Abstract

   Type 2 immunity is associated with adipose tissue (AT) homeostasis and
   infection with parasitic helminths, but whether AT participates in
   immunity to these parasites is unknown. We found that the fat content
   of mesenteric AT (mAT) declined in mice during infection with a
   gut-restricted helminth. This was associated with the accumulation of
   metabolically activated, interleukin (IL)-33, thymic stromal
   lymphopoietin (TSLP) and extracellular matrix (ECM)-producing stromal
   cells. These cells shared transcriptional features, including the
   expression of Dpp4 and Pi16, with multipotent progenitor cells (MPC)
   that have been identified in numerous tissues and are reported to be
   capable of differentiating into fibroblasts and adipocytes.
   Concomitantly, mAT became infiltrated with resident T helper 2 (Th2)
   cells that responded to TSLP and IL-33 by producing stromal
   cell-stimulating cytokines, including TGFβ[1] and Amphiregulin. These
   Th2 cells expressed genes previously associated with type 2 innate
   lymphoid cells (ILC2), including Nmur1, Calca, Klrg1 and Arg1, and
   persisted in mAT for at least 11 months following anthelmintic
   drug-mediated clearance of infection. We found that MPC and Th2 cells
   localized to ECM-rich interstitial spaces that appeared shared between
   mesenteric lymph node, mAT and intestine. Stromal cell expression of
   epidermal growth factor receptor (EGFR), the receptor for Amphiregulin,
   was required for immunity to infection. Our findings point to the
   importance of MPC and Th2 cell interactions within the interstitium in
   orchestrating AT remodeling and immunity to an intestinal infection.

One-Sentence Summary:

   Crosstalk between Th2 cells and stromal multipotent progenitor cells
   drives adipose tissue remodeling during intestinal infection

Introduction

   There is a growing appreciation that lasting inflammation-induced
   changes in tissue homeostasis are mediated by non-hematopoietic cells
   in addition to resident immune cells. Communication between these two
   compartments is essential for host protection and subsequent
   reinforcement of tissues against secondary infections ([64]1). In type
   2 immune responses, cooperation between tissue resident immune cells
   and non-hematopoietic cells ensures successful defence against
   pathogens ([65]2, [66]3), but when dysregulated can drive allergic and
   asthmatic inflammation, resulting in permanent fibrotic remodeling of
   the tissue ([67]4). Non-immune cells, such as epithelial and stromal
   cells, release signals including IL-25, IL-33 and TSLP, which are
   critical for the activation and maintenance of type 2 immune cells
   ([68]3, [69]5).

   Type 2 immune processes underlie aspects of tissue homeostasis. This is
   particularly well recognized in adipose tissue (AT), where eosinophils
   and type 2 innate lymphoid cells (ILC2) contribute to the maintenance
   of alternative macrophage activation, and are implicated in AT beiging
   and the prevention of metabolic disease ([70]6–[71]9). Regulatory T
   (Treg) cells are also important for AT homeostasis ([72]10, [73]11).
   Stromal cells within visceral white AT produce IL-33, which activates
   and sustains ILC2 and Treg populations during homeostatic conditions
   ([74]12–[75]15). The stromal cell population in AT is heterogeneous,
   containing precursor cells in various stages of commitment to the
   adipocyte lineage ([76]16). Recent work has identified multipotent
   progenitor cells (MPC), present in numerous tissues, that are capable
   of giving rise to multiple fibroblast subsets ([77]17). MPC share a
   transcriptional signature, marked by Dpp4, Pi16 and CD55 expression,
   with the cell type that gives rise to committed preadipocytes ([78]16).
   Consistent with this, lineage tracing studies confirmed fibroblasts as
   the cell type that differentiate to adipocytes in vivo ([79]18).
   However, it remains unclear how perturbations such as type 2
   inflammation affect the structure and function of stromal cell
   populations within AT. It is of interest that AT MPC have been
   localized to interstitial tissue at the edge of AT and around the blood
   vessels ([80]16, [81]18), since IL-33 producing stromal cells have also
   been identified to occupy interstitial niches in adventitia around
   blood vessels ([82]19) and in the fascia of the skin ([83]20).

   Parasitic helminths, including the intestinal nematode Heligmosomoides
   polygyrus bakeri (H. polygyrus), induce strong type 2 immune responses
   and resistance to reinfection with this parasite is dependent on
   adaptive type 2 immunity ([84]21). Infection with H. polygyrus, and
   other helminths, prevents the development of obesity in mice fed a high
   fat diet, through an IL-33-dependent pathway linked to AT beiging
   ([85]6, [86]9, [87]22–[88]24). However, little is known about the
   development of adaptive type 2 immunity and Th2 cells in AT, or the
   potential for AT to contribute to protective responses against helminth
   infection. This is of interest since there is a growing understanding
   that AT can be repurposed to participate in host defense against
   infection ([89]25, [90]26).

   Mesenteric AT (mAT) is a white adipose depot that is closely associated
   with the intestinal tract and surrounds the vasculature, lymphatics and
   mesenteric lymph nodes (mLN) that drain the gut ([91]27). When
   examining mice infected with H. polygyrus, we noted gross changes in
   the size and stiffness of mAT that suggested effects distal to the
   enteric site of infection. This prompted us to explore the biology of
   mAT over the course of H. polygyrus infection, recovery after
   treatment, and secondary H. polygyrus infection. Our data revealed
   dynamic changes in type 2 immunity, characterized by the development of
   a mAT-resident population of Amphiregulin and TGFβ[1]-producing Th2
   cells. This was paralleled by activation of the MPC compartment,
   underscored by increased production of the Th2 cell-activating
   cytokines TSLP and IL-33, as well as increased extracellular matrix
   (ECM) secretion. We found that interstitial, ECM-rich niches, which are
   contiguous with the small intestine, became a site of Th2 cell and MPC
   interactions. Our findings implicate stromal cells responses to
   Amphiregulin in resistance to H. polygyrus infection.

Results

Dynamic changes occur in adipose tissue adjacent to the site of infection

   We examined the effects of primary and secondary infection with
   Heligmosomoides polygyrus on mAT biology ([92]Fig. S1A). This infection
   is chronic, but can be cleared by treatment with anthelmintic drugs
   such as pyrantel pamoate, after which mice show increased resistance to
   secondary challenge. We found that infection-induced mLN enlargement
   was accompanied by a reduction in surrounding mAT ([93]Fig. 1A),
   characterized by a decrease in mAT weight and fat content, while food
   intake and total body weights between infected and control mice
   remained comparable ([94]Fig. 1B, [95]C, [96]S1B, [97]C). Although
   serum leptin levels were unaffected by infection, adiponectin levels
   declined, which was in line with the decrease in adiposity ([98]Fig.
   S1D, [99]E). We found that cellularity of the mAT stromal vascular
   fraction, particularly after secondary infection, was increased
   ([100]Fig. 1D). These changes in stromal vascular fraction cellularity
   reflected increases in both immune and stromal cell numbers ([101]Fig.
   1E, [102]F, [103]S1F).

Fig 1. Extensive remodeling of mAT during H. polygyrus infection is
associated with activation of the stromal cell compartment.

   [104]Fig 1.
   [105]Open in a new tab

   (A) H&E-stained sections of mAT and mLN in indicated experimental
   conditions. (B) mAT weights (n = 9 to 23) (C) mAT fat content measured
   by magnetic resonance imaging (n = 10 to 18). (D-F) Counts of: total
   stromal vascular fraction (SVF) cells (D, n = 12 to 20), CD45^+ immune
   cells (E, n = 12 to 20) and PDGFRα^+ Sca1^+ stromal cells (SC) (F, n =
   15 to 17). (G) mAT SC size measured by flow cytometry. (H) mAT stromal
   cells granularity measured by flow cytometry. (I, J) RNAseq of mAT
   stromal cells from control mice or mice with primary or secondary
   infection. Stromal cells were sorted from individual mice. (I) GO
   enrichment analysis of significantly upregulated genes (adjusted p
   value < 0.01 and average log10FC > 0.25) in mice with 1º H. polygyrus
   infection compared with control mice. (J) Heatmap indicating expression
   of selected genes of interest (collagens, cytokines, chemokines) in
   control or infected mice. (K) Masson’s trichrome staining of mAT
   sections from control or infected mice; collagen- enriched areas stain
   blue. Asterisks: fat-associated lymphoid clusters. (L-N) Pro-collagen 1
   (pCol1) (L), IL-33 (M) and TSLP (N) production by mAT SC isolated from
   control or infected mice, measured in cell supernatants after overnight
   culture. (O) Oxygen consumption rates (OCR) of mAT stromal cells from
   indicated conditions at baseline and after sequential oligomycin
   (Oligo), FCCP and rotenone/antimycin (Rot/Ant) injections, measured
   using Seahorse (n = 5 to 10). (P) Baseline OCR of mAT stromal cells (n
   = 5 to 10). (Q) Baseline extracellular acidification rates (ECAR) of
   mAT stromal cells (n = 5 to 10). (R-T) Effects of heptelidic acid on
   baseline ECAR (R), pCol1 (S) and TSLP production (T) of mAT stromal
   cells of control or infected mice. Data combined from two (A, F), three
   (C) four (D, E) experiments, representative of two (A, O-S), three (G,
   H, K, N, T), four (L, M) experiments or from one experiment (I, J).
   Symbols represent biological (B-H) or technical (L-T) replicates. Mean
   ±SEM. *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001. Data were analysed by
   one-way unpaired ANOVA with Bonferroni’s multiple comparison post test
   (B-E, L-N, P, Q, T), two-tailed unpaired t test (R, S) or Mann–Whitney
   test (F-H).

