Chitinase-3-like 1 regulates TH2 cells, TFH cells and IgE responses to helminth infection

Introduction Data from patient cohorts and mouse models of atopic dermatitis, food allergy and asthma strongly support a role for chitinase-3-like-1 protein (CHI3L1) in allergic disease. Methods To address whether Chi3l1 also contributes to TH2 responses following nematode infection, we infected Chi3l1 -/- mice with Heligmosomoides polygyrus (Hp) and analyzed T cell responses. Results As anticipated, we observed impaired TH2 responses in Hp-infected Chi3l1 -/- mice. However, we also found that T cell intrinsic expression of Chi3l1 was required for ICOS upregulation following activation of naïve CD4 T cells and was necessary for the development of the IL-4+ TFH subset, which supports germinal center B cell reactions and IgE responses. We also observed roles for Chi3l1 in TFH, germinal center B cell, and IgE responses to alum-adjuvanted vaccination. While Chi3l1 was critical for IgE humoral responses it was not required for vaccine or infection-induced IgG1 responses. Discussion These results suggest that Chi3l1 modulates IgE responses, which are known to be highly dependent on IL-4-producing TFH cells.


Introduction
Allergens are broadly defined as non-pathogenic proteins that induce specific IgE in sensitized individuals (1). However, sensitized individuals contact allergenic proteins in the context of a complex molecular milieu derived from living organisms. For example, feces of house dust mite (HDM) and cockroach contain major allergens complexed to lipid-binding or chitin-binding proteins, and protease activity is a well-known feature of HDM allergens, mold-derived allergens and the industrial sensitizer papain (2). Thus, individuals are typically sensitized to allergens in the context of inflammatory stimuli that provide adjuvant-like effects. Chitins, which are polysaccharides found within the exoskeleton of arthropods, crustaceans and helminths, and the cell wall of bacteria and fungi (3,4), are thought to promote T H 2 allergic responses (5,6). Chitinases, which degrade chitins, and chitinase-like binding proteins (CLP), which can bind and sequester chitin but cannot degrade chitin (7), are upregulated in the lungs of asthma patients and allergen-exposed mice (7)(8)(9). Chitinase deficient mice (Chit1 -/or Chia -/-) manifest enhanced airway inflammation in a HDM allergic airway disease model (10,11), suggesting that chitinases function to attenuate allergic responses. In contrast, mice deficient in the CLP Chitinase-3-like 1 (Chi3l1 -/-) make impaired allergic airway responses following HDM exposure (8,9) and mice deficient in Chitinase-like protein 3 (Ym1) develop reduced lung eosinophilia and mucus production in an ovalbumin model of allergic airway disease (12). This suggests that CLPs may not prevent allergic responses by sequestering chitin but rather function to facilitate allergic responses.
CHI3L1 has been studied extensively in the setting of human allergic disease. Levels of YKL-40 (the protein product of the human CHI3L1 gene) and expression of CHI3L1 are increased in the lungs and serum of some asthma patient cohorts (13,14). CHI3L1 SNPs are reported to confer risk of asthma development and airway remodeling (14)(15)(16)(17) as well as increased serum IgE and atopy (18) in patient cohorts. Moreover, increased expression of YKL-40 and CHI3L1 have been linked to pathogenesis of allergic rhinitis (19,20), atopic dermatitis (21)(22)(23) and food allergy (24). In agreement with human patient studies, experiments using Chi3l1 -/mice reveal that Chi3l1 regulates type 2 cytokines and IgE levels in mouse models of asthma, atopic dermatitis and food allergy (8,(23)(24)(25)(26). Thus, both mouse and human data support a role for YKL-40/ Chi3l1 in promoting atopy and allergic disease.
Parasitic infections are recognized as the likely impetus for the evolution of T H 2 immunity in mammals, and studies of parasitic infections demonstrate that mediators of protective immunity to parasitic infections often contribute to pathologic responses to allergens (27)(28)(29)(30). Chi3l1 -/mice, which make attenuated allergic responses (8,(23)(24)(25)(26), did not manifest impaired clearance of the nematode Nippostrongylus brasiliensis (Nb) during early neutrophil-mediated stages of infection, in spite of reduced lung IL-17A levels (31). Since the Nb study did not measure T H 2 or IgE responses following Nb infection, we used the helminth Heligmosomoides polygyrus bakerii (Hp) to address whether Chi3l1 is required for type 2 immunity in the setting of a strong T H 2-driven enteric helminth infection. Hp primes a T H 2 response (29,32) and elicits IL-4 producing T FH cells (33)(34)(35)(36)) that initiate polyclonal IgE production by B cells (35,37). Here, we show that, consistent with the prior allergy studies, Chi3l1 regulates the IL-4 + IL-13 + T H 2 response to Hp infection. In addition, we observed that Chi3l1 controls the size of the T FH compartment following both Hp infection and protein immunization. We demonstrate that Chi3l1 -/-T FH cells express significantly lower levels of Inducible T cell costimulator (ICOS)a key regulator of T FH development and maintenance in the B cell follicle (38,39). RNA-seq analysis of the Chi3l1 -/-T FH cells revealed that these cells express the canonical T FH transcriptional program but failed to acquire the normal transcriptional signature of IL-4 producing T FH cells (40). These deficits were associated with decreased IL-4 production by the remaining Chi3l1 -/-T FH cells, decreased germinal center B cell responses and significantly decreased IgE + ASCs and serum IgE following Hp infection or alum-adjuvanted vaccination. Thus, Chi3l1 plays a nonredundant, unexpected and critical role in promoting IL-4 + T FH cells that are known to amplify IgE antibody responses to both pathogens and allergens.

