The T follicular helper/T follicular helper regulatory pathway in FVIII immune responses in mice

Key Points

• •TFR cells have a dual function in regulating FVIII immune responses.
• •The TFH/TFR pathway is crucial in the development of FVIII inhibitors.

Visual Abstract

Abstarct

Developing anti–factor VIII (FVIII) inhibitory antibodies (inhibitors) are a significant complication of FVIII protein replacement therapy in hemophilia A. Our previous study demonstrated that follicular helper T (TFH) cells play a critical role in FVIII inhibitor development. Follicular regulatory T (TFR) cells are a subset of forkhead box protein P3 positive (Foxp3⁺) T cells identified in the germinal center that can modulate TFH cell activation of B cells and antibody development. Here, we report that FVIII immunization significantly increases the TFR cells in the spleens of FVIII inhibitor-producing FVIII^null mice compared with saline-treated controls and non–inhibitor-producing animals. The TFH/TFR ratio significantly increased in FVIII inhibitor-producing mice. The emergence of TFR cells correlated with titers of FVIII inhibitors in FVIII-immunized mice. Using TFR-deficient Foxp3^Cre+Bcl6^fl/fl (Bcl6FC) mice, we found that the loss of TFR cells led to significantly decreased FVIII inhibitors compared with wild-type (WT) mice on FVIII immunization (24 ± 16 and 131 ± 114 Bethesda unit (BU)/mL, respectively) but not total anti-FVIII IgG levels and that TFR cells regulated IgG subclass switching and FVIII-specific B-cell responses. Interestingly, on FVIII immunization, mice with phosphatase and tensin (Pten) deficiency in Foxp3⁺ cells (Foxp3^Cre+Pten^fl/fl), a model with augmented TFR cells, developed markedly lower FVIII inhibitor titers (8.1 ± 8.6 BU/mL) than WT controls. When CD4^Cre+Bcl6^fl/fl mice, a TFH- and TFR-deficient model, were immunized with FVIII, none of the animals developed FVIII inhibitors. In conclusion, FVIII immunization induces TFR cell activation and expansion. TFR cells have a dual function in regulating the development of FVIII inhibitors, and the TFH/TFR pathway is pivotal in FVIII inhibitor development in mice.

Some patients with hemophilia A develop antibodies that block factor VIII (FVIII) replacement therapy. Jing and colleagues used a murine model to discover that FVIII immunization induces follicular regulatory T (TFR) cell activation and expansion. TFR cells regulate FVIII-specific B-cell responses and immunoglobulin G subclass switching during FVIII inhibitor development, and imbalance between follicular helper T cells and TFRs is essential for FVIII inhibitor development. Understanding how immune cells contribute to inhibitor formation could enable prevention in the future.

Introduction

Hemophilia A (HA) is an X-linked bleeding disease resulting from a factor VIII (FVIII) deficiency. Currently, FVIII protein replacement therapy is effective in treating HA. However, 30% of patients develop anti-FVIII inhibitory antibodies (inhibitors) after FVIII infusions.^{1, 2, 3} The hallmark of inhibitor development is that the antibodies neutralize functional FVIII, rendering protein replacement therapy useless, complicating treatment, and increasing morbidity and mortality.^{4, 5, 6, 7}

Antibody responses to foreign antigens can be T helper cell dependent or independent, but most arise from interactions between CD4⁺ T cells and B cells in follicles and germinal centers (GCs).^{8, 9, 10, 11} T follicular helper (TFH) cells, a subset of CD4⁺ T cells, assist B cells and are vital for T-cell–dependent B-cell responses and GC formation in various disease models, including inflammation autoimmunity.^{12, 13, 14} Multiple lines of evidence suggest that the FVIII immune response is dependent on CD4⁺ T cells.^{15, 16, 17, 18, 19, 20, 21} Our previous study²² demonstrated that FVIII infusions activate and accumulate TFH cells in the spleens of HA mice, correlating their emergence with FVIII inhibitor titers. Deleting TFH cells hinders FVIII inhibitor production, confirming their vital role in FVIII immune responses.

T regulatory cells (Tregs) are a subset of T cells that modulate the immune system, regulate responses to both self and nonself antigens, and maintain peripheral tolerance.^{23, 24, 25} Programmed death-ligand 1 (PD-L1)–expressing Tregs contribute to maintaining immune tolerance to FVIII infusion.²⁶ Research has demonstrated that FVIII-specific Tregs can inhibit the development of inhibitors in mice. The adoptive transfer of FVIII-specific Tregs safeguards recipient mice from developing high-titer inhibitors.²⁷ Tregs can be engineered to express a specific T-cell receptor that recognizes HLA-restricted T-cell epitopes in FVIII to generate FVIII-specific Tregs. These Tregs inhibited the proliferation and cytokine secretion of FVIII-specific CD4⁺ T-effector clones, thereby reducing the FVIII immune response,^{28, 29, 30, 31, 32} indicating that Tregs modulate the FVIII immune response.

T follicular regulatory (TFR) cells are a Treg subset found in the GC, expressing Foxp3, Bcl6, and CXCR5. Although TFR cells influence GC reactions, their functions are controversial, and their roles in disease development remain unclear.^{33, 34, 35, 36, 37} Animal model studies indicate TFR cells influence inflammation, autoimmunity, and infections. A study using TFR-DTR (diphtheria toxin receptor) mice revealed that TFR cells regulate antibody responses before GC formation but are less involved afterward.³⁸ However, a TFR-deficient mouse model with Bcl6 deleted in Treg cells, using a Foxp3-Cre system, demonstrated that TFR cells positively regulate antigen-specific antibody responses.^{37,39, 40, 41} These results demonstrate the intricate role of TFR in regulating humoral responses.

