BASIC RESEARCH www.jasn.org

Gut Microbial Metabolites Induce Donor-Specific Tolerance of Kidney Allografts through Induction of T Regulatory Cells by Short-Chain Fatty Acids

Huiling Wu,1,2,3 Julian Singer ,1,2 Tony K. Kwan ,1,2 Yik Wen Loh,1,2 Chuanmin Wang,1,2 Jian Tan ,2,4 Yan J. Li ,1 Sum Wing Christina Lai ,1 Laurence Macia,4,6 Stephen I. Alexander,2,5 and Steven J. Chadban1,2,3

Due to the number of contributing authors, the affiliations are listed at the end of this article.

ABSTRACT Background Short-chain fatty acids derived from gut microbial fermentation of dietary fiber have been shown to suppress autoimmunity through mechanisms that include enhanced regulation by T regulatory cells (Tregs). Methods Using a murine kidney transplantation model, we examined the effects on alloimmunity of a high- fiber diet or supplementation with the short-chain fatty acid acetate. Kidney transplants were performed BASIC RESEARCH from BALB/c(H2d)toB6(H2b) mice as allografts in wild-type and recipient mice lacking the G protein– coupled receptor GPR43 (the metabolite-sensing receptor of acetate). Allograft mice received normal chow, a high-fiber diet, or normal chow supplemented with sodium acetate. We assessed rejection at days 14 (acute) and 100 (chronic), and used 16S rRNA sequencing to determine gut microbiota composition pretransplantation and post-transplantation. Results Wild-type mice fed normal chow exhibited dysbiosis after receiving a kidney allograft but not an isograft, despite the avoidance of antibiotics and immunosuppression for the latter. A high-fiber diet prevented dysbiosis in allograft recipients, who demonstrated prolonged survival and reduced evidence of rejection compared with mice fed normal chow. Allograft mice receiving supplemental sodium acetate exhibited similar protection from rejection, and subsequently demonstrated donor-specific tolerance. 1 Depletion of CD25 Tregs or absence of the short-chain fatty acid receptor GPR43 abolished this survival advantage. Conclusions Manipulation of the microbiome by a high-fiber diet or supplementation with sodium acetate modified alloimmunity in a kidney transplant model, generating tolerance dependent on Tregs and GPR43. Diet-based therapy to induce changes in the gut microbiome can alter systemic alloimmunity in mice, in part through the production of short-chain fatty acids leading to Treg cell development, and merits study as a potential clinical strategy to facilitate transplant acceptance.

JASN 31: 1445–1461, 2020. doi: https://doi.org/10.1681/ASN.2019080852

Kidney transplantation remains the best treatment Received August 28, 2019. Accepted March 22, 2020. for end-stage kidney failure, but life-long immuno- H.W. and J.S. contributed equally to this work. suppression is required to prevent rejection.1 Cur- Published online ahead of print. Publication date available at rent immunosuppressive strategies remain only www.jasn.org. partially effective, increase recipient susceptibility Correspondence: A/Prof. Huiling Wu, Kidney Node Laboratory, to infection and cancer, and paradoxically contrib- Charles Perkins Centre D17, The University of Sydney, Sydney, ute to premature graft failure.2 Establishing donor- New South Wales, 2006 Australia. Email: huiling.wu@sydney. specific allograft tolerance while maintaining edu.au adequate immunity to protect against infection Copyright © 2020 by the American Society of Nephrology

JASN 31: 1445–1461, 2020 ISSN : 1046-6673/3107-1445 1445 BASIC RESEARCH www.jasn.org and cancer would limit the burden of current treatment, but is Significance Statement only rarely achieved in humans. With growing knowledge of the gut microbiomes capacity to influence systemic host im- The gut microbiome is known to affect immune responses in au- mune responses,3 modulation of the gut microbiota, such as toimmunity and cancer, but little is known about its role in trans- by diet, offers a novel pathway to favorably influence the host plant immunity. In a mouse model, the authors observed dysbiosis after kidney transplantation in the absence of antibiotics or other response to alloantigens. drugs. A high-fiber diet prevented dysbiosis and afforded pro- The clinical relevance of the gut microbiota in human im- tection against allograft rejection, as did supplementation with the munity has been demonstrated most profoundly in cancer short-chain fatty acids sodium acetate or sodium butyrate, micro- studies, where the antitumor response of immune check- bial metabolites produced by gut fermentation of dietary fiber. This – point inhibition was revealed to be dependent on specific protection was dependent on the G protein coupled receptor GPR43 and T regulatory cells. These findings show how the mi- gut microbiota, with nonresponsive mice converted to a re- crobiome can be modified to retard alloimmunity in a mouse model – sponder phenotype by gut microbiota transfer.4 6 In experi- of kidney transplantation, and provide a rationale to explore this mental autoimmunity, spontaneous development of diabetes strategy in humans as a means to facilitate transplant acceptance. in a nonobese diabetic mouse model of type 1 diabetes re- quired both a proinflammatory gut microbiota and an intact kidney transplantation model, and explored the role of the toll-like receptor signaling cascade to facilitate development of SCFA acetate and its metabolite-sensing receptor GPR43 in insulitis,7 with susceptibility transferrable from affected to un- mediating this response. affected mice by microbiota transfer.8 Kidney allograft rejec- tion shares common pathways to those causing insulitis in nonobese diabetic mice, including the toll-like receptor METHODS 4–MyD88 pathway that we have previously reported to be re- quired for the development of ischemia reperfusion injury9 Study Design and full development of acute allograft rejection in experi- Animals mental kidney transplantation.10,11 Male BALB/c (H2-Kd) kidney donor, B10Br (H2-Kk)skindo- Several experimental models of allograft rejection have nor, and C57BL/6 (H2-Kb) recipient mice were obtained from 2 2 demonstrated altered rejection kinetics on the basis of micro- the Animal Resource Centre (Perth, Australia). GPR43 / biota composition; however, results have been variable. mice on C57BL/6 backgrounds have been described,18 and Survival of skin allografts in C57BL/6 mice differs between were obtained from our animal facility. Life-sustaining kidney vendors and is both microbiota dependent and transferra- transplants were performed on wild-type (WT) C57BL/6 and 2 2 ble.12 Delayed rejection kinetics of cardiac allografts are seen GPR43 / mice who received a kidney from a BALB/c donor in both gnotobiotic and antibiotic pretreated mice.13 How- as allografts or from a C57BL/6 donor as isografts. Male mice ever, similar antibiotic depletion of gut microbiota accelerated aged 10–16 weeks were used in all experiments and were rejection of murine aortic allografts,14 suggesting greater com- housed in a specific pathogen-free facility within the Univer- plexity to the microbiota–immunity axis. sity of Sydney. Animal care and experiments were conducted These findings suggest that the gut microbiota is likely to following established guidelines for animal care and were ap- play a significant role in the immune responses that determine proved by the Animal Ethics Committee of the University of the fate of a kidney allograft: tolerance or rejection. Several Sydney. potential mechanisms exist: (1) interaction between gut or- ganisms, their metabolites, and immune cells in the gut wall, Survival Experiment mediated by pathogen-associated molecular pattern/innate WT1HF allografts (n528), WT1 sodium acetate (SA) allo- receptor interactions15;(2) disruption of gut homeostasis by grafts (n527), and WT1 sodium butyrate (SB) allografts microbiota leading to an inflammatory milieu that alters im- (n521) were compared with WT allografts (n521) and iso- mune cell maturation16;and(3) production of metabolites, grafts (n513) to establish a survival curve up to day 100 post- including short-chain fatty acids (SCFAs) that modulate cell transplant. function through specificGprotein–coupled receptors For allograft recipients surviving to day 200, a proportion (GPRs)17 or through inhibition of histone deacetylase of WT1SA allografts (n54) and WTallografts (n54) received (HDAC) activity, affecting gene transcription. Diet remains the full-thickness skin grafts from three sources: BALB/c (H2-Kd largest exogenous determinant of gut microbiota composition, kidney donor–matched skin allograft), C57BL/6 (H2-Kb syn- and can predictably and sustainably alter each of the above geneic), and B10Br (H2-Kk third-party allografts). Skin grafts mechanisms. In particular, high intake of dietary fibers, which were observed regularly for signs of rejection up to day 100 are in turn fermented by colonic bacteria to form SCFAs, has after skin transplantation. shown promising outcomes in experimental autoimmunity.18,19 Here, we examined the effects of a high-fiber(HF)dieton Time-Course Study the gut microbiome and the alloimmune response to kidney To examine for intragraft events and mechanisms, groups of transplantation, using a fully MHC-mismatched, murine mice were euthanized early (day 14) and late (day 100) to

1446 JASN JASN 31: 1445–1461, 2020 www.jasn.org BASIC RESEARCH investigate for evidence of acute and chronic rejection, Mouse fecal samples were collected under sterile conditions respectively. immediately after extrusion, frozen on dry ice after retrieval, and stored at 280°C. Diets and Acetate Treatments HF (SF11–029) enriched in guar gum and cellulose, zero fiber Bacteria 16S rRNA Sequencing and Bioinformatics (SF09–28), and normal mouse chow (irradiated mouse chow) DNA was extracted from feces collected under sterile condi- were purchased from Specialty Feeds (Glen Forrest, WA, Aus- tions with the QIAamp DNA Stool Mini Kit (QIAGEN) ac- tralia) (Supplemental Table 1). Mice received normal chow cording to the manufacturer’s instructions. DNA was (NC) upon arrival at our facility, followed by specificdiets sequenced using tagged amplicons spanning the V4 region ad libitum commencing 2 weeks before and continuing of bacterial 16S rRNA gene (515f/806r) on the Illumina MiSeq throughout experiments. platform (23250 bp) at the Ramaciotti Centre for Genomic The dose of SA and butyrate was adapted from Andrade- (University of New South Wales, Sydney, NSW, Australia). Oliveira et al.20 Briefly, 150 mM of either SA or SB solution Data were deposited in the European Nucleotide Archive un- (Sigma-Aldrich) was administered to mice as an intraperito- der accession number PRJEB34109. Demultiplexed reads were neal injection at a dose of 200 mg/kg, 30 minutes before and processed using the QIIME 1.9.1 software.22 Briefly, paired- immediately after transplantation, and then daily for 2 weeks. end reads were joined using the fast-q algorithm with no mis- SA (150 mM) or SB (100 mM) were then administered in their matches allowed. Operational taxonomic units (OTUs) were drinking water ad libitum until 100 days post-transplantation. picked using 97% similarity with taxonomy assigned using the Greengenes v13_8 database, and de novo OTU picking was 1 CD25 Cell Depletion Study performed on sequences that did not match the reference da- 1 1 To deplete CD4 CD25 cells in vivo, mice received rat anti- tabase. Chimeric sequences were identified using ChimeraS- mouse CD25 mAb (clone PC-61, rat IgG1; BioXcell, West layer, and OTUs with total abundance of ,0.01% were filtered Lebanon, NH) or isotype rat IgG, 0.5 mg per mouse, via in- from the table. Rarefaction analysis was used to compare traperitoneal injection on days 22 and 0 post-transplantation the adequacy of sequencing depth to ensure that the micro- 1 (n517). CD25 cell depletion was confirmed between days 14 biota populations were sufficiently sampled such that addi- and 21 after kidney transplantation, by staining white blood tional sampling would produce few additional OTUs cells with anti-CD25 (clone 7D4 or 3C7; BD Biosciences Phar- (Supplemental Figure 1). For a-diversity analysis, samples mingen, CA) as previously described.11,21 Survival was as- were rarefied to a read depth of 46,397 using the Calypso sessed to day 100 post-transplantation. workflow, which was also used to visualize the taxonomic make-up of microbial communities.23 The weighted UniFrac Kidney Transplantation distance matrix was used to compare differences in microbial Heterotopic kidney transplants and skin transplant were per- community composition between sample groups and visual- formed as previously described,11 and the detailed method is ized using a principal coordinate analysis plot,24 with the sig- described in the Supplemental Material. Briefly, the left kidney nificance in the divergence of community structure assessed of the donor animal was removed together with the ureter and by Adonis (9999 permutations). The differential abun- vessels en masse and placed in the left iliac fossa of the recipient dance of microbiota species in response to dietary change animal after an ipsilateral nephrectomy. No immunosuppres- and transplantation was determined using the Kruskal–Wallis sive therapy was administered. The recipient’s right native nonparametric test on libraries normalized by cumulative kidney was removed at day 3–7, rendering the graft to be sum scaling25 through Calypso, and also on complete libraries life-sustaining. Animals with technical graft failure or wound using the negative binomial DESeq226 model (R package, phy- infection became overtly ill (and were euthanized) or died loseq v1.29.0),27 with an false discovery rate–adjusted P value within 4 days of the contralateral nephrectomy and were re- ,0.01 for significance. moved from the study. Assessment of Kidney Function Skin Transplantation Serum creatinine was measured using the modified Jaffe rate Full-thickness tail skin grafts were placed on graft beds pre- reaction by the Biochemistry Department of The Royal pared on recipients at day 200 after kidney allograft transplan- Prince Alfred Hospital (Sydney, NSW, Australia). Total urinary tation. Grafts were covered with protective bandages for 7 protein was assessed by the Bradford method using a commer- days. Rejection was defined as graft necrosis of $90% of the cially available kit (Bio-Rad Laboratories, Gladesville, transplant skin area. Recipients that accepted their allograft Australia) according to the manufacturer’s instructions. demonstrated preserved graft size and hair growth. Histology Sample Harvest Periodic acid–Schiff or Picro-Sirius red staining was per- Blood, kidney tissue, and urine were harvested at day 14 and formed to assess tubulitis, glomerulosclerosis and interstitial day 100 after kidney transplant, as previously described.11,21 fibrosis, and interstitial collagen deposition. Previously

