Islet transplantation is a potential cure for type 1 diabetes (T1D), offering the possibility of insulin independence and improved glycemic control. However, the success of this therapeutic option is often limited by graft rejection mediated by both alloreactive and autoreactive T cells, as well as donor-specific humoral responses. In this context, the CXCR5-CXCL13 chemokine axis plays a significant role. CXCL13, produced by germinal center components, transplanted organs, and inflamed tissues recruits CXCR5⁺ B and T lymphocytes to these sites (34). This axis facilitates B-T cell interactions, promoting B cell maturation and antibody production in lymphoid organs.

In organ transplantation settings, DSA produced in secondary and tertiary lymphoid organs can lead to tissue damage and graft rejection, as observed in pancreatic islet transplantation for T1D (35). To address these challenges, we engineered anti-HLA-A2 CXCR5⁺ CAR-Tregs from human peripheral blood mononuclear cells (PBMCs) to enhance their trafficking to lymphoid structures and improve their therapeutic efficacy in transplantation models. We first cloned the human CXCR5 gene in antisense in a bi-directional lentiviral vector under the control of a minimal cytomegalovirus promoter (mCMV). Next, we employed a second-generation anti-HLA-A2 CAR, which was cloned in sense under the control of a human phosphoglycerate kinase (hPGK) promoter ( Figure 1A ). We isolated human CD4⁺CD25⁺ cells from healthy donor (HD) PBMCs, transduced them with the bi-directional lentiviral vector and expanded them in the presence of IL-2 and rapamycin, as previously described (28) ( Supplementary Figure 1A ). Anti-HLA-A2 CXCR5⁺ CAR-Tregs displayed high expression and co-localization of the two transgenes after 14 days of culture (respectively mean 66.10% ± 12.26% CAR expression and 84.80% ± 2.57% hCXCR5 expression) ( Figure 1B , Supplementary Figure 1B ) Engineered cells showed high purity in terms of Treg enrichment (mean Treg percentage 66.1% ± 12.3% for anti-HLA-A2 CXCR5⁺ CAR-Tregs and 84.8% ± 2.6% for UT Tregs) and an expansion rate comparable to untransduced (UT) Tregs (mean fold increase 34.1 ± 21.8 for CAR-Tregs and 32.2 ± 15 for UT Tregs) ( Figure 1C , Supplementary Figure 1C ). FoxP3⁺ CAR Tregs had >80% CD25 expression and >30% expression of Helios (considered a marker for Treg stability, Supplementary Figures 1D, E ), and markedly higher MFIs of FoxP3, Helios, and CD25 expression ( Supplementary Figure 1F ) as compared to T_conv cells. Helios expression was markedly absent in T_convs.

When functionally tested, anti-HLA-A2 CXCR5⁺ CAR-Tregs retained similar suppressive capacities as UT Tregs upon polyclonal stimulation ( Figure 1D ). Furthermore, anti-HLA-A2 CXCR5⁺ CAR-Tregs suppressed the proliferation of anti-NY-ESO-1 T_conv cells upon antigen-specific stimulation when cultured in the presence of HLA-A2⁺ NY-ESO-1⁺ feeder cells, thus confirming CAR functionality (mean of suppression index of 0.390 ± 0.108 for UT Tregs and 0.798 ± 0.079 for CAR-Tregs, p-value <0.05) ( Figure 1E ).

T follicular helper cells support B cell maturation and class-switch. To assess the effect on Tfh-B cell crosstalk, we isolated circulating HLA-A2⁺ B cells and CXCR5⁺ Tfh cells ( Supplementary Figure 1G ) and co-cultured them with either anti-HLA-A2 CXCR5⁺ CAR- or UT Tregs. Engineered cells significantly reduced the number of IgG⁺ B cells in the presence of Tfh compared to unmanipulated Tregs (IgG⁺ B cell number: 108.7 ± 61.5 in the CAR-Treg group vs 301.0 ± 138.8 in the UT group; p-value 0.03) ( Figure 1F ). To test their safety profile, engineered Tregs were co-cultured with an HLA-A2⁺ cell line. In contrast to anti-HLA-A2 CAR-T_convs, anti-HLA-A2 CXCR5⁺ CAR Tregs exerted a much lower killing capacity toward the target ( Supplementary Figure 1H ).

