Gallbladder-Derived Surfactant Protein D Regulates Gut Commensal Bacteria for Maintaining Intestinal Homeostasis

Gallbladder-derived surfactant protein D regulates gut commensal bacteria for maintaining intestinal homeostasis

Hana Sarashina-Kidaa,b,1, Hideo Negishia,1, Junko Nishioa,1, Wataru Sudac,d,e, Yuki Nakajimaa, Mika Yasui-Katoa, Keiko Iwaisakof, Sujin Kangb,g, Nobuyasu Endoa, Hideyuki Yanaia,h, Masataka Asagirif,i, Hiroshi Kidab, Masahira Hattoric,e,j, Atsushi Kumanogohb, and Tadatsugu Taniguchia,h,2

aDepartment of Molecular Immunology, Institute of Industrial Science, The University of Tokyo, Tokyo 153-8505, Japan; bDepartment of Respiratory Medicine and Clinical Immunology, Osaka University Graduate School of Medicine, Suita, Osaka 565-0871, Japan; cLaboratory for Microbiome Sciences, RIKEN Center for Integrative Medical Sciences, Yokohama, Kanagawa 230-0045, Japan; dDepartment of Microbiology and Immunology, Keio University School of Medicine, Tokyo 160-8582, Japan; eLaboratory of Metagenomics, Department of Computational Biology and Medical Science, Graduate School of Frontier Sciences, The University of Tokyo, Kashiwa, Chiba 277-8561, Japan; fDepartment of Target Therapy Oncology, Graduate School of Medicine, Kyoto University, Kyoto 606-8501, Japan; gDepartment of Immune Regulation, Immunology Frontier Research Center, Osaka University, Suita, Osaka 565-0871, Japan; hMax Planck–The University of Tokyo Center for Integrative Inflammology, Tokyo 153-8505, Japan; iDepartment of Pathobiology, Graduate School of Pharmaceutical Sciences, Nagoya City University, Nagoya, Aichi 467-8603, Japan; and jGraduate School of Advanced Science and Engineering, Waseda University, Tokyo 169-8555, Japan

Contributed by Tadatsugu Taniguchi, August 13, 2017 (sent for review July 21, 2017; reviewed by Ruslan Medzhitov and Satoshi Uematsu) The commensal microbiota within the gastrointestinal tract is essential transmitted pathogens (5–7). SP-D binds carbohydrates, lipids, and in maintaining homeostasis. Indeed, dysregulation in the repertoire nucleic acids with a broad spectrum of specificity (5–7). To date, the of microbiota can result in the development of intestinal immune– physiological role of SP-D in other organs remains totally unknown. inflammatory diseases. Further, this immune regulation by gut micro- In this study, we show that SP-D is produced by the gallbladder biota is important systemically, impacting health and disease of organ and plays a critical role in maintaining symbiotic status of the systems beyond the local environment of the gut. What has not been commensal bacteria. We show SP-D expression in the gallbladder explored is how distant organs might in turn shape the microbiota via epithelial cells and provide evidence that the gallbladder-produced microbe-targeted molecules. Here, we provide evidence that surfac- SP-D, secreted into the bile and delivered to the intestinal lumen, tant protein D (SP-D) synthesized in the gallbladder and delivered is the major source of intestinal SP-D. Our results also indicate into intestinal lumen binds selectively to species of gut commensal that SP-D selectively affects populations of commensal bacterial bacteria. SP-D–deficient mice manifest intestinal dysbiosis and show a species through its direct binding to the bacteria, which interferes susceptibility to dextran sulfate sodium-induced colitis. Further, fecal transfer from SP-D–deficient mice to wild-type, germ-free mice con- with their replication and causes a cascade of events that skew the repertoire of bacterial species in the intestine. veyed colitis susceptibility. Interestingly, colitis caused a notable in- −/− crease in Sftpd gene expression in the gallbladder, but not in the Indeed, mice deficient in the Sftpd gene (Sftpd mice) show lung, via the activity of glucocorticoids produced in the liver. These intestinal dysbiosis and sensitization to dextran sulfate sodium findings describe a unique mechanism of interorgan regulation of (DSS)-induced colitis. The importance of the dysbiosis resulting intestinal immune homeostasis by SP-D with potential clinical implications such as cholecystectomy. Significance