   We noted that infection caused an increase in stromal cell size and
   side scatter, indicating that these cells had become activated and
   potentially secretory ([106]Fig. 1G, [107]H, [108]S1G). To understand
   infection-associated changes in mAT stromal cells, we performed RNAseq
   on sorted CD45^- CD31^- Sca1^+ PDGFRα^+ stromal cells. Pathway
   enrichment analysis revealed an emphasis on ECM and inflammation, with
   expression of collagen, chemokine and cytokine genes during infection
   ([109]Fig. 1I, [110]J). Consistent with this, Masson’s trichrome
   staining revealed the expansion of collagen-rich areas within the mAT
   of infected mice, often in proximity to tertiary lymphoid structures,
   also referred to as fat-associated lymphoid clusters ([111]Fig. 1K).
   Moreover, pro-collagen 1 (pCol1) as well as TSLP and IL-33 were
   constitutively secreted by mAT SC when cultured ex vivo ([112]Fig.
   1L-[113]N). Production of these proteins was greater after secondary
   infection than after primary infection. We did not observe a strong
   infection-linked increase in cytokine or collagen production by stromal
   cells isolated from the gonadal AT (gAT) of the same animals ([114]Fig.
   S1H).

   The increased size, granularity and secretory activity of SC from
   infected mice suggested an increase in anabolic metabolism. In other
   stromal cells, activation to assume a secretory phenotype is supported
   by increased uptake of both glucose and glutamine, and by increased
   glycolysis to facilitate anabolism ([115]28–[116]30). Our analysis
   revealed that mAT stromal cells, but not gAT stromal cells, from
   infected mice were broadly more metabolically active than those from
   control mice, with increased baseline oxygen consumption rates and
   spare respiratory capacity ([117]31), indicative of increased
   mitochondrial respiration ([118]Fig. 1O, [119]P, [120]S1I). Further,
   the extracellular acidification rate, an indicator of lactate release
   as a result of glycolysis, was increased in mAT stromal cells from
   infected animals ([121]Fig. 1Q). Metabolic activation was most
   pronounced in secondary infection ([122]Fig. 1O-[123]Q). Moreover,
   metabolic activation was important for altered stromal cells function
   during infection since the selective Glyceraldehyde-phosphate
   dehydrogenase inhibitor heptelidic acid not only inhibited increased
   extracellular acidification rates but also pCol1 and TSLP production by
   mAT stromal cells ([124]Fig. 1R-[125]T). Together, these results
   indicate that H. polygyrus infection is associated with activation of
   the mAT stromal cell compartment.

Th2 cells with innate immune cell properties take up long term residence in
adipose tissue

   We next explored infection-induced changes in the cellular composition
   of the mAT stromal vascular fraction in greater detail using single
   cell (sc) RNAseq. Unsupervised clustering of stromal vascular fraction
   cell transcriptomes revealed multiple cell clusters (C1-C18; [126]Fig.
   2A, [127]B, [128]S2A, [129]B). Infection-induced changes within the
   immune cell compartment were apparent, with a reduction in myeloid
   cells and an expansion of the CD4^+ T cell compartment ([130]Fig.
   2B-[131]D). These changes were confirmed by flow cytometry ([132]Fig.
   2E, [133]F, [134]S2C-[135]H), which also revealed an increase, after
   secondary infection, of mAT eosinophils ([136]Fig. 2G, [137]S2D,
   [138]H) that was not detected by scRNAseq. To better resolve CD4^+ T
   cells, and in particular Gata3-expressing cells which would be expected
   to be involved in the recruitment of eosinophils, we performed an
   unsupervised re-clustering of C4 and C9 and identified Th2, Th1, Treg,
   ILC2 and NKT cells on the basis of canonical marker expression
   ([139]Fig. S3A, [140]B). The relative proportions of each cell type
   showed that numbers of Gata3^+ Th2 cells increased during infection
   ([141]Fig. S3B).

Fig 2. The mAT Th2[RM] compartment expands during infection.

   [142]Fig 2.
   [143]Open in a new tab

   (A, B) Uniform Manifold Approximation and Projection (UMAP) plots of
   10869 mAT stromal vascular fraction cells from control mouse or mice
   with primary or secondary H. polygyrus infection (one mouse per
   condition). Unsupervised clustering identified 18 cell groups; plots
   are color-coded according to (A) cell cluster (left panel) and broad
   identification of cell types (right panel) or according to experimental
   condition (B). (C, D) Changes in immune cell populations after
   infection (normalized to total cell number in experimental condition).
   Number in brackets indicate cluster ID from (A). (E-G) Frequencies of
   mAT macrophages (F4/80^hi SiglecF^low of CD45^+ CD11b^+ cells) (E),
   CD4^+ T cells (CD4^+ TCRβ^+ of CD45^+ cells in lymphocyte gate, n = 10
   to 16) (F) and eosinophils (F4/80^low SiglecF^hi of CD45^+ CD11b^+
   cells) (G) in control or infected mice, measured by flow cytometry.
   (H-K) Flow cytometry plots (H), frequencies (I, K), and numbers (J) of
   GATA3^+ IL-33R^+ and GATA3^+ TSLPR^+ Th2 cells (gated on FOXP3^-CD4^+
   TCRβ^+ CD45^+ cells) in mAT from control or infected mice. (L) Whole
   mount immunofluorescent images of mAT of infected mice stained for CD3
   and CD31. Arrow: interstitial space. Asterisks: FALC. (M-O) Flow
   cytometry plots (M) and frequencies of Areg (N, n = 8 to 9) and IL-5
   (O, n = 10 to 14) expression by mAT CD4^+ T cells (gated on FOXP3^-
   CD4^+ TCRβ^+ CD45^+ cells). (P) TGFβ[1]-LAP expression by mAT Th2 cells
   (gated on FOXP3^- CD4^+ TCRβ^+ CD45^+ cells in lymphocyte gate). Data
   combined from two (I, N) or three (F, O) experiments, representative of
   two (E, G, J-L), experiments or from one experiment (A-D, M). Symbols
   represent biological replicates (E-G, I-K, N, O). Mean ±SEM. **p ≤
   0.01, ***p ≤ 0.001. Data were analysed by one-way unpaired ANOVA with
   Bonferroni’s multiple comparison post test (E-G, I, J, N, O) or
   Mann–Whitney test (K).

   FACS analysis confirmed that the majority of mAT CD4^+ T cells in
   infected mice were GATA3^+ FOXP3^- and therefore Th2 cells ([144]Fig.
   2H, [145]S4A). mAT Th2 cells were characterized by high expression of
   receptors for the stroma derived cytokines IL-33 and TSLP, whereas
   cells with these characteristics constituted only a small percentage of
   mLN cells ([146]Fig. 2H-[147]K, [148]S4B-[149]E). Indeed, in secondary
   infection up to 80% of all mAT CD4^+ T cells were FOXP3^- GATA3^+
   IL-33R^+ Th2 cells ([150]Fig. 2I, [151]S4A). We found that mAT Th2
   cells were CD69^+ CD44^+ and CD62L^-, suggesting that they were
   resident memory T cells (T[RM]) ([152]Fig. S4F). H. polygyrus infection
   also increased the frequency of IL-33R^+ Th2 cells within the CD4^+ T
   cell compartment of gAT, although the overall frequency of CD4^+ T
   cells did not change ([153]Fig. S4G-[154]J). Expansion of mAT deposits
   by high fat diet feeding prior to infection did not affect the
   establishment of resident mAT Th2 populations during infection
   ([155]Fig. S4K).

   Prior studies located AT-resident T cells in fat-associated lymphoid
   clusters ([156]26, [157]32, [158]33). In line with this, whole mount
   mAT confocal microscopy revealed that numerous fat-associated lymphoid
   clusters, evident as dense clusters of non-adipocyte cells, were
   enriched in CD3^+ GATA3^+ (Th2) cells in the infected mice ([159]Fig.
   S5A, [160]B). These structures were rare and smaller in mAT from
   uninfected mice. Moreover, in infected mice Th2 cells were also present
   in areas outside of the fat-associated lymphoid clusters, scattered
   among adipocytes and PDGFRα^+ stromal cells ([161]S5B, C), but also
   along the interstitial spaces (the adventitia and reticular
   interstitium), that appeared to connect with these lymphoid structures
   ([162]Fig. 2L, [163]S6).

   Functionally, the majority of mAT GATA3^+ Th2 cells from infected mice
   were capable of making the eosinophil survival factor IL-5, but also
   the tissue modulatory cytokines Amphiregulin and TGFβ[1] ([164]Fig.
   2M-[165]P, [166]S7A, [167]B); cells with these attributes were less
   frequent in the mLN from the same animals ([168]Fig. S7-[169]E),
   consistent with the view that terminal Th2 cell differentiation occurs
   within peripheral tissues ([170]34). Our data also indicated that as a
   result of infection, Amphiregulin production in mAT shifted from the
   innate compartment, where it occurred mostly in ILC, to the T cell
   compartment, where it was primarily a function of Th2 cells, although
   ILC2 still remained a source of Amphiregulin ([171]Fig. S7F-[172]I).
   scRNASeq data additionally suggested that Th2 cells also became a major
   source of TGFβ[1] in mAT during secondary infection ([173]Fig.
   S7J-[174]L). We did not detect strong expression of Areg, the gene
   encoding Amphiregulin, in myeloid cells ([175]Fig. S7J-[176]L). Lastly,
   we found that the increase in the Th2[RM] population was also
   paralleled by the progressive infection-associated decline in the
   frequencies of mAT Treg cells and IFN-γ producing CD4^+ T cells
   ([177]Fig. S8A-[178]E), suggesting that Th2 cells may compete with Treg
   and Th1 cells for available space in the mAT lymphoid niche. In this
   context, we noted that IL-33R was more strongly expressed by mAT Th2
   cells than mAT Treg cells in infected mice, and that IL-33 i.p.
   injection into naïve wild type mice resulted in greater expansion of
   the Th2 compared to Treg population in mAT ([179]Fig. S8F-[180]I),
   suggesting that Th2 cells are more responsive to IL-33 than Treg cells.