JASN 31: 1445–1461, 2020 Gut Microbiota and Kidney Rejection 1447 BASIC RESEARCH www.jasn.org described scoring systems11,28 were applied for each histologic multiple groups were compared using one- or two-way AN- parameter and have been included in the Supplemental OVA with post hoc Tukey correction, using GraphPad Prism Material. 7.0 software. A P value of ,0.05 was considered to be statis- tically significant. Data are presented as mean6SEM. Immunohistochemistry and Immunofluorescence The detailed method for immunohistochemistry and immu- nofluorescence for C4d, including antibody clones and quan- RESULTS tification, is included in the Supplemental Material. Briefly, acetone-fixed frozen sections were prepared, blocked, and ex- Transplantation-Associated Dysbiosis Was Prevented posed to a primary antibody for 60 minutes with by an HF Diet concentration-matched IgG used as an isotype-negative con- We first examined the effect of kidney transplantation, in the trol, followed by incubation with the appropriate biotinylated absence of antibiotics or immunosuppression, on the compo- secondary antibody or anti-rat IgG conjugated with Alexa sition of the gut microbiota and determined the effect of di- Fluor 488. Vector stain ABC kit (Vector Laboratories, Burlin- etary fiber content on this effect. WT mice were fed either NC game, CA) was applied to the tissue followed by or an HF diet enriched with guar gum and cellulose. After 3,39diaminobenzidine substrate-chromogen solution (Dako 2 weeks on their respective diet, groups of mice then received North America) and counterstained. Quantification was per- a kidney allograft (NC1Allo n510, HF1Allo n516) or iso- formed by scoring the number of positive cells in 20 consec- graft (n59), and were single-housed and observed on their utive high-power fields (HPFs), or by percentage of positive allocated diets for a further 2 weeks. Fecal samples were ob- staining per HPF using a digital image analysis program. tained immediately before transplantation and at 2 weeks post-transplantation to enable microbial analysis by sequenc- RNA Extraction and Complementary DNA Synthesis ing of the V4 region of the 16S rRNA gene. After quality trim- Total RNA was extracted from kidney tissue and cells using ming, filtering, and removal of nonbacterial sequences, a total TRIzol (Invitrogen, Mulgrave, VIC, Australia). Complemen- of 6,398,328 reads were generated from 68 fecal collections, tary DNA was synthesized using oligo d(T)16 (Applied Biosys- with an average of 94,093628,665 reads per sample. tems) primers and the SuperScript III Reverse transcription Gut microbiota a-diversity was evaluated using the non- kit (Invitrogen) as per the manufacturer’s instructions. parametric Shannon diversity index, reflecting both the rich- ness of species and their relative abundance. After transplant, HDAC Activity NC fed mice experienced a significant loss in microbial diver- To determine the activity of HDAC in kidney allograft tissue, sity (Shannon diversity index: NC, 4.0460.04; NC1Allo, equal quantities of nuclear fraction proteins (20.0 mg) were 3.6060.08; P,0.01), with isograft controls exhibiting a sim- analyzed using a Fluorometric HDAC Activity Assay Kit ilar although less pronounced change (Shannon diversity in- (k330–100; BioVision) as per the manufacturer’s instructions. dex: Pre-Isograft, 3.9360.12; isograft, 3.4560.12; P,0.01). In contrast, diversity was modestly expanded post-transplant in Real-Time PCR HF-fed mice (Shannon diversity index: HF, 3.5460.11; Specific TaqMan primers and probes for IFN-g,TNF,TGF-b, HF1Allo, 3.8860.05; P,0.01) (Figure 1A). The richness of IL-4, IL-6, IL-10, IL-12, IL-23, granzyme A, granzyme B, per- gut microbial communities was similarly reduced in NC-fed forin, CCL2, CCL5, CXCL9, CXCL10, indoleamine 2,3-diox- allograft and isograft recipients, yet remained unchanged in ygenase, MMP-2, MMP-9, TIMP-1, TIMP-2, fibronectin, and HF-fed allograft recipients (Supplemental Figure 2). glyceraldehyde-3-phosphate dehydrogenase (GAPDH) have The UniFrac distance matrix was used to compare the ef- all been described previously.9,11,29 Complementary DNA fects of both diet and transplantation on the community struc- was amplified in 13 Universal Master Mix (Applied Biosys- ture of the gut microbiota, and weighted UniFrac distances tems) with gene-specific primers and probes using either a were plotted using principal coordinate analysis, as shown in Rotor-Gene 6000 system (Corbett Life Science) or Roche Figure 1B. The fecal microbiota of mice clustered sepa- LightCycler 480 (Roche Applied Science, Penzberg, Ger- rately according to diet, with subsequent shifts in microbial many). Acquired data sets were analyzed using the accompa- community structure occurring in both groups after trans- nying Rotor-Gene 6000 Analysis Software v1.7 or LightCycler plantation. Adonis (9999 permutations) confirmed that both 480 SW v1.5.1, respectively. All results are expressed as a fold transplant and diet accounted for the significant variation in expression normalized to GAPDH expression. bacterial communities encountered between groups (R250.49; P,0.001). In comparison, NC-fed isograft controls Statistical Analyses exhibited no change in b-diversity after surgery (P50.14; Survival curves were constructed via the Kaplan–Meier Figure 1B), suggesting that shifts in the microbial landscape method, and comparisons between graft survival times were seen in allografts are predominantly driven by the host–donor analyzed with the log-rank test. Statistical differences between immune response, rather than by ischemia-reperfusion two groups were analyzed by unpaired, two-tailed t tests and injury.

1448 JASN JASN 31: 1445–1461, 2020 www.jasn.org BASIC RESEARCH

A B HF NC Pre-Iso C HF+Allo NC+Allo Isografts Verrucomicrobia Firmicutes Actinobacteria Proteobacteria Bacteroidetes Other 4.5 **** 0.04 4.0 0.02 3.5 0.00

3.0 Axis.2 [16.8%] Shannon Index −0.02 2.5

HF NC HF 0.00 0.02 NC −0.06 −0.04 −0.02 Pre-Iso HF+Allo NC+Allo Isograft Axis1[23.7%] NC+Allo HF+Allo Pre-Iso Isograft

D HF HF + Allo NC NC + Allo

Unclassified.Ruminococcaceae Oscillospira 010203040 Ruminococcus Sutterella Value Unclassified.YS2 Unclassified.Rikenellaceae Unclassified.Lachnospiraceae Lactobacillus Bacteroides Akkermansia Allobaculum Bifidobacterium Unclassified.S247 Unclassified.Clostridiales Prevotella

E Bifidobacterium F 20 ** NC vs. HF diet NC Allo vs. HF Allo 15 * o__RF39 Phylum g__Turicibacter Actinobacteria f__Peptococcaceae g__Rikenella 10 Bacteroidetes

Relative g__C. Arthromitus o__RF39 5 Deferribacteres

Abundance (%) f__Clostridiaceae Firmicutes g__Staphylococcus 0 f__Erysipelotrichaceae Proteobacteria HF HF+Allo NC NC+Allo g__C. Arthromitus Tenericutes g__Dorea f__Clostridiaceae Clostridiales Verrucomicrobia 18 o__RF32 f__Peptococcaceae g__Coprococcus g__Akkermansia 16 ** 14 Taxa g__Prevotella g__Ruminococcus 12 g__Ruminococcus f__Rikenellaceae Relative 10 g__Adlercreutzia o__Clostridiales Abundance (%) 8 g__Lactobacillus g__Bifidobacterium HF HF+Allo NC NC+Allo g__Bacteroides g__Parabacteroides logbaseMean g__Bifidobacterium Akkermansia muciniphila 4 g__Bacteroides g__Akkermansia g__Mucispirillum **** 6 15 ** g__Mucispirillum 8 g__Blautia 10 −5.0 −2.5 0.0 2.5 5.0 −20 −10 0 10 log2 fold change log2 fold change Relative 5

Abundance (%) 0 HF HF+Allo NC NC+Allo

Figure 1. HF diet altered the gut microbial community structure and prevented transplant-associated dysbiosis. Fecal DNA analysis was performed on mice fed an NC (n512) or HF (n512) diet for 2 weeks. Mice then received a kidney allograft and continued on a NC diet (NC1Allo, n510) or HF diet (HF1Allo, n516) with fecal microbiota analyzed 14 days post-transplant. (A) Shannon diversity index demonstrated a change in bacterial diversity from pre- to post-transplant, with a reduction in diversity in NC-fed mice after transplant (P,0.01) compared with an increase in HF-fed mice (P,0.01). Isograft controls fed NC pre- and post-transplant (n59 for both groups)

JASN 31: 1445–1461, 2020 Gut Microbiota and Kidney Rejection 1449 BASIC RESEARCH www.jasn.org