surfactant protein D | gut microbiota | gallbladder | colitis | glucocorticoids The imbalance in the intestinal microbiota repertoire, called dysbiosis, underlies the development of inflammatory diseases in he composition of the gut commensal microbiota reflects the the intestine and distant organ systems. Whether and how dis- Tcoevolution of host and microorganisms to achieve intestinal tant organs regulate the commensal gut bacteria and intestinal homeostasis that is symbiotic or mutually beneficial (1, 2). In this homeostasis via microbe-targeted molecules has remained poorly context, local interactions between commensal microbiota and in- explored. In this study, we show that the gallbladder regulates testinal cells are central to the regulation of the intestinal immune intestinal immune homeostasis through the production of sur- system (1, 2). Imbalances in the microbiota, termed dysbiosis, un- factant protein D (SP-D) that is delivered to the intestine via bile. derlie the development and/or aggravation of immune-related dis- The importance of the gallbladder-derived SP-D was underscored eases even in distant organs from the gastrointestinal tract (1, 2). by an intestinal dysbiosis in SP-D–deficient mice and susceptibility Given the importance of this connection to human health, there has of these mice to dextran sodium sulfate sodium-induced colitis. been much focus on the elucidation of the mechanisms by which the Our study thus revealed a unique interorgan regulation of in- symbiosis is maintained. testinal homeostasis by SP-D with potential clinical implications One focus on this process has been the study of soluble, im- such as cholecystectomy. munoregulatory molecules produced in the intestinal lumen by Author contributions: H.S.-K., H.N., J.N., A.K., and T.T. designed research; H.S.-K., H.N., immune and epithelial cells that maintain intestinal immune ho- J.N., W.S., Y.N., M.Y.-K., K.I., S.K., and N.E. performed research; H.K. and M.H. contributed meostasis (1, 3). However, in this context, little attention has been new reagents/analytic tools; H.S.-K., H.N., J.N., W.S., Y.N., M.Y.-K., K.I., S.K., N.E., H.Y., paid to the role of immunoregulatory molecules produced by dis- M.A., H.K., M.H., A.K., and T.T. analyzed data; and H.N., J.N., and T.T. wrote the paper. tant organs. During the course of our study on the role of the IRF3 Reviewers: R.M., Yale University School of Medicine; and S.U., Chiba University. transcription factor in intestinal homeostasis (4), we observed the The authors declare no conflict of interest. presence of surfactant protein D (SP-D) in the intestinal lumen. Data deposition: The sequence reported in this paper has been deposited in the GenBank SP-D belongs to the collectin family of soluble innate immune database (accession no. BioProjectID PRJDB5950). receptors and is a component of lung surfactant proteins that are 1H.S.-K., H.N., and J.N. contributed equally to this work. continuously and specifically produced in the lungs by alveolar type 2To whom correspondence should be addressed. Email: [email protected]. – II epithelial cells (5 7). Thus, SP-D protein is constitutively se- This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. creted into the alveolar fluid for the protection against aerially 1073/pnas.1712837114/-/DCSupplemental.

10178–10183 | PNAS | September 19, 2017 | vol. 114 | no. 38 www.pnas.org/cgi/doi/10.1073/pnas.1712837114 Downloaded by guest on October 1, 2021 from the absence of SP-D is further underscored by fecal transfer OTU00840) and Lactobacillus animalis (OTU01291) (Fig. 2 B and − − − − experiments, wherein germ-free mice receiving feces from Sftpd / C), whose relative amounts were increased in Sftpd / mice (Fig. mice are more susceptible to DSS-induced colitis compared with S1), whereas such binding was not observed with OTUs closely + − those that received the feces of Sftpd / mice. Of note, there was a related to Clostridia IV and XIVa (OTU00194 and OTU00272) significant reduction of Foxp3 expression in regulatory T cells (Fig. 2C), which showed a decrease in these mice (Fig. S1). Ad- − − (Tregs) in the intestinal lamina propria of Sftpd / mice, which ditionally, OTU00009 (Lactobacillus johnsonii) and OTU00069 may account for the increased susceptibility to colitis. (Lactobacillus reuteri), which were unchanged in mutant mice, In addition, we also found a notable elevation of SP-D expres- did not show significant binding to SP-D (Fig. 2C). sion during the development of DSS-induced colitis in the gall- When we next examined the effects of recombinant SP-D (rSP-D) bladder, but not in the lung. We provide evidence that colitis on the growth of L. murinus in vitro, a dose-dependent inhibition enhances synthesis of glucocorticoids in the liver, which is secreted was observed, suggesting that SP-D directly interferes with the into the bile to induce Sftpd mRNA expression in gallbladder growth of L. murinus in the intestine (Fig. 2D, Left). Growth in- epithelial cells. Taken together, our findings reveal the hitherto hibition was also observed against another L. murinus strain unknown role of SP-D in the interorgan regulation of intestinal ATCC35020 (Fig. S4A). Of note, we found features of cell lysis of commensal microbiota. We discuss the significance of these find- L. murinus cells incubated with rSP-D using analysis by scanning ings as well as the clinical implications such as cholecystectomy. electron microscopy (Fig. S4B). In contrast, no growth inhibition was observed for L. johnsonii, which is not bound by SP-D (Fig. 2 C Results and D, Right). In this regard, it is interesting that L. murinus in- Expression of SP-D in Gallbladder. When we examined relative Sftpd terferes with the growth of the Clostridia species in vitro (Fig. S5). gene expression in various organs by quantitative RT-PCR (qRT- PCR), a notably high expression level of Sftpd mRNA was ob- Development of Intestinal Immune Cells in the Absence of SP-D. As served in the gallbladder compared with other digestive organs, changes in the composition of intestinal microbiota can affect the including the stomach, intestines, appendix, liver, and pancreas immune system (1, 2), we next asked whether immune cell pop- (Fig. 1A). Further, SP-D protein expression was detected by im- ulations in the intestinal lamina propria are altered by the absence munohistochemical staining in the gallbladder epithelium, but not of SP-D. No overt change was observed in the frequencies of − − in lamina propria (Fig. 1B). As shown in Fig. 1C, SP-D protein was immune cells in the intestinal lamina propria of Sftpd / mice (Fig. − − measured in the bile, suggesting that it reaches the intestinal lu- S6). However, the majority of Tregs in Sftpd / mice showed sig- men through its secretion into the bile. Consistent with this, bile nificantly lower expression levels of Foxp3 compared with Tregs duct ligation performed to inhibit bile flow into the intestine from control mice (Fig. 3A). On the other hand, no abnormality resulted in a marked reduction in SP-D protein levels in the fecal was observed in Tregs in the spleen of the mutant mice (Fig. 3B). suspension (Fig. 1D), indicating that gallbladder is the main These data suggest that intestinal dysbiosis caused by the absence source of SP-D found in the intestinal lumen. of SP-D selectively affects the development of intestinal Tregs, which are shown to be critical to the regulation of intestinal in- Alteration of Intestinal Bacterial Population by the Absence of SP-D. flammatory responses (12). Since SP-D has been reported to play an important role in the

