BAFF promotes heightened BCR responsiveness and manifestations of chronic GVHD after allogeneic stem cell transplantation

Chronic graft-versus-host disease (cGVHD) remains a life-altering, potentially fatal immune toxicity in patients otherwise cured of hematolymphoid cancers through allogeneic hematopoietic stem cell or bone marrow transplantation (allo-BMT).¹  Studies of patients and mice suggest that the underlying pathophysiology of cGVHD occurs along a spectrum that starts early after allo-BMT,²  well before the clinical signs and symptoms manifest. Alloreactive T cells are known to incite acute and cGVHD, but T-cell elimination to prevent GVHD compromises antitumor effects.³  Clinically active cGVHD relies on a coordinated T- and B-cell response,^4,5  and antibody production is required for development and perpetuation of cGVHD.^6-11  Therefore, B-cell targeting to treat or avert cGVHD is an area of active investigation.

Unfortunately, elimination of CD20⁺ B cells with rituximab is only sporadically effective, likely related to an inability to achieve B-cell homeostasis.^12,13  This hypothesis is supported by the observation that CD27⁺ memory B cells persist¹⁴  and are capable of constitutively producing immunoglobulin G (IgG).^12,14  B cells from patients with clinically active cGVHD display increased B-cell receptor (BCR) responsiveness and are preferentially killed after BCR inhibition.^14,15  Importantly, constitutive BCR signaling is critical for survival of both CD27⁺ and CD27⁻ B-cell subsets in both patients with cGVHD and mice, an effect autoamplified by enhanced NOTCH inputs.^16,17  Blocking BCR-activated B cells with SYK or BTK inhibitors attenuated disease in mouse models,^7,18,19  and subsequent clinical trials documented clinical benefit and led to US Food and Drug Administration approval of the BTK inhibitor ibrutinib for cGVHD.²⁰ 

Targeting B cells does not address the extrinsic factor B-cell activating factor (BAFF), known to determine B-cell homeostasis^21,22  and BCR responsiveness.²³  Intriguingly, B cells from patients with cGVHD are activated and primed for survival via BAFF-associated pathways.²⁴  Increased plasma BAFF levels are associated with disease severity, treatment response, and outcome in patients with cGVHD.^12,25-29  Evidence supporting a pathologic role for BAFF in cGVHD is lacking,³⁰  because high plasma BAFF levels after allo-BMT are potentially a consequence of low B-cell number related to poor B-cell production.^31-35  Without peripheral B cells, cGVHD might not develop,^5,6,8,9  but without BAFF, naïve/mature B-cell recovery after BMT does not occur, and B-cell homeostasis cannot be achieved.³⁶  Thus, studying and potentially targeting BAFF after BMT are not straightforward, given the concentration-dependent dual role played by BAFF in conferring B-cell immune homeostatic vs pathologic effects.^37,38 

To address whether BAFF plays a pathologic role in cGVHD, we employed BAFF knockout (KO) and BAFF transgenic (Tg) donor mice in allo- and syngeneic BMT (Syn-BMT). We found posttransplantation increases in BAFF production occurred in mice that developed cGVHD manifestations. Elevated BAFF promoted NOTCH2 expression on B cells and maintained SYK protein after BCR engagement. Further corroborating a pathologic role for BAFF, we demonstrate that excess BAFF and alloantigen operate together to promote a circulating BCR-responsive B-cell pool and alloantibody production. Thus, we demonstrate that BAFF is a pivotal extrinsic regulator of B-cell pathology in cGVHD.