The relative abundance of the dominant bacteria at the rate–adjusted P value was ,0.01 (Figure 1F, Supplemental phylum level is shown in Figure 1C. Post-transplant, alter- Tables 2, 3, and 4). ations in the fecal microbiota were relatively consistent within the treatment groups, despite the single-housing of transplan- HF Diet Prolonged Kidney Allograft Survival ted mice (Figure 1, C and D). Bacteroidetes and Firmicutes To determine the effect of an HF diet on allograft survival, we were the dominant phyla both pre- and post-transplant, re- transplanted BALB/c (H2-Kd) kidneys into C57BL/6 (H2-Kb) gardless of diet. The next most-dominant phylum differed nephrectomized recipients as WT allografts, whose survival significantly according to diet. Verrucomicrobia (represented was dependent upon sustained function of the allograft. Re- by a single genus, Akkermansia muciniphila),wassignificantly cipients were fed HF or NC, commencing 2 weeks before expanded in NC-fed mice post-transplant. This pattern, and transplantation and continuing throughout the experiments. in particular the expansion of Verrucomicrobia seen post- Survival to 100 days was observed in all C57BL/6 isografts. WT transplant (mean relative abundance: NC, 0.3%60.11%; mice on NC diets rejected their allografts with a mean graft NC1Allo, 13.72%61.77%), closely resembled that seen in survival of 40 days, whereas WT1HF allografts displayed pro- nontransplant mice on a zero-fiber diet, previously reported longed survival (P,0.05; Figure 2A). as dysbiotic18,30 (Supplemental Figure 3). In contrast, HF-fed mice yielded only a small population of Verrucomicrobia (rel- HF Diet Protected against Acute Allograft Rejection ative abundance 2.78%61.09%), but displayed marked ex- As allograft rejection in this model characteristically occurs in pansion of Actinobacteria both pre- and post-transplantation the acute phase (day 14), and in surviving mice, the chronic (Figure 1, C and D). phase (day 100), further transplant experiments were under- Differential abundance testing revealed significant differ- taken to examine the effect of diet at these timepoints. ences in OTU population sizes between dietary groups, which At day 14, WTallografts exhibited kidney dysfunction, with was further modified after transplantation (Figure 1E). The a four-fold increase in serum creatinine as compared with iso- microbiota of mice fed an HF diet exhibited significant in- grafts (WT, 61.169.3 mmol/L; isografts, 14.463.4 mmol/L), as creases in SCFA producing bacteria, including Bifidobacte- well as severe tubulitis on histologic examination of the trans- rium spp. (phylum Actinobacteria) and Bacteroides spp. planted kidneys (mean tubulitis score: WT, 131.6612.9; iso- Similarly, an unclassified commensal Clostridiales sp. was grafts, 0.860.4; Figure 2, B and C). WT1HF allografts were more abundant in the HF group compared with NC-fed protected against acute rejection, with lower serum creatinine mice, before and particularly after transplantation. The in- (WT1HF,29.363.6 mmol/L; Figure 2B) and a 30% reduction in crease in Verrucomicrobia seen in transplanted NC-fed mice tubulitis score compared with WT allografts (WT1HF, 89.9% was exclusively because of the expansion of the mucus- 67.1%; Figure 2C). degrading bacteria A. muciniphila. In contrast, isograft con- Immunohistochemical assessment of allografts at day 14 1 1 trols demonstrated relatively stable bacterial communities, revealed significant accumulation of CD4 and CD8 1 1 with no significant shifts at the phylum level (Figure 1C) and T cells, CD68 macrophages, Foxp3 T regulatory cells 1 differential abundance testing at the genus level revealing (Tregs) and CD11c dendritic cells in WT allografts as com- only a single Parabacteroides genus to have an altered relative pared with isografts (Figure 2D). Compared with WT allo- 1 abundance in response to isograft surgery (Supplemental grafts, WT1HF allografts showed an increase in CD4 1 Table 2). T cells, including an increase in Foxp3 Tregs (P,0.01), 1 The univariate differential abundance of recorded OTUs and a reduction in CD68 macrophages (P,0.01). was confirmed using a negative binomial distribution model Assessment of gene expression relevant to rejection by real- as implemented by DESeq2. Taxa were considered to be have time PCR of kidney tissue obtained 14 days after transplanta- differed significantly in their abundance if their false discovery tion revealed significant upregulation of proinflammatory

also exhibited reduced diversity after transplantation by Shannon diversity index. (B) Principal coordinate analysis of the weighted UniFrac distance demonstrated significant modulation of the microbiota community postallograft (Adonis: NC versus NC1Allo, R250.28; P,0.001; HF versus HF1Allo R250.16; P50.002), with significant dissimilarities between treatment groups (Adonis: NC1Allo versus HF1Allo, R250.31; P,0.001), but no significant change in isografts pre- versus post-transplant. (Adonis: preisograft versus isograft, R250.1; P50.14). (C) Relative abundance of the dominant phyla in treatment groups and isograft controls, demonstrating shifts in microbial composition postallograft, and stability after isograft surgery. (D) Heatmap of the dominant microbiota genera of allograft animals. (E) Relative abundance after cumulative sum squaring (CSS) normalization of significant SCFA-producing OTUs and the mucin- degrading genus Akkermansia. (F) DESeq2 analysis demonstrating differential abundance of OTUs (false discovery rate–adjusted P,0.01) between dietary groups pre- and post-transplant. OTUs were assigned to their lowest described classification (y axis) and color-coded by phylum. Bubble size represents a log fold-change in the log base mean of the recorded OTU, with the x axis values demonstrating log2 fold-change in relative abundance. Fecal microbiota composition was assessed by 16S rRNA sequencing of the V4 region. Data are shown as the mean6SEM. Statistical analysis by ANOVA with Tukey post hoc analysis (A), and Kruskal–Wallis non- parametric testing (E). *P,0.05; **P,0.01; ****P,0.0001.

1450 JASN JASN 31: 1445–1461, 2020 www.jasn.org BASIC RESEARCH

A B 100 80 **

60

50 * 40 Isografts

% Graft survival WT+ HF allografts 20 WT allografts 0 0 0 50 100 Serum creatinine (µmol/L) WT WT+HF Iso Days post transplantation

C WT allograft WT + HF allograft Isograft 200 * 150

100 PAS 50 Tubulitis score 0 WT WT + HF Iso D 150 **** 100 + field 50 CD4 CD4+ cells/ 400x 0 WT WT + HF Iso 100 80 60 +

field 40 C8D 20 CD8+ cells/ 400x 0 WT WT + HF Iso 50 ** 40

+ 30

field 20 Foxp3 10

Foxp3+ cells/ 400x 0 WT WT + HF Iso 25 ** 20 + 15

CD68 10 (400x) 5 CD68 % positive 0 WT WT + HF Iso 20

+ 15

10 CD11c (400x) 5 CD11c % positive 0 WT WT + HF Iso

Figure 2. HF diet prolonged renal allograft survival and attenuated acute allograft rejection at day 14 after kidney transplant. (A) Survival to day 100 was observed in all isografts (n513), but in only eight WT allografts (n521), with a mean survival of 40 days. HF-fed mice showed enhanced survival (HF n520 of 28). (B) At day 14, impaired renal function was evident in WT allografts (n57) with an

JASN 31: 1445–1461, 2020 Gut Microbiota and Kidney Rejection 1451 BASIC RESEARCH www.jasn.org cytokines (TNF-a, IL-6), Th1 (IFN-g), and Th2 (IL-4) cyto- reduction in chemokine CXCL9 expression in HF allografts kines, chemokines (CCL2, CCL5, and CXCL10), and cytotoxic (WT, 171649; WT1HF, 140620 CXCL9/GAPDH*1000; molecules (perforin and granzyme), in addition to regulatory P,0.05) no other significant difference in mRNA expression cytokines (TGF-b and IL-10), in WTallografts versus isografts was evident between the WT1HF and WT groups (Figure 3). Compared with WT allografts, WT1HF allografts (Supplemental Figure 4). exhibited significantly greater upregulation of IL-10 (WT, 10726119 IL-10/GAPDH31000; WT1HF, 1595667 IL-10/ Acetate Supplementation Prolonged Allograft Survival GAPDH31000; P,0.001) and diminished production of IL- by Promoting Donor-Specific Tolerance 23 (WT, 20006250 IL-23/GAPDH31000; WT1HF, 760676 As dietary supplementation with HF has been shown to in- IL-23/GAPDH31000; P,0.001). crease colonic and serum SCFA concentrations,18 we hypoth- esized that SCFAs were key mediators of the protective effects HF Diet Protected against Chronic Allograft Rejection of the HF diet in our kidney allograft model. We sought to test Groups of transplanted mice who survived to day 100 were this by examining the effect of direct SA supplementation. We euthanized to assess for evidence of chronic allograft rejection. transplanted BALB/c kidneys into C57BL/6 recipients who A minority of WT allografts survived to day 100 and these received 200 mg/kg SA via intraperitoneal injection on the exhibited severe chronic rejection, manifest by marked in- day of transplantation and for 14 days post-transplantation, creases in serum creatinine (WT, 48.869.5 mmol/L; isograft, followed by oral 150 mM SA solution via drinking water in- 13.061.8 mmol/L; Figure 4A), and proteinuria compared with definitely (SA allografts) and compared outcomes with WT isografts (WT, 1.8160.4 mg/16 h; isograft, 0.1560.05 mg/16 allografts. As SA is not the only SCFA reported to mediate h; Figure 4B). Severe histologic changes of chronic rejection proregulatory immune responses, the experiment was also were evident in WT allografts, including glomerulosclerosis, performed with SB as described above (SB allografts). interstitial fibrosis and tubular atrophy, and extensive chronic Both SA allografts and SB allografts experienced superior inflammatory cell infiltration and deposition of collagen survival to WT allografts, similar to the effects of an HF diet shown by Picro-Sirius red staining (Figure 4C). The majority but without reaching significance (Figure 5A). SA allografts of WT1HF allograft mice survived to day 100 and exhibited euthanized at post-transplant day 14 showed preservation of significantly less kidney dysfunction, with serum creatinine kidney function (creatinine: WT, 61.169.3 mmol/L; WT1SA, (28.462.1 mmol/L) and urine protein (0.6260.12 mg/16 h) 32.061.3 mmol/L; P,0.01; Figure 5B) and significantly less not statistically different to isografts (Figure 4, A and B). tubulitis than WT allografts (mean tubultis score: WT, WT1HF allograft histology revealed modest degrees of in- 118.469.9; WT1SA, 76.364.8; P,0.001; Figure 5C). Immu- flammation, glomerulosclerosis, tubulointerstitial fibrosis, nohistochemical staining revealed greater accumulation of 1 and collagen deposition, all of which were reduced in severity CD4 TcellsinWT1SA allografts as compared with WT 1 compared with WT allografts (Figure 4C). At day 100 post- allografts (P,0.0001), although not CD8 cells, along with 1 1 1 transplant, there was no difference in the accumulation of an increase in CD4 Foxp3 Tregs and CD11c cells (P,0.01 1 1 1 1 CD4 T cells, CD8 T cells, CD68 macrophages, CD11c and P,0.05, respectively), with no significant difference in the 1 1 dendritic cells, or Foxp3 Tregs (WT, 30.5463.3 cells/HPF; abundance of CD68 -stained cells (Figure 5D). WT1HF, 27.5062.3 cells/HPF; P50.67) in the allografts of Measured at day 14 post-transplant, donor-specificanti- WT and WT1HF fed mice, as determined by immunohis- body levels in both WT and WT1SA allograft mice were sig- tochemistry (Figure 4D). Prominent C4d staining was pre- nificantly elevated compared with isograft counterparts, sent diffusely in almost all cross-sections of peritubular although they did not differ significantly from each other capillaries (PTCs) with a bright, broad, linear pattern in (Figure 5E). WT allografts, and this remained unaltered in WT1HF Examination of allograft tissue by RT-PCR revealed a sim- mice (Figure 4E). ilar pattern of gene expression to that observed in HF allo- Day 100 WT allografts demonstrated significant upregula- grafts, including upregulation of IL-10 (WT, 683652 IL-10/ tion of cytokines (IFN-g, IL-4, IL-6, IL-10, IL-12, IL-23, TNF, GAPDH31000; WT1SA, 993698 IL-10/GAPDH31000; TGF-b), chemokines (CCL5, CXCL9, CXCL10), and genes P,0.05), but a marked reduction in IL-6 expression (WT, involved in tissue re-modeling (fibronectin, TIMP-1, TIMP- 725361323 IL-6/GAPDH31000; WT1SA, 306651 IL-6/ 2, MMP-2, MMP-9) as compared with isografts. Despite a GAPDH31000; P,0.0001) (Figure 6). increase in serum creatinine as compared with isografts (n55). Mice fed an HF diet (n59) were partially protected from allograft dysfunction with lower serum creatinine (P,0.01). (C) Tubulitis was present in both WT and HF-fed allograft mice, although it was significantly attenuated in the HF group (tubulitis score P,0.05). (D) Representative photomicrographs of WT and WT1HF allograft sections at day 14 post-transplant demonstrate significant inflammatory cell infiltrates as compared with isografts. Increased numbers of 1 1 1 CD4 and Foxp3 cells and lesser numbers of CD68 macrophages were evident for WT1HF allografts compared with WT allografts. Scale bar, 50.0 mm. Data are shown as mean6SEM. Statistical analysis by log-rank test and ANOVA with Tukey post hoc analysis. *P,0.05; **P,0.01; ****P,0.0001.