elimination of pathogenic bacteria such as Pseudomonas aeruginosa Susceptibility of the SP-D–Deficient Mice to Experimental Colitis. INFLAMMATION IMMUNOLOGY AND and Staphylococcus aureus within the alveoli of the lungs (5, 7), we Given the intestinal dysbiosis that is associated with abnormal − − asked if expression of SP-D affects populations of commensal Treg development in Sftpd / mice, we next examined the ability bacteria in the intestine. To address this question, we first per- of these mice to control colonic inflammation. A model of DSS- − − formed a comparative microbiome analysis on feces from wild-type induced colitis showed that Sftpd / mice had a marked increase in − − (WT) and Sftpd / mice. the susceptibility to colitis compared with WT mice as evidenced Interestingly, significant differences in the bacterial population by a rapid decrease in body weight (Fig. 3C) and shortened length − − and diversity were observed between WT and Sftpd / mice (Fig. of the colon (Fig. 3D). Consistent with this, disease activity index − − 1E and Fig. S1 A and B). Since the sequence analysis of 16S ri- (DAI) was significantly higher in Sftpd / mice (Fig. 3E). bosomal RNA (rRNA) revealed that most of the altered bacteria To further examine the link between intestinal dysbiosis and − − species have been poorly characterized, these data are presented aggravated colitis in Sftpd / mice, we next performed a fecal + − using operational taxonomic units (OTUs) (Fig. 1F and Fig. S1C). microbiota transplant experiment in which feces from Sftpd / or − − In this analysis, we observed that species more closely related to Sftpd / mice were transplanted to germ-free WT mice and then Clostridia cluster IV and XIVa were markedly decreased while challenged with DSS. As shown in Fig. 3F, a significant loss of the species more closely related to the Lactobacillus salivarius cluster body weight was observed in the mice that received the feces from − − − − + − were increased in Sftpd / mice. Among the latter cluster, Sftpd / mice compared with those that received feces from Sftpd / OTU00007 is Lactobacillus murinus as the sequence showed a mice. The different phenotype observation in the germ-free mice similarity of 100% (8). On the other hand, commensal fungi, supports the notion that, in this experimental condition, the which have also been implicated to be bound by SP-D (9, 10), were status of microbiota remains unchanged in the germ-free mice, not detected in our specific-pathogen-free (SPF) mice, an obser- which have the intact Sftpd gene. These results indicate that the − − vation similarly reported in the SPF mice (11) (Fig. S2). dysbiotic microbiota present in Sftpd / mice indeed accounted for the susceptibility to the colitis development, at least in part. SP-D Binding to Intestinal Bacteria and Its Effect on Bacterial Growth. To gain further insight into the intestinal dysbiosis resulting from Interorgan Regulation of Sftpd Gene Expression in Colitis Development. the absence of SP-D, we next examined SP-D binding to com- In view of the importance of gallbladder-derived SP-D to control mensal bacteria in the feces from WT mice. For this, fecal bacteria DSS-induced colitis, we then examined whether intestinal in- from WT mice were stained with anti–SP-D antibody and sub- flammation affected Sftpd gene expression in the gallbladder. jected to flow cytometric analysis. As shown in Fig. 2A, commensal Interestingly, we found that Sftpd mRNA and SP-D protein were microbiota contained about 2% of SP-D binding bacteria and 16S markedly enhanced in the gallbladder but not lungs of WT mice rRNA sequencing further revealed that 61 OTUs were increased made to develop DSS-induced colitis (Fig. 4A and Fig. S7A). in SP-D–positive fraction with significant difference compared Since the Sftpd gene is known to be induced by fibroblast growth with SP-D–negative fraction (Fig. S3). Among them, significant factors (FGFs), glucocorticoids, and endotoxin in the lung epi- SP-D binding was observed with L. murinus (OTU00007 and thelial cells (13–15), we tested whether these signaling pathways

Sarashina-Kida et al. PNAS | September 19, 2017 | vol. 114 | no. 38 | 10179 Downloaded by guest on October 1, 2021 ACB *** 1.5 *

1.0 Control 0.2 IgG

*** 0.1

NS anti-SP-D SP-D (ng/μL) Relative expression mAb 0.0 Lung St SI LIApp Pan Liver GB WT Sftpd -/- DE Sham BDL 60 0.3 0.7 PrePost Pre Post 0.2 *** *