Mice
Animals were bred and maintained in the UAB animal facilities. All procedures were approved by the UAB IACUC and were conducted in accordance with the principles outlined by the National Research Council. BALB/cByJ mice (WT) and BALB/c CD45.1 + congenics (CByJ.SJL(B6)-Ptprc a /J) were purchased from The Jackson Laboratory. BALB/c BRP-39 (Chi3l1 -/-) mice (8) were provided by Dr. Allison Humbles (MedImmune). Bone marrow (BM) chimeras were generated by irradiating recipients with 850 Rads from a high-energy X-ray source (split dose 5 hours apart), and then reconstituting the recipients with 5x10 6 total BM cells delivered i.v. To generate 50:50 chimeras we reconstituted irradiated recipients (BALB/cByJ) with 2.5x10 6 BALB/c CD45.1 + (CByJ.SJL(B6)-Ptprc a /J) + 2.5x10 6 Chi3l1 -/-BM cells. Chimeras were used in experiments 8-12 weeks post-reconstitution. Both male and female mice were used and animals were matched for age and sex within an experiment. No gender-specific differences were observed.

Cell sorting
14 days after oral gavage with 200 L3 Hp (D14), infected BALB/ c ByJ and Chi3l1 -/mice were sacrificed and msLN were collected and pooled. Single cell suspensions of msLN cells were incubated with FcBlock (BD), enriched with CD4 microbeads (Miltenyi), and s t a i n e d w i t h fl u o r o c h r o m e -c o n j u g a t e d a n t i b o d i e s . CD4 + CXCR5 + PD-1 hi Lin neg cells were sorted (BD Aria, UAB Flow Cytometry Core), pelleted, lysed in TRIzol (ThermoFisher), and stored at -80°C until used.

Production of recombinant influenza NS1 antigen and NS1 tetramers
The influenza NS1 gene, modified to contain a 3' in frame addition of the BirA enzymatic biotinylation site and the 6X-His purification tag (GeneArt), was mutated at R38A and K41A (to prevent aggregation of NS1 at high concentrations (42)), then cloned into the pTRC-His2c expression vector (Invitrogen) and expressed in the BirA-enzyme containing E. coli strain CVB101 (Avidity). Biotinylated recombinant NS1 was purified by FPLC and then tetramerized to fluorochrome-conjugated streptavidin (Prozyme). To detect NS1-specific B cells, PE-labeled NS1 tetramers (1:100) were incubated with cells for 30 min at 4°C. NP(15)-OVA or NS1 immunization 50 mg biotinylated NS1 protein or 50 mg (4-hydroxy-3nitrophenyl)-acetyl(15)-OVA (NP-OVA) was diluted to 1 mg/mL in PBS and adsorbed to 100 mg alhydrogel alum (InVivoGen) in a total of 200 mL per mouse for 30 min at room temperature, then injected i.p. on day 1 and analyzed on day 12 (D12) (splenic T and B cell responses) or day 14 (D14) (serum antibody).

IgG1 detection
Serum from uninfected mice was analyzed for total IgG1 using a mouse clonotyping kit and standards (Southern Biotech) according to manufacturer's recommendations. Hp-specific IgG1 was detected as previously described (43) using plates coated with Hp extract and detected using rat anti-mouse IgG1-HRP antibody (Southern Biotech). NP-OVA specific IgG1 was detected as previously described (44) using plates coated with NP(5)-BSA and detected using goat anti-mouse IgG1-HRP (Southern Biotech).

RNA-seq analysis
RNA was isolated (RNeasy micro column (Qiagen)) from T FH cells sorted into TRIzol as previously described (45) and then enriched with Oligo(dT) beads. Sequencing libraries (NEBNext Ultra II RNA Library Prep Kit for Illumina (NEB, Ipswitch, MA, USA)) were prepared by GeneWiz LLC (South Plainfield, NJ, USA) and sequenced with a 2x150bp Paired End configuration on an Illumina HiSeq 4000. Image analysis and base calling were conducted using Hiseq Control Software (HCS). Raw sequence data (.bcl files) generated from Illumina HiSeq was converted into fastq files and de-multiplexed using Illumina's bcl2fastq 2.17 software. One mismatch was allowed for index sequence identification. Adapter content was removed from fastq files using Skewer 0.2.2 (46) and data aligned with STAR 2.5.3a (47) to the ENSEMBLE BALB/c/j GCA_001632525 reference mouse genome and transcriptome. PCR duplicate reads were flagged using PICARD MarkDupilicates 1.127 (http://broadinstitute.github.io/ picard), gene counts for each sample were computed using GenomicRanges 1.34.0 (48), and reads per kilobase per million (RPKM) normalized in R 3.5.2. Genes with at least 3 reads per million (RPM) in all samples of either the WT and/or Chi3l1 -/-T FH groups were considered detected. After confirming that the first 8 exons of the Chi3l1 gene, which were deleted in the Chi3l1 -/mouse strain, were not expressed in the Chi3l1 -/-T FH samples, we removed the Chi3l1 gene (also called Chil1) from the RNA-seq analysis as exons 8-10 of the Chi3l1 gene, which are directly downstream from the inserted promoter + neomycin cassette (8), were expressed, presumably due to the insertion of the neomycin deletion cassette.
9853 expressed genes (defined as genes with at least 3 reads per million (RPM) in all samples of either the WT and/or Chi3l1 -/-T FH ) were identified. Differential expression between groups was analyzed using DESeq2 1.26.0 (49) and 1465 genes were identified as significantly different between WT and Chi3l1 -/-T FH cells using an FDR cutoff of q<0.05. See Table S1 for gene expression levels and detailed methods. 193 genes with average reads per kilobase of transcript per million (RPKM) > 1 in either group and meeting a criteria of FDR p < 0.05 and threshold of ±0.3785 log 2 fold-change (FC) (1.3-fold) were submitted to Ingenuity Pathway Analysis (IPA, QIAGEN Digital Insights), of which 182 were used by IPA to identify significantly enriched pathways. For gene set enrichment analysis (GSEA), expressed genes from WT and Chi31l -/-T FH cells were ranked by multiplying the -log 10 of the P-value from DESeq2 (49) by the sign of the fold change and then used as input in the GSEA (50) PreRanked analysis program (http:// software.broadinstitute.org/gsea/index.jsp). RNA-seq data sets were deposited in the NCBI Gene Expression Omnibus (GSE203113). RNA-seq processing code is available at https:// github.com/cdschar/Curtiss_Tfh_RNAseq/.