To validate the functionality of the hCXCR5 transgene, we tested anti-HLA-A2 CXCR5⁺ CAR-Treg migratory capacity at different concentrations of CXCL13. Engineered Tregs showed superior migratory capacity compared to UT Tregs, directly proportional to the CXCL13 concentration (p-value <0.05 at CXCL13 0.5 μg/mL and p-value <0.01 at CXCL13 1 μg/mL) ( Figure 1G ). Collectively, these results demonstrate that anti-HLA-A2 CXCR5⁺ CAR-Tregs acquire superior migratory properties compared to UT Tregs in vitro, while retaining antigen-specific suppressive capacities against both T and B cells without eliminating their cognate target.

To evaluate the in vivo safety and trafficking of anti-HLA-A2 CXCR5⁺ CAR-Tregs, we used the kidney capsule islet transplantation model. NSG (HLA-A2 negative) streptozotocin (STZ)-induced diabetic animals were transplanted with pancreatic islets derived from HLA-A2 transgenic NSG mice under the kidney capsule ( Figure 2A ). After the establishment of normoglycemia, transplanted animals were infused with either anti-HLA-A2 CXCR5⁺ CAR-Tregs or CAR-T_conv. To assess graft rejection, blood glucose was monitored every 5 days after cell injection ( Figure 2B ). UT Tregs served as a control group. Mouse weight was also recorded to monitor for signs of graft-versus-host disease (GvHD) ( Figure 2C ).

Islet rejection was observed in 5 out of 5 mice treated with anti-HLA-A2 CAR-T_conv, within 25 days after treatment ( Figure 2B ). In contrast, none of the 5 mice that received anti-HLA-A2 CAR- or UT Tregs exhibited islet rejection. 30 to 35 days after cell infusion, the experiment was interrupted to confirm the functionality of the graft by removing the pancreatic islet bearing-kidney. All the mice returned to hyperglycemia few days after nephrectomy ( Figure 2B ).

Next, we assessed the frequency of circulating human leukocytes (hCD45⁺ cells) in the animals injected with engineered cells at different timepoints by flow cytometry, finding no differences ( Figure 2D ). At the end of the experiment, we evaluated the human infiltrate in the spleen and the transplanted pancreatic islets by flow cytometry. CAR-Treg treated mice showed a preferential accumulation of hCD45⁺ cells in the graft compared to the spleen, which were mainly composed by CXCR5⁺ CAR^- cells. Conversely, CAR-T_conv treated mice showed a similar frequency of hCD45⁺ and CXCR5⁺ cells between the graft and the spleen, with a high percentage of CAR⁺ cells in some animals ( Figures 2E–G ).

Altogether, these data demonstrated the good safety profile of anti-HLA-A2 CXCR5⁺ CAR-Tregs, which did not reject the transplanted pancreatic islets compared to CAR-T_convs. In addition, the accumulation of engineered cells in the graft confirmed their migratory properties

Next, to generate the FVIII TRuCε synthetic receptor, an scFv with specificity to the immunogenic C2 domain of human FVIII (clone BO2C11) was synthesized, fused to murine CD3ε and cloned into a retroviral backbone that co-expressed the reporter protein mScarlet as described (27) ( Figure 3A ). The murine CXCR5 coding sequence was expressed downstream of the auto cleaving P2A peptide to generate the FVIII TRuCε CXCR5 receptor ( Figure 3B ). Transduced FoxP3-GFP Tregs maintained >98% GFP expression indicating no outgrowth by contaminating T_conv and ~40% Tregs were mScarlet⁺, indicating FVIII TRuCε expression following retroviral transduction and ex vivo expansion ( Figure 3C ). FVIII TRuCε CXCR5 transduction increased the frequency of CXCR5 expressing Tregs from 63.36±1.45% in polyclonal Tregs to 88.6±2.75% in engineered Tregs ( Figure 3D ). CXCR5 MFIs were also increased in engineered Tregs as compared to polyclonal Tregs (1044±206 vs 383±22, Figure 3E ). FVIII TRuCε CXCR5 Tregs (mScarlet⁺) bound Fc-conjugated FVIII in a highly specific manner as detected by fluorochrome labeled anti-Fc antibody ( Figure 3F ).