C57BL/6, µMT, and BALB/c mice were obtained from The Jackson Laboratory. G.K.M. (University of North Carolina) provided BAFF KO mouse splenocytes and BM cells, and J.C.R. (Duke University) provided BAFF Tg mice. Female 10- to 12-week-old age-matched C57BL/6 (H-2K^b), BAFF Tg (C57BL/6 background; H-2K^b),³⁹  BAFF KO (H-2K^b),⁴⁰  or μMT (H-2K^b) donor mice and BALB/c (H-2K^d) recipient mice were used in allo-BMT experiments. T cell–depleted BM was used as previously described.⁴¹  Mice were housed at the Duke Cancer Center Isolation Facility. All animal experiments were approved by the Institutional Animal Care and Use Committee of Duke University. Eye pathology was scored using a Stemi 2000-C stereo microscope (Zeiss) with camera (Nikon Coolpix P5100) as described.⁴¹  Lung and liver fibrosis was assessed by Masson’s trichrome staining and quantitated by ImageJ software (National Institutes of Health).

Detailed material and methods can be found in the data supplement, which includes the following: BAFF measurements, flow cytometric analyses with antibodies used to examine B-cell subsets, BAFF receptor (BAFF-R) occupancy determination, BCR stimulation assay with B-cell phosphoflow and intracellular staining, carboxyfluorescein diacetate succinimidyl ester (CFSE) dilution assay, immunohistochemical staining, quantitation of images by ImageJ, stromal cell isolation and cell sorting, conjunctiva tissue cell preparation, quantitative polymerase chain reaction of BAFF transcript measurement, in vitro NOTCH2 stimulation assay, analysis of SYK protein stability and degradation, western blot, autoantibody detection, and statistical analyses.

In patients, before and at the time of clinically apparent cGVHD, plasma BAFF levels are significantly increased.^25,28  High BAFF is associated with an altered and activated peripheral B-cell compartment in these patients.^14,25,42  We first examined whether our mouse model recapitulated these patient findings. We substantiated cGVHD manifestations,⁴³  including ocular GVHD and weight loss, starting a median of 30 days after allo-BMT.⁴¹  These cGVHD manifestations were associated with collagen deposition and lymphocyte infiltration in tissues (supplemental Figure 1, available on the Blood Web site).⁴¹  As shown in Figure 1A, mice receiving BM and splenocytes (BM + Sp) had persistent and significantly increased plasma BAFF levels compared with mice that did not develop disease (BM only) and with B6 Syn-BMT mice early after allo-BMT, starting on day 8 after transplantation. Because high BAFF is found when B-cell numbers are low after allo-BMT,⁴⁴  we used μMT donors to ensure equally low B-cell numbers after allo-BMT before measuring BAFF levels (Figure 1B). We found that BAFF levels in µMT donor cell recipients were further and significantly increased if they also received T cells (Figure 1C), suggesting high BAFF levels in BM + Sp mice were not merely due to a paucity of B cells wielding BAFF-R, which removes soluble BAFF from plasma.

After Syn-BMT, BAFF is required for B-cell recovery and is derived from recipient stromal cells.³⁶  To determine whether BAFF production by recipient stromal cells is required for cGVHD development, we used BAFF KO mice as donors. We found that in the absence of donor-derived BAFF, BAFF levels remained significantly higher in cGVHD mice (Figure 1D). Despite high BAFF levels after allo-BMT, profound B-cell lymphopenia was observed in recipients of BAFF KO BM + Sp (supplemental Figure 2A), resulting in significantly higher BAFF/B-cell ratios (supplemental Figure 2B). Recipients of BAFF KO BM + Sp developed measurable weight loss and eye manifestations indistinguishable from wild-type (WT) BM + Sp recipients (Figure 1E-F), suggesting that recipient-derived BAFF was sufficient to promote cGVHD development. Consistent with a previous report,⁴⁵  we found that the secondary lymphoid organs (SLOs) were significantly enlarged early after allo-BMT in mice that went on to develop cGVHD compared with controls (Figure 2A). We examined nonhematopoietic stromal cells by gating on CD45⁻ cells and then PDPN⁺ and CD31⁻ cells as previously described (Figure 2B).⁴⁶  Among stromal subsets, PDPN⁺CD31⁻MadCaM⁻ FRCs, previously established as major BAFF producers,⁴⁷  were significantly proportionally and numerically increased in BM + Sp recipients that later developed cGVHD (Figure 2C-D). Importantly, FRCs from cGVHD mice had significantly more BAFF transcripts (Figure 2E), suggesting an increased capacity to produce more BAFF. By contrast, no BAFF transcripts could be detected in CD31⁺ stromal cells (data not shown). Thus, recipient FRCs are a potential early source of soluble BAFF in mice that develop cGVHD.