Statistical analysis
Statistical details of all experiments including tests used, n, and number of experimental repeats are provided in figure legends. FlowJo (version 9, Tree Star) was used for flow cytometric analyses. Prism Graphpad (version 9) was used for statistical analyses of flow cytometry experiments. Statistical analysis of RNA-seq experiments is summarized within the text of the RNA-seq experimental design in Table S1.

Results
Chi3l1 regulates IL-4 production by restimulated LN CD4 + T effectors from Hpinfected mice Murine allergic airway disease models revealed that type 2 cytokine responses are blunted in Chi3l1 -/animals (8) and that Chi3l1 -/-CD4 T cells primed in vitro with antibodies to CD3 and CD28 in the presence of T H 2-polarizing conditions are impaired in IL-4 production (51). To test whether Chi3l1 regulates T H 2 responses to helminth infection, we measured mesenteric LN (msLN) CD4 T cell responses in wild-type (WT) BALB/c and Chi3l1 -/-BALB/c mice that were orally infected with Hp. Consistent with previous studies (51), uninfected Chi3l1 -/mice did not exhibit changes in the numbers of total msLN cells or CD19 B cells ( Figures S1A-C). However, the msLN total CD4 and activated CD44 hi CD62L lo CD4 T cell responses were attenuated in Hp-infected Chi3l1 -/mice (Figures S1D-F). To determine whether the remaining Chi3l1 -/-CD4 T cells were competent to produce T H 2 cytokines, we analyzed intracellular cytokine levels in anti-CD3 restimulated msLN CD44 hi CD4 T cells from D8 Hpinfected mice. We observed that the percentage and number of
To assess whether the reduction in ICOS expression in Chi3l1 -/-T FH cells was likely to be cell intrinsic, we activated splenic CD4 T cells from uninfected WT and Chi3l1 -/mice under early T FH differentiation conditions (41) with platebound anti-CD3 + anti-CD28 and IL-6 plus IL-2 blocking antibodies. We observed decreased ICOS upregulation in the in vitro activated Chi3l1 -/-CD4 T cells ( Figure 1I). Together, these data suggest that Chi3l1 does not control T cell commitment to the T FH lineage but might be important for the expansion, maintenance, or function of this T cell subset.

ICOS expression by Foxp3 + CD25 + T REG does not require Chi3l1
Since ICOS levels were decreased in the Chi3l1 -/-T FH cells, we asked whether ICOS expression was altered in Foxp3 + CD25 + CD4 + cells, since ICOS function is critical in both T FH and T REG development during Hp infection (59). We examined mice on D25 post Hp-infectiona timepoint when the CD4 T cell response is contracting and T REG responses are induced to facilitate establishment of a chronic infection (29,60,61). The numbers of CD25 + Foxp3 + T REG (Figures S1G-H) were similar between the groups. Moreover, we did not observe reduced ICOS expression by Chi3l1 -/-CD25 + Foxp3 + T REG s (Figures S1I-J). Thus, Chi3l1 modulates ICOS expression by Hp-infection elicited T FH cells but not T REG cells.