Flow cytometric phenotyping of FVIII TRuCε CXCR5 Tregs at 48h following FVIII stimulation showed significant antigen specific upregulation of canonical Treg activation markers such as CD69, CD25, LAP, CTLA-4, CD44, ICOS, Ki67, and PD1 as compared to freshly isolated polyclonal UT FoxP3-GFP Tregs or stimulation with an irrelevant antigen, factor IX (FIX) ( Figure 4A ). Interestingly, we noted lower expression levels of CXCR5 despite constitutive overexpression in FVIII stimulated TRuCε CXCR5 Tregs as compared to FIX stimulation ( Figure 4A ). This was likely due to internalization of CXCR5 upon activation as has been previously reported (36). Staining for transcription factors confirmed acquisition of an activated phenotype upon FVIII stimulation, with upregulation of IRF4, TBet, and GATA3 ( Figure 4B ). Both FoxP3 and Helios were increased in FVIII stimulated TRuCε CXCR5 Tregs ( Figure 4B ), indicative of an activated, effector Treg like phenotype (37–39).

To validate the functionality of CXCR5 co-expression, we performed an in vitro transwell cell migration assay. For this we tested both a 2^nd generation FVIII CAR described previously (27), FVIII CAR co-expressing CXCR5, or FVIII TRuCε CXCR5 transduced Tregs. Both engineered CAR and TRuCε Tregs co-expressing CXCR5 showed a superior migratory capacity in response to CXCL13 as compared to cells that did not co-express CXCR5 ( Figure 5A ). Migration was CXCL13 dependent as ~1.4 fold lower migration was observed in response to a CXCL12 gradient ( Figure 5A ). We then tested to see if CXCR5 co-expression improved localization to secondary lymphoid organs by i.v. injecting 4x10⁶ FVIII TRuCε or FVIII TRuCε CXCR5 T_conv cells into HA mice. This was followed by 2 IU recombinant FVIII injections administered every 3 days to initiate ADA formation to FVIII. Spleens were harvested on Days 1, 2, 4, and 7 post adoptive transfer, and frequencies of mScarlet⁺ engineered cells were quantified. Frequencies of FVIII TRuCε CXCR5 T_conv cells were significantly increased at all timepoints, which was 1.57-1.66 fold higher than FVIII TRuCε T_conv ( Figure 5B , Supplementary Figure 2 ). Next, we confirmed enhanced in vivo persistence in both spleens and inguinal lymph nodes (ILN) of HA mice that received 4x10⁶ FVIII TRuCε CXCR5 Tregs. We selected Day 7 as a timepoint to confirm that cells that trafficked into the SLO at earlier timepoints persisted in the organs. FVIII TRuCε CXCR5 Tregs were 1.59 fold higher in the spleen ( Figure 5C ) and 2.72 fold higher in the ILNs ( Figure 5D ) as compared to FVIII TRuCε Tregs, collectively indicating improved in vivo persistence of both engineered T_conv and Treg in SLOs.

Finally, we investigated whether FVIII TRuCε CXCR5 Tregs maintained a suppressive phenotype in vivo. HA recipient mice were i.v. injected with 0.5 × 10⁶ sorted FVIII TRuCε or FVIII TRuCε CXCR5 Tregs followed by 8 weekly i.v. injections of 1 IU recombinant FVIII ( Figure 6A ). Formation of ADAs in plasma was measured longitudinally at 4, 6, and 8 weeks by the Bethesda assay, which measures functional inhibition of clotting in the presence of neutralizing ADAs, and anti-FVIII IgG1 ELISA corresponding to human IgG4, the most prominent subtype in hemophilia A patients with high ADA titers (19, 40). At 4 and 6 weeks, lower inhibitory antibody titers were observed in animals that received FVIII TRuC (0.79±0.48 and 3.69±1.5 BU/mL respectively) and FVIII TRuC CXCR5 Tregs (0.47±0.47 and 0.67±0.42 BU/mL respectively) as compared to animals that received only FVIII weekly injections at 4 and 6 weeks (2.4±0.71 and 8.37±3.9 BU/mL respectively, Figure 6B ). A reduction in ADA titers was particularly evident at the 6-week timepoint in FVIII TRuCε CXCR5 Treg recipients as compared to FVIII control animals ( Figure 6B ). ADA titers rebounded by 8 weeks, which correlated with low persistence of adoptively transferred Tregs in immunocompetent recipients as observed earlier (27). Anti-FVIII IgG1 antibody levels corroborated these findings ( Figure 6C ). To confirm that antibodies were suppressed and not skewed toward a different sub-type, we performed ELISA for anti-FVIII IgG2a, which is considered a functional equivalent to IgG1 in humans (40, 41). Anti-FVIII IgG2a antibody levels were much lower than IgG1 for all groups ( Figure 6D ), indicating that antibodies were not skewed to a different subtype.