Just before disease manifested, BAFF levels in WT BM + Sp recipients trended higher than in BAFF KO BM + Sp recipients (mean, 96.7 ng/mL [Figure 1A] vs 83.3 ng/mL [Figure 1D], respectively), suggesting donor cells may also be sources of BAFF in cGVHD. Therefore, we examined BAFF transcript levels in SLOs and donor-derived blood cells. Because follicular helper T cells (T_FH) from healthy mice are known to produce BAFF in the germinal center (GC) niche postimmunization and insufficient CD4⁺CXCR5⁺ T_FH cells were found in SLOs in our diseased mice on day 37,⁴⁸  we examined total CD4⁺ T cells from the spleens of cGVHD vs control mice. In cGVHD mice, we found increased BAFF messenger RNA in cGVHD CD4⁺ T cells (Figure 2F). Interestingly, T_FH from patients with clinically active cGVHD also had significantly increased BAFF transcripts compared with those from patients who never had cGVHD (Figure 2G). By contrast, BAFF transcript results in diseased and nondiseased neutrophils, CD8⁺ T cells, B cells, and macrophages/monocytes from mice (supplemental Figure 2C) and patients were similar (data not shown). Together our data indicate that excess BAFF protein levels are not merely a reflection of low peripheral B-cell counts and that alloreactive T cells also account for lesional tissue increases in BAFF production.

Although steady B-cell recovery occurred after BM only or Syn-BMT,³⁶  cGVHD mice maintained relatively low total blood and splenic B-cell numbers despite high BAFF levels (supplemental Figure 3A-C). Thus, BAFF concentration per B cell (BAFF/B-cell ratio) after allo-BMT in mice with cGVHD manifestations was significantly increased compared with mice without cGVHD (BM only) or control Syn-BMT mice (Figure 3A). In the absence of alloantigen, BAFF promotes late CD93⁺ T2 to T3 populations.^49-51  After allo-BMT, we found that cGVHD mice had significant proportional increases in transitional CD19⁺CD93⁺ B cells and a significant enrichment of T3 B cells (supplemental Figure 3D; Figure 3B-C). T3 cells are generally anergic and unable to mature.^51-53  Therefore, our data suggest that BAFF promotes B-cell anergy after allo-BMT in cGVHD mice.

To examine if BAFF affected BAFF receptors on B cells in cGVHD, we capitalized on the known decrease in BAFF-R detection after occupancy by soluble BAFF.^12,54  Significantly lower BAFF-R was detected on B cells from cGVHD mice (Figure 3D). To determine whether BAFF-R was occupied by BAFF, we performed dissociation and acid elution assays. As shown in Figure 3E, BAFF was readily removed from BAFF-R under both conditions, affirming that low BAFF-R detectability in cGVHD mice reflected receptor occupation by BAFF.⁵⁴  Additionally, we found that the BAFF receptor with known negative regulatory capacity, transmembrane activator and CAML interactor, was decreased,⁵⁵  whereas B-cell maturation antigen trended higher but did not reach statistical significance (P = .41; supplemental Figure 4A-B). These data suggest that BAFF promotes activation rather than survival of most B cells in cGVHD.