   To ask whether mAT Th2 cells have tissue-specific attributes, we used
   scRNAseq to compare them to Th2 cells sorted from anatomically related
   sites affected by infection, namely the small intestine and mLN.
   Clustering and similarity analysis with VarID ([181]35) revealed
   distinct groupings of T cell populations by tissue of residence
   ([182]Fig. 3A, [183]B). Cd44^+, Cd69^+, Cd62l^- (Sell), Klf2^- Th2[RM]
   cells were present in mAT and small intestine but nevertheless
   clustered separately from each other (C7 vs. C11) ([184]Fig. 3A,
   [185]C), indicating location-dependent functional distinctions.
   Previous work has described Th2[RM] cells in small intestine and
   peritoneal cavity of H. polygyrus infected mice, but not in mAT
   ([186]36). We observed that mAT, but not small intestine Th2[RM] cells,
   upregulated CD25 (Il2ra), and expressed high levels of Il1rl1 (IL-33R,
   ST2), while small intestine Th2[RM] cells displayed higher expression
   of Il17rb (IL-25R), reminiscent of tissue-specific alarmin receptor
   expression reported for ILC2 ([187]37) ([188]Fig. 3D). However, Th2[RM]
   cells from mAT, to an equal or greater extent than small intestine
   Th2[RM] cells, expressed several genes previously associated with ILC2:
   Nmur1, Calca (or Cgrp), Klrg1 and Arg1, indicating that Th2[RM] cells
   acquire an innate-like phenotype when residing in mAT ([189]Fig. 3D).

Fig 3.

   [190]Fig 3.
   [191]Open in a new tab

   The mAT Th2 cell population has innate cell properties, expands
   independently of T cell recruitment, and is long lived. (A, B) UMAP
   plots of CD4^+ TCRβ^+ T cells from mLN and mAT of control (n = 2 per
   condition), primary infected (n = 1 per condition) and secondary
   infected (n = 1 per condition) mice, and from small intestine lamina
   propria of control and primary infected mice (n = 2 per condition).
   Unsupervised clustering distinguished 16 cell groups; plots are
   color-coded according to cell identity (A) or tissue of origin (B).
   (C-F) UMAP plots indicating log2 normalized expression of selected
   genes of interest (upper panels) and cluster-specific gene expression
   shown as dot plots (lower panels), where color represents the z-score
   of the mean expression across clusters and dot size represents the
   fraction of cells in the cluster expressing the selected gene. (G-J).
   Mice were given secondary infections and treated with FTY720 where
   indicated. Flow cytometry plots (G), frequencies (H) and numbers (I) of
   GATA3^+ IL-33R^+ Th2 cells (gated on FOXP3^- TCRβ^+ CD45^+ cells). Worm
   counts from small intestine (J). (K, L) Flow cytometry plots (K) and
   frequencies (L) of mAT IL-5^+ Amphiregulin (Areg)^+ T cells (gated on
   FOXP3^- CD4^+ TCRβ^+ CD45^+ cells). (M-O) Frequencies (M) and numbers
   (P) of mAT GATA3^+IL-33R^+ Th2 cells, frequencies of Treg cells
   (FOXP3^+ CD4^+ TCRβ^+ CD45^+) (N) and ILCs (gated on Lin- CD45^+ Thy1^+
   live cells in lymphocyte gate; see Methods) (O) after 11 months of
   recovery following treatment of primary infection. Data combined from
   two experiments (A-F, J) or representative of two experiments (G-P).
   Symbols represent biological replicates (H-J, L-P). Mean ±SEM. *p ≤
   0.05, **p ≤ 0.01, ***p ≤ 0.001. Data were analysed by one-way unpaired
   ANOVA with Bonferroni’s multiple comparison post test (H, I) or
   Mann–Whitney test (J, L-P).

   In contrast to small intestine Th2[RM] cells, mAT Th2[RM] cells
   expressed Ccr2 ([192]Fig. 3E), pointing towards a selective role for
   CCR2 ligands in Th2[RM] cell localization in mAT. Moreover, analysis of
   integrin expression by both scRNASeq and flow cytometry revealed that
   mAT Th2[RM] cells did not express Itgae (CD103), a marker for many
   mucosal resident T cell populations ([193]38), but did express Itga4
   (CD49d), unlike small intestine Th2 cells ([194]Fig. 3E, [195]S9A).
   Both Th2 populations expressed Itgb1 (CD29), Itgb3 (CD61) and Itgb7
   ([196]Fig. 3E, [197]S9A). This integrin expression pattern indicates
   that mAT Th2[RM] cells can form functional heterodimeric integrin
   receptors that enable interactions with the stromal cells and ECM
   components. These findings, along with the fact that Th2 [RM] cells are
   found within the mAT stroma, suggest that they are motile within the
   tissue. Both mAT and small intestine Th2 cells expressed Il5, Il6, Areg
   and Tgfb1, but small intestine Th2 cells expressed these genes, as well
   as Il4 and Il13, more strongly than their mAT equivalents ([198]Fig.
   3F). This distinct cytokine expression pattern is again reminiscent of
   reported differences in ILC2, where intestinal ILC2 preferentially
   express IL-13 ([199]37).

   Consistent with mAT Th2 cells being T[RM], Th2 cell accumulation during
   secondary infection was unaffected by treatment with the
   sphingosine-1-phosphate receptor 1 (S1PR1) agonist FTY720 ([200]Fig.
   3G-[201]I), indicating that after primary infection the mAT Th2[RM]
   population persisted independently from the recruitment of cells from
   secondary lymphoid organs through the S1PR1 dependent pathway. FTY720
   treatment also had no effect on resistance to reinfection ([202]Fig.
   3J). Furthermore, mAT remained enriched in IL-33R^+ Th2[RM] cells
   capable of making both IL-5 and Amphiregulin for up to 11 months post
   treatment ([203]Fig. 3K-[204]M, [205]S9B, [206]C), while Treg and ILC2
   populations within this tissue were less frequent than those in naïve
   control mice throughout this time ([207]Fig. 3N, [208]O,
   [209]S9D-[210]H). Indeed, the mAT Th2[RM] population not only
   persisted, but continued to expand over time in the absence of
   infection ([211]Fig. 3P). In summary, infection with H. polygyrus, an
   intestinal helminth parasite, led to the accumulation of Th2 cells in
   mAT. These cells were phenotypically Th2[RM]-like, expressed receptors
   for the tissue alarmins IL-33 and TSLP, and made a range of tissue
   modulatory mediators including Amphiregulin and TGFβ[1].

Resident Th2 cells and stromal cells activate each other during infection

   We asked how mAT Th2 [RM] cells influenced mAT remodeling during
   infection. Prevention of Th2 cell development through the deletion of
   IL-4Rα on T cells in Il4ra^fl/fl Cd4-Cre mice resulted in a reduction
   in resistance to secondary infection, and the loss of associated
   components of the response such as tissue eosinophilia ([212]Fig.
   4A-[213]C, [214]S10A-[215]C), confirming the importance of Th2 cells
   for orchestrating type 2 immunity and host defence during secondary
   infection. When we examined mAT stromal cell activation in infected
   Il4ra^fl/fl Cd4-Cre mice, we found that cytokine and pCol1 production
   was diminished compared to infected controls ([216]Fig. 4D, [217]E),
   despite ILC2 and Treg cells still being present in mAT ([218]Fig. S10A,
   [219]D-[220]G). Furthermore, stromal cell activation was also
   diminished when CD4^+ T cells were depleted, an intervention that also
   resulted in the loss of resistance to infection ([221]Fig. 4F-[222]H,
   [223]S10H-[224]K). Th2 cells therefore played a critical role in mAT
   stromal cell activation during infection.

Fig 4. Th2 [RM] cells and stromal cells activate each other during H.
polygyrus infection.

   Fig 4.
   [225]Open in a new tab

   (A-E) Il4ra^fl/fl Cd4-Cre and Il4ra^fl/fl mice were subjected to
   secondary infection. Frequencies (A) and numbers (B) of mAT Th2 cells
   (IL-33R^+ GATA3^+ FOXP3^- CD4^+ TCRβ^+). (C) worm numbers in small
   intestine. Production of pCol1 (D) and IL-33 (E) by mAT SC isolated
   from Il4ra^fl/flCd4-Cre and Il4ra^fl/fl mice. (F-H) pCol1 (F), TSLP (G)
   and IL-33 (H) production from isolated mAT stromal cells during
   secondary infection with anti-CD4 depletion where indicated. (I, J)
   Flow cytometry plot (I) and cell number (J) of sorted Th2 cells after 4
   days of co-culture with mAT stromal cells isolated from uninfected or
   infected mice as indicated. (K) IL-5 production by Th2 cells after 4
   days of culture with mAT stromal cells as in (I). (L, M) Th2 cells
   (CD4^+ TCRβ^+ IL4-eGFP^+ FOXP3RFP^-) and non-Th2/non-Treg T cells
   (CD4^+ TCRβ^+ IL4-eGFP^- FOXP3RFP^-) were sorted from mAT and mLN of
   infected mice and cultured for 3 – 6 days with addition of IL-33
   [50ng/ml] or TSLP [50ng/ml]. Flow cytometry plots of mAT Th2 and
   non-Th2/non-Treg cells with frequencies of live cells after 6 days of
   culture (L). Levels of indicated cytokines in the supernatants of mAT
   Th2 cells after 3 days of culture (M). (N-Q) TSLP signalling was
   blocked with anti-IL-7Rα antibody treatment during recovery after
   primary infection. Frequencies (N, n = 9) and numbers (O, n = 9) of Th2
   cells (GATA3^+ FOXP3^- CD4^+ TCRβ^+) in mAT or mLN as indicated. Flow
   cytometry plots (P) and numbers (Q, n = 9) of Ki67^+ Th2 cells in mAT.
   Data combined from two experiments (C, N, O, Q), representative of two
   (A, B, D, I, K-M), three (F, H) experiments or from one experiment (E,
   G, J). Symbols represent biological (A-C, M-Q) or technical (D-K)
   replicates. Mean ±SEM. *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001. Data were
   analysed by one-way unpaired ANOVA with Bonferroni’s multiple
   comparison post test (J, K, M-O, Q), two-tailed unpaired t test (D-H)
   or Mann–Whitney test (A-C).