40 0.1 0.6 2 SP-D 0 PCo 20 -0.1 0.5 Pixel density

-0.2

Unweighted Unifrac distance 0 0 -0.3 PrePost Pre Post -0.3 -0.2 -0.1 0 0.1 0.2 0.3 KO KOWT F Sham BDL PCo1 KO WTWT 0.01 OTU00011 Clostridium sp. ASF356 (95.87%) Clostridium sp. ASF356 OTU00529 Clostridium hathewayi (85.37%) x OTU00155 Clostridium bolteae (88.82%) x OTU00128 Clostridium bolteae (87.96%) x OTU01895 Clostridium bolteae (87.65%) x Clostridium hathewayi OTU00283 Clostridium symbiosum (93.52%) x Clostridium symbiosum WAL-14163 Clostridium bolteae ATCC BAA-613 Clostridium bolteae Fig. 1. Gallbladder-specific expression of the Sftpd −/− OTU00194 Eubacterium coprostanoligenes (84.4%) x gene in digestive organs and dysbiosis in Sftpd Eubacterium coprostanoligenes mice. (A) Quantitative RT-PCR (qRT-PCR) analysis of OTU00272 (93.21%) x Oscillibacter valericigenes Sftpd mRNA expression in lungs, stomach (St), small Oscillibacter valericigenes Sjm18-20 OTU01169 Lactobacillus johnsonii (96.46%) intestine (SI), large intestine (LI), appendix (App), Lactobacillus johnsonii FI978 pancreas (Pan), liver, and gallbladder (GB) from wild- OTU01291 Lactobacillus animalis (96.99%) type (WT) mice (n = 5). (B) Immunohistochemical OTU00840 (96.69%) Lactobacillus murinus analysis of SP-D protein in gallbladder from WT mice. Lactobacillus animalis KCTC 3501 μ OTU00007 Lactobacillus murinus (100%) (Scale bar, 100 m.) (C) Concentration of SP-D pro- Lactobacillus murinus tein in the bile from WT (n = 5) or Sftpd-deficient OTU00288 Alistipes senegalensis (90.57%) (Sftpd−/−)mice(n = 4). (D) Immunoblot analysis of SP-D Alistipes senegalensis JC50 OTU00121 Rikenella microfusus (86.35%) protein in the feces from WT mice with sham oper- Rikenella microfusus ation or bile duct ligation (BDL) (Left). Fresh fecal Barnesiella viscericola suspensions from these mice were subjected to im- Barnesiella intestinihominis munoblot analysis after immunoprecipitation (IP) OTU01148 (83.18%) Barnesiella intestinihominis – OTU00201 Prevotella sp. oral taxon 317 (79.94%) with anti SP-D antibody. Relative band intensity is OTU00285 Prevotella oris (79.94%) shown (Right). (E) Principal coordinate analysis OTU00728 Barnesiella viscericola (80.31%) (PCoA) of unweighted UniFrac distance in the fecal OTU02031 Bacteroides acidifaciens (83.86%) = −/− OTU00183 (99.37%) microbiota from WT (blue, n 8) and Sftpd mice Bacteroides acidifaciens = Bacteroides acidifaciens (red, n 9) (Left). Mean unweighted UniFrac dis- Prevotella oris F0302 tances within and between the groups is shown Prevotella sp. P2A FAAD4 (Right). (F) Phylogenetic tree analysis of operational OTU00402 (84.91%) Prevotella sp. P2A FAAD4 taxonomic units (OTUs) whose read counts are sig- OTU00151 Prevotella genomosp. C1 (85.89%) OTU00181 Prevotella shahii (86.12%) nificantly increased (red) or decreased (green) in − − Prevotella shahii Sftpd / mice compared with WT mice. Bacteria on Prevotella sp. oral taxon 317 str. F0108 the yellow background belong to Firmicutes and Prevotella genomosp. C1 those on the green background belong to Bacter- x LQGLFDWHVEDFWHULDVSHFLHVFODVVLILHGLQWRClostridia Cluster,;RU;,9D oidetes.*P < 0.05, ***P < 0.001. NS, not significant.

could induce Sftpd mRNA in a gallbladder-derived epithelial cell increase (Fig. 4C, Left). This was also accompanied by a significant line, OCUG-1. As shown in Fig. 4B and Fig. S7B, we found Sftpd increase in the volume of bile (Fig. 4C, Right). Further, cortico- mRNA is induced in a dose-dependent manner by synthetic sterone levels were also increased in the liver under conditions of glucocorticoids, but neither by FGFs nor LPS. Interestingly, this intestinal inflammation, but not within the gallbladder itself (Fig. induction was not observed in a hepatocyte cell line Huh7 (Fig. 4D), suggesting the local production of corticosterone occurs S7C), suggesting the cell-type-specific Sftpd gene induction by specifically in the liver. Indeed, corticosterone levels were en- glucocorticoids. hanced in the supernatants of liver tissues cultured from DSS- Consistent with this, DSS-induced colitis enhanced the con- treated mice, which was suppressed by addition to the cultures centrations of corticosterone, a major glucocorticoid in mouse, of metyrapone, an inhibitor for corticosterone-producing enzyme, in the bile, while serum concentrations exhibited only a slight Cyp11b1 (16) (Fig. 4E). Thus, these findings suggest a feedback

10180 | www.pnas.org/cgi/doi/10.1073/pnas.1712837114 Sarashina-Kida et al. Downloaded by guest on October 1, 2021 A Ctrl IgG anti-SP-D *** 3 96.6 4.78e-3 70.8 2.85

2 FL1 1 % in all bacteria

0 APC Ctrl IgG anti-SP-D B #1 #2 #3 #4 #5 #6 #7 #8 #9 #10 OTU00007 Lactobacillus murinus (100%) ** OTU01291 Lactobacillus animalis (96.99%) * OTU00840 Lactobacillus murinus (96.99%) ** OTU00201 Prevotella sp. oral taxon 317 (79.94%) OTU00402 Prevotella sp. P2A_FAAD4 (84.91%) OTU02031 Bacteroides acidifaciens (83.86%) OTU00288 Alistipes senegalensis (90.57%) OTU00155 Clostridium bolteae (88.82%) OTU00728 Barnesiella viscericola (80.31%) OTU00121 Rikenella microfusus (86.35%) Fig. 2. Binding of SP-D protein to intestinal bacteria OTU01169 Lactobacillus johnsonii (96.49%) and its effect on their growth. (A) Flow cytometric analysis of SP-D binding bacteria in the feces from 100 101 102 103 (Read counts) WT mice. SP-D binding bacteria are stained with control (Ctrl) IgG or anti–SP-D antibody (Left). Sta- C -/- -/- Up in Sftpd mice Unchange in Sftpd mice tistical result is shown (Right)(n = 6). (B and C)16S OTU00007 OTU01291 OTU00840 OTU00009 OTU000069 ribosomal RNA (rRNA) sequencing analysis on SP-D– 4000 ** 15 * 4000 ** 2000 2000 positive or -negative bacteria sorted from fecal suspension of WT mice (n = 10). The binding to SP-D of 3000 3000 1500 1500 − − 10 bacteria (OTUs) significantly increased in Sftpd / 2000 2000 1000 1000 mice (shown in Fig. S1) are displayed in the heat map. 5 Colors represent difference in read counts of each 1000 1000 Read counts 500 500 OTU between SP-D–positive and -negative bacteria. The difference was calculated by deducing read 0 0 0 0 0 SP-D SP-D SP-D SP-D SP-D SP-D SP-D SP-D SP-D SP-D counts of the OTU in SP-D–negative bacteria from