1452 JASN JASN 31: 1445–1461, 2020 www.jasn.org BASIC RESEARCH

1500 5000 600 1500 4000 1000 400 1000 3000

2000 500 200 500 1000 IFN/GAPDH*1000 IL-4/GAPDH*1000 IL-6/GAPDH*1000 TNF/GAPDH*1000 0 0 0 0 WT WT+HF Iso WT WT + HF Iso WT WT + HF Iso WT WT + HF Iso

2500 *** 1500 2000 *** 1000 2000 1500 800 1000 1500 600 1000 1000 400 500 500 500 200 TGF/GAPDH*1000 IL-10/GAPDH*1000 IL-12/GAPDH*1000 IL-23/GAPDH*1000 0 0 0 0 WT WT+HF Iso WT WT + HF Iso WT WT + HF Iso WT WT + HF Iso

2000 1000 2000 1500

1500 800 1500 1000 600 1000 1000 400 500 500 200 500 IDO/GAPDH*1000 CCL5/GAPDH*1000 CCL2/GAPDH*1000 0 0 0 CXCL9/GAPDH*1000 0 WT WT+HF Iso WT WT + HF Iso WT WT + HF Iso WT WT + HF Iso

400 1500 1500 1500

300 1000 1000 1000 200 500 500 500 100 Perforin/GAPDH*1000 CXCL10/GAPDH*1000 0 0 0 0 GranzymeA/GAPDH*1000 GranzymeB/GAPDH*1000 WT WT+HF Iso WT WT + HF Iso WT WT + HF Iso WT WT + HF Iso

Figure 3. HF diet resulted in altered mRNA gene expression in transplanted kidneys at day 14 post-transplant. Inflammatory cytokine mRNA was upregulated in both WT (n57) and WT1HF (n59) allografts compared with isografts (n55). Expression of the regulatory cytokine IL-10 was significantly increased and expression of proinflammatory cytokine IL-23 was decreased in WT1HF compared with WT allografts. Data are shown as the mean6SEM. Statistical analysis by ANOVA with Tukey post hoc analysis. ***P,0.001.

At day 100 post-transplant, SB allograft mice displayed pro- allografts (P,0.01 and P,0.0001, respectively, Figure 7C). longed survival similar to SA supplemented mice, but exhibi- Gene expression of key cytokines in SA allografts revealed a ted higher markers of allograft injury (serum creatinine); three-fold increase in IL-10 expression compared with WTallo- therefore, mechanistic studies were explored on SA- grafts (WT,347670 IL-10/GAPDH31000; WT1SA, 10906180 supplemented mice alone. IL-10/GAPDH31000; P,0.01; Figure 7D). C4d staining of both SA and WTallografts revealed prominent staining of peritubular Acetate Protected against Chronic Allograft Rejection capillaries with no difference between groups (Figure 7E). The majority of SA allograft mice survived to day 100, at which Donor-specific antibody levels were assessed in the serum time their serum creatinine and urinary protein excretion of transplanted animals harvested at day 100 (Figure 7F). Both were not significantly different to isografts and were markedly WT and WT1SA allograft mice demonstrated similar eleva- lower than WT allografts (creatinine: WT, 48.869.5 mmol/L; tions of donor-specificIgG1bymeanfluorescene intensity WT1SA, 28.561.8 mmol/L; P,0.05; and proteinuria: WT, (MFI) (MFI: WT, 106160.5; WT1SA, 988.16215.8; 1.2160.14 mg/16 h, WT1SA, 0.3960.09 mg/16 h; P50.95) and IgG2a (MFI: WT, 519.3696.4; WT1SA, P,0.0001; Figure 7A). Histologic studies showed a significant 394.5651.6; P50.46), greater than those seen in isograft reduction in glomerulosclerosis, interstitial fibrosis and tubu- mice but not significantly divergent from each other. lar atrophy, and collagen deposition in SA allografts com- pared with WTallografts (Figure 7B). Immunohistochemistry SA Supplementation Promoted Donor staining revealed a reduction in inflammatory infiltrate in SA Antigen–Specific Tolerance 1 1 allografts, with modestly reduced CD4 T cell and CD11c To test for dominant immune tolerance, a subset of both dendritic cell accumulation when compared with WT WT allograft and SA allograft survivors received skin grafts

JASN 31: 1445–1461, 2020 Gut Microbiota and Kidney Rejection 1453 BASIC RESEARCH www.jasn.org

A B 80 2.5 ** * 60 2.0 1.5 40 1.0 (µmol/L)

20 Proteinuria

(mg/16 hours) 0.5 Serum creatinine 0 0.0 WT WT+HF Iso WT WT+HF Iso

60 **** 3 ****

C WT allograft WT + HF allograft Isograft 40 2

20 1 sclerosed % Glomeruli 0 atrophy (grade) 0 Fibrosis and tubular PAS WT WT+HF Iso WT WT+HF Iso 40 **** 30 20

Collagen 10

PSR 0 (% points positive) WT WT+HF Iso 80 D 60 40 +

400x field 20 CD4+ cells/ CD4 0 WT WT + HF Iso 40 30 + 20 400x field CD68 10 CD8+ cells/ 0 WT WT + HF Iso 5 + 4 3

CD11c 2 1 % field CD11c positive (400x) 0 WT WT + HF Iso E 3

+ 2 C4d

(grade) 1 C4d staining 0 WT WT+HF Iso

Figure 4. HF diet attenuated chronic allograft damage at day 100 after kidney transplant. (A) Serum creatinine and (B) urinary protein excretion were significantly reduced in WT1HF (n512) allografts compared with WT allografts (n58) at day 100 post-transplant. (C) Representative photomicrographs of the histologic changes on periodic acid–Schiff (PAS); Picro-Sirius Red (PSR); immunochemistry staining for CD4, CD8, CD68, and CD11c; and fluorescence staining for C4d in WT allografts, WT1HF allografts, and isografts (n55). PAS and PSR sections showed significant glomerulosclerosis, interstitial fibrosis and tubular atrophy, and interstitial collagen deposition in WT allografts compared with isografts, all of which were significantly reduced in WT1HF allografts compared to WT allografts. (D) CD4, CD8, and CD11c infiltrates, and (E) C4d deposition, were significantly and equally increased in WT and WT1HF allografts compared with isografts. Photomicrographs at 3400; scale bar, 50.0 mm. Data are shown as the mean6SEM. Statistical analysis by ANOVA with Tukey post hoc analysis. *P,0.05; **P,0.01; ****P,0.0001.

1454 JASN JASN 31: 1445–1461, 2020 www.jasn.org BASIC RESEARCH

A B 100 WT + SB allografts WT + SA allografts 80 WT allografts ** 60

50 40

% Grafts survival 20

0 Serum creatinine ( mol/L) 0 0255075100 WT WT + SA Isografts Days post transplantation

C WT allograft SA + allograft Isograft 150 *** 100

50 Tubulitis score

PAS 0 WT WT + SA Isografts

150 D **** 100

+ 50 CD4 0 CD4+ cells/ 400x field WT WT + SA Isografts

40 **

30 + 20

Foxp3 10

0 Foxp3+ cells/ 400x field WT WT + SA Isografts

E WT allografts 800 WT + SA allografts Isografts 80 30 20 * 600 60 15 20 400 40 10 field

10 intensity (MFI) 200 20 5 Mean fluorescence % field CD68 % field CD11c positive (400x) positive (400x) CD8+ cells/ 400x 0 0 0 0 WT WT + SA Isografts WT WT + SA Isografts WT WT + SA Isografts IgG1 IgG2a DSA isotype

Figure 5. SCFA supplementation attenuated acute allograft rejection. (A) WT1SA allografts (n527) and WT1SB allografts (n521) exhibited a trend toward improved survival compared with WT allografts (n521), without reaching significance. At day 14 post- transplant, WT1SA allografts (n59) demonstrated improved graft function with lower serum creatinine (P,0.05) (B), and less tubuli- tis (C) compared with WT allografts (n57) (P,0.01). (D) Representative photomicrographs and analysis of cellular infiltrate by IHC 1 1 1 showed a marked increase in CD4 and Foxp3 T cells, as well as CD11c cells in WT1SA allografts compared with WT allografts (isografts n55). (E) Serum donor-specific antibody titers were evaluated at 14 days post-transplantation. WT (n513) and SA- supplemented (n59) allograft mice demonstrated elevated donor-specific IgG1 and IgG2a compared with isografts (n55). Scale bar, 50.0 mm. Data are shown as the mean6SEM. Statistical analysis by one-way ANOVA with Tukey post hoc analysis. *P,0.05; **P,0.01; ***P,0.001; ****P,0.0001. from BALB/c (donor-matched allograft), C57BL/6 (iso- were performed. SA allografts accepted donor-matched graft), and B10Br (third-party allograft) mice at least skin allografts (H-2d) for over 100 days, rejected B10Br 200 days post-kidney transplant (Figure 8A). Skin iso- (H-2k) third-party allografts by day 15, and survived for grafts remained intact on both sets of animals. WT mice over 100 days after placement of skin grafts, at which time rejected donor-matched and third-party skin allografts at kidney function remained normal (serum creatinine 12–14 days and died within 34–107 days after skin allografts 2163 mmol/L).

JASN 31: 1445–1461, 2020 Gut Microbiota and Kidney Rejection 1455 BASIC RESEARCH www.jasn.org

1500 2500 10000 **** 2000 2000 8000 1500 1000 1500 6000 1000 1000 4000 500 500 2000 500 IL-4/GAPDH*1000 IL-6/GAPDH*1000 TNF/GAPDH*1000 IFN γ /GAPDH*1000 0 0 0 0 WT WT + SA Isografts WT WT + SA Isografts WT WT + SA Isografts WT WT + SA Isografts

2000 2000 1500 1500 * 1500 1500 1000 1000 1000 1000 500 500 500 500 IL-10/GAPDH*1000 IL-12/GAPDH*1000 IL-23/GAPDH*1000 0 0 0 TGF β /GAPDH*1000 0 WT WT + SA Isografts WT WT + SA Isografts WT WT + SA Isografts WT WT + SA Isografts

2500 2000 2500 * 2000 2000 1500 2000 1500 1500 1500 1000 1000 1000 1000 500 500 500 500 IDO/GAPDH*1000 CCL2/GAPDH*1000 CCL5/GAPDH*1000

0 0 0 CXCL9/GAPDH*1000 0 WT WT + SA Isografts WT WT + SA Isografts WT WT + SA Isografts WT WT + SA Isografts

1500 2500 2500 2000

2000 2000 1500 1000 1500 1500 1000 1000 1000 500 500 500 500 0 0 0 0 Perforin/GAPDH*1000 CXCL10/GAPDH*1000 GranzymeB/GAPDH*1000 WT WT + SA Isografts WT WT + SA IsograftsGranzymeA/GAPDH*1000 WT WT + SA Isografts WT WT + SA Isografts

Figure 6. SCFA supplementation altered allograftmRNAexpressionatD14.WT1SA allografts (n59) showed upregulation of most Th1 (IFN-g) and Th2 (IL-4) cytokines in a pattern consistent with WT allografts (n57). InflammatorycytokineIL-6wassig- nificantly reduced whereas expression of the regulatory cytokine IL-10 was significantly increased in WT1SA allografts compared with WT allografts (isografts n55). Data are shown as the mean6SEM. Statistical analysis by one-way ANOVA. *P,0.05; ****P,0.0001.