   We next asked whether stromal cells reciprocally stimulated Th2[RM]
   cells, by purifying and co-culturing these populations and measuring
   Th2 cell survival and cytokine production. We found that both were
   significantly enhanced in the presence of mAT stromal cells from
   infected or naive mice ([226]Fig. 4I-[227]K). These effects were seen
   when mAT Th2[RM] cells and mAT stromal cells were cultured in a
   trans-well system, indicating that soluble factor(s) from stromal cells
   could drive Th2[RM] cell activation ([228]Fig. S10L). The expression of
   TSLPR and IL-33R on mAT Th2 cells, and the recognized relationship of
   TSLP and IL-33 with type 2 immunity, suggested that it could be these
   cytokines that were responsible for the observed effects. To examine
   this possibility, we sorted both Th2 cells (CD4^+ Foxp3^-IL-4^+) and
   non-Th2/non-Treg cells (CD4^+Foxp3^-IL-4^-) from mAT and mLN of
   infected animals and cultured them in the presence or absence of IL-33
   and TSLP. We found that mAT Th2[RM] cells were more capable of
   surviving in vitro than non-Th2/non-Treg cells or mLN Th2 cells, and
   that they proliferated extensively in the absence of added growth
   factors ([229]Fig. 4L, [230]S10M). This distinction was further
   enhanced by the addition of IL-33 or TSLP ([231]Fig. 4L, [232]S10M). We
   found that, without the ex-vivo addition of a T cell receptor stimulus,
   TSLP and, in particular, IL-33, activated purified mAT Th2 cells to
   produce Amphiregulin, IL-5, IL-6, IL-13 and TGFβ[1] ([233]Fig. 4M,
   [234]S10N). We noted that while cytokine secretion by mAT Th2 cells was
   driven more strongly by IL-33 ([235]Fig. 4M), survival was enhanced
   more strongly by TSLP ([236]Fig. 4L). These findings are consistent
   with the reported properties of these cytokines ([237]39, [238]40).

   TSLP signals through a heterodimeric receptor consisting of CRLF2 and
   IL-7Rα, the latter of which, when paired with γ[c] (encode by Il2rg) is
   also a component of the bona fide IL-7R ([239]41). Th2[RM] cells
   expressed Crlf2, Il7ra and Il2rg suggesting that they could use TSLP or
   IL-7 for survival in the mAT niche ([240]Fig. 2H, [241]K,
   [242]S11A-[243]C). Consistent with this, antibody-mediated blockade of
   IL7Rα for one week during post-infection recovery resulted in
   significant decreases in Th2 cells, and in Ki67^+ Th2 cells in mAT, but
   not in mLN, confirming a requirement for TSLP/IL-7 signaling for mAT
   Th2[RM] cell proliferation and persistence ([244]Fig. 4N-[245]Q,
   [246]S11D), although this treatment does not exclude indirect effects
   of IL-7Rα inhibition on other mAT-resident immune populations. Since we
   could not detect IL-7 in stromal cell culture supernatants and Il7 was
   not strongly expressed by Pdgfra^+ stromal cells in the scRNAseq data
   ([247]Fig. S11E), these results likely reflected the inhibition of
   TSLP-mediated effects. Together, our data point to the existence of a
   positive feedback loop in which Th2[RM] cells in the mAT activate
   stromal cells to secrete IL-33 and TSLP, which in turn promote mAT
   Th2[RM] cell expansion, survival and cytokine production.

Activated stromal multipotent progenitor cells accumulate and secrete
collagen and immunostimulatory cytokines

   To more fully explore the mAT stromal response to infection we
   re-clustered the scRNAseq stromal cell transcriptomes and identified 6
   distinct cell groups (C0-C5, [248]Fig. 5A, [249]S12A, [250]B). Using
   published transcription profiles to delineate adipocyte differentiation
   stages, we identified Dpp4^+ Pi16^+ CD55^+ multipotent progenitor cells
   (MPC, C2), intermediate uncommitted cells (C0), as well as Fabp4^+
   Pparg^+ committed preadipocytes (C1) ([251]16, [252]42, [253]43). Other
   clusters were enriched in CD9^+ matrix fibroblasts (C3) ([254]44) and
   Ccl19^+ immunofibroblasts (C4) ([255]45) ([256]Fig. 5A, [257]S12A,
   [258]B). By pseudotemporal ordering using Monocle ([259]46) with MPC
   (C2) set as the origin, our data conformed with the model proposed by
   others in which, within AT, committed preadipocytes capable of giving
   rise to mature adipocytes differentiate from Dpp4+ uncommitted
   progenitor cells ([260]16) ([261]Fig. 5B). A quantitative assessment of
   cluster sizes revealed an ~25% increase in MPC (C2) and a decrease in
   intermediate uncommitted cells (C0), in infected versus control mice
   ([262]Fig. 5C), suggesting a reduction in the differentiation of the
   MPC towards the adipocyte lineage.

Fig 5. The mAT interstitial MPC population expands during infection.

   Fig 5.
   [263]Open in a new tab

   (A) UMAP of mAT stromal cells from uninfected (control) mice (2457
   cells), and mice with primary (2640 cells) or secondary (962 cells) H.
   polygyrus infections. One mouse per condition. Unsupervised clustering
   distinguished 6 cell clusters (A); plots are colour-coded according to
   cell cluster. Cell identities were established based on expression of
   the following markers: C1 - committed preadipocytes (expressing Icam1,
   Apoe, Lpl, Fabp4, Pparg), C2 - multipotent progenitor cells, MPC (Dpp4,
   Anxa3, Cd55, Pi16, Dpt), C3 - CD9^+ profibrotic cells (Cd9, Wnt6, Eln,
   Mgp, Col1a1, Col15a1), C4 - immunofibroblasts (Cd9, Ccl19). (B) UMAP
   plot as in A, showing pseudotemporal ordering of cells, setting origin
   at the center of MPC (C2). (C) Contributions of each cluster as in (A)
   to a total cell pool, split by experimental condition. (D) UMAP plots
   showing expression of indicated genes. (E, F) Flow cytometry plots (E)
   and frequencies (F) of mAT stromal cells populations from control and
   infected mice: Ly6c^+ MPC, Ly6c^- CD9^+ matrix fibroblasts and Ly6c^-
   CD9^- preadipocytes (gated on CD45^- CD31^- PDGFRα^+ cells). (G)
   Frequencies of Thy1^hi mAT stromal cells in control and infected mice
   (gated on CD45^- CD31^- PDGFRα^+ cells). (H). mAT stromal cells were
   sorted into Ly6c^+ MPC, Ly6c^- CD9^+ matrix fibroblasts and Ly6c^-
   CD9^- preadipocytes and cultured in adipogenic conditions (see
   Methods). Representative images on day 6 showing lipid droplet
   accumulation (red staining). (I, J) pCol1 (I) and TSLP (J) production
   measured in supernatants from overnight culture of cells sorted as in
   (H). (K) Whole mount immunofluorescent images of mAT from secondary
   infection stained for DPP4, Pi16 and nuclear staining (Hoechst). (L)
   Whole mount immunofluorescent images of mAT from control or secondary
   infection stained for Pi16, CD3, GATA3 and nuclear staining (Hoechst).
   Data combined from two experiments (F), representative of two (G, I-L)
   or from one experiment (A-D). Symbols represent biological (F, G) or
   technical (I, J) replicates. Mean ±SEM. *p ≤ 0.05, **p ≤ 0.01, ***p ≤
   0.001. Data were analysed by one-way unpaired ANOVA with Bonferroni’s
   multiple comparison post-test (I), two-tailed unpaired t test (J) or
   Mann–Whitney test (F, G).

   We used flow cytometry to further investigate quantitative changes in
   mAT stromal cell populations. We gated on CD45^- CD31^- Sca1^+ PDGFRα^+
   stromal cells and used antibodies against surface molecules shown
   previously ([264]16, [265]43) to discriminate between MPC (Ly6c and
   Thy1), committed matrix fibroblasts (which more strongly express CD9
   but not Ly6c), and committed preadipocytes (which are CD9^loLy6c ^lo)
   ([266]Fig. 5D). Based on these criteria, infection led to increased
   frequencies of MPC and reduced frequencies of committed subsets
   ([267]Fig. 5E-[268]G). Furthermore, MPC numbers were increased
   significantly during infection ([269]Fig. S12C). This increase was
   dependent on Th2 cells since it was diminished in infected Il4ra^fl/fl
   Cd4-Cre mice, ([270]Fig. S12D). Moreover, stromal cells from infected
   Il4ra^fl/fl Cd4-Cre mice were less granular than those from infected
   controls, suggesting that stromal cells activation is generally Th2
   cell-dependent ([271]Fig. S12E, [272]F). To verify functional
   differences between identified stromal subpopulations we sorted them
   based on Ly6c and CD9 expression ([273]Fig. 5E) and asked which had the
   potential to become adipocytes under adipogenic culture conditions
   ([274]16). We found that the committed preadipocytes had the highest
   adipogenic potential, evident by extensive lipid droplet development
   (evident as increased staining with Oil Red), while MPC showed
   intermediate adipogenic potential. The matrix fibroblast subpopulation
   contained few cells that were able to differentiate into mature
   adipocytes ([275]Fig. 5H, [276]S12G). Infection did not affect the
   inherent ability of cells within the different stromal cells
   subpopulations to differentiate into mature adipocytes in culture
   ([277]Fig. 5H, [278]S12G). These data confirmed the functional
   relatedness of the clusters identified in our study to previous
   descriptions of adipocyte differentiation ([279]16, [280]43, [281]44).

