, Jurkat lines in RPMI containing 10% FBS were seeded in aCD3-coated plates (Invitrogen; #16003785; concentration as each experiment indicated) with 1 µg/ml aCD28 (Invitrogen; #16028985) in suspension for 24hr. For TCR-independent stimulation, Jurkat lines in RPMI containing 10% FBS were cultured in the presence of 10 ng/ml PMA (InvivoGen; # tlrl-pma) and 500 ng/ml Ionomycin (InvivoGen; # inh-ion) for 24hr. Cells were then collected to test the activation marker expression and GFP levels via flow cytometry.

TAGAP KO Jurkat cells were transfected with constructs encoding WT TAGAP E147, the K147 variant, or the A123 variant, each tagged with a Myc-DDK tag at the C-terminus and cloned into the CD811A backbone vector (System Biosciences, #CD811A-1). An empty vector was used as a control. Transfections were performed using the SE Cell Line 4D-Nucleofector X Kit (Lonza, #V4XC-1032) following the manufacturer’s protocol. Briefly, 1 × 10⁶ cells were resuspended in 20 μl of Nucleofector Solution mixed with Supplementary Data Sheet 1 and 1 μg of plasmid DNA per reaction. Electroporation was carried out using the 4D-Nucleofector System (Lonza), and cells were immediately transferred to pre-warmed RPMI 1640 supplemented with 10% FBS. Transfected cells were rested for 48 hours prior to stimulation and downstream assays as described above.

Purified human pan T cells were transfected as described above to generate NT and TAGAP KO cells. Cells in RPMI containing 10% FBS and β-ME were seeded in 1 ug/ml aCD3-coated plates (Invitrogen; #16003785) with 1 µg/ml aCD28 (Invitrogen; #16028985) in suspension. Cells and supernatant were collected at 24-hr- and 72-hr-post-stimulation to test the activation marker expression via flow cytometry and cytokine levels via ELISA.

Th17 cell composition in small intestines (SI) from anti-CD3-challenged mice was determined by flow cytometry as described by Esplugues et al. previously (16). In short, WT and TAGAP KO mice were injected with anti-CD3 (15 µg; BioXCell; #BE0001-1) intraperitoneally (i.p.) three times at an interval of 2 days between injections and sacrificed 4 hr after the final injection. For the controls, isotype control antibodies (15 µg; BioXCell; # BE0091) were injected. The serum and the distal 2 cm of SI for tissue homogenate were collected for cytokine/chemokine measurement. The IEL and LPL from SI were collected with the lamina propria dissociation kit (Miltenyi Biotec; #130-097-410) according to the manufacturer’s manual. The dissociated IEL and LPL cells were loaded onto a Percoll gradient and centrifuged at 2000 rpm for 20 min. The cells between 32.5% and 80% Percoll were collected and used as purified IEL and LPL. Both purified IEL and LPL were stimulated with the cell stimulation cocktail (Invitrogen; #00-4970-93) in the presence of the protein transport inhibitor (Invitrogen; #00-4980-03) at 37°C for 4 hr prior to the intracellular staining for flow cytometry.

CD4⁺ T cells from spleens of WT or TAGAP KO mice were collected and purified by EasySep mouse CD4⁺ T cell isolation kits (STEMCELL; #19852). After the enrichment, total CD4⁺ T cells were FACS sorted and 5×10⁵ CD45RB^hi T cells were adoptively transferred (i.p.) into female Rag2^-/- recipient mice. Body weight and survival were monitored daily for 60 consecutive days. Harvested SI were processed for flow cytometry analysis as described above.

Single-cell suspensions were blocked, Live/Dead stained, surface stained, washed, fixed, and permeabilized then intracellularly stained as previously described (17). All flow antibodies used in the study are listed in Supplementary Table 2 . Annexin V staining was performed per the manufacturer’s recommendation (Biolegend; #640906). Flow cytometry was performed on BD Symphony A3 Cytometer.

V-PLEX Proinflammatory Panel 1 Human Kit (MSD; #K15049D-1), Human IL-2 ELISA kit (Invitrogen; BMS2212), and Humun IFN-γ quantikine ELISA kit (R&D Systems; #DIF50C) were used to measure the cytokine/chemokine levels from the supernatant of cultured human macrophages and T cells. U-PLEX Proinflam Combo 1 mouse kit (MSD: #K15713K-1) and mouse CCL20 quantikine ELISA kit (R&D Systems; #MCC200) were used to measure the cytokine/chemokine levels from mouse serum and SI tissue homogenate following the instruction from the manufacture.

Differentiated NT and TAGAP KO human macrophages were seeded into poly-D-lysine-coated glass-bottom culture slides (Corning, #354632) and stimulated with 100 ng/ml LPS for 24 hr. Following stimulation, cells were fixed with 4% paraformaldehyde (PFA) and permeabilized with 0.1% Triton X-100. After blocking with 1% bovine serum albumin (BSA) for 1 hr at room temperature, cells were stained with a cocktail of Phalloidin-FITC (Invitrogen, #F432) and DAPI (Invitrogen, #D1306) for 1 hr. Slides were washed three times with PBS, air-dried, and mounted using anti-fade mounting medium. Fluorescent images were acquired using a Leica SP8 confocal microscope.

Expression data from the cell lines were extracted from QIAGEN OmicSoft Cell Line release CellLine_B38_GC33, and expression data from primary human immune cells were extracted from Body Map release Blueprint_B37. Data were analyzed and visualized using QIAGEN OmicSoft Studio software, version 12.4 (QIAGEN, Redwood City, CA).

The data are shown as mean ± S.D. from at least three independent experiments including technical and biological replicates. The p values noted in the corresponding figure legends were calculated using Student’s t-test (among two groups) or one-way ANOVA (among multiple groups) in Prism (GraphPad Software 8.0, La Jolla, CA).

The involvement of TAGAP in IBD has been demonstrated in both animal and human studies (8–10, 18). Common non-coding variants in TAGAP locus have been identified in GWAS to be associated with the risk of Crohn’s disease (CD) (19), but it is still unclear whether the association is driven by gain or loss of function of TAGAP. A recent exosome sequencing study identified a rare missense variant rs41267765 C/T in TAGAP associated with CD risk (20). The protective allele (T) introduces a Glu (E) to Lys (K) substitution at residue 147, yielding an odds ratio of 0.786 and suggesting that TAGAP E147K is protective against CD. Given the evidence that TAGAP deletion exacerbates DSS-induced colitis severity, we hypothesized that the E147K mutation reduces IBD risk by altering TAGAP function.

To investigate this hypothesis, we acquired three human recombinant TAGAP proteins: wild-type (WT; E147), K147, and a negative-control A123. TAGAP A123 has decreased TAGAP activity as it harbors an Arg-to-Ala mutation at residue 123, a critical site for Rho protein interaction. We then assessed TAGAP-facilitated GTP hydrolysis by measuring the release of inorganic phosphate (Pi) over time. Compared with the WT protein, the K147 mutant facilitated GTP hydrolysis with an approximately 14% increase in Vmax ( Figures 1A, B ). Conversely, the A123 mutation significantly impaired TAGAP function, reducing Vmax by ~20% ( Figures 1A, B ). These findings indicated that the E147K mutation enhances TAGAP-dependent GTP hydrolysis activity, conferring a gain-of-function phenotype.

Analysis of publicly available RNA expression data from QIAGEN Omicsoft Body Map release Blueprint_B37 showed that TAGAP is exclusively expressed in leukocytes, with neutrophils exhibiting the highest RNA transcript levels ( Figure 2A ). However, at the protein level, TAGAP is more abundant in human ( Figure 2B ) and murine lymphocytes ( Figure 2C ) than in myeloid cells, with moderate expression in monocytes and undetectable levels in neutrophils.