– INFLAMMATION -/- those SP-D positive bacteria (B). Read counts of rep- Down in Sftpd mice D IMMUNOLOGY AND L. murinus L. johnsonii resentative bacteria that increased (up), unchanged, OTU00194 OTU00272 * *** *** or decreased (down) in Sftpd−/− mice compared with 25 15 1.0 1.5 WT mice (Fig. 1E and Fig. S1), are shown as dot blot 20 (C). (D) Relative viability of L. murinus cultured in the 10 1.0 15 presence of BSA or recombinant SP-D (rSP-D) (45, 80, 0.5 μ 10 and 134 g/mL) for 8 h (Left). Relative viability of 5 0.5 L. johnsonii cultured in the presence of BSA or rSP-D Read counts 5 (80 μg/mL) for 8.5 h (Right). Relative cell viability was 0 0 Relative cell viability 0 0 assessed by 5-cyano-2,3-ditolyl tetrazolium chloride (µ SP-D SP-D SP-D SP-D 45 80 134 80 g/mL) (CTC) assay. *P < 0.05, **P < 0.01, ***P < 0.001.

regulation between intestine, liver, and gallbladder in which in- which is then delivered to the intestine to regulate the repertoire testinal inflammation can enhance corticosterone production by of commensal bacteria. Although it is speculative at this stage, the liver that induces SP-D expression by the gallbladder, which in our study also suggests there is an additional immunoregulatory turn can contribute to maintain symbiotic status of commensal molecule(s) in the bile involved in this process. bacteria in the intestines. The importance of SP-D as an interorgan mediator of immune − − homeostasis was underscored by intestinal dysbiosis in the Sftpd / Discussion mice (Fig. 1 E and F and Fig. S1). In fact, our results suggest SP-D has long been described as a surfactant protein involved in SP-D mediates the growth inhibition of commensal bacteria via a maintaining lung homeostasis (5–7). In this study, we revealed a direct binding to L. murinus and related bacteria (Fig. 2). This hitherto unknown role of SP-D in shaping microbiota in the in- was further confirmed by scanning electron microscopic analysis, testines. In this model, SP-D is constitutively produced in gall- which showed the lysis of L. murinus in the presence of SP-D bladder, but not in other digestive organs, and then delivered to (Fig. S4B). In this context, it is interesting that replication of two − − the intestines where it regulates the symbiotic status of commensal Clostridia species, whose numbers were increased in Sftpd / bacteria. mice, was inhibited by L. murinus in a dose-dependent manner Numerous antibacterial peptides that control commensal gut (Fig. S5). Thus, we can offer one explanation for the dysbiosis in microbiota via their inhibition of certain sets of bacteria have that SP-D inhibits replication of L. murinus, which otherwise been described (1, 3). The conventional wisdom has been that suppresses the growth of Clostridia species. While Lactobacillus they are synthesized locally by immune or epithelial cells in the species are generally thought to be “beneficial” bacteria due to gut (1, 3). In contrast to this view, our findings reveal a facet of their health-promoting effects (17–20), our results suggest that intestinal microbiota regulation in which an immunoregulatory their overgrowth, e.g., by the absence of SP-D, affects other com- molecule, SP-D, is produced in a distant organ, i.e., gallbladder, mensal bacteria, thereby causing dysbiosis.

Sarashina-Kida et al. PNAS | September 19, 2017 | vol. 114 | no. 38 | 10181 Downloaded by guest on October 1, 2021 ABfeedback regulation between the intestines, liver, and gallbladder CD4+T cell-gated 1.5 1.5 organs in response to intestinal immune inflammation. SP-D is 100 *** Sftpd+/- Sftpd+/- involved in this feedback mechanism, as its expression is enhanced 80 Sftpd-/- Sftpd -/- 1.0 1.0 by glucocorticoids such as corticosterone produced by the liver and 60 delivered to gallbladder in response to colitis (Fig. 4 B–D). In fact, 40 0.5 0.5 an elevation of biliary corticosterone was observed showing there 20 is a cross-talk between these organs (Fig. 4C). To our knowledge, Relative MFI of Foxp3 Relative MFI of Foxp3 0 0 0 glucocorticoid synthesis by the liver was previously unknown. En- 0102 103 104 105 α Foxp3 dotoxin and inflammatory cytokines, such as LPS, TNF- , and IL- 1β, are known to activate local corticosterone production in skin C 105 D * Water DSS 100 * and lung (24). We found LPS administration in the mouse indeed WT Sftpd-/- WT Sftpd-/- 100 * * caused an enhancement of liver corticosterone accompanied by 95 Sftpd mRNA expression in the gallbladder (Fig. S8). Therefore, 90 50 the enterohepatic circulation-mediated delivery of LPS and/or 85 Body weight (%) other inflammatory molecules produced during the development 80 Colon length (%) 01234567891011 0 of colitis may account for the enhancement of liver corticosterone (day) WT Sftpd-/- production. E 12 ** F At present, the physiological significance of the feedback reg- ** * 100 ** ulation of SP-D during colitis development remains unclear. In ** ** 8 this context, it is interesting to note that in WT mice, the number 90