   We next asked whether infection induced increases in collagen
   production could be attributed to a particular subpopulation of mAT
   stromal cells. We found that pCol1 was produced by sorted matrix
   fibroblasts and MPC, and that both of these populations produced more
   pCol1 when sorted from infected mice ([282]Fig. 5I). However, the MPC
   made more pCol1 than did the matrix fibroblasts, despite indications
   from the scRNAseq data that the opposite would be the case. By
   comparison, committed preadipocytes from infected mice made little
   pCol1 ([283]Fig. 5I). Additionally, MPC were the major source of TSLP,
   production of which was greatly increased as a result of infection
   ([284]Fig. 5J). We additionally assessed IL-33 expression in stromal
   cells populations using IL-33eGFP reporter mice. Comparison between
   matrix fibroblasts, MPC and preadipocyte populations showed that
   infection increased IL-33 expression in all subsets, however expression
   was the highest in MPC, with almost 90% of cells being IL-33eGFP^+(
   [285]Fig. S13A, [286]B). IL-33eGFP MFI increased in MPC from infected
   mice, indicating increased IL-33 expression per cell ([287]Fig. S13C).
   Comparison of stromal cell subsets in the scRNASeq dataset confirmed
   that MPC expressed Il33 more strongly than any of the other stromal
   populations, in line with previous reports ([288]7, [289]8) ([290]Fig.
   S14A, [291]B). MPC also expressed the chemokine-encoding genes Ccl2,
   which encodes the ligand for CCR2, which is expressed on Th2[RM] cells,
   and Ccl11, which encodes an eosinophil attractant ([292]Fig. S14B).
   Furthermore, MPC also expressed genes encoding ECM components and
   modifying enzymes, including Fn, Postn, Ugdh, Pcolce2, and pCol1a2
   ([293]Fig. S14C, [294]D), although expression of the latter, as well as
   pCol1a2, pCol1a1, pCol3a1 and pCol6a1and Eln was strongest in matrix
   fibroblasts ([295]Fig. S14D).

   Previous reports localized IL-33-producing cells to the niches around
   blood vessels (known as adventitia) within AT and lung tissue
   ([296]19), and within the mesothelium of AT ([297]12), which we
   speculate is anatomically overlapping with interstitial areas at the
   tissue edges ([298]16). We observed that these spaces in mAT were
   enriched in collagen, as indicated by immunofluorescent staining for
   Col1 on whole tissue mounts ([299]Fig. S15). Using Pi16 and DPP4 as
   markers of MPC ([300]Fig. S16, cluster 2) we localized these cells
   primarily within interstitial spaces, including adventitia ([301]Fig.
   5K, [302]S17). Blood vessels, visualized by staining with antibodies to
   CD31([303]Fig. S18), were apparent within adventitia. MPC localization
   was distinct from that revealed by staining for PDGFRα, a broad stromal
   cell marker, which additionally showed interdigitating cells between
   adipocytes throughout the tissue ([304]Fig. S5C). MPC localized to
   interstitial spaces in mAT from both control and infected animals, but
   in infection these niches appeared more densely infiltrated with cells,
   including with CD4^+ T cells ([305]Fig. S18), consistent with the
   observed increases in MPC and CD4^+ T cells during infection ([306]Fig.
   2F, [307]S12C). Indeed, we found mAT adventitia to be heavily
   infiltrated by GATA3^+ Th2 cells during infection, where Th2 cells
   could be seen forming aggregates with MPC at the edges of this
   interstitial space ([308]Fig. 5L, [309]S19A). Lastly, utilizing
   IL-33eGFP reporter mice, we found DPP4^+ IL-33^+ stromal cells in
   proximity to CD4^+ T cells in the mAT interstitium during infection
   ([310]Fig. S19B).

   Using a complementary immunohistochemistry approach on tissue sections,
   with CD55 as a marker for MPC ([311]Fig. S16, cluster 2), we confirmed
   that CD55^+ MPC localized to interstitial spaces, which often connected
   with fat-associated lymphoid clusters and were more apparent in
   infected mice ([312]Fig. 6A). CD55^+ cells were also detectable
   lymphoid clusters ([313]Fig. 6A). Interstitial areas were clearly
   marked by dense collagen networks, as shown by Masson’s trichrome
   staining ([314]Fig. 6B), providing a context for our observation that
   pCol1 secretion by MPC was increased during infection ([315]Fig. 5I).
   Immunohistochemistry additionally revealed details of mAT interactions
   with the intestine, showing that in control mice a clear line of
   distinction, marked by CD55^+ cells, was maintained at contact points
   between the two tissues ([316]Fig. 6C, [317]S20A). These areas were
   free of CD4^+ T cells ([318]Fig 6C, [319]S20A). During infection
   however, they were often marked by adipocyte-free tissue enriched with
   collagen, within which CD4^+ T cells and CD55^+ MPC were numerous and
   interspersed ([320]Fig. 6C, [321]S20A, [322]B). This was also true at
   the contact points with the intestinal granulomas ([323]Fig. 6D),
   altogether indicating that interstitial tissue expands at the
   mAT/intestine interface during infection. Moreover, continuity of
   CD55^+ MPC-infiltrated areas between mLN, mAT and intestine was
   apparent ([324]Fig. S20B), consistent with previous reports that
   interstitial spaces between organs are connected ([325]47). To further
   explore the relatedness of these findings to the granulomatous
   response, we used RNAseq to compare isolated granulomas and adjacent
   intestinal tissue from infected mice, with intestinal tissue from
   uninfected mice. This data showed transcriptional signatures consistent
   with the presence of MPC and Th2 cells (but not Treg or Th1 cells) in
   the granulomas ([326]Fig. S21). Furthermore, it was apparent that
   expression of multiple ECM genes, including collagens, was increased in
   the granuloma tissue and mirrored the expression pattern observed in
   mAT stromal cells during infection ([327]Fig. S21; compare with
   [328]Fig. 1J, [329]S14C).

Fig 6. MPC and Th2 cells co-localize to collagen-rich interstitial spaces in
mAT during infection.

   Fig 6.
   [330]Open in a new tab

   (A, B) mAT sections from uninfected (control) mice and mice with a
   secondary H. polygyrus infection. Multiplex IHC staining for CD55/brown
   and CD4/purple (A) and corresponding images stained with Masson’s
   trichrome (collagen/blue) (B). (C, D) Masson’s trichrome staining and
   corresponding images stained for CD55 and CD4 showing interface between
   mAT and small intestine in control and infected mice. Squares highlight
   the areas shown with a higher magnification in neighboring panels. Data
   representative of two experiments with one to two mice per experiment
   (A-D).

   Together, these results support the view that there is an
   infection-associated expansion of the MPC population within the
   ECM-rich interstitial spaces where Th2 cells are also present. These
   areas expand at the interface of mAT and small intestine. The data
   further suggest that activation of MPC to make ECM, driven by
   stimulatory interactions with Th2 cells, is involved in immunity to H.
   polygyrus.

Activated stromal cells are critical for host protective immunity to
infection

   Given that our data pointed to reciprocal interactions between Th2[RM]
   cells with mAT stromal cells, we examined the role of cytokines
   produced by Th2[RM] cells in mAT stromal cells activation. We focused
   on Amphiregulin, TGFβ[1], IL-4 and IL-13, and first measured metabolic
   rates as a sensitive indicator of stromal cells activation. We found
   that Amphiregulin and TGFβ[1] induced increased baseline oxygen
   consumption rates and spare respiratory capacity, while TGFβ[1,] but
   not Amphiregulin, induced aerobic glycolysis, evident by increased
   extracellular acidification rates, and increased lactate accumulation
   in stromal cell supernatants ([331]Fig. 7A-[332]C, [333]S22A, [334]B).
   In contrast, IL-4 and IL-13 alone had minimal effects on lactate
   production or spare respiratory capacity ([335]Fig. S22A, [336]B), but
   synergized with Amphiregulin to increase aerobic glycolysis ([337]Fig.
   S22C). We also found that Amphiregulin potentiated the increase in
   aerobic glycolysis induced by TGFβ[1] ([338]Fig. 7A, [339]S22C). These
   two cytokines also worked additively to promote oxygen consumption
   rates ([340]Fig. 7B, [341]C). Consistent with this, they individually
   and additively promoted increases in cellular ATP ([342]Fig. 7D).
   Further, tracing incorporation of carbon from 13C-glucose into
   metabolic intermediates showed that incorporation into serine and
   glycine was significantly increased following stimulation with
   Amphiregulin and TGFβ[1] ([343]Fig. 7E, [344]F). Serine is a precursor
   for glycine, which is the most abundant amino acid in collagens. Thus,
   Amphiregulin and TGFβ[1], two cytokines made by mAT Th2[RM] cells,
   worked together to promote mAT stromal cell cellular metabolism, a
   prerequisite for these cells to assume an enhanced secretory function
   during infection ([345]Fig. 1S, [346]T). Our data point to a role for
   Amphiregulin both individually, and as a potentiator of the activity of
   other type 2 cytokines, in the metabolic activation of mAT stromal
   cells.

Fig 7. T cell cytokine-driven activation of stromal cells is important for
immunity to H. polygyrus.