Hematopoietic cell expression of Chi3l1 regulates T H 2 and T FH responses to Hp
Transgene-directed Chi3l1 expression by epithelial cells is reported to restore lung T H 2 cytokine levels in allergen-exposed Chi3l1 -/mice (8). To assess whether the attenuated T H 2 and T FH responses to Hp were due to Chi3l1 expression by radiation resistant cells, like epithelial cells, or to Chi3l1 expression by radiation sensitive cells, like bone marrow (BM)-derived immune cells, we analyzed Hp-elicited T cell responses in BALB/c and Chi3l1 -/mice that were lethally irradiated and reconstituted with either BALB/c BM or Chi3l1 -/-BM. msLN cells of BM chimeras that were competent to express Chi3l1 in all cell types (WT donor/WT recipient), lacked Chi3l1 in all cell types (KO donor/KO recipient), lacked expression of Chi3l1 specifically in the hematopoietic compartment (KO donor/WT recipient) or lacked expression of Chi3l1 in the radiation resistant compartment (WT donor/KO recipient) were compared 8 days after Hp infection. We found no statistically significant differences in total numbers of msLN cells, B cells, or CD4 T cells in any of the groups ( Figures  S3A-C). However, and consistent with our analysis of non-chimeric Chi3l1 -/mice, we observed a significant reduction in the numbers of total activated CD62L lo CD44 hi CD4 cells ( Figure S3D) in the KO donor/KO recipient mice compared to the WT donor/WT recipient mice. Similarly, the day 8 Hp-infected KO donor/KO recipient mice had decreased frequencies and numbers of CXCR5 + PD-1 + T FH cells ( Figures S3E-G), which expressed lower levels of ICOS ( Figure  S3H). Moreover, the T cells from the KO donor/KO recipients produced significantly less IL-4 and IL-13 following anti-CD3 restimulation when compared to the T cells from the WT donor/ WT recipients ( Figures S3I-K). When we compared the data from these chimeras to the other two groups of chimeras, we found no differences in any of the T cell responses between the KO donor/WT recipients and the KO donor/KO recipients ( Figures S3D-K). These results indicated that Chi3l1 expression by BM-derived cells was necessary for optimal CD4 T cell responses to Hp infection. Consistent with this idea, the CD4 T cell responses were largely, although not completely, rescued ( Figures S3D-K) in Chi3l1 -/mice reconstituted with WT BM (WT donor/KO recipient). Thus, while we could not completely rule out a role for Chi3l1 expressing nonhematopoietic cells in regulating CD4 T cell responses to Hp, our data indicated that CD4 T cell responses to Hp minimally require hematopoietic cell expression of Chi3l1.
To further assess the contribution of Chi3l1-expressing hematopoietic cells to Hp-induced CD4 T cell responses, we directly compared Hp-elicited T cell responses in BALB/c and Chi3l1 -/mice that were lethally irradiated and reconstituted with either BALB/c BM or Chi3l1 -/-BM ( Figure 2A). We observed a significant reduction in the numbers of total activated CD62L lo CD44 hi CD4 cells ( Figure 2B) and CXCR5 hi PD-1 hi T FH cells ( Figures 2C-E) in the Hp-infected BALB/c recipients reconstituted with Chi3l1 -/-BM relative to animals reconstituted with BALB/c BM. Expression of ICOS on CXCR5 + PD-1 hi T FH cells was reduced in mice reconstituted with Chi3l1 -/-BM ( Figure 2F). We confirmed the effector potential of CD4 T cells from BALB/c mice reconstituted with Chi3l1 -/-BM chimeras was impaired, as the frequencies and numbers of IL-4 + single producers (Figures 2G-I) and IL-4 + IL-13 + double producers (Figures 2J, K) were decreased following anti-CD3 stimulation. Thus, Chi3l1 expression within BM derived radiation sensitive immune cells regulates ICOS expression by T FH cells and is required for development of T FH responses and T H 2 cells with effector potential following Hp infection.
Cell intrinsic expression of Chi3l1 regulates B cell development but appears not critical for T FH development or expansion CHI3L1 is a secreted protein (62,63) and may regulate T cell responses to Hp via autocrine and/or paracrine mechanisms. To address whether T cell-intrinsic expression of Chi3l1 regulates T H 2 and T FH responses to Hp, we measured CD4 T cell responses in Hpinfected BM chimeras ( Figure 3A Figure S3. Data representative of ≥ 3 independent experiments displayed as mean ± SD of each group with cells from individual animals depicted as circles or triangles. Unpaired 2-tailed student's t-test was used to assess statistical significance. *p≤0.05, **p≤0.01, ***p≤0.001, ****p≤0.0001. frequency ( Figures 3G, H) or number ( Figure 3I) of T FH cells of each genotype present in same host on D0 or 8 days post-infection. While T cell intrinsic expression of Chi3l1 did not appear to be required for the development of the T FH compartment, we did observe that the percentage (Figures 3J, K) and number ( Figure 3L) of Chi3l1 -/-B220 + msLN cells was significantly increased relative to BALB/c B220 + B cells in both uninfected and Hp-infected 50:50 chimeric mice. The same phenotype was also observed in the spleen  Figure S4. Data representative of ≥3 independent experiments. Data displayed as mean ± SD in triplicate shown as bars (C) or as cell populations derived from each genotype from individual animals shown as paired lines (E-L). Statistical analysis was performed with unpaired 2-tailed student's t-test (C) or paired 2tailed student's t-test (all others). *p≤0.05, **p≤0.01, ***p≤0.001, ****p≤0.0001. *p≤0.05, **p≤0.01, ***p≤0.001, ****p≤0.0001.
( Figures S4C-E), even after normalizing for the input BM cells in the 50:50 chimeras ( Figure S4F). To see if this effect was limited to mature B cells in the periphery, we examined B cell development in the BM of the 50:50 chimeras. We found that the BM of these chimeras was heavily skewed toward the Chi3l1 -/-B cell progenitors at multiple stages of B cell development, including the pre-B cell stage (Fraction C-C') in the BM ( Figures S4G-K), and between Fraction E of the BM and development of mature follicular and marginal zone B cells in the spleen (Figures S4L-N). Since the number of mature B cells was not altered in non-chimeric Chi3l1 -/mice ( Figure S1C), or in reciprocal BM chimeras ( Figure S3B) we concluded that a B cell intrinsic role for Chi3l1 during B cell development was revealed when WT and Chi3l1 -/-B lineage precursors were forced to compete in the 50:50 BM chimeras.

CD4 T cell-intrinsic Chi3l1 is required for functional T cell responses to Hp
Although T cell intrinsic expression of Chi3l1 did not appear to be required for development or expansion of the CD4 T cell compartments following Hp infection, we considered the possibility that T cell intrinsic expression of Chi3l1 might instead be required for the functional or effector attributes of the T cells. To test this hypothesis, we analyzed cytokine responses by the CD4 cells from the Hp-infected 50:50 chimeras. We observed that the proportion of Chi3l1 -/-CD44 hi CD4 cells that were competent to produce IL-4 following restimulation was significantly decreased relative to the frequency of IL-4 producing BALB/c CD4 T cells derived from the same animals ( Figures 4A, B). Moreover, consistent with a prior publication reporting a T H 1 differentiation bias in Chi3l1 -/mice (51), we observed that the proportion of CD44 hi Chi3l1 -/-CD4 cells that produced IFNg following anti-CD3 restimulation was significantly increased relative to the frequency of IFNg-producing BALB/c CD4 T cells derived from the same animals ( Figures 4C, D). Next, we assessed ICOS expression by the T FH cells present in the Hp-infected 50:50 chimeras. We observed significantly decreased ICOS expression ( Figures 4E, F) in the Chi3l1 -/-T FH cells compared to the BALB/c T FH cells from the same infected animal. This was not a transient defect as it was seen even at the peak of the T FH response on D14 post-infection ( Figure 4G). These results therefore suggest that T cell intrinsic expression of Chi3l1 was important for the functional potential of the CD4 cells from Hp-infected mice.
One potential caveat to the 50:50 chimera experiments was that the CD45.2 + hematopoietic compartment would include Chi3l1 -/cells as well as any radiation resistant hematopoietic cells from the CD45.2 + WT host. The presence of these radiation resistant WT cells could potentially mask Chi3l1 -/cell intrinsic deficits. Although this is not a major concern for B cells, which are radiation sensitive (64), some T cell subsets are reported to be more resistant to radiation than others (reviewed in (65)). Given that we observed no differences in the frequencies or numbers of WT and Chi3l1 -/-T FH cells in the Hp-infected 50:50 chimeras (Figures 3D, E), yet we observed a reduction in ICOS levels in CD45.2 + T FH cells from these same animals ( Figures 4E-G), we suspected that the defect in ICOS upregulation was likely a Chi3l1-dependent T cell-intrinsic defect. Our in vitro data ( Figure 1I) examining ICOS upregulation in activated Chi3l1 -/-T cells also suggested a T cell intrinsic role for Chi3l1 in regulating ICOS expression. To further confirm this conclusion, we co-cultured purified splenic CD4 + T cells from uninfected BALB/c CD45.1 + mice at a 1:1 ratio with either purified BALB/c CD45.2 + splenic CD4 + T cells or with Chi3l1 -/-CD45.2 + CD4 + T cells in the presence of platebound anti-CD3 + anti-CD28. Again, we observed that ICOS expression by Chi3l1 -/-CD4 T cells was significantly decreased and was not rescued in trans by the presence of the WT CD4 + T cells ( Figure 4H). Taken together, these data indicate that T cell intrinsic expression of Chi3l1 does play a role in ICOS upregulation early after TCR activation, in ICOS expression by Hp-induced T FH cells and in IL-4 production by restimulated Hp-elicited CD4 T cells.