DAI of SP-D binding bacteria is significantly increased during DSS- 4 80 Sftpd+/- → GF induced colitis (Fig. S9A). Interestingly, however, the population Sftpd-/- → GF Body weight (%) of L. murinus was significantly decreased (Fig. S9B), which is 0 70 0123456789 012345678 consistent with the growth inhibition of L. murinus by SP-D (Fig. (day) (day) 2D). Thus, although further work will be required, a view emerges Fig. 3. Critical role of SP-D in the control of intestinal homeostasis. (A) Flow in which corticosterone-mediated induction of SP-D in the gall- + + cytometric analysis of Foxp3 protein expression in CD4 Foxp3 cells from bladder may be important for the proper balancing of intestinal + − − − colonic lamina propria of Sftpd / (n = 7) or Sftpd / mice (n = 6). Repre- bacteria for maintaining homeostasis. sentative histogram (Left) and pooled data of relative mean fluorescence intensity (MFI) of Foxp3 (Right) are shown. Relative MFI was calculated by dividing the MFI of each mouse by the average of MFI of Sftpd+/− mice. (B) Relative MFI of Foxp3 in CD4+Foxp3+ cells from spleen of Sftpd+/− (n = 4) and AB −/− = – Sftpd mRNA SP-D Sftpd mRNA Sftpd mice (n 3). (C E) Colitis was induced by oral administration of ** * 20 ** −/− 0.25 1000 3–3.5% of DSS. Body weight (WT, n = 5; Sftpd , n = 6) (C), pictures of colons ** 0.20 * 800 ** 15 on day 0 and 8 and reduction rate of colon length (n = 4pergroup)(D)and − − 0.15 600 disease activity indices (DAIs) (WT, n = 5; Sftpd / , n = 4) (E)areshown.(F)WT 10 0.10 400 germ-free mice were inoculated with the fecal suspension from Sftpd+/− (n = 5) SP-D (ng/mL) 5 −/− 0.05 200 = Relative expression

or Sftpd (n 5) mice. Colitis was induced by administration of 2.5% of DSS Relative expression 0 0 0 in drinking water for 6 d. *P < 0.05, **P < 0.01, ***P < 0.001. 0 48 0 48 0 244872244872244872244872 24 48 72 (h) (day) (day) (-) Dex FGF7 FGF10 EGF C On the other hand, we did not see a strong correlation between 15000 ** 30 *

SP-D binding to bacteria species and reduced prevalence of the µ g/μL) bacteria in vivo (Figs. S1 and S3). As such, a complex cascade of 10000 20 – *** events, involving interbacterial interactions and/or host bacterial 5000 10 NS interactions, may bring about intestinal dysbiosis in the absence Bile volume (μL)

Corticosterone ( 0 0 of SP-D. Although unlikely, it is not excluded that the absence of Water DSS Water DSS Water DSS SP-D may also affect bile composition, thereby contributing to the Serum Bile intestinal dysbiosis. DE * ** ** The composition of commensal bacteria in the gut is known to 300 10 100 affect development of specific immune cell subsets. This was the 200 case for our study in which we observed alterations of lamina 5 50 − − NS propria Tregs in Sftpd / mice. Indeed, Tregs were not significantly 100

reduced in absolute number, but showed lower expression of Foxp3 Corticosterone (ng/μL) 0 0 0 Water DSS Water DSS PFMET(-) (-) MET PF (Fig. 3A and Fig. S6). Since it has been shown that an attenuated Liver Gallbladder Water DSS Foxp3 expression in Tregs results in a loss of immunosuppressive Liver tissue culture activity of these cells (21, 22), exacerbation of DSS-induced co- −/− Fig. 4. Feedback regulation of Sftpd gene expression by colitis. (A) Sftpd litis found in Sftpd mice may be attributed, at least in part, to mRNA expression in the gallbladder (Left) and concentration of SP-D protein the abnormality of intestinal Tregs in these mice. It is interesting in the bile (Right) from WT mice (n = 4) treated with 3.5% of DSS for 4 or 8 d. in this context that functional Tregs are induced by Clostridia (B) Sftpd mRNA expression in gallbladder epithelial cell line OCUG-1 treated cluster IV and XIVa species in the intestine (23) and that some of with dexamethasone (Dex; 1 μM), FGF7 (100 μg/mL), FGF10 (100 μg/mL), or − − these species are reduced in Sftpd / mice (Fig. 1F and Fig. S1). EGF (100 μg/mL) for indicated time periods. (C) Corticosterone concentration The question for how this abnormal Treg development, within in the serum or bile from WT mice (n = 5) treated with 3.5% DSS for 7 d and the lamina propria, might contribute to the increased suscepti- water for an additional 1 d (Left). Volume of bile is shown at Right. −/− (D) Concentration of corticosterone in the liver (Left) or gallbladder (Right) bility to the DSS-induced colitis in Sftpd mice will be an im- from WT mice (n = 5) treated with 3.5% DSS for 9 d (Left). (E) Concentration portant issue to address in the future. of corticosterone in the tissue culture supernatant of liver from mice (n = 5) It is interesting that Sftpd gene expression is further enhanced by treated with 3.5% DSS for 8 d (Right). MET; 200 μg/mL of metyrapone, PF; DSS-induced colitis in the gallbladder (Fig. 4A), suggesting a 100 μM of PF-915275. *P < 0.05, **P < 0.01, ***P < 0.001. NS, not significant.