   [347]Fig 7.
   [348]Open in a new tab

   (A-C) Purified mAT stromal cells were cultured in adipogenic conditions
   for 3 days in the presence of indicated cytokines; baseline
   extracellular acidification rate (ECAR) (A), baseline oxygen
   consumption rates (OCR) (B) and OCR at baseline and after sequential
   addition of oligomycin (Oligo), FCCP and rotenone/antimycin (Rot/Ant)
   (C, n = 2 to 4). (D-F) Purified mAT stromal cells were cultured in
   adipogenic conditions for 3 days and, during the final 6h, ^12C-glucose
   was exchanged for ^13C-glucose. Isotopologue distribution was assessed
   by targeted mass spectrometry: ATP pools (D), serine (E) and glycine
   (F) plotted to show fractional contributions from newly metabolized
   ^13C or remaining ^12C glucose carbons (n = 2 to 3). (G-I) Count of:
   stromal vascular fraction cells (G), immune cells (CD45^+ CD31^- cells)
   (H) and stromal cells (CD45^- CD31^- PDGFRα^+) (I) in mAT of Egfr^fl/fl
   and Egfr^fl/fl-Pdgfra-Cre mice. (J) Expression of selected genes that
   mark MPC and committed stromal cells, measured by RNAseq of sorted mAT
   CD31^- CD45^- PDGFRα^+ stromal cells from infected Egfr^fl/fl and
   Egfr^fl/fl-Pdgfra-Cre mice. Each column represents stromal cells from
   an individual mouse. (K) Isolated mAT stromal cells from infected
   Egfr^fl/fl and Egfr^fl/fl-Pdgfra-Cre mice were cultured overnight and
   TSLP levels measured in the supernatants. (L, M) Eggs in the ceacum (L)
   and worms in the small intestine (M) were enumerated in Egfr^fl/fl and
   Egfr^fl/fl-Pdgfra-Cre mice with primary H. polygyrus infections. Data
   combined from two (H, I, K-M) or three (G) experiments, representative
   of two experiments (A-C), or from one experiment (D-F, J). Data
   represent biological (D-I, K-M) or technical (A-C) replicates. Mean
   ±SEM. *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001. Data were analysed by
   one-way unpaired ANOVA with Bonferroni’s multiple comparison post test
   (A-C, E, F), two-tailed unpaired t test (K) or Mann–Whitney test (G-I,
   L, M).

   Amphiregulin plays an important role in immunity to intestinal
   helminths ([349]48, [350]49). While overexpression of Amphiregulin in
   white AT was reported to result in the loss of AT mass ([351]50),
   relatively little is known about its function in AT biology. To begin
   to explore this, we examined the effects of Amphiregulin on mAT stromal
   cell activation ex vivo, and found that it was unique amongst the Th2
   cytokines examined in being able to induce TSLP production ([352]Fig.
   S22D). Next, we examined the acute effects of Amphiregulin on mAT by
   directly injecting naïve mice with this cytokine. We found increased
   numbers of immune cells and stromal cells ([353]Fig. S22E-[354]G),
   which is broadly consistent with the effects of H. polygyrus infection
   ([355]Fig. 1D-[356]F). We next generated mice which lack the
   Amphiregulin receptor EGFR on stromal cells using Pdgfra-Cre
   (Egfr^fl/fl-Pdgfra-Cre mice); Pdgfra is broadly expressed in stromal
   cells ([357]Fig. S11E) and anti-PDGFRα was used here as part of the SC
   sorting strategy. Based on our data showing an important role for
   Amphiregulin as a potentiator of stromal cell activation in response to
   IL-4, IL-13 and TGFβ[1] ([358]Fig. 7A-[359]F, [360]S22C), we reasoned
   that stromal cells that could not respond to Amphiregulin would be
   compromised in their ability to respond fully to other cytokines made
   by mAT Th2[RM] cells. Consistent with a role for Amphiregulin in
   modulating mAT, we found fewer stromal cells and immune cells in the
   mAT of Egfr^fl/fl-Pdgfra-Cre mice than in control mice ([361]Fig.
   7G-[362]I). Further, EGFR deficient mAT stromal cells exhibited
   diminished responsiveness to Amphiregulin ex vivo, failing to become
   metabolically activated in response to this cytokine, and making less
   TSLP, confirming that major Amphiregulin effects on stromal cells are
   mediated through EGFR ([363]Fig. S22H, [364]I). While EGFR is the
   receptor for Amphiregulin, it is additionally the receptor for EGF,
   HB-EGF, TGFA, EREG and BTC, but Amphiregulin was the most expressed
   EGFR ligand in the mAT stromal vascular fraction during infection
   ([365]Fig. S23A, [366]B), and in this setting Th2 cells were a major
   source of this cytokine ([367]Fig. S7L).

   To gain insight into the role of EGFR in mAT stromal cell biology
   during infection, we performed RNAseq on mAT stromal cells isolated
   from Egfr^fl/fl-Pdgfra-Cre mice. We confirmed that Egfr expression was
   decreased in stromal cells from Egfr^fl/fl-Pdgfra-Cre mice compared to
   Egfr^fl/fl controls ([368]Fig. 7J). Lack of EGFR resulted in reduced
   expression of genes characteristic of MPC, including Dpp4 and Cd55, and
   the concomitant upregulation of genes that mark committed subsets (e.g.
   Apoe) ([369]16, [370]17) ([371]Fig. 7J). Moreover, EGFR-deficient mAT
   stromal cells from infected mice produced less TSLP and pCol1 ex vivo
   than did stromal cells from infected controls ([372]Fig. 7K,
   [373]S23C), and Egfr^fl/fl-Pdgfra-Cre mice had fewer Th2 cell in mAT
   ([374]Fig. S23D-[375]F) and were more susceptible to H. polygyrus
   infection ([376]Fig. 7L, [377]M). In contrast, deletion of EGFR in
   mature adipocytes in Egfr^fl/fl-Adipoq-Cre mice had no effect on
   susceptibility to H. polygyrus ([378]Fig. S23G, [379]H). Together,
   these data provide an example of how a Th2 derived cytokine can
   influence mAT stromal cells and underscore the role of Amphiregulin
   through EGFR signaling for the maintenance of MPC and immunity against
   an intestinal parasite.

Discussion

   Here, we showed that the mAT response to an enteric parasitic infection
   was marked by coordinated, interactional changes in the immune and
   stromal compartments. Most strikingly, a population of Th2[RM] cells
   expanded and permanently dominated the mAT lymphocyte niche. This
   Th2[RM] population, in response to signals from stromal cells, produced
   activating cytokines that drove the functional reprograming of the
   stroma that was important for resistance to infection.

   Pathways of adipocyte stromal cell differentiation are defined in
   detail, but how these are modulated by physiological perturbations has
   mostly been studied in the context of obesity ([380]13, [381]16,
   [382]18, [383]43). We found that the Th2 response associated with H.
   polygyrus infection had a significant impact on the composition of mAT
   stromal cell populations. This was characterized by a shift towards
   Dpp4^+ Pi16^+ CD55^+ MPC, which became the primary producers of
   cytokines and collagen, and which had the ability to support Th2[RM]
   activation and survival. Dpp4^+ Pi16^+ stromal progenitors are a
   universal reservoir population containing cells capable of giving rise
   to differentiated fibroblast subsets ([384]17) and adipocytes ([385]16,
   [386]18). These cells are localized to interstitial tissue, which is a
   fibroelastic connective tissue that envelops internal organs, blood
   vessels (where it is called the adventitia) and which is also found
   beneath the skin (where it is referred to as the fascia) ([387]51).
   Evidence suggests continuity of interstitial tissue within and between
   organs, possibly providing a pathway for cell and antigen movement that
   is alternative to vascular or lymphatic routes ([388]47, [389]52,
   [390]53).

   Studies in the skin show that mobilization of cells within the fascia
   is critical for wound healing ([391]54). Recent work exploring changes
   in stromal cells and immune cells in the skin after transient postnatal
   Treg cell depletion identify the accumulation of stromal cells with the
   characteristics of MPC in fibrous bands in the skin and the parallel
   accumulation of persistent Th2-cells in the tissue, both of which
   associate with skin healing ([392]20). The authors refer to these
   stromal cells as Th2-interacting fascial fibroblasts (TIFFs), linking
   previous evidence that cells with MPC characteristics are specialized
   for supporting type 2 immune responses ([393]19, [394]55) with the
   localization of IL-33 producing cells to the adventitia and fascia
   ([395]13, [396]16, [397]17, [398]19). Our findings are consistent with
   these reports and taken together point to a general phenomenon in which
   expansion and activation of the MPC population is a hallmark of the
   physiological response to type 2 inflammation that is shared across
   tissues. We speculate that the benefit of such a response is linked to
   the plasticity of this stromal cells subset to assume new supportive
   functions in response to signals received from the immune system. In
   the case of infection with H. polygyrus, we favor the view that the
   accumulation of MPC reflects a block in their differentiation into
   committed preadipocytes, a process that could contribute to the
   reduction in AT mass associated with infection. Alternative
   explanations for the accumulation of MPC are that adipocytes
   de-differentiate into these cells during infection ([399]56, [400]57),
   or that cells of this type migrate into mAT from other tissues. The
   reported continuity of the reticular interstitium between tissues
   ([401]47) would provide a pathway for MPC to move in this way. This
   anatomical link provides a possible connection between immunological
   events in the mAT and adjacent mLN and intestine, suggesting that
   infection driven activation of interstitial niches could facilitate
   cell migration in order to support immunity. Our imaging results, which
   showed the presence of MPC at the interface between the mAT and the
   small intestine, are consistent with this being the case. Thus, the
   interstitial space within mAT is emerging as a specialized location for
   interactions between immune cells and stromal cells.

