Distinct roles for LTalpha3 and LTalpha1beta2 produced by B cells contribute to their multi-faceted impact on ileitis

B lymphocytes may facilitate chronic inflammation through antibody production or secretion of cytokines, including lymphotoxin (LT)-a1b2 associated with development of lymphoid tissue. Tertiary lymphoid structures (TLS) characterize human and murine ileitis by suppressing outflow from the ileum. Here, we show that B cell-derived secretory IgA protected against ileal inflammation, whereas B cell-derived LTa guarded against ileitis-associated loss of body mass. We initially hypothesized this protection resulted from formation of TLS that suppressed lymphatic outflow and thereby restrained systemic spread of inflammatory signals, but B cell-selective deletion of LTb did not exacerbate weight loss, despite eliminating TLS. Instead, weight loss driven by the cachectic cytokine TNF was exacerbated when LTa3, another ligand for TNF receptors, was selectively neutralized. Thus, B cells’ multi-faceted impact on ileitis includes generating secretory IgA, expressing LTa1b2 to drive formation of TLS, and producing LTa3 for protecting against weight loss in the presence of TNF.


INTRODUCTION
Local intestinal in ammation is a major feature of Crohn's disease (CD), one of the most common forms of in ammatory bowel disease (IBD) 1,2 .While gastrointestinal in ammation is a hallmark, up to 46% of CD patients present with systemic manifestations of disease, such as arthritis and low body mass index 3,4 .In the gastrointestinal tract, Crohn's disease involves all layers of the intestine, from the mucosal surface to the underlying mesentery, and most frequently affects the ileum 2 .Beginning with the work of B. B. Crohn, who referred to the disease as regional ileitis before it was named after him, striking alterations in the lymphatic vasculature in CD patients were implicated in disease pathology 5,6 .However, advances in the eld of lymphatic biology were needed before in-depth studies on the role of lymphatics were feasible.In the last several years, we and others have identi ed that B cell-rich tertiary lymphoid structures (TLS) arise adjacent to or within dramatically remodeled mesenteric lymphatic collecting vessels draining the ileum in humans with ileal CD and in the TNF ΔARE mouse model of ileitis 7,8,9,10 .In TNF ΔARE mice, deletion of the AU-rich response element (ARE) prolongs the half-life of tumor necrosis factor-a (TNF) mRNA 11 .The increased TNF, whose expression is initiated by signals from the microbiome, promotes ileitis. 11,12,13 Te TLS that form during disease progression obstruct the tra cking of immune cells and molecular cargo to draining lymph nodes and promote both leakage of lymph as well as back ow of lymphatic cargo towards adjacent regions of the intestine. 9We are left to wonder whether preventing TLS formation would reduce ileitis.Here, we set out to investigate the mechanisms of lymphatic-associated TLS formation in ileitis and to test their impact on disease progression.
B cells are especially enriched in TLS of humans and mice, 7,9 leading us to hypothesize that B cells may play a pivotal role in TLS development in ileitis.B cell production of lymphotoxin (LT) is vital for appropriate secondary lymphoid organ organization, 14,15 remodeling of the in amed lymph node 16,17,18 and TLS organization in cancer, 19 prompting our curiosity of whether B cells might critically promote formation of TLS that arise along the mesenteric lymphatic vessels in the ileitis of TNF ΔARE/+ mice.Furthermore, recent literature that calls for re-examining the role of B cells in IBD 20 provides a broader context for considering the role of B cells in ileitis.Secretory immunoglobulin [sIg], synthesized by gutresident plasma cells mainly as dimeric IgA, shifts towards IgG in IBD patients, who often have an aberrant anti-microbial antibody responses. 21,22,23,24,25,26,27 Whethr these alterations in antibody actively contribute to disease pathogenesis or are simply downstream of other in ammatory processes that truly drive disease is somewhat unclear.Additionally, B cells can act as antigen-presenting cells to T cells and produce pro-or anti-in ammatory cytokines to modulate disease severity. 28,29 previous studies, total B cell de ciency did not prevent in ammation in the ileum of TNF ΔARE/+ mice 30 but other features of disease, like TLS formation or systemic manifestations like weight loss were not reported.Thus, we revisited the role of B cells in ileitis, beginning with studies involving B cell de ciency and proceeding to inclusion of additional experimental strategies to re ne mechanistic insight.TNF ΔARE/+ mice without B cells developed ileitis, as previously reported, but they failed to develop mesenteric TLS and presented with greater loss of body weight and muscle mass compared to TNF ΔARE/+ mice with B cells.De ciency of polymeric Ig receptor (pIgR) in intestinal epithelial cells to reduce IgA in the intestinal lumen modestly exacerbated ileitis but did not in uence TLS formation or body weight.On the other hand, B cell-selective de ciency in LTβ or treatment of TNF ΔARE/+ mice with LTβR-fc to disrupt B cellproduced LTα 1 β 2 interfered with TLS development, but did not lead to the additional weight loss observed in the absence of B cells.However, TNF ΔARE/+ mice with a B cell-selective loss of LTα, which eliminates their expression of both membrane bound LTα 1 β 2 and secreted LTα 3  31, 32 , not only lacked mesenteric TLS but also presented with enhanced loss of lean mass.We go on to reveal that selective blockade of LTα 3 33 , a ligand that does not bind LTbR but instead binds to TNFRs I and II with high a nity and HVEM with lower a nity 32 , has an independent role in controlling body weight of mice with ileitis in a manner that suggests that it may compete with TNFa to attenuate some of the disease-promoting pathology that result from chronic overproduction of TNFa.Altogether, we identify multiple roles for B cells in ileitis that include a key role in production of locally protective sIg, a role in TLS formation that relies on B cellderived LTα 1 β 2 , and a highly unanticipated role for B cell-derived LTα 3 in defending body mass.