2 2 SA-Mediated Donor-Specific Tolerance of Kidney transplanted BALB/c WT kidneys into GPR43 / mice on 1 1 1 Allografts Was Dependent on a CD4 CD25 Foxp3 a C57BL/6 background, who received an HF diet or SA sup- 2 2 Regulatory Mechanism plementation as previously described (as GPR43 / 1HF 2 2 As a greater number of Treg cells, both proportionally and and GPR43 / 1SA allografts, respectively). Both HF and 2 2 absolute, were observed in HF allografts and SA allografts at SA supplementation was ineffective in GPR43 / allograft day 14 post-transplant, we hypothesized that long-term al- recipients, with abrogation of the survival advantage seen in lograft tolerance may be dependent on immune regulation HF and SA allograft mice (Figure 8C), demonstrating that 1 1 mediated by CD4 CD25 Tregs. Toexamine this, we treated signaling of acetate via GPR43 was critical to mitigate graft SA allografts with an anti-CD25 mAb (clone: PC61). rejection. Graft survival was unchanged in NC-fed 1 1 1 2 2 CD4 CD25 Foxp3 cells were depleted by .90% com- GPR43 / allograft recipients, as compared with NC-fed pared with control antibody–treated mice, and the allograft WT allograft recipients, suggesting maintenance of the al- 2 2 survival benefit conferred by SA was abrogated (Figure 8B). loimmune response in GPR43 / mice. Administration of anti-CD25 antibody to WT allograft re- As SCFAs are known to epigenetically alter gene expression cipients did not alter allograft rejection kinetics. by inhibiting HDACs, we next determined whether HDAC activity was upregulated in SA supplemented allografts. Com- GPR43 Was Critical for HF and Acetate-Mediated pared with WTallografts, SA supplemented allografts did not Protection against Allograft Rejection demonstrate significant changes in HDAC activity measured Acetate preferentially binds the metabolite-sensing GPRs, at day 14 post-transplant (P50.27; Supplemental Figure 5), such as GPR43, to elicit cellular responses. To determine indicating that the beneficial effects of acetate in our trans- whether GPR43 was required to mediate the benefitseen plant model were not attributable to alternations in histone in HF and SA supplemented allograft mice, we next acetylation alone.

1456 JASN JASN 31: 1445–1461, 2020 www.jasn.org BASIC RESEARCH

A 80 1.5 **** * 60 1.0 40

( μ mol/L) 0.5

20 Proteinuria (mg/16 hours) Serum creatinine 0 0.0 WT WT + SA Isografts WT WT + SA Isografts

B * 50 3 ** 40 ** 40 30 2 30 20 20

1 Collagen 10 10 atrophy (grade) (% points positive) Fibrosis and tubular

% glomeruli sclerosed 0 0 0 WT WT + SA Isografts WT WT + SA Isografts WT WT + SA Isografts

C 80 ** 40 ** 5 4 **** 60 30 3 40 20 2

20 400x field 10

Foxp3+ cells/ 1 % field CD11c positive (400x)

CD4+ cells/ 400x field 0 0 0 WT WT + SA Isografts WT WT + SA Isografts WT WT + SA Isografts

D E F WT allografts 1500 ** 3 1000 WT + SA allografts Isografts 1000 2 500 500 1 intensity (MFI) Mean fluorescence IL-10/GAPDH*1000 0 C4d staining (grade) 0 0 WT WT + SA Isografts WT WT + SA Isografts IgG1 IgG2a DSA isotype

Figure 7. SA supplementation protected against chronic allograft injury at day 100 post-transplant. (A) SA supplementation resulted in asignificant decrease in serum creatinine and urinary protein excretion in WT1SA allograft mice (n511) at day 100 post-transplant, compared with WT allograft mice (n58). (B) Glomerulosclerosis, interstitial fibrosis and tubular atrophy, and interstitial collagen de- position were markedly decreased in WT1SA allografts compared with WT allografts. (C) Immunohistochemistry staining showed a 1 1 1 decrease in CD4 , Foxp3 , and CD11c infiltrates in WT1SA allografts compared with WT allografts. (D) Expression of the regulatory cytokine IL-10 was significantly increased in WT1SA allografts compared with WT allografts. (E) Prominent C4d staining was present diffusely in peritubular capillaries of both WT and WT1SA allografts compared with isografts (n55). (F) At day 100 post-transplant, serum donor-specific antibody titers were evaluated in WT (n512) and SA-supplemented (n511) allograft mice, with all groups demonstrating elevated donor-specific IgG1 and IgG2a compared with isografts (n55). Data are shown as the mean6SEM. Statistical analysis by one-way ANOVA. *P,0.05; **P,0.01; ***P,0.001; ****P,0.0001.

DISCUSSION rejection, exhibiting prolonged survival and reduced evidence of both acute and chronic allograft rejection. Diet-related pro- Using an established and reproducible kidney transplant duction of SCFAs by the gut microbiome has been shown to model, we found considerable alteration in gut microbiota modify pathogenic immune responses in several experimental composition after allograft transplantation, which occurred settings.18,30,31 Our findings that dietary supplementation in the absence of immunosuppression or antibiotic interven- with SA afforded similar protection against allograft rejection tion and did not occur in isograft recipients. This transplant- to HF diet, and that deficiency of GPR43, a receptor for ace- induced dysbiosis was ameliorated by maintenance on an HF tate, rendered mice resistant to the protective effects of both diet. Mice fed HF were partially protected from allograft HF diet and acetate supplementation, strongly suggests that

JASN 31: 1445–1461, 2020 Gut Microbiota and Kidney Rejection 1457 BASIC RESEARCH www.jasn.org

A B WT + SA allograft 100 100 WT + SA allograft + CD25Ab * WT allograft 75 75 WT allograft + CD25Ab WT + SA allograft + BALB/c skin

50 ** 50 WT allograft + BALB/c skin 25 WT + SA allograft + B10Br skin 25 WT allograft + B10Br skin % Skin graft survival 0 % Renal graft survival 0 0 255075100 0 25 50 75 100 Day post skin transplantation Day post transplantation

C 100

WT + HF allograft * **** 75 GPR43-/- + HF allograft WT + SA allograft 50 GPR43-/- + SA allograft WT + NC allograft 25 -/- GPR43 + NC allograft NS

% Renal graft survival 0 0 25 50 75 100 Day post transplantation

1 1 1 Figure 8. SA promoted donor antigen–specific tolerance of kidney allografts, dependent on a CD4 CD25 Foxp3 regulatory mechanism and signaling via GPR43. (A) Skin graft survival curves for skin challenges in WT and WT1SA kidney allograft acceptors. A subset of WT (n54) and SA-supplemented (n54) allograft mice who survived to over day 200 after kidney transplant received skin grafts from donor-matched (BALB/c), isografts (C57BL/6), and third-party (B10Br) mice. Skin isografts remained intact on all animals, whereas third-party skin allografts were promptly rejected by day 15 by WT and WT1SA kidney allograft acceptors. WT kidney allograft ac- ceptors rejected donor-matched skin grafts at 12–14 days, whereas WT1SA kidney allograft acceptors accepted donor-matched skin grafts for .100 days. (B) A depleting mAb to CD25 was given to WT (n56) and SA-supplemented (n517) allograft mice on days 22and 1 0 post-transplant. The survival benefitseeninSA-supplementedallograftrecipients(n527) was annulled after depletion of CD25 cells (P,0.05), whereas administration of anti-CD25 Ab did not alter allograft rejection kinetics in WT allograft recipients. (C) SA supple- 2 2 mentation and HF diet was ineffective in GPR43 / recipients of WT allografts (n513 for SA, n517 for HF), with no survival advantage 2 2 compared with WT recipients of kidney allograft (n521), whereas GPR43 / allograft recipients fed NC (n510) had graft survival comparable to WT mice fed NC. Statistical analysis by log-rank test. *P,0.05; **P,0.01; ****P,0.0001. gut-derived acetate is important in retarding alloimmunity in immunosuppression, surgery, and resolution of the uremic the context of kidney transplantation. Enhanced immune reg- environment in causing this shift. ulation is another proposed mechanism of immune- The shift in bacterial community structure in our model modulation by the gut microbiome, and this also appears to after transplantation confirmed the development of be important in transplantation as the protective effects of transplant-induced dysbiosis. In the absence of antibiotics 1 1 SCFA were negated by depletion of CD4 CD25 T-cells. and other drugs, potential causes of dysbiosis in our model The gut microbiome has a modulatory role in directing were reduced to transplant surgery and the alloimmune re- host systemic immune responses and has coevolved with the sponse. That isograft controls demonstrated no significant mammalian immune system to form a necessary symbiotic change in the gut bacterial landscape and only a transient re- relationship.15,32 The process of transplantation is highly im- duction in a-diversity postsurgery identifies the allograft im- munogenic, provoking an innate response to surgical tissue mune response as the prime driver of transplant dysbiosis. injury and ischemia-reperfusion injury to the implanted or- Diet is the dominant exogenous factor known to affect gan, followed by T cell recognition of nonself HLA, causing a composition of the gut microbiome and is the key determi- powerful adaptive response.9 Given this profound stimulus to nant of its diversity and richness.36 Murine host microbial host immunity, it is predictable that transplantation would communities are diet-, colony-, and facility-specific; however, have reciprocal effects on the gut microbiota.33 Although clin- the composition of the microbiota in our NC-fed mice were ical studies have demonstrated shifts in microbial composition consistent with previous observations of C57BL/6 gut micro- after kidney transplantation with possible links to allograft biota.4,12 After transplantation, shifts in microbiota structure outcomes,34,35 it is unclear what contribution exposure to al- were seen in both NC- and HF-fed mice. Two key differences logenic antigen plays in addition to the role of antibiotics, between NC- and HF-fed mice were notable and potentially