Chi3l1 regulates B cell responses to Hp infection and immunization
Our data showed that Chi3L1 regulates expression of ICOS by Hp infection elicited T FH cells and we know that ICOS is a key costimulatory molecule that modulates T FH -B cell interactions, particularly during the germinal center B cell (GCB) response (38,39). We therefore hypothesized that Chi3l1 -/mice would mount impaired B cell responses to Hp. To test this, we analyzed B cell subsets in BALB/c and Chi3l1 -/-msLNs before and on D14 after Hpinfection mice ( Figure 5A; Figures S5A-C). The numbers of total msLN cells ( Figure S5D) and B220 + CD138 -B cells ( Figure S5E) were decreased in the D14 Hp-infected Chi3l1 -/mice relative to WT mice. This reduction was not due to changes in the number of naïve msLN B cells ( Figure S5F). Instead, antigen-experienced isotypeswitched B cells ( Figure 5B), isotype-switched GCB cells ( Figure 5C), and antibody secreting cells (ASCs) ( Figure 5D) were decreased in the D14 Hp-infected Chi3l1 -/-msLNs. Interestingly, and unlike the phenotype of GCB cells from spleen ( Figure S7C), msLN GCB cells in infected mice included both a CD38 lo PNA + and CD38 int PNA + population ( Figure S5G). Both msLN GCB populations expressed similarly high levels of PNA ( Figure S5G) and both GCB populations were present in Hp-infected WT and Chi3l1 -/-msLNs ( Figure S5C).
Since our data using 50:50 BM chimeras revealed that B cell intrinsic expression of Chi3l1 can globally alter BM B cell development ( Figure S4) in the setting of competition, we addressed whether B cell intrinsic expression of Chi3l1 might be necessary for optimal GCB and ASC responses to Hp by analyzing D14-infected 50:50 BM chimera msLNs. In contrast to our results showing attenuated GCB and ASC responses in Hp-infected Chi3l1 -/mice ( Figures 5C, D), we observed that Chi3l1 gating approach shown in Figures S6I-K). This result suggested that the reduction in the B cell responses to Hp in the Chi3l1 -/global knockout mice was not due to an intrinsic inability of the Chi3l1 -/-B cells to enter the GCB pathway or differentiate into ASCs. Rather the defect in the Chi3l1 -/-Hp-elicited B cell response in animals with a global deficiency in Chi3l1 was likely due to the loss of Chi3l1 by other cell types. Next, we assessed whether the defective B cell response observed in the Hp-infected Chi3l1 -/mice was limited to the setting of helminth infection. We therefore immunized mice with recombinant protein (influenza NS1) that was adsorbed to the T H 2-biasing adjuvant alum and used flow cytometry to measure splenic polyclonal CD4 T cell and B cell responses as well as NS1specific B cell responses on D12 post-immunization (B cell subset flow gating strategies shown in Figures S7A-C). We first confirmed that numbers of splenocytes ( Figure S7D) and splenic CD4 T cells ( Figure 5E) did not differ in unvaccinated wildtype and Chi3l1 -/control mice or at D12 post-vaccination with NS1. However, the number of splenic T FH cells in immunized Chi3l1 -/mice was significantly reduced (Figures 5F, G) and the remaining Ch3l1 -/-T FH cells expressed decreased levels of ICOS ( Figure 5H). The number of splenic ASC ( Figure S7E), percentage and number of isotype-switched B cells ( Figures S7F-G), and number of GCB cells (Figures S7H-I) did not differ between WT and Chi3l1 -/unvaccinated control mice. Likewise, the numbers of total B cells ( Figures 5I, J), ASC ( Figure S7E) and switched B cells ( Figure 5K) did not differ in vaccinated mice. In contrast, the percentages and numbers of splenic GCB cells were reduced in vaccinated Chi3l1 -/mice ( Figure 5L; Figures S7H-I). Finally, using fluorochromelabeled NS1 protein tetramers, we identified NS1-specific switched B cells ( Figure S7J) and GCB cells ( Figure S7L) in control and vaccinated WT and Chi3l1 -/mice. Although the frequency of NS1-specific isotype-switched ( Figure S7K) and GCB cells ( Figure S7L) was unchanged between the vaccinated WT and Chi3l1 -/mice, the numbers of NS1-specific isotype-switched B cells ( Figure 5M) and NS1-specific GCB cells ( Figure 5N) were significantly decreased in vaccinated Chi3l1 -/mice. Thus, Chi3l1 regulates the magnitude of T FH and antigen-specific B cell responses to both infection and immunization.