Characterization of B cells and other immune cells in in the mesentery of Crohn's patients
Since B cells in the in amed ileal-draining mesentery of CD patients always localized to TLS 7 , pro ling B cells from single cell suspensions of the mesentery would allow us to characterize B cell phenotypes within TLS.Using ow cytometry on such cell suspensions, we con rmed that B cells were signi cantly enriched in the ileal-draining mesentery of CD patients compared to nearby nonin amed mesentery from the same patients or control ileal-draining mesenteric tissue derived from organ donors or patients with gastrointestinal malignancy (Fig. 1a; gating strategy, Extended Data Fig. 1).In amed versus nonin amed regions of the small bowel from Crohn's patients was determined by a pathologist.Consistent with our past ndings, 7 Crohn's disease-affected mesentery yielded signi cantly more B cells than nondiseased controls, with the greatest numbers observed in the mesentery associated with in amed Crohn's small bowel (Fig. 1a).Focusing on paired samples of in amed and nonin amed regions, we con rmed that, within the same patient, CD19 + B cells in the in amed regions were elevated in 11 out of 13 CD patients compared to the non-in amed draining regions (Fig. 1b).
We employed single-cell RNA sequencing [scRNAseq] to characterize B cells and all CD45 + cells in the mesentery in 3 CD patients wherein paired nonin amed and in amed ileal draining mesentery was obtained.A uniform manifold approximation and projection [UMAP] plot of live, CD45 + cells recovered revealed 19 clusters corresponding to a range of different immune cells (Fig. 1c).In amed mesenteric samples included increased proportions of naïve B cells, in particular, as well as memory B cells (Fig. 1d, e; Extended Data Fig. 2a,b).The top 5 cluster-de ning genes are shown in Extended Data Fig. 2c.In addition to the proportional expansion of B cell subpopulations, mesentery from the in amed ileum of CD patients had an increase in the proportion of proliferating immune cells, compared to non-in amed ilealdraining mesentery from the same patients (Fig. 1d).While this cluster was not exclusively B cells, a meaningful proportion matched the signature of activation-induced proliferating B cells in TLS.Plasma cells were also detected in small numbers (Fig. 1d).Macrophages were the major producers of TNF among the hematopoietic cells in the mesentery, and they also expressed TNFR1 and TNFR2 most robustly.Memory CD4 + T cells and naïve and memory B cells had the highest expression of both LTα and LTβ transcripts (Fig. 1e).Numerically, within the in amed mesentery, naïve and memory B cells outnumbered memory T cells, underscoring B cells as an important source of LTa and LTb in the region of the mesentery where TLS form.These ndings reinforced evidence from the literature that B cells may orchestrate TLS generation or maturation via expression of LT proteins.Examination of BCR sequences and Ig isotype expression patterns revealed evidence of oligoclonal expansion of B cell clones in the mesentery in 2 of 3 of the CD patients (Fig. 1g) and these same patients also had an increase in the number of clonal TCRs in their mesentery and ileum (Extended Data Fig. 2d).In those CD patients with an expansion of clonal B cells, the B cells in their mesentery were dominated by production of IgG, not only compared to a control patient, but also to the upstream ileum of those same patients, which were predominately IgA + (Fig. 1g).The one CD patient without B or T cell clonal expansion instead had a large number of naïve B cells in the mesentery suggesting there is heterogeneity in clonal selection in B cell-rich TLSs of CD patients (Fig. 1g,h).