A second generation anti-HLA-A2 CAR construct was generated by cloning the antigen-specific single chain fragment variant (scFv) into a CAR encoding bi-directional lentiviral vector (LV) backbone, containing an NGFR-derived spacer (46), a CD28 transmembrane and co-stimulatory domain and a CD3 zeta endodomain under the control of the human phosphoglycerate kinase (hPGK) promoter. In anti-sense, the human CXCR5 gene was cloned under the control of a minimal cytomegalovirus (mCMV) promoter. LV backbones were kindly provided by L. Naldini’s group. Third generation replication-defective and self-inactivating LV were produced by transient transfection of HEK-293T cells. Briefly, a solution containing the packaging plasmids pMDLg/pRRE (containing HIV-1 gag/pol genes), the pILVV01-Rev plasmid, the envelope plasmid encoding the vesicular stomatitis virus glycoprotein (VSV-G), pAdvantage (that enhances transient protein expression) and plasmids carrying the transgene of interest were transfected in sub-confluent 293T cells using the calcium chloride precipitation method. Supernatants containing lentiviral particles were collected 48 hours later and filtered using a 0.22 μm filter, ultracentrifuged at 8000 rpm for 18 h at 4°C (Beckman Optima XL-100K Ultracentrifuge), aliquoted, and cryopreserved.

Human buffy coats from de-identified healthy donors were used according to protocols approved by the Institutional Review Board (IRB). Peripheral blood mononuclear cells (PBMCs) were isolated by Ficoll (GE Healthcare, Chicago, IL) density gradient centrifugation. Tregs were enriched using the CD4⁺CD25⁺ human Treg cell isolation kit (Miltenyi Biotec, Auburn, CA), as per the manufacturer’s instructions. For some studies, both CD4⁺CD25^- T_conv and CD4⁺CD25⁺ Tregs were used. Tregs were expanded and transduced as previously described (28). Briefly, Tregs were stimulated with anti-CD3/CD28 stimulation beads with a Treg: bead ratio = 1:3 (Dynabeads, ThermoFisher) and cultured in rapamycin. At Day +2 after stimulation, activated Tregs were harvested and transduced with a lentiviral vector at a multiplicity of infection (MOI) of 10. After 24 hours, infection was stopped by adding fresh culture medium. Starting from Day +3 and every 3 days 500 IU/ml of IL2 were added to the culture. Starting from Day +14, engineered cells were employed for functional assays. Transduction efficiency was assessed at Day +14 by flow cytometry. CAR expression was identified as the percentage of NGFR⁺ cells. Antibodies utilized for flow cytometry are summarized in Supplementary Table 1 .

Anti-HLA-A2 CXCR5⁺ CAR-Treg suppressive capacity was tested upon polyclonal stimulation. Autologous T_conv were co-cultured with either engineered or unmanipulated Tregs in decreasing Treg: T_conv ratio (1:1 to 1:16) in the presence of anti-CD3/CD28 stimulation beads (Dynabeads, ThermoFisher) at a bead: T cell ratio = 1:10. To detect proliferating cells, T_conv were labeled with a cell trace proliferation dye (VioBlue, ThermoFisher), according to the manufacturer’s instructions. T_conv proliferation was assessed after 7 days of culture. Suppression Index (SI) was calculated as 1 – (% of proliferating T_conv in each condition/% of proliferating T_conv alone).

To test their antigen specific suppression, autologous T_conv were transduced to express the anti-NY-ESO-1 TCR and were labeled with VioBlue, according to the manufacturer’s instructions. Anti-NY-ESO-1 TCR T_conv were co-cultured with either anti-HLA-A2 CXCR5⁺ CAR- or unmanipulated Tregs in a decreasing Treg: T_conv ratio (1:1 to 1:16). To provide a stimulatory signal to anti-NY-ESO-1 TCR T_conv, HLA-A2⁺ NY-ESO-1⁺ U266 tumor cells were added at a fixed T_conv: U266 ratio of 2:1. Proliferation of labeled anti-NY-ESO-1 T_conv was assessed after 3 days of culture, as previously described (47).

To assess in vitro migration, anti-HLA-A2 CXCR5⁺ CAR- or UT Tregs were cultured in 24 well/plate transwells with 3 μm pores (ThermoFisher) in IMDM medium + 0.1% BSA, with increasing concentrations of hCXCL13. Cells were incubated at 37°C with 5% of CO₂ and migration was assessed after 4 hours. Migrating cells were collected, counted and their CXCR5 expression was assessed by flow cytometry.

To test the killing activity, HLA-A2⁺ U266 tumor cells were cultured in a 96 well/plate in RPMI 1640 (Lonza) together with Tregs (engineered or unmodified), or with anti-HLA-A2 CAR-T_convs at a fixed ratio of 1:5 (T cell: U266). The number of live U266 was assessed after 3 days by flow cytometry. Cells were labeled with DAPI to assess vitality and were counted with flow-count fluorospheres (Beckman Coulter) using the formula provided by the manufacturer.

Human HLA-A2⁺ B cells were isolated from healthy donors’ PBMCs through MACS cell separation (B Cell Isolation Kit II, Miltenyi). HLA-A2⁺ peripheral follicular helper T cells were isolated as CD4⁺ CXCR5⁺ elements from healthy donors’ PBMCs by pre-enriching the cells for CD4⁺ T lymphocytes by magnetic cell separation (CD4 T cell isolation kit human, Miltenyi) and further refining the sorting by MACSQuant Tyto Cell Sorter.

Autologous HLA-A2⁺ B cells and peripheral follicular helper T cells were cultured in a 96 well/plate in RPMI 1640 (Lonza). To induce B and Tfh cell activation and differentiation, cells were stimulated with 5 ug/mL of anti-human IgM F(ab′)2 fragment as previously reported (48), (Sigma-Aldrich) and 2 ug/mL of anti-human CD3 antibody, functional grade (Miltenyi Biotec), respectively. To avoid the unspecific activation of Tregs, after 24 hours the plate was centrifuged, and the supernatant was aspired to remove the anti-CD3 antibody. Engineered or unmodified Tregs were then added and anti-human IgM was replenished to restore the initial concentration. The number of IgG⁺ cells was assessed by flow cytometry 6 days later.

Female or male NSG mice were rendered diabetic by a single intraperitoneal (i.p.) injection of streptozotocin (STZ) at 220 mg/kg and islets were transplanted 72-96 hours later. Blood glucose levels were monitored 2-3 times per week using a glucometer. Only mice with blood glucose levels above 300 mg/dl were used for transplantation. Pancreatic islets from NSG.HLA-A2 transgenic mice (NSG-HLA-A2/HHD, Jackson Laboratories, Bar Harbor, ME) were isolated as previously described (49). A total of 500 mouse islets were transplanted under the kidney capsule. Blood glucose levels of <250mg/dl on two consecutive days were defined as successful islet engraftment. Anti-HLA-A2 CXCR5⁺ CAR-Tregs or T_conv cells or UT Tregs were infused intravenously at the dose of 2x10^6 cells/mouse in transplanted mice. All NSG mouse experiments were performed according to San Raffaele Institutional Animal Care and Use Committee approved protocol (IACUC#1019).

The FVIII scFv, directed to the C1 and C2 domains of FVIII, was derived from an Epstein-Barr virus (EBV)-transformed B cell clone obtained from a hemophilia A (HA) patient (originally developed by Jacquemin and colleagues, kindly provided to us by Dr. David Scott, Uniformed Services University, Bethesda, MD, USA). The scFv was constructed from the variable heavy (V_H) and light (V_L) antibody sequences (Creative Biolabs, Shirley, NJ, USA) and complexed to the N terminus of murine CD3ϵ by a flexible linker (G4S) X3 to generate the FVIII TRuCϵ construct. Murine CXCR5 was co-expressed separated by a P2A sequence (GenScript, Piscataway, NJ, USA). FVIII TRuCϵ CXCR5 was inserted into the pMYs-IRES-mScarlet retroviral backbone. Transfer plasmids were transfected into the PlatE ecotropic retroviral packaging cell line (Cell Biolabs, San Diego, CA) using polyethylenimine (PEI) transfection reagent, and retrovirus containing supernatants were collected after 48 h and passed through a 0.22μM vacuum filter.