FVIII inhibitors are antibodies that inactivate functional FVIII and develop after repeated exposure. To date, the role of TFR in FVIII inhibitor development remains unclear. Here, we investigated TFR cell activation and their involvement in FVIII immune responses in mice following immunization to better understand the pathophysiology of FVIII inhibitor development.

Materials and methods

Antibodies and reagents

Details of antibodies, reagents, methods, and statistics used in this study are provided in the supplemental Data (available on the Blood website).

Mice

Animal studies were approved by the Institutional Animal Care and Use Committee of the Medical College of Wisconsin. Animal models, as summarized in Table 1, included the following: (1) a FVIII-deficient (FVIII^null) model⁴²; (2) a FVIII-deficient with Foxp3^GFP+ model^43,47; (3) a von Willebrand factor (VWF) and FVIII double-deficient (VWF^nullFVIII^null) model⁴⁴; (4) a TFR-deficient Foxp3^YFP-CreBcl6^fl/fl (Bcl6FC) model³⁹; (5) a TFR-abundant Foxp3^YFP-CrePten^fl/fl (PtenFC) model^45,46; (6) a TFH/TFR-deficient CD4^CreBcl6^fl/fl (CD4-BCL6–cKO) model⁴⁶; and (7) wild-type (WT) C57BL/6J mice.

Table 1.

Mouse models/strains used in this study

Model/strain name	Description	Deletion/deficiency	Source and reference	Genetic background
FVIIInull	Deficient in FVIII	mFVIII	Originally obtained from Haig Kazazian (University of Pennsylvania)42 and crossed onto a C57BL/6J background by the Shi laboratory	C57BL/6
FVIIInullFoxp3GFP+	FVIII deficient in a Foxp3-GFP knock-in model	mFVIII	A generous gift from David Scott (Uniformed Services University of the Health Sciences)43	C57BL/6
VWFnullFVIIInull	Deficient in both FVIII and its carrier protein VWF	mVWF and mFVIII	Generated by the Shi laboratory44 by crossing FVIIInull onto VWFnull	B6/129S
Foxp3YFP-Cre Bcl6fl/fl (Bcl6FC)	With the deletion of Bcl6 in Foxp3+ cells	mBcl6 in Foxp3+ cells	Generated by the Dent laboratory39 by crossing Foxp3YFY-Cre onto Bcl6fl/fl	C57BL/6
Foxp3YFP-Cre Ptenfl/fl (PtenFC)	With the deletion of the phosphatase Pten in Foxp3+ cells, resulting in dysfunction in Treg cells and augmented TFR responses	mPten in Foxp3+ cells	Originally established by Shrestha et al45 and obtained from the Dent laboratory46	C57BL/6
CD4Cre Bcl6fl/fl (CD4-BCL6–cKO)	With the deletion of Bcl6 in CD4+ cells	mBcl6 in CD4+ cells	Generated by the Dent laboratory46 by crossing CD4Cre onto Bcl6fl/fl	C57BL/6
WT	Wild-type	—	Purchased from the Jackson Laboratory (Bar Harbor, ME)	C57BL/6

FVIII immune response study

FVIII-deficient mice were immunized IV with recombinant human B-domain–deleted FVIII at 50 to 100 U/kg per week for 4 to 6 weeks. Blood samples were collected 5 to 7 days after the last immunization. C57BL/6J-WT, Bcl6FC, PtenFC, and CD4-BCL6-cKO mice, with normal endogenous murine FVIII, were immunized with recombinant human full-length FVIII (rhFL-FVIII) at 200 U/kg per week IV for 4 weeks. Blood samples were collected 1 week later for assays. Anti-FVIII inhibitors and total IgG titers were measured using a modified Bethesda assay and enzyme-linked immunosorbent assay (ELISA), as reported.^{47, 48, 49} The apparent affinity of FVIII-specific IgG was assessed using competitive ELISA, as reported.⁵⁰

OVA immune response study

C57BL/6J-WT and Bcl6FC mice were immunized with 100 μg ovalbumin (OVA) IV and boosted 2 weeks later. One week after the boost, blood samples were collected, and anti-OVA total IgG levels were determined by ELISA, as reported.⁵¹

Flow cytometry (fluorescence-activated cell sorting) analysis

Spleens were isolated, and single-cell suspensions of splenocytes were prepared as previously described.^22,26 Cells were stained for CD19, D220, CD3, CD4, Foxp3, CD38, CD44, CD62L, ICOS (inducible T-cell costimulator), CD25, PD-1 (programmed cell death protein 1), CXCR5, Fas, GL7, CD21/35, IgM, IgD, Ki-67, CTLA4 (cytotoxic T-lymphocyte-associated protein 4), Bcl6, and FVIII-Alexa 647, and analyzed by flow cytometers.