Absence of B cells in murine ileitis leads to loss of mesenteric TLS but minimal impact on ileal disease Wild type (WT) recipients given TNF ΔARE/+ bone marrow (BM) develop ileitis 30 .To investigate whether and how B cells affect ileitis, we crossed TNF ΔARE/+ mice rst to CD45.1 mice, so that both CD45.1 and CD45.2 mice in these congenic strains could be used to trace BM cells after transplant.We subsequently crossed the TNF ΔARE/+ mice to B cell-de cient µMT mice.We then used WT, TNF ΔARE/+ , and µMT-TNF ΔARE/+ mice as bone marrow (BM) donors for lethally irradiated, CD45.2 + congenic WT recipients (Fig. 2a).Four months after BM transplant, we quanti ed ileal in ammation in the recipients using three separate methods.First, we examined the distal 2 cm of the ileum by histology and calculated the extent of in ammation using a semi-quantitative scoring approach (Fig. 2b-c). 34Second, we quanti ed the numbers of neutrophils (Fig. 2d) and T cells (Fig. 2e) in the ileum using the ow cytometry gating strategy in Extended Data Fig. 2a.Finally, fecal lipocalin-2, a sensitive biomarker for murine intestinal in ammation, 35 was measured by ELISA (Fig. 2f).Ileal in ammation developed to a quantitatively similar extent, with a modest increase in disease when assessed blindly by histological scoring (Fig. 2b), in recipients of B cell-de cient µMT-TNF ΔARE/+ BM compared to recipients of TNF ΔARE/+ BM, consistent with previous ndings. 30In ltration of neutrophils was mostly restricted to the ileum, as gating of neutrophils in the blood (Extended Data Fig. 3b) revealed minimal to no neutrophilia, such that few TNF ΔARE/+ and µMT-TNF ΔARE/+ mice had increased neutrophils in the proximal portion of the small intestine or in the blood compared to WT mice (Extended Data Fig. 4a,b).Monocytes were increased in the blood of recipients that received TNF ΔARE/+ or µMT-TNF ΔARE/+ BM compared to those receiving WT BM (Extended Data Fig. 4c), with a shift in the proportion of monocyte subsets towards more Ly6C lo non-classical monocytes (Extended Data Fig. 4d).
Flow cytometry in the blood and ileum con rmed that B cells were missing in mice receiving µMT-TNF ΔARE/+ BM (Extended Data Fig. 4e,f).Tissue-resident plasma cells can be radioresistant. 36,37 ordingly, although IgA + plasma cells were reduced in µMT-TNF ΔARE/+ BM recipients (Extended Data Fig. 4g), they were not entirely absent, and those remaining were of recipient rather than donor BM origin (Extended Data Fig. 4h) We next examined the mesentery of mice receiving µMT-TNF ΔARE/+ BM, because mesenteric in ammation and TLS formation along out ow lymphatic vessels are features of progressive ileal in ammation in this model 9 .In mice receiving TNF ΔARE/+ BM, compared to those receiving WT BM (Fig. 3a), many B cell-rich TLS were observed (Fig. 3b).In contrast, although intense CD115 + myeloid cell-rich mesenteric in ammation was present in recipients of µMT-TNF ΔARE/+ BM, organized TLSs were absent (Fig. 3c).Flow cytometry con rmed marked accumulation of neutrophils in the mesentery of recipients of either TNF ΔARE/+ and µMT-TNF ΔARE/+ BM compared to those receiving WT BM (Fig. 3d), leading us to conclude that B cells were required for TLS-associated lymphatic remodeling in the mesentery but that the loss of mesenteric TLSs did not prevent accumulation of in ammatory cells in the mesentery.
When we assessed lymphatic out ow through the ileum-draining mesentery after injection of 1-1.5 µl of 2,000 kDa uorescein isothiocyanate [FITC]-dextran into the most distal Peyer's patch (Fig. 3e-h), the tracer owed into mesenteric collecting lymphatic vessels to the mesenteric lymph node [MLN] within ve minutes of injection in recipients bearing WT BM (Fig. 3e, h, Supplementary video 1).However, in 5 out of 6 mice harboring TNF ΔARE/+ BM (16-17 weeks after BMT), lymphatic ow to the MLN was not apparent for up to 60 minutes post-injection (Fig. 3f, h, Supplementary video 2), tting with previous ndings that TLSs located along LYVE-1 − mesenteric collecting lymphatic vessels obstructed lymph out ow from the ileum 9 .Furthermore, in the one mouse with TNF ΔARE/+ BM that had ow of the tracer to the MLN, the dextran tracer still leaked out of the lymphatics, indicating that all mice that received TNF ΔARE/+ BM possessed functional lymphatic de ciencies (Supplementary video 3).By comparison, lymphatic ow to the MLN was completely intact in all mice harboring µMT-TNF ΔARE/+ BM (Fig. 3g, h), even though the mice receiving µMT-TNF ΔARE/+ BM retained ileal in ammation (Fig. 2b-f, Supplementary video 4).These ndings con rm that B cells are critical to the development of blocked out ow of ileal lymph in TNF ΔARE/+ mice.