1458 JASN JASN 31: 1445–1461, 2020 www.jasn.org BASIC RESEARCH underpin the protective capacity of the HF diet: first, the rel- mucous-degrading bacterial communities offers a second ative abundance of bacteria known to produce SCFAs; and mechanism by which the HF diet may have retarded rejection. second, the balance between mucous-producing and Overall, our findings with regard to the effect of diet and the mucous-degrading bacteria. microbiota on kidney allograft rejection do not suggest that Therelativeabundanceofbacteriaknowntoproduce any one specific bacterial species may be sufficient to deter- SCFAs, including Bifidobacterium, Clostridium,andBacteroi- mine transplant survival or rejection, but raise the possibility detes spp., were significantly increased in transplanted mice on that the balance between beneficial and nonbeneficial species an HF diet compared with NC. Bifidobacteria spp. have been may be the most important determinant of a favorable clinical shown to offer protection from kidney ischemia-reperfusion outcome.5 Furthermore, our findings support the notion that injury in a murine model.20 Furthermore, our results are greater overall diversity may not be as important for main- consistent with the recent findings of Bromberg et al.,37 who taining health as the balance between specificbacterial demonstrated that lone transfer of a Bifidobacteria sp. was species.45 sufficient to improve outcomes in murine cardiac allografts. The known capacity of the gut microbiome to enhance We demonstrated that the ability of acetate to limit allograft immune regulation was also evident in our studies. The pro- rejection was dependent on its action on the metabolite- tection afforded by both HF and direct supplementation of 2 2 sensing receptor GPR43, with transplanted GPR43 / mice acetate was associated with an influx of Tregs and elevated IL- not protected by diet. Our findings are consistent with a grow- 10 gene expression within allografts. SCFAs have the potential ing body of data highlighting the importance of GPR43 in to regulate tissue inflammation through their effects on mul- mediating immunomodulatory effects of SCFA.17,38 We have tiple cell types such as Tregs, DCs, and macrophages.46 Sup- previously demonstrated that acetate can ameliorate murine plementation with acetate caused pronounced suppression of dextran sulfate sodium (DSS) induced colitis via GPR43 acti- IL-6 expression in allograft kidneys, which, coupled with up- vation on nonhematopoietic cells,18 and more recently, signal- regulation of IL-10, likely represents the establishment of a ing via GPR43 was shown to be a critical determinant of the local immunoregulatory environment. Similarly, allograft severity of experimental graft versus host disease.39 The ben- gene profiles obtained from HF-fed, WT mice indicated an efitofSCFAinbothmodelswasdependentonGPR43- environment conducive to Treg generation or expansion. mediated ERK phosphorylation and activation of the NLRP3 The expansion of Tregs found with HF diet and SCFA supple- inflammasome in intestinal epithelial cells, rather than direct mentation in our studies is in keeping with earlier observa- activation of GPR43 on hemopoietic cells. GPR43 is known to tions that SCFA can induce colonic and peripheral Treg be expressed by multiple cell types, including antigen- expansion acting through GPR43 expressed on colonic presenting cells, polymorphonuclear cells, B cells, and T cells.38,47 Clostridium species, which were augmented by T cells, in addition to epithelial cell types, including renal HF and diminished after transplantation on NC, have been tubular cells, and thus may modulate both local and systemic identified as important promoters of immune regulation with immunity through numerous pathways.17,40,41 Although the some commensal clusters shown to bias naïve T cell develop- effects of microbiota-derived acetate on local resident cells is ment toward a regulatory phenotype.48 In our model, early 1 of significance in experimental colitis and graft versus host depletion of CD25 T cells was sufficient to abrogate the sur- disease, the importance of circulating Treg cells in mediating vival advantage conferred by acetate, suggesting that natural alloimmunity in transplantation highlights the influence of Tregs are required for the protective effect. Additional actions acetate on modulating adaptive immunity to affect distant of dietary metabolites on Treg induction and functionality in organ sites. our model are likely, and further experiments examining Tregs In health, the mucous-degrading actions of A. muciniphila at numerous post-transplant timepoints and the adoptive 1 1 1 are counterbalanced by bacterial species, including Lactoba- transfer of CD4 CD25 Foxp3 Tcellpopulationstoun- cillus and Bifidobacterium, which protect the colonic mucous treated allograft recipients may provide important insight barrier from diet-induced, microbiota-related degrada- into the mechanistic role of SCFA-enhanced Tregs in kidney tion.42,43 In NC-fed mice, we found this balance to be dis- transplantation. turbed because of an excess of A. muciniphila and reduction Our data suggest dietary modification of the microbiome in the relative abundance of Lactobacillus and Bifidobacterium, may be a simple and safe means to enhance Treg generation in as opposed to HF-fed mice, where balance was preserved. The vivo as a strategy to enhance organ tolerance. Whether such colonic mucosal layer is critical for gut–immune interactions, modifications can be made in humans and what their effect is providing a protective barrier against harmful gut bacteria as remains to be seen. Research in the future may determine well as an environment that enables the trafficking of T cells whether modifying the microbiome with diet-based prebiotic required for maturation. Disruption of the mucous barrier therapies, probiotics, a combination of the two (synbiotics), or leads to translocation of microbiota, pathobionts, and their directed therapy with fecal microbiota transplant or specific products to the epithelial surface, causing inflammation and SCFA supplementation could be a useful adjunct to current directing immune cell maturation toward a proinflammatory treatment regimens. Recent reports suggest that even small phenotype.44 Thus, the promotion of mucous-protective over changes in gut microbial communities can enhance or impair

JASN 31: 1445–1461, 2020 Gut Microbiota and Kidney Rejection 1459 BASIC RESEARCH www.jasn.org the response to immunomodulatory treatments.4,5,49 Given the Supplemental Table 2. DESeq2 analysis of differential microbial toxic off-target effects of current immunosuppressive treatments, abundance in NC-fed isograft pre- and post-transplant. achieving an enhanced therapeutic response with reduced drug Supplemental Table 3. DESeq2 analysis of differential microbial exposure would therefore be advantageous in the clinic. abundance in NC- and HF-fed mice. In summary, our results demonstrate that dietary supple- Supplemental Table 4. DESeq2 analysis of differential microbial mentation with HF promotes renal allograft survival and lim- abundance in NC1Allo and HF1Allo mice. its transplant-induced dysbiosis. The protective effect is mediated through increased microbial SCFA production and is dependent on Tregs through GPR43 signaling. REFERENCES

1. Nankivell BJ, Alexander SI: Rejection of the kidney allograft. NEngl JMed363: 1451–1462, 2010 2. Halloran PF: Immunosuppressive drugs for kidney transplantation. ACKNOWLEDGMENTS NEnglJMed351: 2715–2729, 2004 3. Belkaid Y, Hand TW: Role of the microbiota in immunity and in- fl – We gratefully acknowledge the expertise of Dr. Brian Howden who ammation. Cell 157: 121 141, 2014 4. Sivan A, Corrales L, Hubert N, Williams JB, Aquino-Michaels K, performed the kidney transplant surgeries, and Dr. Jin Ma who Earley ZM, et al.: Commensal Bifidobacterium promotes antitu- performed the histone deacetylase assay. mor immunity and facilitates anti-PD-L1 efficacy. Science 350: Dr. Singer and A/Prof. Wu contributed equally to this paper. 1084–1089, 2015 A/Prof. Wu, A/Prof. Macia, and Prof. Chadban conceived and de- 5. Matson V, Fessler J, Bao R, Chongsuwat T, Zha Y, Alegre M-L, et al.: The fi signed the study. Dr. Singer, Dr. Kwan, Dr. Loh, Dr. Lai, and Dr. Li commensal microbiome is associated with anti-PD-1 ef cacy in meta- static melanoma patients. Science 359: 104–108, 2018 performed the experiments. Dr. Wang performed transplant sur- 6. Routy B, Le Chatelier E, Derosa L, Duong CPM, Alou MT, Daillère R, geries. Dr. Singer, A/Prof. Wu, and Prof. Chadban analyzed the data. et al.: Gut microbiome influences efficacy of PD-1-based immuno- Dr. Singer constructed the figures. Dr. Singer and Dr. Tan performed therapy against epithelial tumors. Science 359: 91–97, 2018 microbiota analysis. Dr. Singer, A/Prof. Wu and Prof. Chadban 7. Burrows MP, Volchkov P, Kobayashi KS, Chervonsky AV: Microbiota regulates type 1 diabetes through Toll-like receptors. Proc Natl Acad drafted and revised the paper. Prof. Alexander revised the paper. All – fi Sci U S A 112: 9973 9977, 2015 authors approved the nal version of the manuscript. 8. Wen L, Ley RE, Volchkov PY, Stranges PB, Avanesyan L, Stonebraker AC, et al.: Innate immunity and intestinal microbiota in the develop- ment of Type 1 diabetes. Nature 455: 1109–1113, 2008 DISCLOSURES 9. Wu H, Chen G, Wyburn KR, Yin J, Bertolino P, Eris JM, et al.: TLR4 ac- tivation mediates kidney ischemia/reperfusion injury. J Clin Invest 117: 2847–2859, 2007 None. 10. Kwan TK, Chadban SJ, Wu H: Toll-like receptor 4 deficiency improves short-term renal function but not long-term graft survival in a fully MHC- mismatched murine model of renal allograft transplantation. Trans- FUNDING plantation 100: 1219–1227, 2016 11. Wu H, Noordmans GA, O’Brien MR, Ma J, Zhao CY, Zhang GY, et al.: Absence of MyD88 signaling induces donor-specifickidneyallograft This work was supported by the Australian National Health and Medical tolerance. J Am Soc Nephrol 23: 1701–1716, 2012 Research Council (project grant APP1162764). The project grant was awarded 12. McIntosh CM, Chen L, Shaiber A, Eren AM, Alegre M-L: Gut microbes to Prof. Chadban, A/Prof Wu and A/Prof Macia. contribute to variation in solid organ transplant outcomes in mice. Microbiome 6: 96, 2018 13. Lei YM, Chen L, Wang Y, Stefka AT, Molinero LL, Theriault B, et al.: The SUPPLEMENTAL MATERIAL composition of the microbiota modulates allograft rejection. JClin Invest 126: 2736–2744, 2016 14. Rey K, Manku S, Enns W, Van Rossum T, Bushell K, Morin RD, et al.: This article contains the following supplemental material online at Disruption of the gut microbiota with antibiotics exacerbates acute http://jasn.asnjournals.org/lookup/suppl/doi:10.1681/ASN.2019080852/-/ vascular rejection. Transplantation 102: 1085–1095, 2018 DCSupplemental. 15. Rakoff-Nahoum S, Paglino J, Eslami-Varzaneh F, Edberg S, Medzhitov Supplemental Figure 1. Multiple sample rarefaction curves. R: Recognition of commensal microflora by toll-like receptors is re- – Supplemental Figure 2. Richness of gut microbial communities in quired for intestinal homeostasis. Cell 118: 229 241, 2004 16. Lennon G, Balfe A, Earley H, Devane LA, Lavelle A, Winter DC, et al.: allograft and isograft mice. Influences of the colonic microbiome on the mucous gel layer in ul- Supplemental Figure 3. Dominant phylum of WT C57Bl/6 mice cerative colitis. Gut Microbes 5: 277–285, 2014 fed a zero-fiber diet. 17. Maslowski KM, Vieira AT, Ng A, Kranich J, Sierro F, Yu D, et al.: Reg- Supplemental Figure 4. Cytokine and chemokine mRNA expres- ulation of inflammatory responses by gut microbiota and chemo- – sion in allografts at D100. attractant receptor GPR43. Nature 461: 1282 1286, 2009 18. Macia L, Tan J, Vieira AT, Leach K, Stanley D, Luong S, et al.: Metabolite- Supplemental Figure 5. HDAC activity in renal allografts. sensing receptors GPR43 and GPR109A facilitate dietary fibre-induced Supplemental Table 1. Nutritional parameters of HF and normal gut homeostasis through regulation of the inflammasome. Nat mouse chow used in experiments. Commun 6: 6734, 2015