Results

Identification and characterization of activated TFR cells in FVIII-immunized mice

Our previous study demonstrates that activated TFH cells in FVIII^null mice are critical for FVIII inhibitor development.²² However, the role of TFR cells in controlling FVIII inhibitor development is unclear. To investigate this, we assessed TFR cell activation in the spleens of FVIII-immunized HA mice. Figure 1A illustrates our gating strategy for determining the activated TFR population by fluorescence-activated cell sorting (FACS) analysis. Starting with CD4⁺ (CD3⁺CD4⁺CD19⁻) live cells, we identified Tregs via Foxp3 expression and effector Tregs (eTregs) by CD44 and CD62L (CD44⁺CD62L^–). Activated TFR cells were confirmed by coexpression of CXCR5 and PD-1. We measured the expression of BCL6, an essential transcription factor for the formation of both TFH and TFR cells,^{52, 53, 54} to validate CXCR5^hiPD-1^hi eTregs as activated TFR cells, finding significantly higher BCL6 levels in them compared with naïve Tregs (nTregs) and CXCR5^loPD-1^lo eTregs (Figure 1B-C). ICOS expression was also evaluated and increased only on TFR cells. Both eTregs and activated TFR cells exhibited higher CTLA-4 levels and lower CD25 levels than nTregs (Figure 1B-C). Ki-67 expression was significantly higher in activated CXCR5^hiPD-1^hi TFR cells than in nTregs and eTregs (Figure 1D), indicating TFR cells' greater proliferation. Compared with activated TFH cells, activated TFR cells displayed higher PD-1 and lower BCL6 expression. Expression levels of CXCR5, ICOS, and Ki-67 were comparable between activated TFH (CD4⁺Foxp3^-CD44⁺CD62L^-CXCR5^hiPD-1^hi) and TFR cells (supplemental Figure 1). Additionally, CD38 expression, a marker of GC-resident induced TFR from TFH cells,⁵⁵ in FVIII-immunized mice was significantly higher than in saline controls (Figure 1E), indicating the presence of induced TFR cells from TFH cells.

Figure 1.

Characterization of activated TFR cells in the spleen of FVIII-immunized mice. FVIII^null mice were given 5 to 6 rounds of weekly IV injections of recombinant human B-domain–deleted FVIII (rhBDD-FVIII). At 5 to 7 days after the last immunization, plasma samples were collected for inhibitor assay. Representative inhibitor-producing mice were euthanized, and splenocytes were collected for flow cytometry analysis. Splenocytes were stained for CD3, CD4, CD19, CD44, CD62L, Foxp3, CXCR5, PD-1, Ki-67, and CD38, and analyzed by flow cytometry. (A) Gating strategy for the identification of activated TFR cells. Treg cells were identified by Foxp3 staining inside the singlet live CD19^-CD3⁺CD4⁺ cell population. Effector Treg cells were identified by CD44 and CD62L expression (CD44⁺CD62L^-). Effector Treg cells were analyzed for CXCR5 and PD-1 coexpression to identify activated TFR cells. Representative histograms (B), quantification of BCL6, ICOS, CTLA4, and CD25 expression (C), and cell proliferation marker Ki-67 expression on naïve, CXCR5^-PD-1^- effector Treg (non-TFR eTreg), and activated TFR cells (D) from FVIII-immunized inhibitor-producing mice (n = 5-9). (E) CD38 expression in activated TFR cells from FVIII-immunized mice (n = 4-5). ∗∗P < .01, ∗∗∗P < .001.

FVIII immunization increases the frequency of activated TFR cells in mice that produce FVIII inhibitors

After FVIII immunization, representative FVIII^null mice with and without inhibitors were sacrificed for TFR analysis. FVIII immunization did not significantly affect Treg frequency (Figure 2A) but significantly increased CD44⁺CD62L^- eTregs in inhibitor-producing mice compared with controls (Figure 2B). The CXCR5^hiPD-1^hi TFR cell percentage within the eTreg population markedly increased in inhibitor-producing mice compared with saline-treated and non–inhibitor-producing mice (21.3% ± 3.9%, 11.5% ±4%, and 12.3% ± 3.1%, respectively) (Figure 2C). The percentages of activated TFR cells among total CD4 cells (Figure 2D) and the absolute activated TFR numbers per spleen (Figure 2E) were significantly higher in inhibitor-producing mice vs controls. The TFH/TFR ratio also increased significantly in inhibitor-producing mice (Figure 2F), indicating preferential TFH cell expansion relative to TFR cells.

Figure 2.

FVIII immunization significantly increases the frequency of activated TFR cells in mice-producing inhibitors. FVIII^null mice were immunized using a similar protocol as used for the studies presented in Figure 1. Representative animals with inhibitors and no inhibitors were sacrificed for TFR analysis. Age-matched saline-treated FVIII^null mice were used as controls. Splenocytes were stained for CD3, CD4, CD19, CD44, CD62L, Foxp3, CXCR5, PD-1, and Ki-67, and analyzed by flow cytometry. (A) Percentage of regulatory T cells among CD4 T cells in saline-injected control mice, non–inhibitor-producing mice, and inhibitor-producing mice. (B) Percentage of effector regulatory cells (CD44⁺CD62L^-) among Foxp3⁺CD4 T cells in saline-injected control mice, non–inhibitor-producing mice, and inhibitor-producing mice. (C) Percentage of CXCR5⁺PD-1⁺ TFR cells among effector regulatory T (eTreg) cells. (D) Percentage of activated TFR cells among total CD4 T cells. (E) The total number of activated TFR cells per spleen. (F) TFH/TFR ratio. (G) Percentage of Ki-67 positive in activated TFR cells. (H) Percentage of annexin-V⁺7AAD⁺ apoptotic cells in naive, non-TFR effector regulatory T cells, and TFR cells as well as naïve CD4, non-TFH effector CD4, and TFH cells from FVIII-immunized mice. ∗P < .05, ∗∗P < .01, ∗∗∗P < .001. (A-H) Each data point represents 1 mouse.