Absence of B cells is associated with more severe systemic weight loss and reduced muscle mass in murine ileitis Many CD patients are underweight, and even CD patients with a normal BMI may have loss of muscle mass, a condition called sarcopenia, which seems to arise from a combination of malnutrition, lower activity, and unfavorable protein synthesis or degradation in muscle 38,39 .TNF ΔARE/+ mice were often stunted pre-weaning before the onset of ileal in ammation.Our use of BM chimeras, where all recipients were WT mice that developed normally to adulthood before BM transplant, allowed for a rigorous approach to study the alterations in body composition in response to ileitis disconnected from problems in earlier development.WT mice that received WT BM increased body weight by 17% between 2 and 16 weeks after BM transplant (Fig. 4a) In contrast, mice receiving TNF ΔARE/+ BM initially lost weight before experiencing a moderate increase that plateaued 7 weeks later, such that these mice had a mean 6% increase from their initial weight (Fig. 4a).Recipients of µMT-TNF ΔARE/+ BM fell behind in weight compared to those getting TNF ΔARE/+ BM, exhibited a net loss at 16 weeks post-BM of -2% instead of the gains observed in the other groups (Fig. 4a; Extended Data Fig. 5a).However, B cell de ciency alone without in ammation did not lead to weight loss as µMT BM donors not carrying the TNF ΔARE/+ genotype had an increased their weight by 17%, mirroring WT mice (Fig. 5a).We used an MRI-based body composition analyzer to evaluate mice receiving WT, TNF ΔARE/+ , µMT-TNF ΔARE/+ BM at 14 weeks after transplant; fat mass did not signi cantly change between the groups, but lean mass declined in mice with µMT-TNF ΔARE/+ BM compared to both other groups (Fig. 4b).
We then set out to evaluate the whole-body metabolism of these mice 14 weeks after BM transplant by placing them in metabolic cages and observing them over 24 hours.We found that mice receiving µMT-TNF ΔARE/+ BM had a decrease in core body temperature compared to those receiving WT BM (Fig. 4c), making it unlikely that the observed weight difference in recipients of µMT-TNF ΔARE/+ BM is driven primarily by brown adipose tissue thermogenesis, as the core temperature of mice undergoing thermogenesis should be maintained or increased.Mice are nocturnal and therefore most active during the dark phase; however, mice that received TNF ΔARE/+ BM were signi cantly less active in the dark phase than mice that received WT BM (Fig. 4d).Energy expenditure (heat) was decreased in recipients of either TNF ΔARE/+ BM and µMT-TNF ΔARE/+ BM during both the light and dark phase compared to those receiving WT BM (Fig. 4e; Extended Data Fig. 5b), although the effect was more pronounced in mice that received µMT-TNF ΔARE/+ BM.Contrary to our expectations, food intake in the dark phase was increased in mice bearing B cell-de cient TNF ΔARE/+ BM compared to the other groups, indicating that food intake did not explain the increased weight de cit (Fig. 4f; Extended Data Fig. 5c).There was no difference between the groups in water intake (Extended Data Fig. 5d) or in whole-body hydration as assessed by body composition analyses (Extended Data Fig. 5e).The loss of lean mass evident in mice receiving µMT-TNF ΔARE/+ BM might re ect a decrease in muscle mass, bone mass, solid organ weight, or a combination of these.Indeed, the weight of the gastrocnemius muscle at 16-17 weeks post-BM transplant was decreased in mice bearing µMT-TNF ΔARE/+ BM compared to the other two groups (Fig. 4g).We also measured bone mineral density in the paws of mice 16-17 weeks after BM transplant but found no difference between the groups (Fig. 4h).TNF ΔARE/+ mice, in addition to ileitis, also develop rheumatoid arthritis-like disease in their joints 11 , and we observed evidence of arthritis in mice with TNF ΔARE/+ BM.
When B cells were not present, arthritis worsened; bone porosity was decreased (Extended Data Fig. 5f), while bone surface to bone volume, indicative of bone pitting, was increased (Extended Data Fig. 5g).
These data indicate that systemic features of ileitis, including weight loss and reduced muscle mass, were exacerbated in conditions wherein B cells were lacking.Because these conditions correlated with a loss of TLSs that block lymphatic out ow from the ileum, we began to consider the possibility that the TLS blockade might safeguard against systemic dissemination of in ammatory mediators that originated from the intestine, a concept consistent with recent work that linked valve patency to systemic in ammatory status 40 .However, a rigorous consideration of this concept required that we dissect the role of B cells in ileitis in conditions when B cells were not completely absent.