To better understand the biology of TAGAP in leukocytes, first we sought to determine TAGAP’s function in myeloid cells. HL-60 cells, a promyeloblast line, were selected due to their relatively high TAGAP expression compared with THP-1 cells ( Supplementary Figure 1A ), a finding confirmed at the protein level via western blot ( Supplementary Figure 1B ). To explore whether TAGAP regulates myeloid cell functions, CRISPR-mediated knockout (KO) of TAGAP was performed in HL-60 cells ( Supplementary Figure 1B ), followed by transwell assays to assess cell migration. Compared to controls, TAGAP KO cells displayed significantly increased migration in response to both fMLP and LTB4 ( Figure 2D ). Phagocytic activity was also enhanced in KO cells, as indicated by 2- to 4-fold higher fluorescence from pHrodo-labeled E. coli BioParticles over time ( Figure 2E ). These observations suggested that TAGAP constrains both migration and phagocytosis in this myeloid cell model.

Human PBMC-derived macrophages were subsequently used to further investigate TAGAP’s role in primary myeloid cells ( Supplementary Figure 2A ). Following knockout and stimulation with PAM3CSK4 (TLR1/2 ligand), poly(I:C) (TLR3 ligand), LPS (TLR4 ligand), or Curdlan (Dectin1 ligand), KO macrophages released higher levels of IL-6, MCP-1, TNF-α, and IP-10 than non-targeting (NT) controls, especially upon LPS stimulation ( Figure 2F ). Although macrophages also produced IL-1β and IL-23 in response to curdlan stimulation, their secretion of these cytokines was lower in KO than in NT cells, consistent with the prior findings ( Supplementary Figure 2B ) (12).

To further examine the involvement of TAGAP in cytoskeletal remodeling, we performed fluorescent confocal imaging of TAGAP KO and NT control macrophages stained with DAPI (nuclei) and Phalloidin-FITC (F-actin). Under the unstimulated condition, no major differences in morphology were observed. Following LPS stimulation, TAGAP KO macrophages exhibited a heterogeneous morphology characterized by a mixture of spindle-shape and dendrite-like cells, whereas NT macrophages primarily displayed a rounded morphology with radiating dendritic extensions ( Supplementary Figure 3 ). These changes suggested that TAGAP modulates cytoskeletal architecture in macrophages, consistent with previous findings that TAGAP regulates thymocyte adhesion and cytoskeletal structure through integrin activation (5). Collectively, these results indicated that TAGAP acts as a negative regulator of human myeloid cell functions and possibly contributes to cytoskeleton dynamics during activation.

Next, we examined the role of TAGAP in T cell responses. TAGAP knockout in Jurkat cells led to increased TCR-mediated activation, indicated by ~1.2-fold higher CD25 and CD69 expression ( Figure 3A ). This finding was corroborated in a Jurkat-NFAT-GFP reporter line, where KO cells exhibited stronger GFP signals in response to both anti-CD3 (~1.3-fold) and PMA/Ionomycin (~1.4-fold) stimulation ( Supplementary Figures 1B , 4A ).

A similar trend was also observed in primary human T cells ( Supplementary Figure 4B ). Knocking out TAGAP significantly increased CD25 and CD69 levels in both CD4⁺ ( Figure 3B ) and CD8⁺ ( Figure 3C ) T cells upon anti-CD3 stimulation. In addition, KO pan T cells secreted higher amounts of IL-2 and IFN-γ at Days 1 and 3 post anti-CD3 stimulation ( Figure 3D ).

To further investigate whether the biochemical phenotype of TAGAP variants correlates with functional outcomes, we transfected TAGAP-deficient Jurkat cells with constructs encoding TAGAP WT E147, K147, or negative control A123. While overexpression of TAGAP E147 and A123 had a marginal impact on cell activation, overexpression of the K147 variant significantly suppressed CD25 and CD69 expression in Jurkat cells following TCR activation ( Supplementary Figure 4C ). This finding corroborated our biochemical assay data, indicating that TAGAP K147 is a GoF variant with enhanced suppressive function in a cellular context. These data together demonstrated a suppressive role of TAGAP in human T cell activation.