   Similar to immune cells, stromal cells can engage different metabolic
   modules depending on their differentiation stage and function. ECM
   remodeling regulates glucose metabolism ([402]58), but how these cells
   adapt metabolically to inflammation is not well understood ([403]59).
   We found that metabolic activation of mAT stromal cells was a hallmark
   of the response to H. polygyrus infection and critical for increased
   ECM production. Metabolic activation in stromal cells resulted in
   increased incorporation of glucose into serine and glycine, and in
   increased ATP levels. This is consistent with a requirement for glycine
   to support increased synthesis of collagens, in which ~30% of amino
   acids are glycine, and for increased cellular ATP to meet the energetic
   demands of protein translation in cells which have assumed a highly
   biosynthetic role, producing ECM, cytokines and chemokines for export
   following stimulation by Th2-derived cytokines. As for collagen
   synthesis and cytokine production, metabolic changes were more
   pronounced in mAT stromal cells from mice responding to a secondary
   versus primary infection. This raises the possibility that stromal
   cells may exhibit innate memory analogous to that described for innate
   immune cells, in which primary exposure to a stimulus results in
   metabolism-dependent epigenetic changes around responsive genes that
   allow the cells to respond more strongly upon secondary stimulation
   ([404]60, [405]61). Alternatively, this may reflect ongoing effects of
   stronger in vivo stimulation related to larger numbers of Th2[RM] cells
   in the mAT. The presence of Th2[RM] cells in gAT from infected mice, in
   the absence of activated stromal cells, and the dependence of stromal
   cells activation on Th2 cells, suggests that the activation status of
   Th2 [RM] cell is dictated by the anatomical proximity to the site of
   infection and related to stromal cells activation.

   mAT activation in infected mice shares some features, including the
   activation of stromal cells to make Col1A1, with the creeping fibrotic
   mAT of Crohns disease. Creeping fat serves to prevent the systemic
   spread of intestinal bacteria which translocate across the gut wall due
   to loss of epithelial integrity associated with Crohns disease
   ([406]62). H. polygyrus are not thought to penetrate the gut wall, but
   based on the creeping fat model we speculate that increased collagen
   deposition within mAT may reflect a defensive process aimed at
   increasing the strength and resilience of the intestine and its
   associated vasculature to minimize the possibility and consequences of
   perforation. Despite the presence within the mAT of CD9^+ matrix
   fibroblasts, which strongly expressed collagen genes, the MPC were the
   main source of pCol1 protein during infection, and also appeared to be
   the dominant producers of Fibrinogen (Fn1), Fibrillin (Fbn1) and
   UDP-glucose 6-dehydrogenase (Udgh), which together are critical
   components of the ECM. Increased mechanical stiffness of the ECM can
   potentiate cell migration ([407]63, [408]64) and therefore may also
   play a role in Th2 cell and other immune cell movement into and through
   the interstitial spaces. Moreover, it can influence stem cell fate
   determination ([409]65), so it is feasible that the changes in ECM
   during infection observed here could contribute to the accumulation of
   MPC. Perhaps related to this, MPC identity is maintained by TGFβ[1]
   ([410]16), and in this context it is notable that mAT Th2[RM] cells
   produced this cytokine.

   Persistence of a large population of mAT Th2[RM] cells almost a year
   post-clearance of infection was striking and consistent with reports of
   the longevity of lung Th2[RM] cells ([411]66). Th2[RM] cells within mAT
   expressed Arg1, Nmur1 and Calca, which have previously been considered
   to be primarily expressed by ILC2 ([412]67–[413]70). This pattern of
   gene expression, together with their ability to become activated
   directly by cytokines, supports the view that innate reprograming of
   Th2 cells is an integral part of terminal differentiation driven by
   exposure to tissue-derived cytokine such as TSLP and IL-33 ([414]34,
   [415]40, [416]71). Similarities between mAT Th2 cells and mAT ILC2
   indicate that tissue-residency is promoting convergent transcriptional
   patterns in different type 2 immune cells, an idea already seen with
   similarities between Th2 cells and ILC2 in lungs and small intestine
   ([417]34). In this context, it is of interest that interactions between
   ILC2 and stromal cells, like those shown here for Th2 cells and stromal
   cells, also occur within the adventitia ([418]12, [419]19).

   In addition to classic type 2 cytokines, mAT Th2[RM] cells also
   produced Amphiregulin, a cytokine previously implicated in immunity to
   intestinal helminths ([420]48, [421]49). Additionally, Amphiregulin
   produced by Treg cells has been linked with tissue repair ([422]72),
   but our data indicate that during and after H. polygyrus infection such
   reparative role could be falling largely to the persistent Th2[RM]
   population. Relatively little is known of roles for Amphiregulin and
   EGFR signaling in AT physiology. However, previous work indicated that
   autocrine signaling via EGFR in Th2 cells is important for immunity to
   intestinal nematodes, because coordinated signaling through EGFR and
   the IL-33 receptor is essential for Th2 cells to make IL-13 ([423]49).
   We observed that in addition to this pathway, Amphiregulin was able to
   drive TSLP production by mAT stromal cells. This crosstalk between Areg
   and TSLP production could stabilize the lymphoid niche within the
   tissue and therefore have implications for the persistence of Th2[RM]
   cells in mAT. The fact that Amphiregulin can release TGFβ[1] from
   latent TGFβ[1] through integrin-av activation ([424]73) indicates that
   polyfunctional Th2[RM] cells capable of making both cytokines may be
   particularly potent sources of active TGFβ[1.] In our experiments,
   synergy between Amphiregulin and TGFβ[1] was apparent in vitro, where
   Amphiregulin was able to prime stromal cells for enhanced metabolic
   responses to TGFβ[1]. This is consistent with previous reports of cross
   talk between EGFR and TGFβ[1]-induced events in the development of
   kidney fibrosis ([425]74). Deletion of Egfr in stromal cells emphasized
   the importance of Amphiregulin signaling for modulating stromal
   vascular fraction cellularity in mAT, and for protective immunity
   against an enteric infection. Our results support an emerging view of
   EGFR signaling in stromal cells in providing a critical component of
   tissue immunity against helminth infection. While our experiments did
   not allow identification of stromal cells specifically in mAT as those
   critical for immunity, we found MPC at the interface between mAT and
   small intestine in what we believe are conduits between the two organs,
   and adjacent to granulomas. This raises the possibility that these
   cells can directly contribute to strengthening of the intestinal wall
   and supporting granuloma formation.

   Our findings on mAT contribute to the growing realization that AT can
   provide help to tissues fighting infection or recovering from wounding
   ([426]26, [427]75–[428]77). These findings warrant broader
   consideration of the function of AT during disease. The extent to which
   changes in populations of resident immune cells affect the helper
   activity of AT has been unclear, but our findings indicate that this
   may be of major significance since immune cells and AT stromal cells
   have evolved powerful dynamic mechanisms for reciprocal activation and
   regulation.

Materials and Methods

Study design

   The objective of this study was to investigate the role of mAT
   remodelling that is driven by intestinal infection with H. polygyrus in
   mice. We studied changes in immune and stromal cell populations using
   RNAseq and scRNAseq technologies, flow cytometry, ELISA based
   measurements, metabolic profiling, and immunofluorescence and
   immunohistology imaging. Studies in which primary H. polygyrus
   infection was analysed had an endpoint between days 11 and14, which
   permitted the assessment of adult worm burden and adaptive immune
   responses. To study memory responses, H. polygyrus infection was
   repeated 5 weeks after the clearance of initial infection and analysis
   was again performed between day 11 and 14. We used age- and sex-matched
   mice. Group sizes were determined on the basis of experimental purposes
   using previous experience or estimated on the basis of preliminary
   data. This study was not blinded. Sampling and experimental replicates
   are indicated in the figure legends.

Mouse models

   C57BL/6J, (JAX:000664), Balb/cJ (JAX:000651), Cd4-Cre (JAX: 022071),
   Pdgfra-Cre (JAX:013148), Il4eGFP Foxp3RFP Il10Bit (generated by
   crossing Il4^tm1Lky ([429]78), Foxp3^tm1Flv ([430]79) and
   Tg(Il10-Thy1a) ([431]80) mice), Il4ra^tm2Fbb (IL-4Rfl/fl) ([432]81),
   Egfr^tm1Dwt (EGFRfl/fl) ([433]82), B6(129S4)-Il33^tm1.1Bryc/J (JAX:
   030619), B6;FVB-Tg(Adipoq-cre)1Evdr/J (JAX: 028020) mice were used.
   Mice were maintained at the Washington University School of Medicine in
   St. Louis, Max Planck Institute for Immunobiology and Epigenetics or at
   the Bloomberg-Kimmel Institute for Cancer Immunotherapy at Johns
   Hopkins. All corresponding animal protocols were approved by the animal
   care committee of the Regierungspraesidium Freiburg, the Animal Care
   and Use Committee (ACUC) of Washington University in St. Louis, or the
   ACUC of Johns Hopkins University. Mice were bred under specific
   pathogen–free standards. Animals used for tissue harvest or
   experimental procedures were aged between 7 to 12 weeks at the start of
   the experiment and were aged and sex-matched. Both female and male mice
   were used in the study.