To investigate the accumulation of TFR cells in inhibitor-producing mice, we examined their proliferation and survival. Ki-67 expression on TFR cells showed that activated TFR cells undergoing proliferation were similar in inhibitor-producing and control mice (Figure 2G). We immunized FVIII^nullFoxp3^EGFP mice⁴⁷ with FVIII and measured the apoptosis of splenic Foxp3^GFP+ Tregs. We noted a significant increase in Annexin-V⁺ apoptotic cells in eTregs and activated TFR cells compared with nTregs, TFH, and effector cells vs naïve CD4 T cells (supplemental Figure 2A-F,H).

TFR activation induced by FVIII immunization was also confirmed in VWF and FVIII double-deficient (VWF^nullFVIII^null) mice on recombinant human B-domain–deleted FVIII infusion. Some mice developed high inhibitor titers, prompting a comparison of TFR cell activation in mice with high titers (≥250 BU/mL) vs low-moderate titer. Activated TFR cells in the spleens of inhibitor mice were significantly higher than in saline controls (Figure 3A). The activated TFR percentage increased in the high-titer group vs the low-titer group. The high-inhibitor group had the highest TFH/TFR ratios, followed by low-modest, no-inhibitor, and control groups (Figure 3B). The frequency of TFR cells correlated positively with TFH cells (Figure 3C), although TFH expansion was greater, as suggested by increased TFH/TFR ratio. We assessed the correlation between TFR activation and FVIII-specific immune response, finding a stronger correlation of TFR and TFH/TFR ratio with FVIII inhibitors than with total IgG. The correlation between the TFH/TFR ratio and inhibitor titers was most pronounced (Figure 3D-G).

Figure 3.

The emergence of activated PD-1⁺CXCR5⁺ TFR cells correlates with the development of FVIII inhibitors. FVIII and VWF double-knockout mice were given weekly IV recombinant human B-domain–deleted FVIII (rhBDD-FVIII) immunizations for 4 to 5 times. One week after the last immunization, mice were euthanized. Splenocytes and plasma were collected for analysis. Age-matched saline-treated mice were used as controls. Plasma FVIII inhibitor titers were determined by Bethesda assay. Splenocytes were analyzed by flow cytometry. (A) Percentage of activated TFR cells among CD4 T cells. Anti-FVIII inhibitor titers >250 BU/mL were defined as the high-titer inhibitor group. (B) TFH/TFR ratio. (C) Scatter diagram analyses show a positive correlation between the percentage of activated TFH and TFR cells. (D) The correlation between the titer of FVIII inhibitors and the percentage of activated TFR cells among CD4 T cells. (E) The correlation between the titer of anti-FVIII total IgG and the percentage of activated TFR cells among CD4 T cells. (F) The correlation between the titer of FVIII inhibitors and the TFH/TFR ratio. (G) The correlation between the titer of anti-FVIII total IgG and the TFH/TFR ratio. Statistical significance was analyzed by linear regression. The R and P values are indicated in the figure. ∗P < .05, ∗∗P < .01, ∗∗∗P < .001. (A-D) Each data point represents 1 mouse.

Taken together, these results demonstrate that TFR cells increase in FVIII-inhibitor–producing mice. Activated TFR cells in FVIII-immunized mice are proapoptotic, and their expansion in the spleens of inhibitor-producing mice is not due to long-term survival. The emergence of TFR cells correlates with the development of FVIII inhibitors.

TFR cells promote anti-FVIII inhibitor development and B-cell responses

Using TFR-deficient Bcl6FC mice,³⁹ we explored the role of TFR cells in the generation of FVIII inhibitors. Bcl6FC and WT mice underwent immunizations with rhFL-FVIII, and anti-FVIII antibody titers were measured. As shown in Figure 4A, the loss of TFR cells significantly decreased FVIII inhibitor titers. The inhibitor titers in FVIII-immunized Bcl6FC mice (22 ± 12 BU/mL) were lower than in WT mice (145 ± 82 BU/mL). The total anti-FVIII IgG titers were similar between the Bcl6FC and WT groups (Figure 4B), indicating that TFR cells contribute to the generation of neutralizing anti-FVIII antibodies. Both groups also received OVA antigen immunization, with no significant difference in anti-OVA total IgG titers (Figure 4C).

Figure 4.

The development of FVIII inhibitors is attenuated in TFR-deficient mice. Bcl6FC mice and control WT C57BL/6J animals were immunized with recombinant human full-length FVIII at a dose of 200 U/kg per week for 4 weeks via retro-orbital venous injections. One week after the last immunization, plasma samples were collected for Bethesda assay to determine the FVIII inhibitor titers (A) or ELISA to determine the total anti-FVIII immunoglobulin (Ig) G titers (B). Bcl6FC mice and control WT C57BL/6J were immunized with full-length ovalbumin (OVA) protein by IV injection at a dose of 100 μg/mouse and boosted with the same dose 2 weeks later. (C) One week after the second immunization, plasma samples were collected and used to determine the titers of anti-OVA total IgG by ELISA. ∗∗∗P < .001. (A-C) Each data point represents 1 mouse.

We asked whether inefficient FVIII inhibitor generation in Bcl6FC mice stemmed from defective TFH cell activation after FVIII immunization. We compared spleen TFH and TFR cell activation between FVIII-immunized WT and Bcl6FC mice. Effector CD4⁺ T-cell frequency increased significantly in FVIII-immunized vs naïve WT mice. Naïve Bcl6FC mice exhibited a higher frequency of effector CD4⁺ T cells, which did not increase after FVIII immunization (Figure 5A). The percentage of TFH in effector CD4 cells (Figure 5B) and activated TFH cell numbers per spleen (Figure 5C) were significantly higher in both immunized WT and Bcl6FC mice compared with their naïve controls. These results indicate that the TFH cell response is unaffected by the absence of TFR cells during FVIII immunization, and impaired FVIII inhibitor development in Bcl6FC mice is not due to defective TFH activation. Frequencies of Tregs and eTregs in naïve WT and Bcl6FC spleens were similar, showing no significant change after FVIII immunization, indicating total Tregs remain unaffected by TFR absence (Figure 5D-E). WT mice experienced robust TFR expansion after FVIII immunization, whereas TFR activation was nearly absent in Bcl6FC mice, confirming their TFR deficiency (Figure 5F-G).

Figure 5.

FVIII immunization effectively induces TFH cell activation but does not elicit GC–B-cell responses in TFR-deficient individual mice. Bcl6FC mice and control WT C57BL/6J animals were immunized with recombinant human full-length FVIII at a dose of 200 U/kg per week for 4 weeks via retro-orbital venous injections. One week after the last immunization, mice were euthanized. Splenocytes were collected for analysis. Splenocytes were stained for CD3, CD4, CD19, CD44, CD62L, Foxp3, CXCR5, PD-1, Fas, and GL-7, and analyzed by flow cytometry. (A) Percentage of effector T cells (CD44⁺CD62L^-) among Foxp3^-CD4 T cells in FVIII-immunized WT and Bcl6FC mice. Naïve nonimmunized mice were used as controls. (B) Percentage of CXCR5⁺PD-1⁺ TFH cells among effector T cells. (C) Total number of activated TFH cells per spleen. (D) Percentage of regulatory T cells among CD4 T cells. (E) Percentage of effector regulatory cells (CD44⁺CD62L^-) among Foxp3⁺ CD4 T cells. (F) Percentage of CXCR5⁺PD-1⁺ TFR cells among effector regulatory T cells. (G) The total number of activated TFR cells per spleen. (H) Percentage of activated TFH cells expressing GL-7 in the spleens of FVIII-immunized WT or Bcl6FC mice. Naïve nonimmunized mice were used as controls. (I) Percentage of CD19⁺ B cells bearing a GC phenotype (Fas⁺GL-7⁺). (J) Ratio of GL-7⁺ TFH/TFR in FVIII-immunized WT mice. n = 4-6. ∗P < .05, ∗∗P < .01, ∗∗∗P < .001. ns, no significant difference between the 2 groups.

To investigate whether the absence of TFR cells influenced the GC response, we assessed the frequency of GC-TFH cells and GC-B cells in the spleen of FVIII-immunized mice. We analyzed the GC-TFH cell subpopulation by measuring its GL-7 expression. The percentage of GL-7⁺ TFH cells in the spleen was significantly lower in Bcl6FC mice compared with WT mice after FVIII immunization (15.9% ± 2.8% vs 35.2% ± 1.8%, respectively) (Figure 5H). CD19⁺Fas⁺GL-7⁺ cells were used to define antigen-activated GC-B cells. The percentage of CD19⁺Fas⁺GL-7⁺ cells was also significantly reduced in the Bcl6FC group compared with the WT group (Figure 5I). The GL-7⁺-TFH/TFR ratio significantly increased in WT mice after FVIII immunization compared with naïve controls (Figure 5J). Collectively, these results suggest that TFR cells enhance GC responses during the development of FVIII inhibitors.

TFR cells regulate anti-FVIII IgG subclass switching and antibody affinity development

To investigate how FVIII inhibitor development, rather than anti-FVIII total IgG, is impaired in Bcl6FC mice following FVIII immunization, we characterized the subclasses of anti-FVIII IgG, their affinity, and FVIII-specific B-cell subsets. To eliminate the potential influence of endogenous murine FVIII in nonhemophilic Bcl6FC mice on antibody class switching, we transplanted bone marrow from either Bcl6FC (Bcl6FC bone marrow transplantation [Bcl6FC-BMT]) or C57BL6/J-WT (WT-BMT) mice into FVIII^null recipients. After bone marrow reconstitution, the animals were immunized weekly with rhFL-FVIII for 4 weeks, and the production of anti-FVIII antibodies was analyzed. Like Bcl6FC mice (Figure 4A-B), the FVIII inhibitor titer in the Bcl6FC-BMT/FVIII^null group was significantly lower than in the WT-BMT/FVIII^null group, whereas anti-FVIII total IgG showed no significant difference (Figure 6A-B). The Bcl6FC-BMT/FVIII^null group indicated lower IgG1 titers but higher IgG2a and IgG2b compared with WT-BMT/FVIII^null, although not statistically significant (Figure 6C-E). Notably, anti-FVIII IgG3 titers were significantly higher in the Bcl6FC-BMT/FVIII^null group (Figure 6F). Incidences of detectable anti-FVIII IgG2b and IgG3 antibodies, along with the ratios of IgG2a, IgG2b, and IgG3 to IgG1, were significantly higher in the Bcl6FC-BMT/FVIII^null group than in WT-BMT/FVIII^null (supplemental Figure 3A-G).

Figure 6.