TAGAP has been implicated in T cell differentiation, especially in the generation of Th17 cells; however, conflicting evidence has emerged. Some studies suggest that TAGAP is critical for intrinsic Th17 development (11), whereas others attribute its influence to innate immune signaling in myeloid cells (10, 12). Another line of research indicates that TAGAP interference mitigates CIA severity in rats by downregulating the Th17 response (7).

To reconcile these discrepancies and ensure biological relevance, an in vivo anti-CD3 (aCD3) i.p. injection model was used. In this system, aCD3 depletes circulating T cells but expands the Th17 population in the small intestine (SI) when T cells repopulate. As expected, mice receiving aCD3 possessed more RORγt⁺CCR6⁺ cells in both SI lamina propria (LP) and intraepithelial layer than animals injected with an isotype control antibody ( Figure 4A ; Supplementary Figure 5A ; the corresponding gating strategies shown in Supplementary Figure 6 ). Notably, KO mice exhibited nearly doubled proportion and absolute count of RORγt⁺CCR6⁺ cells in the SI LP compared with their WT counterparts under aCD3 treatment ( Figure 4A ). This difference was confirmed by stimulating total SI lamina propria lymphocytes (LPL) ex vivo, revealing ~2-fold higher proportions and absolute numbers of IL-17⁺ cells in KO mice subjected to aCD3 ( Figure 4B ). These results suggested that TAGAP constrains aCD3-mediated Th17 cell expansion in vivo.

To identify the underlying mechanism, we measured CCL20 levels in the SI, as CCR6/CCL20 signaling is one of the key drivers of Th17 recruitment (21, 22). Although tissue CCL20 level was elevated by the aCD3 challenge, there were no significant genotype-specific differences ( Supplementary Figure 5B ). These findings indicated that chemokine-dependent cell migration and development may contribute to the overall Th17 response in the SI upon TCR stimulation, but do not fully explain the heightened Th17 cell increase observed in KO mice. Therefore, the possibility of genotype-related differences in T cell apoptosis was explored next. Further analyses of SI LP revealed that the percentage of Annexin V⁺/PI^- cells among CD4⁺ cells was ~50% lower in KO mice than in WT controls ( Supplementary Figure 5C ). These findings implied that TAGAP may restrict Th17 cell populations in part by modulating T cell apoptosis within the SI.

While TAGAP deficiency is known to exacerbate acute colitis (10), its role in chronic colitis remains unclear. To address this gap, a chronic colitis model with CD4⁺CD45RB^hi T cell transfer was established in Rag2^-/- recipient mice. Recipients were infused with either WT or TAGAP KO donor T cells and monitored over 55 days. Body weight changes were not significantly different between groups ( Supplementary Figure 7 ), yet the survival rate was reduced (67%) in mice receiving KO cells compared with those receiving WT cells (87%) ( Figure 5A ), aligning with prior findings in acute colitis models.

Th17 cell frequencies were also examined in the colon given the crucial role of Th17 cells in chronic colitis (corresponding gating strategies shown in Supplementary Figure 8 ). While little difference was noted in LP Th17 cells ( Supplementary Figure 9A ), we focused on intestinal intraepithelial lymphocytes (IEL), which serve as a frontline defense against luminal microbes and help maintain mucosal homeostasis by rapidly responding to epithelial stress and infections (23, 24). Moreover, dysregulation of IEL populations has been implicated in various models of colitis and may potentiate local inflammation (25, 26). Notably, the proportion of RORγt⁺ cells in colon IEL from KO cell recipients was ~2-fold higher than in those receiving WT cells ( Figure 5B ). Correspondingly, the absolute number of RORγt⁺ in colon IEL ( Figure 5B ), as well as IL-17A levels in serum ( Supplementary Figure 9B ) and colonic tissue ( Supplementary Figure 9C ), were elevated by ~1.5- to 2-fold in KO cell recipients. Collectively, these results indicated that TAGAP is essential for limiting colitis severity, potentially by restraining Th17 cell populations in chronic inflammatory settings.