1460 JASN JASN 31: 1445–1461, 2020 www.jasn.org BASIC RESEARCH

19. Furusawa Y, Obata Y, Fukuda S, Endo TA, Nakato G, Takahashi D, et al.: 36. Xu Z, Knight R: Dietary effects on human gut microbiome diversity. Br Commensal microbe-derived butyrate induces the differentiation of JNutr113[Suppl]: S1–S5, 2015 colonic regulatory T cells. Nature 504: 446–450, 2013 37. Bromberg JS, Hittle L, Xiong Y, Saxena V, Smyth EM, Li L, et al.: Gut 20. Andrade-Oliveira V, Amano MT, Correa-Costa M, Castoldi A, Felizardo microbiota-dependent modulation of innate immunity and lymph node RJ, de Almeida DC, et al.: Gut bacteria products prevent AKI induced remodeling affects cardiac allograft outcomes. JCI Insight 3: e121045, by ischemia-reperfusion. JAmSocNephrol26: 1877–1888, 2015 2018 21. Kwan T, Chadban SJ, Ma J, Bao S, Alexander SI, Wu H: IL-17 deficiency 38. Smith PM, Howitt MR, Panikov N, Michaud M, Gallini CA, Bohlooly-Y M, attenuates allograft injury and prolongs survival in a murine model of et al.: The microbial metabolites, short-chain fatty acids, regulate co- fully MHC-mismatched renal allograft transplantation. Am J Transplant lonic Treg cell homeostasis. Science 341: 569–573, 2013 15: 1555–1567, 2015 39. Fujiwara H, Docampo MD, Riwes M, Peltier D, Toubai T, Henig I, et al.: 22. Caporaso JG, Kuczynski J, Stombaugh J, Bittinger K, Bushman FD, Microbial metabolite sensor GPR43 controls severity of experimental Costello EK, et al.: QIIME allows analysis of high-throughput commu- GVHD. Nat Commun 9: 3674, 2018 nity sequencing data. Nat Methods 7: 335–336, 2010 40. Kobayashi M, Mikami D, Kimura H, Kamiyama K, Morikawa Y, Yokoi S, 23. Zakrzewski M, Proietti C, Ellis JJ, Hasan S, Brion M-J, Berger B, et al.: et al.: Short-chain fatty acids, GPR41 and GPR43 ligands, inhibit TNF- Calypso: A user-friendly web-server for mining and visualizing a-induced MCP-1 expression by modulating p38 and JNK signaling microbiome-environment interactions. Bioinformatics 33: 782–783, 2017 pathways in human renal cortical epithelial cells. Biochem Biophys Res 24. Lozupone C, Knight R: UniFrac: A new phylogenetic method for com- Commun 486: 499–505, 2017 paring microbial communities. Appl Environ Microbiol 71: 8228–8235, 41. Covington DK, Briscoe CA, Brown AJ, Jayawickreme CK: The G-pro- 2005 tein-coupled receptor 40 family (GPR40-GPR43) and its role in nutrient 25. Paulson JN, Stine OC, Bravo HC, Pop M: Differential abundance analysis sensing. Biochem Soc Trans 34: 770–773, 2006 for microbial marker-gene surveys. Nat Methods 10: 1200–1202, 2013 42. Schroeder BO, Birchenough GM, Ståhlman M, Arike L, Johansson ME, 26. Love MI, Huber W, Anders S: Moderated estimation of fold change and Hansson GC, et al.: Bifidobacteria or fiber protects against diet- dispersion for RNA-seq data with DESeq2. Genome Biol 15: 550, 2014 induced microbiota-mediated colonic mucus deterioration. Cell Host 27. McMurdie PJ, Holmes S: phyloseq: an R package for reproducible in- Microbe 23: 27–40.e27, 2018 teractive analysis and graphics of microbiome census data. PLoS One 8: 43. den Besten G, van Eunen K, Groen AK, Venema K, Reijngoud D-J, e61217, 2013 Bakker BM: The role of short-chain fatty acids in the interplay between 28. McWhinnie DL, Thompson JF, Taylor HM, Chapman JR, Bolton EM, diet, gut microbiota, and host energy metabolism. J Lipid Res 54: Carter NP, et al.: Morphometric analysis of cellular infiltration assessed 2325–2340, 2013 by monoclonal antibody labeling in sequential human renal allograft 44. Shan M, Gentile M, Yeiser JR, Walland AC, Bornstein VU, Chen K, et al.: biopsies. Transplantation 42: 352–358, 1986 Mucus enhances gut homeostasis and oral tolerance by delivering 29. Wu H, Craft ML, Wang P, Wyburn KR, Chen G, Ma J, et al.: IL-18 con- immunoregulatory signals. Science 342: 447–453, 2013 tributes to renal damage after ischemia-reperfusion. J Am Soc Nephrol 45. Zhao L, Zhang F, Ding X, Wu G, Lam YY, Wang X, et al.: Gut bacteria 19: 2331–2341, 2008 selectively promoted by dietary fibers alleviate type 2 diabetes. Sci- 30. Tan J, McKenzie C, Vuillermin PJ, Goverse G, Vinuesa CG, Mebius RE, ence 359: 1151–1156, 2018 et al.: Dietary fiber and bacterial SCFA enhance oral tolerance and 46. Crespo-Salgado J, Vehaskari VM, Stewart T, Ferris M, Zhang Q, Wang protect against food allergy through diverse cellular pathways. Cell Rep G, et al.: Intestinal microbiota in pediatric patients with end stage renal 15: 2809–2824, 2016 disease: A Midwest Pediatric Nephrology Consortium study. Micro- 31. Berer K, Martínez I, Walker A, Kunkel B, Schmitt-Kopplin P, Walter J, biome 4: 50, 2016 et al.: Dietary non-fermentable fiber prevents autoimmune neurologi- 47. Arpaia N, Campbell C, Fan X, Dikiy S, van der Veeken J, deRoos P, et al.: cal disease by changing gut metabolic and immune status. Sci Rep 8: Metabolites produced by commensal bacteria promote peripheral 10431, 2018 regulatory T-cell generation. Nature 504: 451–455, 2013 32. Round JL, Mazmanian SK: The gut microbiota shapes intestinal immune 48. Atarashi K, Tanoue T, Shima T, Imaoka A, Kuwahara T, Momose Y, et al.: responses during health and disease. Nat Rev Immunol 9: 313–323, Induction of colonic regulatory T cells by indigenous Clostridium spe- 2009 cies. Science 331: 337–341, 2011 33. Maynard CL, Elson CO, Hatton RD, Weaver CT: Reciprocal interactions 49. ZhangZ,LiuL,TangH,JiaoW,ZengS,XuY,etal.:Immunosuppressive of the intestinal microbiota and immune system. Nature 489: 231– 241, effect of the gut microbiome altered by high-dose tacrolimus in mice. 2012 Am J Transplant 18: 1646–1656, 2018 34. Fricke WF, Maddox C, Song Y, Bromberg JS: Human microbiota characterization in the course of renal transplantation. Am J Transplant 14: 416–427, 2014 35. Lee JR, Muthukumar T, Dadhania D, Toussaint NC, Ling L, Pamer E, et al.: Gut microbial community structure and complications after kid- See related editorial, “Can Diet Induce Transplantation Tolerance?” on pages ney transplantation: A pilot study. Transplantation 98: 697–705, 2014 1417–1418.

AFFILIATIONS

1Kidney Node Laboratory, The Charles Perkins Centre, Camperdown, New South Wales, Australia 2Sydney Medical School, Faculty of Medicine and Health, University of Sydney, Sydney, New South Wales, Australia 3Department of Renal Medicine, Royal Prince Alfred Hospital, Sydney, New South Wales, Australia 4Nutritional Immunometabolism Laboratory, The Charles Perkins Centre, Camperdown, New South Wales, Australia 5Centre for Kidney Research, The Children’s Hospital at Westmead, Westmead, New South Wales, Australia 6School of Medical Sciences, Faculty of Medicine and Health, University of Sydney, Sydney, New South Wales, Australia

JASN 31: 1445–1461, 2020 Gut Microbiota and Kidney Rejection 1461 SUPPLEMENT

Gut Microbial Metabolites Induce Donor Specific Tolerance of Kidney Allografts through SCFA

Induction of T Regulatory Cells

Huiling Wu, Julian Singer, Tony K Kwan, Yik Wen Loh, Chuanmin Wang, Jian Tan, Yan J Li, Sum Wing

Christina Lai, Laurence Macia, Stephen I Alexander, and Steven J Chadban

Table of Contents:

Methods

• Kidney Transplantation

• Histology

• Immunohistochemistry

• Immunofluorescence

Results

• Supplemental Figure 1. Multiple sample rarefaction curves

• Supplemental Figure 2. Richness of gut microbial communities in allograft and isograft mice.

• Supplemental Figure 3. Dominant phylum of WT C57Bl/6 mice fed a zero-fiber diet

• Supplemental Figure 4. Cytokine and chemokine mRNA expression in allografts at D100

• Supplemental Figure 5. HDAC activity in renal allografts

• Supplemental Table 1. Nutritional parameters of high-fiber and normal mouse chow used in

experiments

• Supplemental Table 2. DESeq2 analysis of differential microbial abundance in NC fed isograft pre

and post-transplant

• Supplemental Table 3. DESeq2 analysis of differential microbial abundance in NC and HF fed mice

• Supplemental Table 4. DESeq2 analysis of differential microbial abundance in NC+Allo and

HF+Allo mice

Methods:

Kidney Transplantation:

Heterotopic kidney transplants were performed with the left kidney of the donor animal flushed with heparinized saline and removed together with the ureter and vessels en mass, including a small (1-2mm) bladder cuff attached to the distal ureter. The recipient animal underwent a left sided nephrectomy and the transplanted kidney was placed heterotopically in the left iliac fossa on day zero. Urinary tract reconstruction was established by either inserting the ureter into the bladder (day 14 experiment only) or by suturing the bladder patch to a cystotomy located on the bladder dome (a bladder-to-bladder anastomosis) for survival study and day 100 experiments. All mice received induction and maintenance anesthesia with inhaled isoflurane and were monitored throughout procedures. All recipient allograft mice received a single, intra- peritoneal injection of ampicillin at the time of transplant surgery with the exception of mice on diet experiments or microbiota analysis, which did not receive any antibiotics. No immunosuppressive therapy was administered. The recipient’s right native kidney was removed at day 3-7, rendering the graft to be life- sustaining. Animals with technical graft failure or wound infection became overtly ill (and were euthanized) or died within 4 days of the contralateral nephrectomy and were removed from the study.

Histology

Periodic acid-Schiff (PAS) staining was performed on 3 µm paraffin embedded kidney sections to assess tubulitis (day 14 group only), glomerulosclerosis and interstitial fibrosis. Picro-Sirius red (PSR) staining was performed on 5 µm paraffin embedded sections of the kidney (D100 group only) to assess for interstitial collagen deposition. Scoring systems for each histological parameter have been previously described in detail, we summarize them briefly below. All histological analysis was performed in a blinded manner.

Tubulitis was examined on 250 tubular cross-sections per animal. Each tubular cross section was assessed as either, i) normal, ii) mild tubulitis (one infiltrating mononuclear cell per tubular cross-section), iii) moderate tubulitis (two or three infiltrating mononuclear cells per tubular cross-section and disruption of the basement membrane), or iv) severe tubulitis (defined as ≥ four infiltrating mononuclear cells per tubular cross-section).

A score for the degree of tubulitis was calculated for each animal, whereby each normal tubule received a score of 0, with mild tubulitis assigned a value of 1, and the number of tubules affected with mild and severe tubulitis was multiplied by 2 or 3 respectively. The total tubulitis score for each animal was the sum of these figures.

Glomerulosclerosis was quantitated by the presence of PAS-positive staining material involving >30% of each glomerulus. All glomeruli per section were scored to determine the percentage of glomeruli displaying glomerulosclerosis.

Interstitial fibrosis and tubular atrophy was graded following the Banff 97 scoring criteria on a scale of 0 to

3: 1 = mild interstitial fibrosis and tubular atrophy (<25% of cortical area); 2 = moderate interstitial fibrosis and tubular atrophy (26–50% of cortical area); 3 = severe interstitial fibrosis and tubular atrophy/loss (>50% of cortical area). If changes were minimal but not absent, the score of 0.5 was applied. Using an ocular grid, the score of each sample was counted in at least 15-25 consecutive fields across a full section (x 400 magnification) and was averaged for each graft.

Interstitial PSR staining for collagen was assessed by point counting using an ocular grid in at least 15 consecutive fields (x 400 magnification). Only interstitial collagen was counted, with collagen surrounding vessels and glomeruli excluded. The result was expressed as the number of interstitial grid points positive over the total number of interstitial grid points assessed per field.

Immunohistochemistry staining

Acetone-fixed frozen sections (7 µm) were exposed to 0.06% H2O2 in PBS for 10 minutes, and subsequently blocked with an avidin-biotin blocking system (DAKO North America Inc. Ca., USA.) followed by 20% normal horse serum in PBS. Primary antibody consisting of rat anti-mouse CD68 antibody (clone FA-11, AbD

Serotec MCA1957), CD4 (clone RM4-5, BD Pharmingen 550280), CD8 (clone 53-6.7, BD Pharmingen

550281), FoxP3 (clone FJK-16s eBioscience 14-5773-82), or hamster anti-mouse CD11c (clone HL3, BD

Pharmingen 550283) was applied to the sections for 60 min. Concentration-matched IgG was used as an isotype negative control. Sections were incubated with the appropriate biotinylated secondary antibody: anti- rat IgG or anti-hamster IgG (BD Pharminogen). Vector stain ABC kit (Vector Laboratories Inc.) was applied to the tissue followed by 3,3’diaminobenzidine (DAB) substrate-chromogen solution (DAKO North America

Corporation Inc. CA., USA.) Slides were counterstained with Harris’ haematoxylin.