The anti-FVIII IgG subclass switching and FVIII-specific B-cell responses are impaired in TFR-deficient mice. To avoid the potential influence of endogenous mouse FVIII on FVIII-specific antibody class switching in Bcl6FC mice, bone marrow cells from Bcl6FC or WT mice were transplanted into FVIII^null mice under 1100 R total body irradiation. After at least 12 weeks of bone marrow reconstitution, recipients were immunized with recombinant human full-length FVIII at a dose of 200 U/kg per week for 4 weeks via retro-orbital venous injections. Five days after the last immunization, plasma samples were collected for Bethesda assay to determine the titers of FVIII inhibitors and ELISA to quantitate anti-FVIII total IgG, IgG1, IgG2a, IgG2b, and IgG3. The affinity of FVIII-specific antibodies was determined by competitive ELISA. One week after the last immunization, mice were euthanized. Splenocytes were collected for analysis. Splenocytes were stained for B220, CD21/35, CD93, IgM, IgD, Fas, GL-7, FVIII-Alexa 647, CD44, CD62L, Foxp3, CXCR5, PD-1, Fas, and GL-7 and analyzed by flow cytometry. (A) FVIII inhibitor titers. (B) Anti-FVIII total IgG titers. (C) Anti-FVIII IgG1 titers. (D) Anti-FVIII IgG2a titers. (E) Anti-FVIII IgG2b titers. (F) Anti-FVIII IgG3 titers. (G) Apparent affinities of FVIII-specific total IgG. (H) Apparent affinities of FVIII-specific total IgG1. (I) The total number of FVIII-specific B (B220⁺FVIII⁺) cells per spleen. (J) The total number of FVIII-specific germinal center B (GC-B) (B220⁺Fas⁺GL-7⁺FVIII⁺) cells. (K) The total number of FVIII-specific follicular B (FO-B) (B220⁺FVIII⁺CD93^-CD21/35^-IgM^-IgD⁺) cells. (L) The total number of FVIII-specific marginal zone B (MZ-B) (B220⁺FVIII⁺CD93^-CD21/35⁺IgM⁺IgD^-). ∗∗P < .01, ∗∗∗P < .001, ∗∗∗∗P < .0001. (A-F) and (I-L) Each data point represents 1 mouse. (G-H) A line links data points for high and low affinity from the same mouse.

Using a competition-based affinity ELISA platform,^50,56,57 we assessed the apparent affinities of FVIII-specific IgG antibodies in plasma from recipients following FVIII immunization. The apparent affinity of anti-FVIII total IgG showed 2 pools in all 5 Bcl6FC-BMT/FVIII^null mice, with high affinity (K_A^high = 2.49 ± 1.41 × 10⁹/M^-1) and low affinity (K_A^low= 3.61 ± 1.55 × 10⁷/M^-1), and in 3 of 4 WT-BMT/FVIII^null animals with 2 pools (K_A^high = 3.31 ± 1.79 × 10⁹/M^-1 and K_A^low = 6.07 ± 4.49 × 10⁷/M^-1) (Figure 6G). Notably, in IgG1, 3 of 5 Bcl6FC-BMT/FVIII^null recipients displayed a single pool of low-affinity antibodies, whereas all 4 WT-BMT/FVIII^null animals exhibited 2 pools of antibodies with K_A^high= 4.30 ± 3.51 × 10⁹/M^-1 and K_A^low = 1.08 × 10⁸ ± 7.61 × 10⁷/M^-1 (Figure 6H).

To examine FVIII-specific B-cell responses, FVIII-specific B cells were stained with rhFL-FVIII conjugated with the Alexa Fluor 647 following the reported protocol,²⁶ and analyzed by flow cytometry, as shown in supplemental Figure 4. The total number of FVIII-specific B cells in the Bcl6FC-BMT/FVIII^null group was significantly lower than that in the WT-BMT/FVIII^null group (Figure 6I). When subsets of B cells were analyzed, TFR deficiency resulted in a similar reduction effect on FVIII-specific B-cell responses in GC, follicle, and marginal zone (MZ) in Bcl6FC-BMT/FVIII^null mice, with total numbers of FVIII-specific GC-B (B220⁺Fas⁺GL-7⁺FVIII⁺), follicular-B (B220⁺FVIII⁺CD93^-CD21/35^-IgM^-IgD⁺), and MZ-B (B220⁺FVIII⁺CD93^-CD21/35⁺IgM⁺IgD^-) cells significantly less than in the WT-BMT/FVIII^null group (Figure 6J-L).

Taken together, these results suggest that TFR cells promote FVIII-specific B-cell responses, IgG1 subclass switching, and help affinity maturation of antibodies in FVIII immune responses.

FVIII inhibitor induction was dampened in TFR-abundant mice

To further investigate the regulatory function of TFR cells on FVIII immune responses, mice with phosphatase Pten deficiency in Foxp3⁺ cells (PtenFC), a model with augmented Treg, TFH, and TFR cells,^45,46 as confirmed by our flow cytometry analysis (supplemental Figure 5A-C), were immunized with rhFL-FVIII weekly for 4 weeks. FVIII inhibitor titers in FVIII-immunized PtenFC mice averaged 8.06 ± 8.6 BU/mL, significantly lower than in WT mice (131.3 ± 114 BU/mL) (Figure 7A). The percentages of Treg, TFH, TFR, and GC-B cells in PtenFC mice immunized with rhFL-FVIII were comparable to unimmunized naïve controls (supplemental Figure 6A-D). After immunization, the PtenFC group had a higher total and percentage of Foxp3⁺ Tregs and TFH cells compared with WT and Bcl6FC groups, which showed no significant differences apart from Treg percentages. TFR cells were also significantly more numerous in the PtenFC group compared with the WT group (Figure 7B-G). TFH/TFR ratios increased in WT mice but not PtenFC after immunization (Figure 7H). Significant correlations were found between FVIII inhibitor titers and total anti-FVIII IgG in the FVIII^null, VWF^nullFVIII^null, WT, and PtenFC groups, but not in the Bcl6FC group after immunization (supplemental Figure 7A-E).

Figure 7.

Comparison of FVIII inhibitor induction in TFR-abundant PtenFC, TFR-deficient Bcl6FC, and TFH/TFR-deficient CD4-BCL6-cKO mice. PtenFC, Bcl6FC, CD4-BCL6-cKO, and control WT C57BL/6J mice were immunized with recombinant human full-length FVIII at a dose of 200 U/kg per week for 4 weeks via retro-orbital venous injections. One week after the last immunization, plasma samples were collected for Bethesda assay to determine FVIII inhibitor titers, and mice were euthanized for splenocyte analysis. Splenocytes were stained for CD3, CD4, CD19, CD44, CD62L, Foxp3, CXCR5, and PD-1 and analyzed by flow cytometry. (A) FVIII inhibitor titers. (B) Percentage of Foxp3⁺ Treg cells among CD4 T cells. (C) The total number of Treg cells per spleen. (D) Percentage of CXCR5⁺PD-1⁺ TFH cells among effector Foxp3^- CD4 T cells. (E) Total number of activated TFH cells per spleen. (F) Percentage of CXCR5⁺PD-1⁺ TFR cells among effector regulatory T cells. (G) The total number of activated TFR cells per spleen. (H) The ratio of TFH/TFR cells. ∗P < .05, ∗∗P < .01, ∗∗∗P < .001, ∗∗∗∗P < .0001. ns, no significant difference between the 2 groups. (A-H) Each data point represents 1 mouse.

To confirm the critical role of the TFH/TFR pathway in FVIII inhibitor development, we used CD4-CreBcl6^fl/fl (CD4-Bcl6-cKO) mice, a model with Bcl6 deficiency in all CD4⁺ T cells.⁴⁶ TFH/TFR cell deficiency in CD4-Bcl6-cKO mice was confirmed by our flow cytometry analysis of their splenocytes (supplemental Figure 5B-C). As expected, none of the CD4-Bcl6-cKO mice developed detectable levels of FVIII inhibitors (Figure 7A) or anti-FVIII total IgG (data not shown), demonstrating that the transcription repressor Bcl6 expression in CD4 T cells is pivotal for FVIII inhibitor development.

Together, these data show that TFR cells may play dual functions in regulating FVIII immune responses. The TFH/TFR pathway is essential for FVIII inhibitor development in mice on FVIII immunization.

Discussion

It is still not fully understood why most patients with HA can successfully tolerate IV FVIII infusion, but other patients develop inhibitors. Naïve T cells can differentiate into immunosuppressive Treg cells through peripheral tolerance mechanisms, helping to maintain tolerance to FVIII through the expression of suppressive cytokines and inhibitory molecules.^58,59 Previous studies demonstrated that FVIII-specific Treg cells could suppress FVIII-inhibitor development.^{27, 28, 29, 30, 31, 32} TFR cells are a subset of Tregs that have specialized roles in regulating TFH and B-cell interactions within the GC. Little is known about the physiological role of TFR cells in inducing FVIII inhibitors. In this study, we demonstrate that FVIII immunization induces the activation and accumulation of TFR cells, along with TFH cells, in the spleens of HA mice. We also demonstrate that TFR cells can promote inhibitor generation, as the production of FVIII inhibitors, but not total IgG, is significantly impaired in TFR-deficient mice following FVIII immunization. Our data strongly suggest that TFR cells play more complex roles in FVIII inhibitor development than we initially hypothesized.

After receiving FVIII immunizations, the spleens of inhibitor-producing mice showed a significant increase in activated TFR cells, accompanied by the expansion of activated TFH cells. The accumulation and expansion of TFR cells likely result from enhanced differentiation and induction of TFR cells; increased proliferation and prolonged survival alone cannot account for this, but CD38 expression can. Although there is an observable increase in TFR activation, we found that the TFH/TFR ratio still significantly increased in FVIII inhibitor-producing mice, suggesting a preferential expansion of TFH cells relative to TFR cells. These data indicate that breaking peripheral tolerance to FVIII occurs when both TFH and TFR cells are activated, yet their activation remains unbalanced. This suggests that the pathogenic immune responses against FVIII are primarily due to an exacerbated immune response rather than a failure in the induction of tolerance involving Treg cells.⁶⁰

The emergence and accumulation of TFR cells correlated with the titers of FVIII inhibitors in FVIII-immunized mice. This indicates the existence of a functional relationship between TFR cell activation and inhibitor production after FVIII immunization. Using TFR-deficient Bcl6FC mice, we examined the in vivo role of TFR cells in humoral immunity against FVIII. Loss of TFR cells resulted in significantly decreased FVIII inhibitor levels but did not affect the total induction of anti-FVIII IgG antibodies. It has been demonstrated that FVIII inhibitors are high-affinity antibodies against FVIII in humans.⁵⁰ The impaired development of inhibitors in Bcl6FC mice following FVIII immunization, as shown in our current study, supports the emerging perspective that TFR cells play a crucial role in high-affinity antibody production against foreign antigens.^39,61,62 Our data from TFR-deficient model studies suggest that TFR cells regulate FVIII-specific B-cell responses. The induction of FVIII inhibitor development requires repeated IV FVIII administration, complicating mechanistic studies related to the cellular response due to a mix of primary and secondary responses. The temporal accumulation of TFR and TFH cells in the GC may differ. Results from our current single-time-point study cannot clarify whether TFR cell activation synchronizes with TFH cell activation or is delayed during FVIII immunization. Further studies on how TFR cells regulate B-cell receptor rearrangement are warranted for a better understanding of FVIII inhibitor development.

FVIII immunization can effectively activate TFH cells in TFR-deficient mice, demonstrating that TFR cells do not influence the TFH population in Bcl6FC mice during FVIII infusions. However, the percentage of effector CD4-T cells and total TFH cells in the spleens of naïve Bcl6FC mice was higher than in WT mice, with no differences in Foxp3⁺ Treg and eTregs between the 2 groups. This indicates that TFR cells regulate the expansion of effector and TFH cells in a steady state in naïve animals. FVIII-specific GC responses were mitigated in TFR-deficient mice, suggesting that TFR cells enhance such responses. Interestingly, in addition to GC-B cells, FVIII-specific MZ and follicular-B cells were also reduced in Bcl6FC mice following FVIII immunization, implying that TFR cells regulate not only GC-B but also atypical B-cell responses in FVIII immune responses. This highlights the importance of TFR cells in regulating FVIII immune responses, as it has been shown that inhibitor development can also occur through a non-GC pathway.⁶³

The functions of TFR cells in the GC are still being elucidated, and the roles of TFR cells in disease development remain incompletely understood.^{33, 34, 35, 36,64} TFR cells were initially described as suppressors of the GC reaction and the antibody response when this subset of cells was first identified.^54,65 Further studies using an inducible TFR-deletion mouse model demonstrated that TFR cells regulate early, but not late, GC responses to restrict autoreactive and antigen-specific antibody responses.⁶⁶ However, studies using noninducible models of Bcl6 deficiency in Tregs suggest that TFR cells have complex roles in regulating antibody responses, displaying both repressive and promotive functions depending on the type of immunogens, antibody subclass, and disease models.^{36,37,39,46,67, 68, 69, 70} For instance, TFR cells function as suppressors in autoimmune disease models,^40,68 but as helpers in maintaining the antigen-specific GC response and high-affinity antibody production.^38,39,67 These divergent results have led to ambiguity in the roles of TFR cells in vivo, which remain elusive.³³

The precise mechanism by which TFR cells are essential for effective FVIII inhibitor development is still unclear. Given that the anti-FVIII total IgG titer in the Bcl6FC group was comparable to that in the WT group after FVIII immunization, and no correlation existed between the titers of anti-FVIII inhibitors and total IgG in FVIII-immunized Bcl6FC mice, we speculate that TFR cells play multiple roles in regulating FVIII immunity responses. Our study suggests that the reduced generation of FVIII inhibitors in TFR-deficient mice after FVIII immunization may arise from impaired IgG subclass switching despite unaffected or increased anti-FVIII total IgG levels. Anti-FVIII total IgG encompasses both inhibitory and noninhibitory antibodies, including low- and high-affinity antibodies that may belong to different subclasses of IgG. In HA mice, anti-FVIII IgG1 is the most abundant subclass of antibodies following FVIII immunization.⁷¹ Our studies demonstrated that in Bcl6FC-BMT/FVIII^null mice, the ratios of IgG2a, IgG2b, and IgG3 subclasses to IgG1 were significantly higher than in WT-BMT/FVIII^null animals following FVIII immunization. Although 40% of TFR-deficient mice could still produce 2 pools (high and low affinity) of IgG1 antibodies, it is evident that all TFR-WT animals developed both pools of antibodies on FVIII immunization. Our findings strongly suggest that TFR cells play a crucial role in promoting IgG subclass switching from other IgG subclasses to IgG1 and in the maturation of antibody affinity during FVIII inhibitor development in mice.

Interestingly, we found that FVIII inhibitor development was mitigated in PtenFC mice on FVIII immunization. This was unexpected because PtenFC mice have augmented TFR cells with excessive GC reactions both under steady state and on immunization with sheep red blood cells, hapten-NP (4-hydroxy-5-nitrophenylacetate)–coupled OVA, or lipopolysaccharides⁴⁵ and have also been shown to promote peanut-specific IgG1 and IgE production in food allergy model studies.⁴⁶ These unexpected results from the PtenFC and Bcl6FC model FVIII immunization studies suggest that TFR cells play a dual role in FVIII immune responses, promoting FVIII inhibitor production while modulating the overall FVIII immune responses. Our results also suggest that the role of TFR cells in the FVIII immune response is likely different from that induced by other immunogens (eg, sheep red blood cells and NP-OVA). FVIII is a trace but unusually large protein.^72,73 In blood circulation, FVIII is bound to its carrier protein VWF, which can impact FVIII immune responses in many aspects.^{74, 75, 76, 77, 78, 79, 80, 81, 82} Thus, the mechanism of FVIII inhibitor development may be distinctive, as it involves both a noninhibitory and an inhibitory antibody development process, along with the influence of VWF. Indeed, it has been reported that MZ B cells play a critical role in initiating FVIII inhibitor formation in HA mice,⁸³ and the antigen trafficking and presentation process for FVIII differs from that of other proteins, such as OVA.⁸⁴ Although the underlying mechanism of how TFR cells play multifaceted roles in FVIII immune responses is still unclear, our study clearly shows that the development of FVIII inhibitors was completely aborted when Bcl6 expression was deleted in both TFH and TFR cells in CD4-Bcl6-cKO mice, confirming that the TFH/TFR pathway is crucial in FVIII immune responses.

In conclusion, the data reported here demonstrate that FVIII immunization induces TFR cell activation and expansion. TFR cells regulate FVIII-specific B-cell responses and IgG subclass switching during FVIII inhibitor development in mice, and the TFH/TFR pathway is essential for FVIII inhibitor development. Further study is warranted to identify the pathway by which TFR cells exert in optimizing inhibitor induction. This may aid in the development of novel therapeutics to restore tolerance to FVIII and reverse FVIII immune responses in patients with HA.

Conflict-of-interest disclosure: The authors declare no competing financial interests.