Chi3l1 regulates IgE responses
Given our data and published reports showing that Hp-induced IgE responses depend on T FH cells (35), we next analyzed IgE responses in the Hp-infected Chi3l1 -/mice. We observed a significant reduction in the frequency and number of IgE + ASCs in the msLNs from Hp-infected Chi3l1 -/mice ( Figures 6A-C) compared to the WT animals. Moreover, the impaired IgE + ASC response was accompanied by a significant reduction in total IgE levels in the serum of D21 Hp-infected Chi3l1 -/mice ( Figure 6D). In contrast, Hp-specific IgG1 serum antibody titers were equivalent between Chi3l1 -/and WT mice ( Figure 6E). To determine whether the loss of the IgE + ASCs in the Chi3l1 -/-Hp-infected mice was due to a B cell intrinsic defect, we enumerated IgE-expressing ASCs within the msLN of D14 Hp-infected 50:50 chimeras. Consistent with the increase in Chi3l1 -/mature B cells in the 50:50 chimeras, we found significantly more Chi3l1 -/-ASCs in the msLN of Hpinfected 50:50 chimeras ( Figure S6L). However, the frequency of IgE ASCs within the Chi3l1 -/-ASC compartment was equivalent to the frequency of IgE + ASCs within the WT ASC compartment ( Figures  S6M-N). These data suggested that Chi3l1 expression by non-B cells likely regulated IgE + ASC responses to Hp.
Next, to assess whether the defective IgE response in Chi3l1 -/mice was restricted to Hp infection, we first examined serum IgE levels in non-infected animals. Although serum IgE levels were low in WT naïve mice, IgE was detected ( Figure 6F). In contrast, IgE levels were undetectable (at least 100-fold lower) in the Chi3l1 -/serum ( Figure 6F). However, total IgG1 in serum from uninfected Chi3l1 -/mice was only decreased~2.5 fold relative to the WT animals ( Figure 6G), again suggesting a more profound deficit in IgE production by Chi3l1 -/mice. To address whether antigenspecific IgE responses are also dependent on Chi3l1, we vaccinated mice with nitrophenyl haptenated ovalbumin (NP-OVA) adsorbed to alum and measured total and NP-specific IgE on D14. Again, total IgE levels were significantly lower in the vaccinated Chi3l1 -/mice ( Figure 6H). Moreover, NP-specific IgE, which was easily measured in vaccinated WT mice, was undetectable in Chi3l1 -/serum ( Figure 6I). In contrast, NP-specific IgG1 levels were not significant different between Chi3l1 -/and WT mice. ( Figure 6J). Thus, Chi3l1 plays a central role in IgE responses but is dispensable for IgG1 responses.
Chi3l1 regulates IL-4 + T FH programming IL-4 producing T FH cells are required for the polyclonal IgE response to Hp (35). Since B cell intrinsic expression of Chi3l1 was not required for the IgE ASC response to Hp (Figures S6L-N) and T FH cells were decreased in the Hp-infected Chi3l1 -/mice, we suspected that the large reduction in IgE responses seen in these mice might reflect additional functional impairments in the Chi3l1 -/-T FH cells. To assess this in an unbiased fashion, we compared the transcriptome of T FH cells from msLNs of WT and Chi3l1 -/mice on D14 post Hp infection. We identified 9853 expressed genes and 1465 differentially expressed genes (DEG) between the two T FH populations that met an FDR q<0.05 cutoff (Table S1). 252 DEG exhibited at least a ±0.3785 log 2 fold-change (log 2 FC) ( Figure 7A), with 193 of these genes expressed at >1 mean RPKM in either BALB/c or Chi3l1 -/-T FH . Principal component analysis based on these 193 DEGs shows clear sample group separation ( Figure 7B). Since the differences in expression levels between most of the genes expressed by WT and Chi3l1 -/-T FH cells was relatively modest, with only 24 of the DEG with RPKM > 1 meeting a ± 1 log 2 FC threshold ( Figures 7A, C), we reasoned that we might observe smaller changes in gene expression across many genes within specific pathways. To test this, we performed Gene Set Enrichment Analysis (GSEA) (50) comparing the rank-ordered WT and Chi3l1 -/-T FH gene set to the 5219 C7 Immunologic Signature Gene Sets from MSigDB. None of the C7 gene sets were preferentially enriched (FDR q<0.05) in the WT or Chi3l1 -/-T FH transcriptomes, including the 54 C7 gene sets derived from T FH cells (data not shown). This result, which was consistent with normal expression of the T FH lineage master regulator Bcl-6 by Chi3l1 -/-T FH cells ( Figures 1D, E), indicated that Chi3l1 is not required for establishment of the core T FH transcriptional program (66). We next performed Ingenuity Pathway Analysis (IPA) with the 193 genes meeting an FDR q<0.05 and ± 0.3785 log 2 FC threshold to assess whether changes in expression of suites of genes might reflect alterations in Chi3l1 -/-T FH cell function or signaling. IPA-defined pathways that were predicted to be most significantly different (B-H p < 0.05) between the WT and Chi3l1 -/-T FH cells included T H 2 and T H 1/T H 2 activation ( Figure 7D).

IL-4 produced by T FH cells is required for B cell class switching
to IgE following Hp infection (35,67) and even modest reductions in the amount of IL-4 present can prevent B cell switching to IgE while having no impact on the IgG1 response (68). Prior studies examining the T FH response after infection with the nematode Nb showed that T FH cells change over time following infection and proceed from producing IL-21 alone (T FH21 cells) to producing primarily IL-4 (T FH4 cells) (40). This change in the T FH cytokine profile from IL-21 to IL-4 is associated with migration of T FH cells into the B cell follicle (69), formation of ICOS-regulated T FH /B cell conjugates that support T FH cell survival and expansion (70), and acquisition of a transcriptionally distinct T FH4 program (40). Given our data, we hypothesized that Chi3l1 might regulate acquisition of the T FH4 transcriptional program. To assess this, we performed GSEA using published lists of genes that are differentially expressed in T FH4 and T FH21 cells from Nb-infected mice (40). We found that DEG that are normally upregulated in Nb T FH4 cells compared to Nb T FH21 cells (40) were significantly enriched in Hp WT T FH cells compared to Hp Chi3l1 -/-T FH cells ( Figure 7E). In contrast, we observed significant enrichment in the transcriptome of the Hpinduced Chi3l1 -/-T FH cells for genes that are increased in Nb T FH21 cells relative to Nb T FH4 cells ( Figure 7F). Finally, to address whether loss of Chi3l1 also altered IL-4 production specifically by T FH cells, we measured IL-4 in PMA+ionomycin-restimulated CXCR5 + PD-1 hi T FH cells from D14 Hp-infected WT and Chi3l1 -/mice. Consistent with the GSEA data, we observed a significant decrease in frequency and number (Figures 7G-I) of Chi3l1 -/-T FH4 cells relative to WT T FH4 cells. Together, the data support the conclusion that Chi3l1 contributes to the development or maintenance of the ICOS-expressing T FH4 subset that facilitates B cell IgE responses.

Discussion
Studies of human allergic diseases reveal that expression of YKL-40 protein and CHI3L1 transcripts are increased in the lungs and serum of some asthma cohorts (13,14). Increased expression of YKL-40/CHI3L1 is positively associated with pathogenesis in allergic rhinitis (19, 20), atopic dermatitis (21)(22)(23) and food allergy (24) and CHI3L1 SNPs confer risk of asthma development and airway remodeling (14)(15)(16)(17) as well as increased serum IgE and atopy (18). Similarly, experiments using Chi3l1 -/mice show that Chi3l1 regulates type 2 cytokines and IgE in models of asthma, atopic dermatitis and food allergy (8,(23)(24)(25)(26). Thus, both mouse and human data support a role for Chi3l1 in promoting atopic disease. Here, we extend these prior studies to show that Chi3l1 also facilitates T H 2 responses to the pathogen Hp. In addition, we identify Chi3l1 as a key regulator of T FH responses following both Hp infection and alum-adjuvanted vaccination and demonstrate that Chi3l1, while not required for T FH development, plays critical roles in T FH expansion, ICOS expression by T FH cells, IL-4 production by T FH cells and T FH -dependent IgE production.
Lower organisms utilize chitinases and chitinase-like proteins (CLPs) in innate host defense against chitin-containing pathogens. In Drosophila, the transport of secreted CLP homologues via the hemolymph as well as expression of CLP by hemocytes, the evolutionary precursors of mammalian leukocytes, mediate host defense (71). Following infection, the CLP homologue is transcriptionally upregulated (72), resulting in increased hemolymph levels (73,74). In parallel, mammalian CLPs are theorized to function as soluble mediators that can be expressed and secreted by activated hematopoietic and damaged mesenchymal cells (75,76). Consistent with this, our data suggested that Chi3l1 can function as a cytokine-like paracrine regulator in some settings. However, we also observed cell-intrinsic roles for Chi3l1 in CD4 T cell function and progenitor B cell development, suggesting that Chi3l1 might signal in an autocrine manner. Consistent with prior reports (51), we observed no difference in the size of the mature CD4 T cell in mice globally deficient in Chi3l1. Likewise, the number of WT and Chi3l1 -/mature msLN CD4 T cells and splenic CD4 T cells, CD8 T cells and myeloid cells were similar in uninfected 50WT:50Chi3l1 -/chimeras, indicating that Chi3l1 expression by T and myeloid cells is not required for their development. In contrast, when Chi3l1 -/-B lineage BM precursors were placed in direct competition with WT B lineage precursors, we observed that the mature peripheral naïve B cell compartment was dominated by Chi3l1 -/-B cells. This effect was evident as early as the pre-B cell stage in the BM, suggesting that Chi3l1 can play a B cell intrinsic role as a negative regulator of B cell development and that this intrinsic property becomes evident when WT and Chi3l1 -/-B cell precursors are forced to compete in the competitive BM chimera setting.
Following Hp infection, we observed significant decreases in size of the CD4 T cell, T FH cell, and GCB cell responses in the Chi3l1 global knockouts. These defects appeared to be rescued in the LNs of Hp-infected 50WT:50Chi3l1 -/-BM chimeras, suggesting that intrinsic expression of Chi3l1 by T cells is not important for the development or expansion of T FH cells. However, given the way that we generated the 50WT:50Chi3l1 -/-BM chimeras, we can't absolutely exclude a T cell-intrinsic role for Chi3l1 in T FH development. What we do know is that Chi3l1 expression by hematopoietic cells was both necessary and sufficient to elicit CD4 T cells responses to Hp. These results, which are complimentary to a study showing that hematopoietic lineage expression of Chi3l1 was sufficient to rescue airway inflammation in Chi3l1 -/mice sensitized to Aspergillus fumigatus conidia (25), suggest that Chi3l1-producing hematopoietic cells support expansion of Hp-induced T FH and CD4 T cells.
Although Chi3l1 expression by T cells does not control the size of the CD4 T cell and T FH response to Hp, our data show that multiple attributes of T FH cells are dependent on T cell intrinsic expression of Chi3l1. Indeed, despite equivalent numbers of WT and Chi3l1 -/-CD4 T cells and T FH cells in Hp-infected 50WT:50Chi3l1 -/-BM chimeras, the frequency of CD4 T cells competent to produce IL-4 within the Chi3l1 -/ -CD4 compartment was decreased by almost 50% relative to the WT CD4 T cell compartment. These data fit well with prior studies showing that Chi3l1 -/-CD4 T cells primed in vitro with antibodies to CD3 and CD28 in the presence of T H 2-polarizing conditions are impaired in IL-4 production (51). Moreover, both our in vitro and in vivo experiments reveal that ICOS upregulation by naïve T cells and T FH cells requires T cell-specific expression of Chi3l1. Given that both ICOS and Chi3l1 are upregulated by WT CD4 T cells following TCR and CD28 engagement (51,(77)(78)(79) and Chi3l1 -/-CD4 T cells are impaired in ICOS expression after in vitro stimulation with plate-bound antibodies to CD3 and CD28, we think it likely that Chi3l1 enhances ICOS expression by modulating signaling downstream of TCR and/or CD28 engagement. Chi3l1 -/-CD4 T cells are reported to be hyperresponsive to anti-CD3+CD28 ligation in vitro (51), suggesting that Chi3l1 functions as a negative regulator. This is intriguing given a recent report showing that low tonic TCR signaling is associated with increased ICOS expression by pre-T FH , enhanced T FH development and more robust GCB cell responses (80).
While T FH responses to Hp infection and vaccination are clearly impaired in Chi3l1 -/mice, we do not think that this is due to a requirement for Chi3l1 during commitment to the T FH lineage as the remaining msLN Chi3l1 -/-T FH cells express normal levels of the master T FH regulator Bcl-6 (66) and the core T FH transcriptional program (66) appears intact. Instead, we find that fewer Chi3l1 -/-T FH cells progress to becoming functional IL-4 producing T FH cells (T FH4 cells), which are normally induced following Hp infection (35). This is due, at least in part, to an inability of these cells to acquire the mature T FH4 transcriptional program (40). Instead, Chi3l1 -/-T FH cells appear to be transcriptionally enriched in genes expressed by IL-21 and the IL-21 plus IL-4 producing T FH cells (40). Prior studies examining the T FH response after infection with the nematode Nippostrongylus brasiliensis showed that T FH cells change over time following infection and proceed from producing IL-21 alone to producing primarily IL-4 (40). This change in the T FH cytokine profile from IL-21 to IL-4 coincides with the migration of T FH cells into the B cell follicle (69) and the formation of T FH /B cell conjugates (70). Our data therefore suggest that Chi3l1 is required for the full maturation of the T FH response from T FH21 to B cellengaged T FH4 cells.
Although our data do not specify the mechanism by which Chi3l1 controls the expansion and maturation of the T FH4 response to Hp infection, we suspect that it is related to decreased expression of ICOS by Chi3l1 -/-CD4 T FH cells. It is well appreciated that ICOS/ ICOSL interactions between B cells and T FH cells is required for the maintenance of the T FH population in the B cell follicle and GC (81) and that blockade of ICOS/ICOSL in established allergic responses is sufficient to ablate ongoing T FH and GC responses (82). Moreover, data from parasite infection models indicate that IL-4 production by T FH cells not only requires B cells (34) but is dependent on ICOS/ ICOSL interactions within the B cell follicle (70). Thus, we think that Chi3l1, which is upregulated in CD4 T cells upon TCR stimulation (51), induces ICOS expression in a cell intrinsic manner. ICOSexpressing T FH cells can then productively engage with ICOSLexpressing B cells, which in turn provide T FH cells with appropriate co-stimulatory signals to support their expansion, maintenance and acquisition of the T FH4 transcriptional program within the B cell follicle (70,(81)(82)(83). Given the known role for ICOS-ICOSL interactions in supporting T FH -dependent B cell responses (81), we argue that the decrease in antigen-specific B cell responses in vaccinated Chi3l1 -/mice is also due, in large part, to the poor expansion of the T FH compartment and the reduction in ICOS expression by the remaining Chi3l1 -/-T FH cells.
One of the most striking defects in naïve, vaccinated and Hpinfected Chi3l1 -/mice is the near total ablation of secreted IgE and IgE-expressing ASCs. The data from the Hp-infected 50WT:50Chi3l1 -/mice indicate that this is not due to the loss of Chi3l1 expression by B lineage cells. We propose the loss of the IgE response in vaccinated and infected Chi3l1 -/mice is most likely due to the impaired IL-4 production by the T FH as it was reported that IL-4 produced specifically by T FH cells is required for B cell class switching to IgE following Hp infection (35,67). We think that poor conversion of T FH21 cells to the T FH4 subset may also explain why Hp-infected Chi3l1 -/mice make relatively normal IgG1 responses. It is reported that IL-21 promotes switching to IgG1 and actively inhibits switching to IgE (84) and that even modest reductions in the amount of IL-4 produced can prevent B cell switching to IgE while having no impact on the IgG1 response (68). Although we think that the loss of the IgE response in Chi3l1 -/mice is due to the impaired T FH4 response, ICOS/ICOSL interactions also contribute to the development of both IL-4 producing T cells and IgE responses following allergen sensitization and challenge (83) and blockade of ICOS/ICOSL in established allergic responses is sufficient to impair IgE production (82). Therefore, it is possible that Chi3l1 tunes ICOS expression levels, which favors IL-4 production by T FH cells and supports isotype switching of the ICOSL-expressing B cells to IgE.
In summary, our data reveal new roles for Chi3l1 in controlling upregulation of ICOS by T FH cells, expansion of the T FH compartment, acquisition of the T FH4 transcriptional program, establishment of T FH -dependent B cell responses and production of secreted IgE. We show that these effects are due to paracrine hematopoietic cell derived Chi3l1 as well as T cell intrinsic Chi3l1. These data provide avenues for future research into the mechanisms by which Chi3l1 exerts its various functions in allergic responses and suggest that directed approaches that block Chi3l1 activity or function specifically within the T cell compartment may attenuate both T H 2 responses and T FH -driven IgE responses in pathologic settings such as atopy and allergic airway disease.

Data availability statement
The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found below: GSE203113 (GEO); https://www. ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE203113.

Ethics statement
The animal study was reviewed and approved by University of Alabama Birmingham Institutional Animal Care and Use Committee (IACUC).