BALB/c FoxP3 IRES-GFP (Foxp3-GFP) mice were purchased from The Jackson Laboratory (Bar Harbor, ME, USA) and bred in house at Indiana University. Both male and female mice were used as Treg donors for in vitro studies, whereas male mice were used as Treg donors for adoptive transfer studies, indicative of the X-linked disorder. CD4⁺CD25⁻ T_conv or CD4⁺CD25⁺ Tregs from Foxp3-GFP mice were magnetically enriched using a mouse CD4⁺CD25⁺ Treg isolation kit (Miltenyi Biotec, Auburn, CA, USA), and further purified by four-way purity cell sorting (FACSAria II or FACSAria SORP, BD Biosciences). Cells were pre-activated for 48 h with a 1:1 bead-to-cell ratio using α-CD3/28 mouse microbeads (Dynabeads, Invitrogen, Carlsbad, CA). T_conv and Tregs were cultured in Immunocult-XF T cell expansion medium (StemCell Technologies, Vancouver, BC) supplemented with 5% fetal bovine serum (FBS; Atlanta Biologicals, Norcross, GA, USA), 10,000 IU/mL penicillin, 10 mg/mL streptomycin, 1× GlutaMAX-1, 1 mM sodium pyruvate, 10 mM HEPES, 1× nonessential amino acids, and 10 μM 2-mercaptoethanol. Clinical-grade recombinant hIL-2 (Proleukin/aldesleukin; Prometheus Therapeutics and Diagnostics, San Diego, CA) was added at a final concentration of 100 IU/mL for T_conv and 1,000 IU/mL for Tregs. Cells were transduced by spinoculation with retrovirus containing supernatants at 1,200 × g for 90 min in non-tissue culture-treated 6-well plates coated with 20 μg/mL retronectin (Takara Bio, Middleton, WI). Transduced cells were ex vivo expanded for 3−4 days in the presence of α-CD3/28 microbeads at a 1:1 bead-to-cell ratio. Cells were rested for 4−6 h prior to functional in vitro or in vivo experiments.

For in vitro activation, 1 × 10⁶ untransduced (UT) or FVIII TRuCϵ CXCR5 Tregs were stimulated with 5 IU/mL of recombinant human B domain deleted (BDD) FVIII (Xyntha; Pfizer, New York, NY, USA) or recombinant human FIX (Benefix; Pfizer, New York, NY, USA) for 48hrs, following which they were subject to flow cytometry analysis. A complete table of all the antibodies used is listed in Supplementary Table 1 .

For in vitro migration, FVIII CAR, FVIII CAR CXCR5 and FVIII TRuCϵ CXCR5 Tregs were serum starved overnight and 0.5 × 10⁶ cells/well placed in the upper chamber of 24 well cell culture plates with 5µm pore polycarbonate membrane inserts (Costar Transwell, Corning, NY). Lower chambers contained 600 μL serum-free medium with 1μg/mL of either recombinant mouse CXCL12 or CXCL13 (PeproTech, Rocky Hill, NJ). Serum-free medium with no added chemokine served as a negative control. Cells in the upper chamber were allowed to migrate for 6 h at 37°C and 5% CO₂. Migrated cells were collected from the lower compartment and quantified on a BD Fortessa flow cytometer.

BALB/c F8e16^−/− hemophilia A (HA) mice were originally provided by Dr. David Lillicrap (Queens University, Ontario, Canada). Animals were housed under specific pathogen-free conditions at Indiana University (Indianapolis, IN) and treated under Institutional Animal Care and Use Committee-approved protocols. Male mice were used for studies involving adoptive transfer or ADA formation. HA mice (n = 5−7) received weekly i.v. administrations of 1 IU B domain deleted (BDD) recombinant FVIII (Xyntha, Pfizer, New York, NY). Mice received 0.5 × 10⁶ FVIII TRuCϵ CXCR5 Tregs 1 day prior to starting FVIII injections. At 4-, 6-, and 8-week time points post-adoptive transfer, ∼200 μL blood was collected from the retroorbital plexus using non-treated capillary tubes into 3.8% sodium citrate, and plasma was analyzed for inhibitor formation by the Bethesda assay (measured on a Diagnostica Stago STart Hemostasis Analyzer; Stago, Parsippany, NJ, USA), anti-FVIII IgG1 and IgG2a ELISA as previously described (50). 1 BU is defined as the reciprocal of the dilution of test plasma at which 50% of FVIII activity is inhibited.

All experiments were independently repeated at least 2 times. Data shown are mean ± SEM. Statistical significance was determined using Mann-Whitney test for non-parametric data and unpaired t test, 1-way or 2-way ANOVA for parametric data, and multiple comparisons were made using Dunnett’s, Tukey’s, Sidak’s, or Brown-Forsythe post-tests as indicated, using GraphPad Prism 10 software (GraphPad, La Jolla, CA, USA). Values at p < 0.05 were deemed significant and indicated as follows: ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001.