Quantification of immunohistochemistry

Analysis of the cellular infiltrates for CD4, CD8 and Foxp3 was performed in a blinded manner, by assessing

20 consecutive high-power fields (HPFs, x 400 magnification) of the cortex in each section. Using an ocular grid, the number of cells staining positively for each antibody was counted and expressed as cells per HPF.

Analysis of CD68 and CD11c infiltrates was performed using a digital image analysis program (Image-Pro

Premier 9.0, Media Cybernetics). An area of cortex was analyzed for interstitial cellular positive staining versus counter-stained area. The results were expressed as percentage of positive staining per HPF.

Immunofluorescence

For C4d immunofluorescent staining, frozen sections were blocked with 1% BSA in PBS for 20 minutes and incubated with rat anti-mouse antibodies to C4d (Abcam plc, Cambridge, UK) for 60 min followed by anti- rat IgG conjugated with AlexaFluor 488 (Molecular Probes, Eugene, OR). Staining for C4d was considered positive when the peritubular capillaries were diffusely (all high-power fields) and brightly stained. Scoring of C4d staining was based on the percentage of stained tissue on immunofluorescence that had a linear, circumferential staining pattern in PTCs following the Banff 97 scoring criteria on a scale of 0 to 3: 0 =

Negative: 0%; 1 = Minimal C4d stain/detection: 1<10%; 2 = Focal C4d stain/positive: 10–50%; 3 = Diffuse

C4d stain/positive: >50%.

Supplemental Figure 1. Multiple sample rarefaction curve based on 16S rRNA gene sequencing.

HF n=12; NC n=12; NC+Allo n=10; HF+Allo n=16

500 NS ** ***

s 400 hnes c 300 Ri

200

HF NC HF+Allo NC+Allo Iso Pre Isograft Supplemental Figure 2. Richness of gut microbial communities in allograft and isograft mice.

15.56% Verrucomicrobia 1.43% Proteobacteria 44.40% Firmicutes 35.01% Bacteroidetes 1.88% Actinobacteria 1.71% Other

ZF

Supplemental Figure 3. Dominant phylum of WT C57BL/6 mice fed a zero-fiber diet. WT mice fed a fiber restricted diet develop dysbiosis with expansion of the pylum Verrucomicrobia. (n=10)

Supplemental Figure 4. Cytokine and Chemokine mRNA expression in WT and WT+HF allografts, and WT isografts at day 100 post-transplant. Similar to WT-allografts, HF fed allograft mice demonstrated a marked upregulation of cytokines, chemokines, and genes involved in tissue remodeling as compared to isografts. WT+HF mice demonstrated a decrease in the expression of chemokine CXCL9 as compared to WT allograft mice (P<0.05). WT n=9, WT+HF n=9, isografts n=5. P values by one-way

ANOVA. *P<0.05

Supplemental Figure 5. HDAC activity in transplanted kidneys was not upregulated by SA supplementation . Compared to WT allograft mice, WT+SA allograft mice did not demonstrate a significant change in HDAC activity (P=0.2731). WT n=12, WT+SA n=9, Iso n=5. P values by one-way

ANOVA.

Nutritional Parameter Normal Chow High-Fiber

Protein (%) 19 13.2

Total Fat (%) 4.6 4.5

Crude Fiber (%) 5.2 35.0

Acid Detergent Fiber (%) - 35.0

Digestible Energy (MJ/kg) 14.2 11.0

Total Calculated Energy from Carbohydrate (%) 59.9 58.7

Total Calculated Energy from Protein (%) 23 19.7

Total Calculated Energy from Lipids (%) 12 15.0

Supplemental Table 1. Nutritional parameters of high-fiber and normal mouse chow used in experiments.

Deseq2: Significant OTUs Pre-Isograft vs Isograft Mice

baseMean log2FoldChange lfcSE stat pvalue padj Rank2 Rank3 Rank4 Rank5 Rank6

37d7 269.7341 -1.4682 0.33194 -4.42308 9.73E-06 0.00035 p__Bacteroidetes c__Bacteroidia o__Bacteroidales f__Porphyromonadaceae g__Parabacteroides Supplemental Table 2. DESeq2 analysis demonstrating differential abundance of significant OTUs at the genus level (FDR adjusted p value < 0.01) between isograft recipients, pre and 2 weeks following isograft-placement. Deseq2: Significant OTUs NC v HF

baseMean log2FoldChange lfcSE stat pvalue padj Rank2 Rank3 Rank4 Rank5 Rank6

180107 337.997638 1.25712714 0.41221822 3.04966421 0.00229097 0.00646863 p__Firmicutes c__Clostridia o__Clostridiales f__Ruminococcaceae g__Ruminococcus

323024 37.5128584 6.44311967 1.27498867 5.0534721 4.34E-07 2.97E-06 p__Tenericutes c__Mollicutes o__RF39 f__ g__

263705 8.84691163 4.82531286 0.83718449 5.76373896 8.23E-09 9.87E-08 p__Firmicutes c__Clostridia o__Clostridiales f__Peptococcaceae g__

363731 4901.67386 -4.0497423 0.66609898 -6.0797906 1.20E-09 1.93E-08 p__Verrucomicrobia c__Verrucomicrobiae o__Verrucomicrobiales f__Verrucomicrobiaceae g__Akkermansia

180869 41.5894473 4.16699017 1.22476158 3.40228683 0.00066824 0.00229112 p__Firmicutes c__Erysipelotrichi o__Erysipelotrichales f__Erysipelotrichaceae g__

444791 719.772719 2.34417242 0.62334656 3.7606246 0.00016949 0.00062581 p__Cyanobacteria c__4C0d-2 o__YS2 f__ g__

780650 72.0221425 4.52100366 0.68796399 6.57157021 4.98E-11 1.19E-09 p__Firmicutes c__Clostridia o__Clostridiales f__Clostridiaceae g__

1684221 283.245117 -1.5515791 0.49489642 -3.1351593 0.00171761 0.00515282 p__Proteobacteria c__Deltaproteobacteria o__Desulfovibrionales f__Desulfovibrionaceae g__Desulfovibrio

OTU220 283.016401 1.89250587 0.55987963 3.38020134 0.00072433 0.00231785 p__Proteobacteria c__Alphaproteobacteria o__RF32 f__ g__

1136443 27.1575525 -4.9111887 1.03285481 -4.7549653 1.98E-06 1.19E-05 p__Deferribacteres c__Deferribacteres o__Deferribacterales f__Deferribacteraceae g__Mucispirillum

g__Candidatus 22668 15.9244237 4.72749781 1.05258323 4.49132923 7.08E-06 3.40E-05 p__Firmicutes c__Clostridia o__Clostridiales f__Clostridiaceae Arthromitus

1107027 5549.48278 -2.5278564 0.54581097 -4.6313772 3.63E-06 1.94E-05 p__Firmicutes c__Bacilli o__Lactobacillales f__Lactobacillaceae g__Lactobacillus

997439 8487.29121 -3.6404419 0.68481853 -5.3159221 1.06E-07 8.49E-07 p__Actinobacteria c__Actinobacteria o__Bifidobacteriales f__Bifidobacteriaceae g__Bifidobacterium

338644 48.4806212 1.26437036 0.30396756 4.15955682 3.19E-05 0.00013914 p__Actinobacteria c__Coriobacteriia o__Coriobacteriales f__Coriobacteriaceae g__Adlercreutzia

589277 5686.25973 -3.5039549 0.41566534 -8.42975 3.46E-17 1.66E-15 p__Bacteroidetes c__Bacteroidia o__Bacteroidales f__Bacteroidaceae g__Bacteroides

839200 200.959355 3.09877931 0.56380796 5.49616099 3.88E-08 3.73E-07 p__Firmicutes c__Clostridia o__Clostridiales f__Lachnospiraceae g__Dorea

372622 158.313804 1.66402245 0.43192768 3.85254873 0.00011689 0.00046758 p__Firmicutes c__Clostridia o__Clostridiales f__Lachnospiraceae g__Coprococcus

Supplemental Table 3. DESeq2 analysis demonstrating differential abundance of significant OTUs at the genus level (FDR adjusted p value < 0.01) between NC and HF fed mice

Deseq2: Significant OTUs NC+Allo v HF+Allo

baseMean log2FoldChange lfcSE stat pvalue padj Rank2 Rank3 Rank4 Rank5 Rank6

696563 76.412471 -26.48346386 3.068907 -8.62960562 6.16E-18 1.48E-16 p__Firmicutes c__Clostridia o__Clostridiales f__Lachnospiraceae g__Blautia

187768 8867.6480 -1.915591554 0.361279 -5.30223721 1.14E-07 6.86E-07 p__Firmicutes c__Clostridia o__Clostridiales f__ g__

180107 337.99763 1.418430243 0.402489 3.52414028 0.000424859 0.0014566 p__Firmicutes c__Clostridia o__Clostridiales f__Ruminococcaceae g__Ruminococcus

264240 1464.7510 -1.681575223 0.363969 -4.6201005 3.84E-06 1.84E-05 p__Bacteroidetes c__Bacteroidia o__Bacteroidales f__Rikenellaceae g__

OTU554 61.153891 9.339921041 0.971742 9.61152050 7.15E-22 3.43E-20 p__Bacteroidetes c__Bacteroidia o__Bacteroidales f__Rikenellaceae g__Rikenella

323024 37.512858 6.678376416 1.301158 5.13263856 2.86E-07 1.52E-06 p__Tenericutes c__Mollicutes o__RF39 f__ g__

263705 8.8469116 2.747841117 0.742125 3.70266291 0.000213348 0.0007877 p__Firmicutes c__Clostridia o__Clostridiales f__Peptococcaceae g__

363731 4901.6738 2.269838261 0.649579 3.49432151 0.000475268 0.0015208 p__Verrucomicrobia c__Verrucomicrobiae o__Verrucomicrobiale f__Verrucomicrobiacea g__Akkermansia

OTU152 115.85461 4.825780024 0.774636 6.22973761 4.67E-10 4.49E-09 p__Firmicutes c__Bacilli o__Bacillales f__Staphylococcaceae g__Staphylococcus

780650 72.022142 3.797212847 0.662794 5.7290925 1.01E-08 6.92E-08 p__Firmicutes c__Clostridia o__Clostridiales f__Clostridiaceae g__

1136443 27.157552 -7.97276023 1.098077 -7.26065628 3.85E-13 4.62E-12 p__Deferribacteres c__Deferribacteres o__Deferribacterales f__Deferribacteraceae g__Mucispirillum

g__Candidatus 22668 15.924423 4.757650907 1.050405 4.52934675 5.92E-06 2.58E-05 p__Firmicutes c__Clostridia o__Clostridiales f__Clostridiaceae Arthromitus

OTU45 40.638197 10.70153101 2.645877 4.0446050 5.24E-05 0.0002096 p__Firmicutes c__Bacilli o__Turicibacterales f__Turicibacteraceae g__Turicibacter

997439 8487.2912 -2.14056513 0.668522 -3.2019329 0.001365088 0.0040952 p__Actinobacteria c__Actinobacteria o__Bifidobacteriales f__Bifidobacteriaceae g__Bifidobacterium

342873 807.37847 -3.130335277 0.509779 -6.14056337 8.22E-10 6.58E-09 p__Bacteroidetes c__Bacteroidia o__Bacteroidales f__Porphyromonadacea g__Parabacteroide

589277 5686.2597 -3.198392568 0.405763 -7.88239948 3.21E-15 5.14E-14 p__Bacteroidetes c__Bacteroidia o__Bacteroidales f__Bacteroidaceae g__Bacteroides

Supplemental Table 4. DESeq2 analysis demonstrating differential abundance of significant OTUs at the genus level (FDR adjusted p value < 0.01) between NC+Allo and HF+Allo mice