Given the known role of BAFF in BCR activation⁵⁶  and that of the proximal BCR signaling molecule SYK in cGVHD development,^15,18,41  we examined relative B-cell responses to surrogate BCR agonist. We found that a subset of B cells from mice with cGVHD manifestations had increased responsiveness to BCR stimulation (Figure 4A). Total blood B cells from cGVHD mice had increased SYK and BLNK activation after ex vivo stimulation through the BCR (supplemental Figure 5A-B). We previously found that mice treated with SYK inhibitors had decreased peripheral blood GL7⁺ B cells.^18,41,57,58  Therefore, we examined whether GL7⁺ cells had increased SYK phosphorylation and found significantly increased phosphorylated SYK in GL7⁺ B cells, which was further significantly increased in the cGVHD setting (Figure 4B). Differential IgM expression alone did not account for differences in BCR response, because most IgM^high B cells were also GL7⁺ (Figure 4C; supplemental Figure 5C-D). Recombinant BAFF promoted BCR activation–driven GL7⁺ B cells and GL7 expression on healthy mouse B cells (Figure 4D; supplemental Figure 5E). Consistent with this, we found the percentage of circulating GL7⁺ B cells in cGVHD mice relative to control mice was significantly and persistently increased (Figure 4E) when BAFF was high in vivo (Figure 1A). Likewise, cGVHD mice with significantly increased clinical eye scores (Figure 4F) had a significant proportional increase in conjunctival GL7⁺ B cells⁵⁹  (Figure 4G). Circulating GL7⁺ B cells are not GC B cells, because they do not express CD95 (supplemental Figure 6A-B). The frequency of GL7⁺ B cells was increased in spleen by day 30 post-BMT (supplemental Figure 6C). Together, these data suggest that BAFF acts synergistically with alloantigen to promote circulating GL7⁺ BCR-activated B cells in cGVHD.

To determine if BAFF after T cell–depleted allo-BMT affected BCR responses, we examined cells from mice undergoing allo-BMT and recipients of BAFF Tg BM only. B cells from recipients of BAFF Tg BM only were more BCR responsive compared with B cells from WT BM only controls (Figure 5A). This increased BCR responsiveness was not related to differences in surface IgM density, which was similar between groups (supplemental Figure 7A). Because anti-NOTCH2 (NRR2) antibody blocked GL7⁺ B cells in another mouse model of cGVHD,¹⁷  we examined NOTCH2 expression on GL7⁺ B cells. We found that diseased mice had a significant increase in a NOTCH2 and GL7 double-positive (NOTCH2⁺GL7⁺) B-cell subset after onset of disease manifestations compared with control allogeneic or syngeneic recipients (Figure 5B-C). B cells from active cGVHD patients maintain higher NOTCH2 expression after ex vivo stimulation and are highly responsive to NOTCH ligand plus BCR stimulation,¹⁶  suggesting that BAFF may promote BCR hyperresponsiveness in part through increases in NOTCH2 expression and activation.

To test whether NOTCH2 expression after BCR NOTCH activation¹⁶  was influenced by BAFF, we examined NOTCH2 expression on B cells after allo-BMT either with or without in vivo exposure to excess BAFF. Strikingly, we found that B cells maintained in an alloantigen-rich, excess BAFF (BAFF Tg) environment had significantly higher NOTCH2 expression after ex vivo engagement of BCR (Figure 5D; supplemental Figure 7B). To determine if BAFF influenced the BCR-NOTCH2 signaling axis in our cGVHD model, we stimulated B cells from BAFF Tg allo-BMT mice with NOTCH ligand and surrogate antigen. B cells were significantly more responsive to BCR agonist when Delta-like 1 was also present (Figure 5E-F). MZ B cells rely on NOTCH2 expression and have a decreased BCR signaling threshold. Extensive analyses of blood and spleen for CD93⁻CD21^highCD23^low MZ B cells revealed that typical MZ B cells were not increased in cGVHD mice, and MZ B cells were not generally detectable in the blood of any BMT mice, including those with BAFF Tg donors (supplemental Figure 8A-B). Only in the spleen when BAFF was overexpressed after BMT, even in the absence of alloantigen (Syn BAFF Tg), were MZ B cells proportionally increased (supplemental Figure 8B; Figure 5G). Thus, although increased BAFF did not promote classical MZ B cells, we found that circulating B cells exposed to high levels of BAFF in vivo were MZ-like, with significantly higher responsiveness to BCR and NOTCH agonists.

To determine if BAFF affected SYK protein, we analyzed B cells from BAFF Tg vs WT mice and found that B cells with high SYK and BLNK levels were maintained after in vitro BCR engagement after high BAFF exposure in vivo (Figure 6A; supplemental Figure 9A). Like cGVHD patient B cells,¹⁵  B cells from cGVHD mice had increased intracellular SYK and BLNK (Figure 6B; supplemental Figure 9B-D). To dampen aberrant BCR responsiveness, SYK protein is typically degraded after BCR engagement.^60,61  Strikingly, B cells from a high BAFF environment maintained higher SYK protein levels 18 hours after in vitro BCR engagement (Figure 6C). These findings in the BAFF Tg B cells were not related to differences in IgM expression (supplemental Figure 9E). We then examined whether SYK protein half-life was prolonged in B cells from high BAFF environments, comparing SYK levels in BAFF Tg and WT B cells after stopping protein synthesis in a classical cycloheximide chase assay. As shown in Figure 6D-E, SYK levels remained significantly higher both before and after ex vivo BCR stimulation if B cells were from BAFF Tg mice, further suggesting that BAFF promotes the proximal BCR signalosome before and after antigen engagement.

To determine if BAFF and alloantigen operate together to promote BCR activation in cGVHD in vivo, we examined B cells from syngeneic vs allogeneic recipients of T cell–depleted (BM only) BAFF Tg. We found that SYK was significantly increased in BAFF Tg allo-BMT mice (Figure 6F; supplemental Figure 9F). Although elevated BAFF in vivo alone did not affect BLNK levels, BLNK was also significantly increased when BAFF and alloantigen were both present (supplemental Figure 9G-H). This increased SYK was not associated with BAFF-R occupancy by BAFF. As shown in Figure 6G, BAFF-R occupancy by BAFF was the same in Syn BAFF Tg and BAFF Tg BM B cells after BMT. Thus, B cells require both BAFF and alloantigen for SYK maintenance and BCR activation.

Finally, we assessed whether excess BAFF and alloantigen promoted cGVHD manifestations. Allo-BMT mice with BAFF Tg donors were compared with Syn-BMT mice with BAFF Tg donors. We found that T cell–depleted BAFF Tg (BM only) recipients had significantly more severe cGVHD manifestations, including ocular disease and weight loss (Figure 7A-B). To also determine whether high BAFF and alloantigen promoted increased GL7⁺ B cells in circulation and cGVHD development, we used T cell–depleted donor cells from BAFF Tg mice in T cell–depleted allo- vs Syn-BMT. BAFF Tg BM + Sp values are included for reference. Importantly, circulating GL7⁺ BCR-activated B cells were significantly proportionately increased in allogeneic BAFF Tg BM only recipients (Figure 7C). As expected, BAFF Tg recipient mice had significantly higher BAFF levels compared with WT recipients (supplemental Figure 10A). Despite high BAFF levels, B-cell numbers remained notably low (supplemental Figure 10B). After ex vivo BCR reengagement, we found increased dead cells in BAFF Tg BM only cultures, suggesting enhanced activation-induced death of B cells in cGVHD mice (supplemental Figure 10C). This is consistent with a proportional increase in T3 cells, which are considered the death niche for anergic B cells that are increased in BAFF Tg recipient cGVHD mice (supplemental Figure 10D-E), possibly explaining low B-cell numbers and persistence of high BAFF/B-cell ratios in mice with cGVHD. Overexpression of soluble BAFF and persistent total B lymphopenia seemed to preferentially promote GL7⁺ B cells (Figure 7C). In parallel with the increase in GL7⁺ B cells, we found increased antibody to recipient antigens in BAFF Tg recipient mice compared with controls, particularly in the context of allo-BMT (Figure 7D-E). Persistent excess BAFF after allo-BMT promoted GL7⁺ B-cell survival, antirecipient antibody production, and cGVHD manifestations. Additionally, even without the addition of splenocytes or T cells, high BAFF and presence of alloantigen significantly increased cGVHD manifestations.

Because immune reconstitution occurs in the setting of ubiquitous foreign antigen, ongoing removal of B cells reactive to recipient tissues is imperative for immune tolerance. The BAFF checkpoint has been well described in de novo autoimmune diseases. Intermediate affinity autoreactive B cells are known to be promoted by excess soluble BAFF when specific antigen is present in BCR Tg models.^62,63  We have long known that B cells play a pathogenic role in cGVHD, and we find high levels of soluble BAFF in patients, but a pathologic role for BAFF after allogeneic hematopoietic stem cell transplantation has remained elusive. We now find that BAFF rescues aberrant, BCR-activated B cells in vivo in mice with cGVHD manifestations.

In the current study, we used a mouse model that recapitulated the aberrant B-cell phenotype of patients and afforded measurable physical manifestations of cGVHD. We revealed the mechanistic links between BAFF and aberrant BCR signaling in the development of cGVHD. Specifically, we found that extrinsic factors, BAFF and alloantigen, operate together to drive a circulating BCR-responsive B-cell compartment in mice with cGVHD manifestations. Strikingly, BCR-activated B cells that we found in cGVHD mouse circulation and in lesional tissue expressed the activation marker GL7, further suggesting extrafollicular B cells may have been incited by GC reactions but can be found outside of the GC microenvironment in cGVHD.⁵  For the first time, we show that BAFF production is increased after allo-BMT in cGVHD. By employing BAFF Tg and BAFF KO donors, we also begin to define a pathologic role for BAFF in cGVHD.

In this study, we found increased BAFF production in mice that developed cGVHD manifestations. Allo-BMT mice that received T cells had increased niche production of BAFF by an increased number of FRCs in SLOs, suggesting that alloreactivity promoted FRCs and BAFF production (Figures 1 and 2). Early interactions between FRCs and T cells through DLL and NOTCH1 are the key for activation of alloreactive T cells and GVHD initiation.⁶⁴  Increased BAFF production in this model in which mice develop subclinical acute GVHD suggests a pathologic role for BAFF in cGVHD genesis. Although BAFF produced by CD4⁺ T cells seemed dispensable for disease development, the importance of this source for later stages of cGVHD remains unknown. Donor T cells have increased BAFF transcript, and further study of these cells as BAFF-producing cells in cGVHD is warranted. Given that FRCs themselves are targets during GVHD⁶⁵  and their loss is accompanied by impaired humoral immunity, it is possible that the T_FH source of BAFF becomes important for perpetuation of pathologic B-cell homeostasis in established cGVHD. We found that during early phases of cGVHD development, FRC numbers were significantly increased and FRCs produced more BAFF. However, a recent acute GVHD study found no difference in FRC cell numbers between GVHD and control,⁶⁶  potentially related to higher-dose T-cell administration in these model systems. Additionally, ongoing B-cell lymphopenia likely also continues to account for, and potentially drives, high soluble BAFF levels after allogeneic transplantation. We now provide novel strong evidence for pathologic BAFF production in cGVHD and characterize an extrafollicular GL7⁺ BCR-responsive B-cell population, circulating in cGVHD.

Importantly, we found that BAFF and alloantigen together promoted an altered peripheral B-cell compartment (Figure 3) with a propensity to immediately respond through BCR. Strikingly, BAFF promoted alloreactive GL7⁺ BCR-activated B-cell recovery and pathology even when GVHD-inducing T cells were not transplanted (Figure 7), although which B cells were responsible for BAFF-associated alloantibody production (Figure 7D) remains unknown. In patients with clinically active cGVHD, the peripheral B-cell pool is prone to BCR hyperresponsiveness, and a subset of CD27⁺ B cells is capable of constitutive IgG production.¹⁴  Low BAFF-R detectability is a hallmark of activated and IgG-producing B cells^67,68  and is observed in human patients and cGVHD mice. Because BAFF is known to induce class switching, it is tempting to speculate that niche IgG production may occur in lesional tissues.^14,69  Our data suggest that after allo-BMT, most B cells encounter alloantigen and become anergic and die,^62,63  leading to profound B lymphopenia. We show that excess BAFF affords maintenance of B cells that become BCR activated after allo-BMT but do not die when they encounter alloantigen. CD27 does not mark a similar population in mice. Using the GC marker GL7⁺, we find that excess BAFF is associated with the presence of constitutively BCR-activated B cells recirculating in cGVHD and alloantibody production.