Polymeric Ig receptor de ciency in the context of TNF ΔARE/+ ileitis points to a protective role for fecal sIg in local intestinal in ammation but no impact on weight loss We began our strategy to dissect distinct roles of B cells by rst seeking to understand the role of secretory Ig (sIg), which would include IgA transported to the intestinal lumen, in TNF ΔARE driven ileitis.As pIgR is expressed on epithelial cells and controls basolateral translocation of polymeric Ig into the lumen of the intestine, we lethally irradiated pIgR −/− mice or pIgR +/+ littermates, reconstituted both genotypes with TNF ΔARE/+ bone marrow, and separately housed them after the transplant.The pIgR −/− mice harboring TNF ΔARE/+ BM had much less IgA in the stool compared to pIgR +/+ TNF ΔARE/+ BM recipients (Fig. 5a), although there was still some detectable IgA present, perhaps because, especially in the context of in ammation, immunoglobulin can reach the lumen of the intestine via pIgR independent mechanisms.No difference in fecal lipocalin-2 was detected between the two groups (Fig. 5b).However, in the ileum lamina propria, pIgR −/− TNF ΔARE/+ BM recipients compared to pIgR +/+ TNF ΔARE/+ BM recipients had an increase in in ammatory in ltrating cells with more neutrophils (Fig. 5c) and in T cells (Fig. 5d).These results suggest that sIg is locally protective in the gastrointestinal tract.By comparison, there was no difference in absolute body weight (Fig. 5e) or in the percent weight change at week 16 post-BM.The number of monocytes in the blood, a surrogate for more systemic in ammation, was also similar between the groups (Fig. 5f).The number of TLSs in the most distal mesenteric fat branch-the location where the blood and lymphatic vessels that drain most of the ileum-was also not different between pIgR +/+ and pIgR −/− TNF ΔARE/+ BM recipients (Fig. 5g).Together, these results indicate that, as anticipated, a protective role of B cells in ileitis is to support plasma cell secretion of sIg.This mechanism is locally protective against in ammation at the intestinal wall, but B cell modulation of body weight, mesenteric TLS organization, and lymphangiogenesis are independent of sIg.induced ileitis, but for these mixtures, all B cells would lack either LTa or LTb.By contrast, all hematopoietic cells except B cells will primarily (approximately 90%) arise from the µMT-TNF ΔARE/+ BM, so the LT de ciency will be B cell speci c.LT can signal either as a LTα 3 homotrimer or a LTα 1 β 2 heterotrimer; therefore, LTα −/− /µMT-TNF ΔARE/+ BM recipients will lack both B cell production of LTα 3 and LTα 1 β 2 , and LTβ −/− /µMT-TNF ΔARE/+ BM recipients will lose B cell production of LTα 1 β 2 but will still retain B cell production of LTα 3 .Accordingly, we rst examined LTβ −/− /µMT-TNF ΔARE/+ BM recipients as they lack only one form of LT instead of both.
At 30 to 32 weeks post-BM transplant, more than a dozen TLS arose in the most distal ileal draining branch of the mesentery of mice that received WT/µMT-TNF ΔARE/+ BM (Fig. 6a).Consistent with the morphology of ileitis-associated TLS 9 , these TLS had with lymphatic capillaries that had expanded to wrap around them (Fig. 6b).However, mice that received LTβ −/− /µMT-TNF ΔARE/+ BM transplants, so that B cells selectively lacked LTb expression, had few TLS present in the distal ileal draining mesentery (Fig. 6a), although some mice did develop occasional undeveloped lymphoid clusters with no evidence of associated lymphomagenesis (Fig. 6c).Both groups of mice had similar neutrophil (Fig. 6d) and T cell (Fig. 6e) in ltration into the ileum, suggesting that loss of LTb in B cells or consequent loss of TLS did not substantially impact the in ammatory component of ileitis.In the blood, the number of neutrophils (Fig. 6f) and T cells (Fig. 6g) circulating in mice that received LTβ −/− /µMT-TNF ΔARE/+ BM signi cantly increased compared to mice given WT/µMT-TNF ΔARE/+ BM.Since LTα 1 β 2 exclusively signals through LTβR, we further tested these nding in the mixed chimera BM transplant setting by dosing with soluble LTβR fused to an Fc domain (LTβR-Fc) to speci cally target LTα 1 β 2 .As observed in mice where B cells were engineered to genetically lack LTβ, there was a reduction in TLS number in the ileal draining mesentery in mice given LTβR-fc compared to isotype (Fig. 6h), but no difference in neutrophilic in ltration into the ileum(Fig.6i).These data fail to support the hypothesis that mesenteric TLS robustly exacerbate local ileal in ammation, though the loss of TLS may be countered by other ways in which LTb de ciency might impact outcomes.
Because TNF and LT are closely related cytokines, we tested whether B cell production of TNF was important for TLS organization in TNF ΔARE ileitis.We generated CD19 cre/+ Td tomato/+ TNF +/+ and CD19 cre/+ Td tomato/+ TNF / mice to generate mixed bone marrow chimeras with µMT-TNF ΔARE/+ BM wherein B cells possessed the gene to express TNF or not.We checked the percentage of B cells in the blood of these BM recipients 10 weeks post-BM chimera and found essentially all of the CD19 + antibodystained cells in the blood were CD45.2 + and therefore came from the CD19 cre/+ Td tomato/+ or CD19 cre/+ Td tomato/+ TNF / donor (Extended Data Fig. 5a, b).Furthermore, we observed that an average of 83.7% of the cells stained by an anti-CD19 antibody expressed Td tomato in the CD19 cre/+ Td tomato/+ /µMT-TNF ΔARE/+ mice, but that there was a signi cant increase in the number of Td tomato + cells in the CD19 cre/+ Td tomato/+ TNF / /µMT-TNF ΔARE/+ BM recipients to 96.7% (Extended Data Fig. 6c).This persisted through the end of the experiment where there was a signi cant increase in the Td tomato + CD19 + B cells (Extended Data Fig. 6d) and IgA + plasma cells (Extended Data Fig. 6e) in mice receiving CD19 cre/+ Td tomato/+ TNF / /µMT-TNF ΔARE/+ BM compared to those receiving CD19 cre/+ Td tomato/+ /µMT-TNF ΔARE/+ BM.It seems that, even though the B cells in these mice do not carry a deletion in the AU-rich regulatory element of TNF mRNA and, therefore, are not producing more TNF protein themselves, that there is a cell survival bene t for B cells that cannot make TNF in this context.Additionally, both mice where the B cells could (Fig. 6j) or could not produce TNF (Fig. 6k) still developed well-organized TLS with distinct B cell follicles along the mesenteric lymphatics.Therefore, B cell production of TNF appears dispensable for TLS.
LTα 3 produced by B cells protects against excessive weight loss in the context of TNF ΔARE/+ ileitis We hypothesized that B cell-speci c LTα de ciency would phenocopy B cell-speci c LTβ de ciency.Like mice given mixed LTβ −/− /µMT-TNF ΔARE/+ BM, mice that received mixed LTα −/− /µMT-TNF ΔARE/+ BM did not generate robust TLS in the ileal-draining mesentery when compared to mice given mixed WT/µMT-TNF ΔARE/+ BM (Fig. 7a).However, we observed more weight loss in mice given mixed LTα −/− /µMT-TNF ΔARE/+ BM, where there is a B cell-speci c LTα de ciency, compared to mixed bone marrow chimeras where B cells were WT or LTb-de cient (Fig. 7b; Extended Data Fig. 7).This nding was reminiscent of the more severe reduction in body weight we earlier observed in B cell-de cient TNF ΔARE/+ BM compared to TNF ΔARE/+ BM (Fig. 4a).To further evaluate whether the differences in weight loss between B cells de cient in LTa versus LTb could be attributed to a distinct role for LTa 3 and LTα 1 β 2 , we turned to a Fcmutant anti-LTα 3 antibody that binds and neutralizes LTα 3 but does not deplete cells producing LTα 3 33 .
While this antibody neutralizes LTα 3 , it has limited ability to bind to the LTα 1 β 2 heterotrimer in vitro, and does not induce splenic architecture abnormalities in vivo, unlike dosing with LTβR-Fc 33 .The impact of the anti-LTa3 mAb on body weight was not observed in WT mice.While TNF ΔARE/+ BM recipients given anti-LTα 3 mAb rapidly lost over 5% of their initial body weight after only 2 days, the weight of WT mice receiving the reagent remained steady (Fig. 7c).This result suggested that neutralizing LTa 3 impacts weight loss only in some contexts, such as a context with TNF signaling was ongoing.To test this idea further, we studied TNF ΔARE/+ BM transplanted mice treated with either (i) isotype mAb only, (ii) anti-TNF mAb, (iii) anti-LTα 3 mAb, or (iv) combined anti-TNF and anti-LTα 3 mAbs.TNF ΔARE/+ BM recipients given anti-LTα 3 again rapidly lost over 5% of their initial body weight after only 2 days, while isotype-treated and anti-TNF treated TNF ΔARE/+ BM recipients maintained their weight, and there was a partial rescue of the anti-LTα 3 weight loss in mice given both anti-TNF and anti-LTα 3 (Fig. 7d), suggesting that the biological impact of anti-LTa3 was directly connected to the biological availability of TNF.When the 4 groups of mice were analyzed by MRI, the anti-LTα 3 treated mice had a signi cant reduction in the fat mass (Fig. 7e).Additionally, if dosing was extended to 5 days, there was a trend towards a decrease in gastrocnemius weight in TNF ΔARE/+ BM-bearing mice treated with anti-LTα 3 (Fig. 7f).No difference was observed in hydration ratio (Fig. 7g), indicating that the weight loss was not primarily driven by dehydration.Changes in activity (Fig. 7h) or food intake (Fig. 7i) also did not explain the weight loss.Loss of LTα 3 in B cells or neutralization with anti-LTα 3 did not prevent absorption, as bomb calorimetric analysis in these conditions did not yield higher caloric retention in feces (Extended Data Fig. 8a,b) and challenge with an oral bolus of triglyceride uncovered no alterations in absorption of fat under steady state (Extended Data Fig. 8c) or in the context of ileal in ammation (Extended Data Fig. 8,d).These data suggest that LTα 3, including LTα 3 produced by B cells protects against in ammation-mediated weight loss in a TNF-driven model of ileitis.

DISCUSSION
Here, we set out to examine the role of B cells in ileitis, in part motivated by the hypothesis that B cells would have a central role in driving the formation of TLSs that obstruct lymph out ow from the ileum.The concept that B cells might have a central role in the genesis or maturation of TLS through expression of LTα1β1 14,15,16,17,18,19 led us to develop experimental scenarios wherein B cells development was blocked entirely or wherein LTb or LTa were separately and selectively deleted from B cells in TNF DARE mice 11 that progress to chronic ileitis due to a mutation in TNF mRNA that increases its stability and thus its capacity to promote in ammatory disease.Our analyses included an evaluation of disease severity based on in ammatory criteria, including extent of in ltrated immune cells like neutrophils or lymphocytes, transmural distribution of the in ltrated cells, shedding of lipocalin-2 into the feces, and overall body weight and lean/fat composition.We did not yet consider other important features of the disease, like Paneth cell phenotype or brosis, important additional clinical features of human Crohn's disease.The most striking outcome was that the loss of B cells was associated with an inability to maintain body weight and lean mass as disease progressed, another feature re ected in many humans with Crohn's disease 3,4 .By contrast, while loss of gut luminal IgA through epithelial pIgR de ciency led to heightened immune cell in ltration into the ileum, weight loss did not ensue.It became clear from our data that weight loss and local in ammation did not always track in parallel, an observation that was previously observed in colitis models 42 .
We identi ed that B cells, in fact, govern the development of TLS that arise at lymphatic valves and inhibit lymphatic out ow in ileitis as it becomes chronic.Deletion of LTα or LTβ in B cells, or lack of B cells altogether, was su cient to halt development of robust TLS, although some mice still had disorganized small foci in regions near where TLS form.However, future studies will be needed to further address the speci c impact of TLS on Crohn's-like ileitis, as the parameters of pathology that we studied, from the local magnitude of in ammatory in ltration to systemic changes in mass, were not invariably associated with the presence of TLS.In particular, the impact of B cell de ciency of LTb, which blocked TLS formation, was not associated with changes in local in ammation or body weight, though it was associated with a striking increase in the number of immune cells in the circulation.
Unexpectedly, however, deletion of LTβ versus LTα in B cells gave rise to divergent results, and the latter more closely re ected outcomes with total B cells de ciency than any other manipulation we made, including studies in pIgR −/− recipient mice that had diminished gut lumen IgA.A previous study revealed that de ciency in LTβ versus LTα in RORgt + cells gave rise to different effects on serum IgA, with LTa but not LTβ de ciency in RORgt + cells being required for IgA production 43 .Further investigation, in this case, pointed to a role for LTα1β1 in induction of T-independent IgA and LTα3 in T-dependent IgA 43 , an outcome where both products generated from the Lta gene were relevant but wherein the presence of LTa 3 partially masked the impact of losing LTα1β1.Indeed, the ability to separate the biology of LTα3 from LTα1β2 is challenging.In our case, like that of Kruglov et al. 43 , we were led to consider a possible role for LTα3 in maintenance of body weight in the face of in ammation by observing discrepancies between phenotypes arising from the loss of LTα versus LTβ.
Past studies have linked LTα3 to in ammatory responses through its binding of TNFR1 and TNFR2, positioning it to possibly ac redundantly with TNF signaling.For example, it has been argued that LTα3 serves as a critical proin ammatory mediator for intestinal in ammation in TNF-de cient mice when epithelial cells lack expression of IBD-associated genes A20 and Abin-1 44 .To support the idea that LTα3 drove TNF-independent epithelial cell death, the authors used a mAb that selectively targets LTα3 33 .
However, it appears that the version of the mAb they used depletes LTa + cells 33 , so while it also neutralizes LTα3, the use of the depleting form of themAb would have a broader effect beyond neutralizing LTα3, as it would eliminate activated T cells 33 , for instance.We utilized a version of the same mAb with a mutated Fc domain so that it does not deplete LTα3 + cells, but simply neutralizes secreted LTα3 33 .Our ndings were in marked contrast to those of Rusu et al. 44 , since in our studies, blocking LTα 3 led to rapid weight loss in TNF DARE mice.Besides employing different versions of the same reagent with different potential consequences related to cell depletion, it is also possible that the difference in impact between our study versus that of Rusu et al. 44 could be related to genetic absence of TNF as a ligand for TNFRs in the former model and the role of TNF in driving TNFR-dependent in ammation in our model.That is, it is possible that LTα 3 engages TNFRs on target cells to promote a less robust in ammatory response than TNF itself.LTα 3 was, early on, viewed as a partial agonist of TNFRs 45 and consistently showed more attenuated in ammatory effects in vitro than TNF on the same target cells 46,47,48,49,50,51 .Our in vivo data raise the possibility, consistent with this historical work, that LTα 3 propagates in ammation when TNF is absent but serves as a partial brake that dampens TNF/TNFR signaling when both ligands are present.More work is needed to reveal the mechanistic underpinnings of the possible interaction between TNF and LTα 3 .
These ndings could contribute to an understanding of why therapeutics like etanercept that neutralize both LTα and TNF failed in IBD, whereas those selectively targeting TNF alone have shown signi cant e cacy 52 .One of the avenues of future investigation will be a search to better understand how LTa is linked to body weight.The link we uncover in our studies, which we could not explain at the level of caloric intake or metabolic activity, is consistent with several other studies that have found possible regulation of BMI or fat mass to polymorphisms or expression of LTα 53,54,55 .Our results put a spotlight on LTα 3 over LTα1β1 in that regard.
Overall, in this study, we took a broad look at the role of B cells in ileitis and uncovered disparate areas where they in uence the disease in positive or adverse ways, from the protective role conferred locally by IgA production and secretion into the lumen, to their role in TLS formation, and, nally, in their unexpected role for B cell production of LTα 3 in protection against a systemic manifestation of disease.These ndings reinforce the concept that therapeutics to eliminate all B cells non-selectively in IBD patients may be a double-edged sword, where B cell depletion ameliorates some in ammatory aspects of disease progression but worsens others, such as weight less.Furthermore, our ndings suggest that targeting speci c functions of B cells may enable the development of improved therapeutics to treat IBD.These ndings should also provide a solid basis for continued research to better understand the mechanistic basis of pathogenesis in ileitis, and the biological role of TLS in chronic in ammation through the manipulation of B cells.
Secretory antibody partially protects against ileal in ammation in TNF ΔARE/+ BM recipients, but this protection is limited only to the ileum.

LTα 1 β
2 produced by B cells is required for TLS development As B cell production of LT has been implicated in TLS organization and lymphangiogenesis under in ammatory conditions16,17, 18  and B cells producing LTα and LTβ are present within the in amed mesentery of CD patients (Fig.1e) we sought to test if B cell synthesis of lymphotoxins in the TNF ΔARE/+ model governed TLS formation.Since the genes for TNFa, LTα, and LTβ are very close to each other within the MHC loci, crossing LTα KO or LTβ KO mice to TNF ΔARE mice was not tractable, as the likelihood of a recombination event would be extremely low41 .Thus, we instead used mixed BM chimeras where 90% of the donor bone marrow was provided by µMT-TNF ΔARE/+ donor marrow and 10% of the bone marrow was supplied by WT, LTα −/− , or LTβ −/− donors.Recipients with WT/µMT-TNF ΔARE/+ mixed BM should still develop TNF ΔARE/+ -induced ileitis, but the WT BM supplies B cells that would be exclusively WT.The resulting mice should closely resemble TNF ΔARE/+ BM recipients.Recipients of 10%/90% mixtures of LTα −/− /µMT-TNF ΔARE/+ and LTβ −/− /µMT-TNF ΔARE/+ BM should also be impacted by TNFΔARE/+

Figure 3 B
Figure 3

Figure 4 B
Figure 4

Figure 6 B
Figure 6