10182 | www.pnas.org/cgi/doi/10.1073/pnas.1712837114 Sarashina-Kida et al. Downloaded by guest on October 1, 2021 × Considering the critical role of SP-D in the protection of the 1mMCaCl2/PBS with 8,000 g and left on ice for 20 min. After washing in 1% lung against pathogens, it is possible that SP-D also plays an BSA/1 mM/CaCl2/PBS twice, it was resuspended and incubated for blocking in 2% important role in the protection against pathogen in the gall- BSA/1 mM/CaCl2/PBS for 20 min on ice. Bacterial concentration was adjusted by resuspending bacterial pellet from 10 mg feces in 100 μL of 2% BSA/1 mM/CaCl2/ bladder. As infection is the main cause of biliary tract in- μ μ – flammation such as cholecystitis and cholangitis (25), this issue PBS. The 100 L of bacterial suspension was incubated with 2 gofantiSP-D antibody (sd13980; Santa Cruz) or rabbit IgG for 15 min on ice. After washing may have clinical importance to be addressed in further studies. with 8,000 × g at 4 °C twice, bacterial suspension was resuspended in 100 μLof In addition, our findings may raise a possible concern for the 2% BSA/1 mM/CaCl2/PBS containing 1 μg of APC-conjugated anti-rabbit IgG treatment of gallbladder-related diseases such as gallstone and (sc8346; Santa Cruz) for 15 min on ice. After being washed, bacterial suspension gallbladder cancer. Usually, gallbladder removal or cholecystec- was subjected for sorting of SP-D–positive or -negative bacteria by FACSAria cell tomy is a common treatment of these diseases. Although the side sorter (BD). effect of cholecystectomy is thought to be modest, significant change of intestinal microbiota is reported in patients treated DSS-Induced Colitis. DSS-induced colitis was performed as described before with this procedure (26). Therefore, our study may have clinical with modification (4). Age-matched animals were administered 3.5% DSS (MP implication on the pathogenesis of intestinal diseases, possibly Biomedicals) in drinking water for 7–9 d. For body weight change studies, the with therapeutic potential of SP-D in intestinal diseases. percentage of body weight was determined by the following equation: 100 × (body weight at the indicated day)/(body weight before DSS treatment). In + − − − Materials and Methods fecal transfer experiment, 0.01 g of feces from Sftpd / or Sftpd / mice was orally inoculated into germ-free mice (C57BL/6N; Sankyo Lab). After 4 d, mice Mice. FVB/Njcl (FVB) were purchased from Clea Japan, Inc. SP-D–deficient − − were treated with 2.5% DSS for 6 d and water for additional 2 d. DAI of DSS (Sftpd / ) mice, described previously (27), were kindly gifted by Dr. Jeffery A. colitis was assessed by a previously described method (28). Whitsett, Cincinnati Children’s Hospital Medical Center, Cincinnati, and maintained on a FVB genetic background in SPF conditions. Germ-free C57BL/6N mice were purchased from Sankyo Labo Service Corporation, Inc. Measurement of Tissue Corticosterone. To examine tissue-contained cortico- or Clea Japan, Inc. All animal experiments were done in accordance with sterone, tissues were homogenized in PBS (liver; 150 mg/mL, gallbladder; 1 tissue guidelines of the University of Tokyo and Osaka University. in 0.5 mL) by Shakeman3 BMS-SMN03 (Biomedical Science) and supernatants were subjected to ELISA with corticosterone ELISA kit (Enzo Life Sciences). For Immunoprecipitation and Immunoblot Analysis for SP-D. The fresh feces from tissue culture, liver was sectioned into small pieces (around 3 mm square) and the mice were suspended in the lysis buffer (1% Nonidet P-40, 10 mM Tris·HCl cultured as 100 mg tissue/mL with 199 medium (Sigma) including 5% charcoaled pH 7.8, 150 mM NaCl, 1 mM EDTA) and subjected to immunoprecipitation with stripped FBS (Gibco) and 1% antibiotic-antimycotic (Gibco). After a 15-h in- anti–SP-D antibody (H-120; Santa Cruz) and Dynabeads Protein G (Invitrogen) cubation at 37 °C, culture supernatants were subjected to ELISA. according to the manufacturer’s protocol. Immunoblot analysis for SP-D was All other information is available in Supporting Information. performed by SDS/PAGE with SP-D antibody (ab17781; Abcam). ACKNOWLEDGMENTS. We thank K. Honda, K. Atarashi, T. Tanoue, H. Takamatsu, Bacteria. L. murinus (ATCC35020) was obtained from American Type Culture and H. Kikutani for helpful discussion and M. Sugahara, M. Taniguchi, and Collection. Clostridium 17 strain mix was described before (23). Endogenous N. Fujii for technical assistance. This work was supported in part by Grant-In-Aid for Scientific Research (S) 15638461 from the Ministry of Education, Culture, Sports, L. murinus and L. johnsonii were isolated from feces from SPF mice and Science; Japan Agency for Medical Research and Development 15656877; the identified by sequencing of 16S ribosomal RNA genome. Nakajima Foundation; and Uehara Memorial Foundation. H.S.-K., Y.N., and N.E. are research fellows of the Japan Society for the Promotion of Science. The INFLAMMATION

Analysis of SP-D Binding Bacteria. Fecal pellets were suspended in 1 mM CaCl2/ Department of Molecular Immunology is supported by BONAC Corporation and IMMUNOLOGY AND PBS, weighed, and filtered by a 40-μm cell strainer. Flowthrough was washed in Kyowa Hakko Kirin Co.

1. Mowat AM, Agace WW (2014) Regional specialization within the intestinal immune 17. Turpin W, Humblot C, Thomas M, Guyot JP (2010) Lactobacilli as multifaceted system. Nat Rev Immunol 14:667–685. probiotics with poorly disclosed molecular mechanisms. Int J Food Microbiol 143: 2. Goldszmid RS, Trinchieri G (2012) The price of immunity. Nat Immunol 13:932–938. 87–102. 3. Hancock RE, Haney EF, Gill EE (2016) The immunology of host defence peptides: 18. Alok A, et al. (2017) Probiotics: A new era of biotherapy. Adv Biomed Res 6:31. Beyond antimicrobial activity. Nat Rev Immunol 16:321–334. 19. Sarao LK, Arora M (2017) Probiotics, prebiotics, and microencapsulation: A review. 4. Negishi H, et al. (2012) Essential contribution of IRF3 to intestinal homeostasis and Crit Rev Food Sci Nutr 57:344–371. microbiota-mediated Tslp gene induction. Proc Natl Acad Sci USA 109:21016–21021. 20. Delucchi L, Fraga M, Perelmuter K, Cella CD, Zunino P (2014) Effect of native Lacto- 5. Wright JR (2005) Immunoregulatory functions of surfactant proteins. Nat Rev bacillus murinus LbP2 administration on total fecal IgA in healthy dogs. Can J Vet Res – Immunol 5:58–68. 78:153 155. – 6. Kishore U, et al. (2006) Surfactant proteins SP-A and SP-D: Structure, function and 21. Rudensky AY (2011) Regulatory T cells and Foxp3. Immunol Rev 241:260 268. receptors. Mol Immunol 43:1293–1315. 22. Wan YY, Flavell RA (2007) Regulatory T-cell functions are subverted and converted – 7. Crouch EC (2000) Surfactant protein-D and pulmonary host defense. Respir Res 1:93–108. owing to attenuated Foxp3 expression. Nature 445:766 770. 8. Sun Z, et al. (2015) Expanding the biotechnology potential of lactobacilli through 23. Atarashi K, et al. (2013) Treg induction by a rationally selected mixture of Clostridia – comparative genomics of 213 strains and associated genera. Nat Commun 6:8322. strains from the human microbiota. Nature 500:232 236. 9. Iliev ID, et al. (2012) Interactions between commensal fungi and the C-type lectin 24. Talabér G, Jondal M, Okret S (2013) Extra-adrenal glucocorticoid synthesis: Immune regulation and aspects on local organ homeostasis. Mol Cell Endocrinol 380:89–98. receptor dectin-1 influence colitis. Science 336:1314–1317. 25. Bornscheuer T, Schmiedel S (2014) Calculated antibiosis of acute cholangitis and 10. Dollive S, et al. (2013) Fungi of the murine gut: Episodic variation and proliferation cholecystitis. Viszeralmedizin 30:297–302. during antibiotic treatment. PLoS One 8:e71806. 26. Sauter GH, et al. (2002) Bowel habits and bile acid malabsorption in the months after 11. Tang C, et al. (2015) Inhibition of dectin-1 signaling ameliorates colitis by inducing cholecystectomy. Am J Gastroenterol 97:1732–1735. Lactobacillus-mediated regulatory T cell expansion in the intestine. Cell Host Microbe 27. Korfhagen TR, et al. (1998) Surfactant protein-D regulates surfactant phospholipid 18:183–197. homeostasis in vivo. J Biol Chem 273:28438–28443. 12. Boehm F, et al. (2012) Deletion of Foxp3+ regulatory T cells in genetically targeted 28. Kim JJ, Shajib MS, Manocha MM, Khan WI (2012) Investigating intestinal in- mice supports development of intestinal inflammation. BMC Gastroenterol 12:97. flammation in DSS-induced model of IBD. J Vis Exp 3678. 13. Mason RJ, et al. (2002) Maintenance of surfactant protein A and D secretion by rat 29. Chiba S, et al. (2014) Recognition of tumor cells by dectin-1 orchestrates innate im- – alveolar type II cells in vitro. Am J Physiol Lung Cell Mol Physiol 282:L249 L258. mune cells for anti-tumor responses. Elife 3:e04177. 14. Rucka Z, et al. (2013) Differential effects of insulin and dexamethasone on pulmonary 30. Martorell P, Querol A, Fernández-Espinar MT (2005) Rapid identification and enu- surfactant-associated genes and proteins in A549 and H441 cells and lung tissue. Int J meration of Saccharomyces cerevisiae cells in wine by real-time PCR. Appl Environ – Mol Med 32:211 218. Microbiol 71:6823–6830. 15. Bachurski CJ, Ross GF, Ikegami M, Kramer BW, Jobe AH (2001) Intra-amniotic endo- 31. Ueno M, et al. (2010) Assessment and improvement of methods for microbial DNA toxin increases pulmonary surfactant proteins and induces SP-B processing in fetal preparation from fecal samples. Handbook of Molecular Microbial Ecology II: sheep. Am J Physiol Lung Cell Mol Physiol 280:L279–L285. Metagenomics in Different Habitats, ed de Bruijn FJ (John Wiley & Sons, Inc., Hoboken, 16. Gold EM, Diraimondo VC, Forsham PH (1960) Quantitation of pituitary corticotropin NJ), pp 191–198. reserve in man by use of an adrenocortical 11-beta hydroxylase inhibitor (SU-4885). 32. Nishio J, et al. (2015) Requirement of full TCR repertoire for regulatory T cells to Metabolism 9:3–20. maintain intestinal homeostasis. Proc Natl Acad Sci USA 112:12770–12775.

Sarashina-Kida et al. PNAS | September 19, 2017 | vol. 114 | no. 38 | 10183 Downloaded by guest on October 1, 2021