The role of BCR-activated B cells in cGVHD pathophysiology is well established,^15,18,19  and novel small molecules that target BTK and SYK are approved for use and being studied in patients.^18,19,70  cGVHD patient B cells have increased protein levels of SYK and BLNK.¹⁵  We extend these findings to the cGVHD environment and document a breakdown of B-cell tolerance control that is directly attributed to B-cell alloantigen exposure and BAFF. SYK has previously been implicated in BAFF-R signaling.⁷¹  We now reveal how BAFF mediates constitutive BCR signaling after allo-BMT. How BAFF mediates these effects on a molecular level had remained largely unknown. Our data suggest that in cGVHD mice, when certain B cells encounter alloantigen, they do not die, and they do not stop signaling via BCR. We also found both that NOTCH2 expression and responsiveness were increased and that SYK protein was not degraded after BCR engagement in B cells after exposure to high levels of BAFF in vivo. Thus, we now show that B cells bathed in vivo with high BAFF experience loss of this negative regulation. Consistent with loss of this negative regulatory feedback loop when challenged with surrogate antigen ex vivo, and in keeping with the previously documented pathogenic role for NOTCH in cGVHD B-cell promotion, we show that BAFF promotes increased BCR responsiveness, because even without T cells in the transplant product, high BAFF promoted activation with surrogate antigen (Figure 5).^16,17 

On the basis of the previously reported association between excess BAFF and cGVHD,^12,14  the anti-BAFF monoclonal antibody belimumab is already being examined for safety and efficacy in allogeneic transplantation (registered at www.clinicaltrials.gov as #NCT03207958). Loss of the T3 B-cell population after use of belimumab in patients with systemic lupus erythematosus⁷²  is consistent with our findings of BAFF regulation of this cellular subset. Although the efficacy of belimumab has been limited in systemic lupus erythematosus, our prior work identified SYK addiction of cGVHD B cells, and current data highlight the essential role of elevated BAFF in control of BCR responsiveness, potentially through SYK protein maintenance. Our data provide a novel mechanistic rationale for SYK or BAFF blockade in cGVHD and other diseases with aberrations in B-cell tolerance.

The authors thank Divino Deoliveira for helping establish the mouse cGVHD model, Lauren Riley for critical reading of the manuscript, Michael Cook, Bin Li, and Lynn Martinek for helping with cell sorting, and Zuowei Su for helping with immunohistochemical staining.

This work was supported by National Institutes of Health (NIH)/National Heart, Lung, and Blood Institute grants 1K08HL145116-01A1 (V.R.), R01HL129061 (S.S.), and NIH/National Cancer Institute Cancer Center support grant P30CA014236.

Contribution: W.J., J.C.P., and S.S. designed the study; S.S., W.J., and J.C.P. wrote the paper; W.J., H.S., S.A., N.J.R., D.M.C., K.I., A.N.S., I.M.C.-C., and R.A.D. performed the experiments, acquired data, and analyzed results; Z.L. performed statistical analysis; G.K.M. provided BAFF KO mice; J.C.R. provided BAFF Tg mice; and V.R., I.M., N.J.R., D.M.C., D.R.S., B.J.C., and N.J.C. interpreted data and edited the manuscript.

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

Correspondence: Stefanie Sarantopoulos, Adult Blood and Marrow Transplant Program, Duke University, Box 3961, 2400 Pratt St, Suite 5000, Durham, NC 27110; e-mail: stefanie.sarantopoulos@duke.edu.