Experimental infections and interventions

   Heligmosomoides polygyrus bakeri L3 stage larvae were prepared at the
   U.S. Department of Agriculture (Beltsville, USA) ([434]83). For H.
   polygyrus infection, mice were each gavaged with 200 L3 stage larvae in
   PBS. For primary infection, mice were left for 11–14 days before being
   sacrificed or treated with the anthelminthic pyrantel pamoate
   (1mg/mouse). For secondary infection, mice were infected at 5 weeks
   post treatment, and sacrificed 11–14 days later. For H. polygyrus egg
   and adult worm counts small intestines were removed, opened
   longitudinally, and placed into a mesh cloth on top of a 50 ml tube
   filled with PBS for 3–4 h in a 37°C water bath. Adult parasites dropped
   through the mesh into the tube and were recovered for counting on a
   dissecting microscope. Parasite eggs were isolated by floatation on
   saturated sodium chloride from caecal contents collected from
   individual mice, and counted under a microscope. For CD4^+ T cell
   depletion, mice were infected and treated, and 5 weeks later injected
   with anti-CD4 monoclonal antibody (mAb, clone GK1.5, BioXCell, 500 μg
   /mouse i.p. per injection) one day before 2º infection, and then again
   at day 6 after infection. Mice were sacrificed on day 11 of secondary
   infection. For IL-7Rα blockade, mice were infected and treated, and
   injected with anti-IL-7Rα mAb (clone A7R34, BioXCell, 500 μg/mouse i.p.
   per injection) on days 3 and 9 post treatment, and sacrificed on day 12
   post treatment. For treatment with FTY720 (Enzo, BML-SL233–0005), mice
   were infected and treated, and 5 weeks later injected with FTY720 every
   second day, starting from one day before secondary infection (6
   injections in total, 10 μg/mouse i.p. per injection). Mice were
   sacrificed on day 11 of secondary infection. For Areg (R&D System,
   989-AR) treatment, naïve mice were injected 3 times with 10 μg/mouse
   Areg i.p. on days 0, 3, and 6 and were sacrificed on day 9. For IL-33
   (Biolegend, #580504) treatment, naïve mice were injected 3 times on 3
   consecutive days with 1μg/mouse IL-33 i.p. and sacrificed on day 9. For
   the magnetic resonance imaging (MRI), an EchoMRI 3-in-1 Body
   Composition Analyzer was used according to the manufacturer’s
   instructions. To assess food intake, mice were housed 3 per cage and
   chow consumption per day was estimated based on the weight of chow
   remaining in the food hopper, and divided by the number of mice in the
   cage to obtain chow consumption /day/mouse measurements. High fat diet
   experiments utilized chow containing 60% fat (Research Diets,
   #D12492i).

Cell isolation and culture

   Stromal cells and immune cells were isolated from tissues obtained from
   euthanized and PBS-perfused animals. For isolation from mAT, gAT and SI
   LP, enzymatic tissue digestion was performed. To purify Th2 cells
   (TCRβ^+ CD4^+ IL4-eGFP^+ FOXP3-RFP^-), non-Th2/ non-Treg cells (CD45^+
   TCRβ^+ CD4^+ IL4-eGFP^- FOXP3-RFP^-) and stromal cell subpopulations
   (based on Ly6c and CD9 expression) cells were isolated via
   fluorescence-activated cell sorting, excluding dead cells and doublets.
   Total mAT and gAT stromal cells were isolated using the Adipose Tissue
   Progenitor Isolation Kit (Miltenyi Biotec, #130–106-639). Stromal cells
   were plated at 2–4×10^5 cells/well and cultured overnight. Supernatants
   were collected and cytokines or pCol1 measured by ELISA. In experiments
   where effects of cytokines on stromal cell biology or adipogenic
   potential of mAT stromal cell subpopulations were assessed, adipogenic
   factors were added to mimic adipose tissue conditions. For details of
   cell isolation and culture see [435]Supplementary Methods.

Flow cytometry

   For analysis of intracellular cytokine production cells were
   re-stimulated for 4 hours with 0.1 μg/ml Phorbol 12-myristate 13-
   acetate (PMA), 1 μg/ml Ionomycin and 10 μg/ml Brefeldin A. Cells were
   surface stained with mAbs diluted in PBS/0.1% BSA and Fc-block
   (Biolegend) for 30 min on ice. Fixable Viability Dye (eBioscience) was
   added to allow the exclusion of dead cells. For intracellular staining,
   cells were fixed and permeabilized using the FOXP3/transcription factor
   staining kit (eBioscience). The list of antibodies and more details on
   the flow cytometry protocols can be found in [436]Supplementary
   Methods.

Microscopy

   Hematoxylin and eosin (H&E), Masson’s trichrome or CD55/CD4 multiplex
   immunostaining were performed on formalin-fixed, paraffin embedded
   tissue sections. Images were acquired using a Zeiss Axio Imager Apotome
   or Hamamatsu NanoZoomer S210 Digital slide scanner. For
   immunofluorescent whole mount staining for CD3, CD4, CD31, DPP4, Pi16,
   GATA3, Perilipin1, PDGFRα, Collagen 1 and IL-33 mAT samples were
   formalin fixed, permeabilized in 1% Triton X-100 in PBS (Sigma),
   blocked with 2.5% BSA, 0.5% Triton X-100 in PBS and incubated with
   primary antibodies, followed by staining with secondary antibodies and
   nuclear staining (Hoechst 33342). Samples were mounted with ProLong^®
   Diamond Antifade Mountant (Thermofisher). Confocal images were acquired
   using a Zeiss spinning disk confocal microscope equipped with a
   Photometrics Prime BSI camera and Apochromat objectives or with Zeiss
   AxioObserver inverted microscope with LSM800 confocal module. For
   details see [437]Supplementary Methods.

RNA sequencing

   Ambion’s RNAqueous Micro Kit (Cat 1931) or Qiagen RNeasy kit (#75144)
   were used for bulk RNA isolation from sort-purified PDGFRα^+ Sca1^+
   stromal cells from C57BL/6J mice ([438]Fig. 1I, [439]J) and PDGFRα^+
   Sca1^+ stromal cells from Egfr^fl/fl-Pdgfra-Cre mice ([440]Fig. 7J).
   RNA isolation from granulomas and unaffected small intestinal tissues
   ([441]Fig. S19) was done using oligo-dT beads (Invitrogen). scRNA
   sequencing of sort-purified mAT stromal vascular fraction cells was
   performed using a 10X Genomics Chromium Controller. scRNA sequencing of
   sort-purified CD4^+ TCRβ^+ T cells from mLN, mAT and small intestine
   lamina propria was performed using the CEL-Seq2 method ([442]84) with
   modifications as described ([443]85). Details for RNA sequencing and
   scRNA sequencing are included in [444]Supplementary Methods.

Metabolic Profiling

   Oxygen consumption rates and extracellular acidification rates were
   measured in XF media under basal conditions and in response to 1 μM
   oligomycin, 1.5 μM fluoro-carbonyl cyanide phenylhydrazone (FCCP) and
   100 nM rotenone + 1μM antimycin A using a 96 well XF or XFe
   Extracellular Flux Analyzer (EFA) (Seahorse Bioscience). For
   stable-isotope tracing, mAT stromal cells isolated from naïve mice were
   cultured in adipogenic differentiation conditions for 3 days and during
   the final 6 h (^12C)-glucose was replaced with uniformly heavy labeled
   (^13C) glucose. Cell metabolites were extracted using 70 μL extraction
   buffer (50:30:20, methanol: acetonitrile: water). Metabolite
   measurements were LC-MS utilizing an Agilent 1290 Infinity II UHPLC in
   line with a Bruker Impact II QTOF operating in negative ion mode.
   Metabolites were quantified using AssayR ([445]86), and identified by
   matching accurate mass and retention time to standards. For details see
   [446]Supplementary Methods.

Quantification and statistical analysis

   With the exception of scRNASeq and RNASeq datasets, statistical
   analyses were performed using Prism 7 software (GraphPad). Comparisons
   for two groups were calculated using rank Mann–Whitney or unpaired t
   tests. Comparisons of more than two groups were calculated using
   one-way ordinary unpaired ANOVA with Bonferroni’s multiple comparison
   tests. Tests were two-sided and α = 0.05 (95% confidence interval).

Supplementary Material

   main supplementary materials

   Fig. S1. H. polygyrus infection model

   Fig. S2. Marker gene expression in mAT stromal vascular fraction
   scRNAseq data.

   Fig. S3. ILC2, Treg and Th2 cells in mAT stromal vascular fraction
   scRNAseq data.

   Fig. S4. Characteristics of T cells in mLN, mAT and gAT in infection.

   Fig. S5. Localization of Th2[RM] cells in mAT in H. polygyrus
   infection.

   Fig. S6. mAT interstitium is enriched in T cells in infection.

   Fig. S7. Th2[RM] cell cytokine production in H. polygyrus infection.

   Fig. S8. mAT Th1 and Treg cells in H. polygyrus infection.

   Fig. S9. Integrin expression on mAT Th2[RM] cells in infection.

   Fig. S10. Interdependence of mAT Th2[RM] and stromal cells.

   Fig. S11. mAT stromal cells support Th2[RM] population.

   Fig. S12. Characterization of mAT stromal cell populations in H.
   polygyrus infection.

   Fig. S13. IL-33 expression in mAT stromal cell populations.

   Fig. S14. Cytokine and ECM gene expression in mAT stromal cell
   populations in H. polygyrus infection.

   Fig. S15. Interstitial spaces are enriched in Collagen 1 in mAT of H.
   polygyrus infected mice.

   Fig. S16. Cd55, Pi16, Dpp4 mark mAT MPC population.

   Fig. S17. DPP4^+ MPC localize to the mAT interstitium.

   Fig. S18. Pi16^+ MPC and CD4^+ T cells localize to the mAT interstitium
   in the H. polygyrus infected mice.

   Fig. S19. DPP4^+ MPC and Th2 cells localize to the mAT interstitial
   spaces in the H. polygyrus infected mice.

   Fig. S20. CD55^+ MPC and CD4^+ T cells localize to the interface
   between mAT and small intestine during H. polygyrus infection.

   Fig. S21. ECM gene expression in the small intestine granulomas in H.
   polygyrus infection.

   Fig. S22. Amphiregulin supports mitochondrial respiration and TSLP
   production in mAT stromal cells.

   Fig. S23. EGFR signalling in mature adipocytes is not required for host
   protection in H. polygyrus infection.
   [447]NIHMS1861112-supplement-main_supplementary_materials.pdf^ (23.9MB,
   pdf)
   raw data

   Table S1 (Microsoft excel format). Raw data.
   [448]NIHMS1861112-supplement-raw_data.xlsx^ (71KB, xlsx)

Acknowledgments: