Extrafollicular CD4+ T-B interactions are sufficient for inducing autoimmune-like chronic graft-versus-host disease

Chronic graft-versus-host disease (cGVHD) is an autoimmune-like syndrome mediated by pathogenic CD4+ T and B cells, but the function of extrafollicular and germinal center CD4+ T and B interactions in cGVHD pathogenesis remains largely unknown. Here we show that extrafollicular CD4+ T and B interactions are sufficient for inducing cGVHD, while germinal center formation is dispensable. The pathogenesis of cGVHD is associated with the expansion of extrafollicular CD44hiCD62loPSGL-1loCD4+ (PSGL-1loCD4+) T cells. These cells express high levels of ICOS, and the blockade of ICOS/ICOSL interaction prevents their expansion and ameliorates cGVHD. Expansion of PSGL-1loCD4+ T cells is also prevented by BCL6 or Stat3 deficiency in donor CD4+ T cells, with the induction of cGVHD ameliorated by BCL6 deficiency and completely suppressed by Stat3 deficiency in donor CD4+ T cells. These results support that Stat3- and BCL6-dependent extrafollicular CD4+ T and B interactions play critical functions in the pathogenesis of cGVHD.

This study from Deng et al provides some novel information about the role of TFH-like cells in chronic graft versus host disease (cGVHD), using a mouse model system. The major claims of the paper are showing that extra-follicular CD4 T cells are involved in cGVHD pathogenesis and that cGVHD can be blocked by inhibiting these cells via an ICOS blockade, or via genetic disruption of the transcription factors Bcl6 or Stat3 in T cells. These finding are important for the understanding of cGVHD and also help to clarify respective roles of germinal center-associated TFH cells and extra-follicular TFH-like cells. However, there are a number of problems in the manuscript that weaken these conclusions, and a number of other concerns.
The following issues should be addressed in a revision: 1) What is the relationship of TFH cells and extra-follicular TFH-like cells in this model? In every mouse experiment, there should be a direct quantitation of TFH cells (CXCR5+ CD62L-low, PSGL-1low) and extra-follicular TFH-like cells (CXCR5-CD62L-low, PSGL-1-low). It also would be useful to compare the expression of PD-1 and ICOS on these two populations. 2) What is the cytokine profile of the extra-follicular TFH-like cells in the cGVHD system? Does the cytokine profile change over the course of the disease? 3) Are the extra-follicular TFH-like cells promoting pathology via helping B cells make auto-Ab or are they making inflammatory cytokines that drive disease independent of Ab? 4) The nomenclature used for the "no GVHD" groups are confusing in Figures 2,6,7. What are in the groups that say, for instance, "B-BCL-6+/+ -noGVHD"? What is the difference between the "B-BCL-6+/+-noGVHD" and "B-BCL-6-/--noGVHD" groups? 5) Figure 3C purports to show a difference in CXCR4 expression but both groups just have black boxes with no apparent differences. 6) Figure 7A is missing a label 7) When are spleens analyzed in Figure 7G? 8) In figure 8, the cytokine arrow between pre-TFH and B cells should be going from the pre-TFH to B cells, not other way as shown 9) It should be noted that deletion of Bcl6 and Stat3 with CD4-cre affects Tregs, and specifically inhibit TFR cell generation. 10) References should be given for the source of the Bcl6-flox mice and the Stat3-flox mice 11) In discussion "T-B boarder" is written twice. It should be "T-B border" Reviewer #3 (ICOS, cytokine signaling, Tfh) (Remarks to the Author): T follicular helper (Tfh) cells have emerged as a distinct subset/lineage of CD4+ T cells that are primarily responsible for providing help to B cells during immune responses to TD Ag and mediating the differentiation of B cells into memory and plasma cells during GC reactions in secondary lymphoid tissues. The critical function of Tfh cells is evident from animal models and human diseases where Tfh cells are reduced/absent and humoral immunity is impaired, while on the other hand excessive numbers/production or function of Tfh cells has been associated with, and likely causes, autoimmune conditions such as SLE etc. however, there are other "types" of CD4 T cells that also provide help to B cells that are not strictly located within the B-cell follicles -these have been coined extrafollicular CD4+ T cells, and were first identified and characterized by Joe Craft in the setting of a murine model of autoimmunity. Since then, other groups have identified analogous cells in normal humoral immune responses, implicating these cells in the early stages of B-cell activation and differentiation (pregerminal center).
In this current study, the authors have used numerous models of graft vs host disease and detailed physiological, developmental, cellular and functional analyses to reveal a critical role for extrafollicular CD4 T cells in disease pathogenesis. Importantly (but perhaps predictably) the generation of pathogenic extrafollicular CD4 T cells could be reduced by blocking ICOS/ICOS-L interactions, as well as genetic ablation of Bcl-6 or STAT3 in CD4 T cells. Overall, this study sheds substantial light on etiology of chronic GVHD, at least in murine models -it will need to be determined whether these findings are directly relevant to human GVHD. This notwithstanding, the study is well and comprehensively performed, and contains some important novel findings. Several comments follow that should be used as a guide to improve the novelty, mechanisms and conclusions of the findings.
1. The finding relating to the effects of STAT3 deficiency on disease need to be extended. First, what was the level of expression of Bcl-6 in Stat3-deficient CD4 T cells (and extrafollicular CD4 T cells specifically)? Second, Stat3 deficiency did not overcome acute GVHD, but greatly improved chronic GVHD. This raises questions about the nature and quality of extrafollicular CD4 T cells during these distinct phases of disease progression. Are these cells less likely to be pathogenic in the early/acute phase of GVHD? Third, Stat3 deficiency not only reduced pathogenic extrafollicular CD4 T cells but also increased Tregs. However, these latter cells were only assessed by phenotype -not function. Were the Stat3-deficient Tregs superior to WT Tregs at suppressing T cell activation and function? Was the increase in Tregs protective ie did disease severity worsen if Tregs were depleted from mice harbouring Stat3-deficient CD4 T cells? 2. several of the Figures contain results from data for gene expression from RNA Seq. these data need to be confirmed by Western blot or FACS. Eg Fig 3C - Fig 5D, it is possible that the reduced level of detected ICOS expressed by the labelled cells by FACS in the mice receiving the anti-ICOS mAb reflects masking by the blocking ICOS mAb that was still bound to the cells. this needs to be determined. Also for this data the treatment regime was extreme -blocking ICOS Ab injected every 2nd day for 45 days. Was this necessary? Was there a shorter time frame in which the ICOS/ICOS-L interaction could be blocked and a physiological effect still be detected?
The authors have clarified some issues.
Reviewer #2 (Remarks to the Author): As I wrote previously, the major claims of the paper are showing that extra-follicular CD4 T cells are involved in cGVHD pathogenesis and that cGVHD can be blocked by inhibiting these cells via an ICOS blockade, or via genetic disruption of the transcription factors Bcl6 or Stat3 in T cells. These findings are important for the understanding of cGVHD and also help to clarify respective roles of germinal center-associated TFH cells and extra-follicular TFH-like cells.
The revised manuscript addresses my most pressing concerns, but I still find it disconcerting that the authors do not have flow cytometry comparing Tfh versus the supposed extra-follicular Tfh-like cells. However, the RNA expression data does address CXCR5 expression and basically addresses the issue.
The legends are better, but in Figure 1, it should be laid out clearly what Group 1 is versus Group 2 versus Group 3. Also, TCD should be defined somewhere in the main manuscript.
In summary, the main conclusions of the paper are significant for the field, and convincingly presented. Statistics seem fine and there is probably enough detail given for reproduction by other researchers.
Reviewer #3 (Remarks to the Author): The authors have adequately addressed most of the concerns raised in teh original review of their manuscript. However, they did not assess/confirm differential gene expression by FACS, as requested -rather they have confirmed these differences by qPCR. This should suffice -tho it would have been preferable to see either protein expression (ie FACS analyses) or simply for the authors to state that they felt the data presented as qPCR was suitable.
Deng et al. report on the influence of BCL6 and STAT3 deficiency in CD4 T cells on germinal center formation and occurrence of graft versus host disease. The manuscript is well written but has several shortcomings in particular with respect to novelty and methods. We regret that we have created a misunderstanding about the main focus of our manuscript. (Fig 2).

We also agree that others have shown amelioration of acute and chronic GVHD by blockade of ICOS interaction with its ligand, and these publications have been incorporated into our text (see page 17). However, by using B-BCL6 deficient transplants that cannot give rise to germinal centers, we have specifically demonstrated that extrafollicular CD4 + T and B cell interactions also depend on ICOS/ICOS ligand interactions. Therefore, blockade ICOS/ICOS ligand interaction amelioration of chronic GVHD in the report by
Flynn et al may not result from prevention of germinal center formation, because our studies showed that there was no germinal center formation in those models. Instead, prevention of chronic GVHD may result from blocking extrafollicular CD4 + T and B interaction. This point has been incorporated into discussion (see page 17).
Taken collectively, the novelty of our report is very high, because it has clarified a highly controversial question about the role of germinal center formation in chronic GVHD pathogenesis. Our report has highlighted novel mechanisms for the pathogenesis of autoimmune-like chronic GVHD that are likely to be relevant to other autoimmune conditions.
2. Another weakness is the use of a well-known acute GVHD model (B6 into BALB/c) while the authors state that they investigate chronic GVHD. They considered the skin inflammation as a typical sign for chronic GVHD but acute GVHD is affecting the skin as well. Only in some initial experiment the authors have used a chronic GVHD model (LP/J to C57BL/6). A previous study by the same authors used the same model but named it acute GVHD model. The survival of the WT is very similar to the model that is described in the manuscript as cGVHD. 2. Typical signs of cGVHD are not studied such as antibody deposition, Th17, sclerodermititis, macrophage infiltration, fibrosis or other.
We understand that antibody deposition, Th17, scleroderma, macrophage infiltration and fibrosis are involved in chronic GVHD pathogenesis. However, those factors are not the focus of current studies. The focus of our studies are on using well described chronic GVHD models to unravel the role of follicular and extrafollicular donor CD4 + T and B interactions in the pathogenesis of chronic GVHD. Therefore, we include those factors only when it is necessary. For example, we compared antibody deposition in the thymus and skin tissues to help reveal the role of specific Stat-3 deficiency in donor CD4 + T cells in prevention of chronic GVHD (see newly added data Fig. S23 and page14, first paragraph).
3. Figure 6: Splenocytes from either WT or B-BCL-6 -/-C57BL/6 donors are used -therefore the authors cannot distinguish between effects caused by the BCL-6 deficient T cells or B-cells or other splenic immune cells of donor origin.
We regret the misunderstanding here. We would like to clarify that comparison of WT and specific BCL6 deficiency in B cells (B-BCL6 -/-) is with Figure 2. In Figure 6, we used spleen cells from WT donors or from donors with BCL6 deficiency specifically in CD4 + T cells, because we intended to study the role of BCL6 in donor CD4 + T cells in regulating follicular and extrafollicular CD4 + T and B cell interactions. Thus, it is appropriate to use whole spleen cells with specific BCL-6 deficiency in CD4 + T cells.

and Srinivasan et al. We have never found enlarged germinal centers in chronic GVHD recipients, as compared to control no-GVHD recipients given TCD-BM cells only. We observed much smaller germinal centers in recipients with very mild chronic GVHD. What's more chronic GVHD was successfully induced in recipiens which lack of germinal center formation (Figure 2)Therefore, our data indisputablely showed that chronic GVHD is associated with destruction or loss of germinal centers and was not associated with enlarged germinal centers (see result section page 6-7 and discuss section page 16).
We agree that mild or non-lethal acute GVHD is likely to damage the lymphoid structure and prevent germinal center formation in chronic GVHD recipients (see discussion section page17, first paragraph).

The adequate control for expansion of PSGL-1 lo CD4 + cells would be a syngeneic transplantation.
The additional transplantation of 1x10 6 splenocytes in GVHD recipients in comparison to no-GVHD recipients raises the question if the increased number that was observed is just by homeostatic proliferation and survival of these cells or by alloantigen based activation. Syngeneic transplantation of 1x10 6 splenocytes would also account for the additional transplanted T cell numbers and the homeostatic proliferation. Figure S14 and the Results on page 10, second paragraph).

We found that there was no expansion of PSGL-1 lo CD4 + T cells in syngeneic transplantation recipients (see newly added data
6. The microarray data are not very convincing with n=2. They have to be confirmed by quantitative RT-PCR using independent samples and on the protein level. Each sample represented the sorted PSGL-1 low CD4 T cells pulled from 8 mice, and two replicate experiments are shown. The related RNA-seq data are validated with real time PCR (see supplemental Fig. S12, S15 and S18) 7. Statistical tests for GVHD score, Body weight and survivals are missing. Figures1,4,6,7 and pages31,34,36,37).

Minor comments: …pathogenic CD4+ T and B cells, but the role of extrafollicular… Chronic GVHD often follows acute GVHD. Stimmt das?
Yes, it is true based on our best knowledge.

…cGVHD recipients… cGVHD patients
We did not understand the question and cannot address this point.
Introduction to much/complicated/detailed.

We intend to prepare sufficient background information for broad readership to help their understanding of this report, since Nature Communications has broader readership as compared to Blood.
BALB/c recipients were injected with TCD-BM plus 1x10 6 or 0.01 x10 6 C57BL/6 spleen cells.

Yes, it is true!
Reviewer #2 (Tfh, Bcl6) (Remarks to the Author): This study from Deng et al provides some novel information about the role of TFH-like cells in chronic graft versus host disease (cGVHD), using a mouse model system. The major claims of the paper are showing that extra-follicular CD4 T cells are involved in cGVHD pathogenesis and that cGVHD can be blocked by inhibiting these cells via an ICOS blockade, or via genetic disruption of the transcription factors Bcl6 or Stat3 in T cells. These finding are important for the understanding of cGVHD and also help to clarify respective roles of germinal center-associated TFH cells and extra-follicular TFH-like cells. However, there are a number of problems in the manuscript that weaken these conclusions, and a number of other concerns.
The following issues should be addressed in a revision: 1) What is the relationship of TFH cells and extra-follicular TFH-like cells in this model? In every mouse experiment, there should be a direct quantitation of TFH cells (CXCR5+ CD62L-low, PSGL-1-low) and extrafollicular TFH-like cells (CXCR5-CD62L-low, PSGL-1-low). It also would be useful to compare the expression of PD-1 and ICOS on these two populations.

We greatly appreciate the very insightful comment. The relationship of TFH and extra-follicular TFH-like cells in chronic GVHD mice remains unclear, although we have uncovered some hints.
Further studies in the future are required. As depicted in summary diagram (Fig. 9) Fig. S17 in result section). In a murine model of allergic pneumonia, classical CXCR5 + BCL6 + or CXCR5 + PD-1 + TFH cells were found in the draining LN, but not found in the inflamed lung tissue. However, CXCR5 -PD-1 + ICOS + CD40L + TFH-like CD4 + (Rao et al: Nature 2017). We would anticipate considerable difficulty in attempting to conduct similar studies in our model, due to the lack of CXCR5 + PD-1 + TFH cells in both lymphoid and GVHD target tissues of mice with chronic GVHD. As an alternative for our future studies, we plan to compare the surface phenotype and function of PSGL1 lo and PSGL1 hi CD4 + T cells in helping B cells in lymphoid tissues and GVHD target tissues of chronic GVHD recipients. (See pages 18).

We appreciate reviewer's suggestion to do a direct quantitation of TFH cells (CXCR5 + CD62L -PSGL-1 lo ) and extra-follicular TFH-like cells (CXCR5 -CD62L -PSGL-1 lo ) in each experiment and to measure PD-1 and ICOS expression by the two populations. These are helpful suggestions for future experiments. It would not be feasible for us to redo all of the experiments with these measures for this revision. In addition, doing those experiments is not critical to support the major conclusions of the current manuscript, namely 1) whether germinal center formation is required for induction of chronic GVHD; and 2) whether extrafollicular PSGL-1 lo CD4 + T cells play a critical role in induction of chronic GVHD. To further strengthen our major conclusion, we have performed add-back experiments. Injection of PSGL-1 lo CD4 + T cells from chronic GVHD recipients to recipients lacking PSGL-1 lo CD4 + T cells and cutaneous chronic GVHD. We found that injection of PSGL-1 lo CD4 + T cells restored cutaneous chronic GVHD (see Supplemental figure S17 on page 13 in result section).
2) What is the cytokine profile of the extra-follicular TFH-like cells in the cGVHD system? Does the cytokine profile change over the course of the disease?
We used intracellular cytokine flow cytometry and found that PSGL-1 lo CD4 + T cells in the spleen of chronic GVHD mice at 21 days after HCT produced IFN-γ, IL-13, IL-21and IL-17 (see Figure 3D and page 9 bottom). We were not able to make similar measurements at later time points due to too few cells resulted from lymphopenia. We would also like to point out that intracellular cytokine staining requires in vitro stimulation with PMA plus Ionomycin of T cells for 4 hours, but this stimulation also changes PSGL-1 expression by CD4 + T cells. Therefore, in order to measure cytokine profile of PSGL-1 lo CD4 + T cells, we had to use flow cytometry sorting to purify the cell subset first, then stimulate them with PMA plus Ionomycin during in vitro culture and then measure their cytokine profile. The procedure requires large numbers of T cells.
3) Are the extra-follicular TFH-like cells promoting pathology via helping B cells make auto-Ab or are they making inflammatory cytokines that drive disease independent of Ab?
We observed that both PSGL-1 lo CD4 + T cells and tissue antibody deposition contribute to chronic GVHD pathogenesis, because blockade of PSGL-1 lo CD4 + T interaction via blockade ICOS/ICOS ligand interaction reduced serum anti-dsDNA concentration and ameliorated chronic GVHD tissue damage (Fig. 4 and 5). Reduction of PSGL-1 lo CD4 + T cell expansion by Stat3 deficiency in donor CD4 + T cells also decreased serum autoantibody production and tissue antibody deposition and ameliorated chronic GVHD tissue damage ( Fig.S22 and newly added S23). These results indicate that PSGL-1 lo CD4 + T cells interact with B cells and augment B cell production of IgG antibodies, resulting in chronic GVHD tissue damage. Whether PSGL-1 lo CD4 + T cells alone can induce tissue damage or whether IgG antibody production from PSGL-1 lo CD4 + T interaction with B cells is required for chronic GVHD tissue damage will need to be addressed in the future studies.
To avoid confusion, we have clarified the labels in corresponding figure legends. Figure 3C purports to show a difference in CXCR4 expression but both groups just have black boxes with no apparent differences.

5)
We agree that the color difference in Figure 3C for CXCR4 was not clear. We have use real-time PCR to compare the difference and found it was statistically significant (see newly added Fig S12) 6) Figure 7A is missing a label The label has been added. Figure 7G? This error has been fixed.

7) When are spleens analyzed in
9) It should be noted that deletion of Bcl6 and Stat3 with CD4-cre affects Tregs, and specifically inhibit TFR cell generation.

11) In discussion "T-B boarder" is written twice. It should be "T-B border"
This error has been corrected.
Reviewer #3 (ICOS, cytokine signaling, Tfh) (Remarks to the Author): T follicular helper (Tfh) cells have emerged as a distinct subset/lineage of CD4+ T cells that are primarily responsible for providing help to B cells during immune responses to TD Ag and mediating the differentiation of B cells into memory and plasma cells during GC reactions in secondary lymphoid tissues. The critical function of Tfh cells is evident from animal models and human diseases where Tfh cells are reduced/absent and humoral immunity is impaired, while on the other hand excessive numbers/production or function of Tfh cells has been associated with, and likely causes, autoimmune conditions such as SLE etc. however, there are other "types" of CD4 T cells that also provide help to B cells that are not strictly located within the B-cell folliclesthese have been coined extrafollicular CD4+ T cells, and were first identified and characterized by Joe Craft in the setting of a murine model of autoimmunity. Since then, other groups have identified analogous cells in normal humoral immune responses, implicating these cells in the early stages of B-cell activation and differentiation (pre-germinal center).
In this current study, the authors have used numerous models of graft vs host disease and detailed physiological, developmental, cellular and functional analyses to reveal a critical role for extrafollicular CD4 T cells in disease pathogenesis. Importantly (but perhaps predictably) the generation of pathogenic extrafollicular CD4 T cells could be reduced by blocking ICOS/ICOS-L interactions, as well as genetic ablation of Bcl-6 or STAT3 in CD4 T cells. Overall, this study sheds substantial light on etiology of chronic GVHD, at least in murine models -it will need to be determined whether these findings are directly relevant to human GVHD. This notwithstanding, the study is well and comprehensively performed, and contains some important novel findings. Several comments follow that should be used as a guide to improve the novelty, mechanisms and conclusions of the findings.
1. The finding relating to the effects of STAT3 deficiency on disease need to be extended. First, what was the level of expression of Bcl-6 in Stat3-deficient CD4 T cells (and extrafollicular CD4 T cells specifically)? Second, Stat3 deficiency did not overcome acute GVHD, but greatly improved chronic GVHD. This raises questions about the nature and quality of extrafollicular CD4 T cells during these distinct phases of disease progression. Are these cells less likely to be pathogenic in the early/acute phase of GVHD? Third, Stat3 deficiency not only reduced pathogenic extrafollicular CD4 T cells but also increased Tregs. However, these latter cells were only assessed by phenotype -not function. Were the Stat3-deficient Tregs superior to WT Tregs at suppressing T cell activation and function? Was the increase in Tregs protective ie did disease severity worsen if Tregs were depleted from mice harbouring Stat3-deficient CD4 T cells?
We greatly appreciate the reviewer's suggestion, and we have conducted new experiments to elucidate the role of Stat3 deficiency in CD4 + T cells in regulating chronic GVHD pathogenesis. We have also addressed this question by citing a previous publication.

Radojcic et al (J. Immunol. 2010) showed that Stat-3 deficiency in donor CD4 + T cells effectively prevented induction of chronic GVHD in a model with B10D2 donors and MHC-matched BALB/c recipients. Prevention of chronic GVHD was associated with protection of the thymus and its ability to generate Treg cells, because depletion of Treg cells by anti-CD25 depletion abolished prevention of chronic GVHD in recipients given grafts from donors with specific Stat-3 deficiency in donor CD4 + T cells. Mechanisms that protect the thymus were not defined in that study.
In the current report, we showed that chronic GVHD in BALB/c recipients given B10D2 transplants lost lymphofollicles and germinal center formation (Fig. S7), as observed with with BALB/c recipients given MHCmismatched C57BL/6 transplants. We found that specific Stat3 deficiency in donor CD4 + T cells protected the thymus and increased the numbers of Foxp3 + Treg cells in the periphery in BALB/c recipients given C57BL/6 transplants ( Fig. S19 and Fig. S24). These findings are consistent with the results reported by Rodojcic et al.
Furthermore, we found that thymus protection in BALB/c recipients given C57BL/6 donor spleen cells with specific Stat3 deficiency in CD4 + T cells (CD4-Stat3 -/-) was associated with a significant reduction of expansion