Epithelial Inflammation Is Associated with CCL28 Production and the Recruitment of Regulatory T Cells Expressing CCR10

This information is current as Bertus Eksteen, Alice Miles, Stuart M. Curbishley, Chris of September 26, 2021. Tselepis, Allister J. Grant, Lucy S. K. Walker and David H. Adams J Immunol 2006; 177:593-603; ; doi: 10.4049/jimmunol.177.1.593

http://www.jimmunol.org/content/177/1/593 Downloaded from

References This article cites 51 articles, 22 of which you can access for free at: http://www.jimmunol.org/content/177/1/593.full#ref-list-1 http://www.jimmunol.org/

Why The JI? Submit online.

• Rapid Reviews! 30 days* from submission to initial decision

• No Triage! Every submission reviewed by practicing scientists

• Fast Publication! 4 weeks from acceptance to publication by guest on September 26, 2021

*average

Subscription Information about subscribing to The Journal of Immunology is online at: http://jimmunol.org/subscription Permissions Submit copyright permission requests at: http://www.aai.org/About/Publications/JI/copyright.html Email Alerts Receive free email-alerts when new articles cite this article. Sign up at: http://jimmunol.org/alerts

The Journal of Immunology is published twice each month by The American Association of Immunologists, Inc., 1451 Rockville Pike, Suite 650, Rockville, MD 20852 Copyright © 2006 by The American Association of Immunologists All rights reserved. Print ISSN: 0022-1767 Online ISSN: 1550-6606. The Journal of Immunology

Epithelial Inﬂammation Is Associated with CCL28 Production and the Recruitment of Regulatory T Cells Expressing CCR101

Bertus Eksteen,* Alice Miles,* Stuart M. Curbishley,* Chris Tselepis,‡ Allister J. Grant,* Lucy S. K. Walker,† and David H. Adams2*

Mucosal tissues require constant immune surveillance to clear harmful pathogens while maintaining tolerance to self Ags. Reg- ␣ ␤ ulatory T cells (Tregs) play a central role in this process and expression of E 7 has been reported to define a subset of Tregs with tropism for inflamed tissues. However, the signals responsible for recruiting Tregs to epithelial surfaces are poorly understood. We have isolated a subset of CCR10-expressing CD25؉CD4؉Foxp3؉ Tregs with potent anti-inflammatory properties from chronically inflamed human liver. The CCR10؉ Tregs were detected around bile ducts that expressed increased levels of the CCR10 ligand CCL28. CCL28 was secreted by primary human cholangiocytes in vitro in response to LPS, IL-1␤, or bile acids. Exposure of ؉ CCR10 Tregs to CCL28 in vitro stimulated migration and adhesion to mucosal addressin cell adhesion molecule-1 and VCAM-1. Downloaded from Liver-derived CCR10؉ Tregs expressed low levels of CCR7 but high levels of CXCR3, a chemokine receptor associated with ؉ ␤ ␣ infiltration into inflamed tissue and contained a subset of E 7 cells. We propose that CXCR3 promotes the recruitment of Tregs to inflamed tissues and CCR10 allows them to respond to CCL28 secreted by epithelial cells resulting in the accumulation of .CCR10؉ Tregs at mucosal surfaces. The Journal of Immunology, 2006, 177: 593–603

hronic inflammation leads to tissue damage and the re- important determinant of Treg function because CTLA-4-deficient http://www.jimmunol.org/ lease of multiple potential autoantigens. Although sec- mice have a similar phenotype to Foxp3 deficiency (8, 7). A recent C ondary immune responses against such Ags can be de- study suggests that Tregs do not express the inhibitory receptor tected, they are not a dominant feature of most chronic programmed cell death-1 (PD-1) on their surface, although it is inflammatory diseases suggesting that mechanisms exist to sup- retained intracellularly. This finding discriminates them from press the adaptive immune response in inflamed tissues. Regula- CD4ϩ/CD25ϩ effector cells that express high levels of cell surface tory T cells (Tregs)3 have evolved to limit the local damage re- PD-1 (9). sulting from infectious challenges to the host. Natural Tregs arise Tregs control inflammation by contact-dependent TGF-␤ and in the thymus and survive as well as operate in the periphery by IL-10 production (10, 11) and are able to control experimental gut by guest on September 26, 2021 responding to a large variety of self-Ags (1–3). Tregs are CD4 inflammation in adoptive transfer models (12). Paradoxically, cells that have a distinct phenotype characterized by expression of Tregs are required to maintain chronic intestinal inflammation in the a subunit of the high affinity IL-2R, CD25, and the glucocor- animal models, presumably by dampening more aggressive acute ticoid-induced TNFR (4). Despite displaying diverse TCR, there is inflammation (13). Epithelial surfaces in particular are vulnerable evidence to suggest Tregs have a higher propensity to recognize Ϫ to invasion by microbes and are a frequent target of chronic in- self peptides than conventional CD25 T cells (3, 5). The most flammatory diseases (14). Although Tregs have been reported in specific Treg marker is the transcription factor Foxp3, which is inflamed peripheral tissues, little is known about their function or critical for Treg function. Retroviral transfer of Foxp3 to naive T the homing mechanisms that localize them to epithelial sites (15). cells converts them into functional Tregs, whereas its deletion ab- The chemokine receptor CCR10 is detected on both T and B lates regulatory function and triggers autoimmunity (6, 7). Tregs lymphocytes at epithelial sites (16, 17) and defines subsets of lym- also constitutively express the negative regulatory receptor phocytes that can be recruited to either mucosal or cutaneous ep- CTLA-4, which binds the ligands CD80/CD86 and may be an ithelial sites. Mucosal homing is driven by the ligand CCL28, which is expressed by columnar epithelia in the gut, lung, breast *Liver Research Laboratories, †Medical Research Council Centre for Immune Reg- and salivary glands (18), whereas homing to the skin is triggered ulation, Institute for Biomedical Research, and ‡Epithelial Research Group, Univer- by the alternative CCR10 ligand, CCL27 (19). Specificity is en- sity of Birmingham, Birmingham, United Kingdom hanced by the coexpression of CCR10 with organ-specific adhe- Received for publication October 14, 2005. Accepted for publication April 12, 2006. sion receptors. Thus mucosal CCR10ϩ lymphocytes also express The costs of publication of this article were defrayed in part by the payment of page ␣ ␤ required for recruitment to the gut, whereas CCR10ϩ lym- charges. This article must therefore be hereby marked advertisement in accordance 4 7 with 18 U.S.C. Section 1734 solely to indicate this fact. phocytes, which show tropism for the skin, coexpress the skin- 1 This work was supported by grants from The Medical Research Council and Core homing receptor CCR4 and the cutaneous lymphocyte Ag (19). Charity, U.K. CCL28 is constitutively expressed in the colon and increased by 2 Address correspondence and reprint requests to Dr. David H. Adams, Liver Re- proinflammatory cytokines (20) and bacterial products, suggesting search Group, Medical Research Council Centre for Immune Regulation, 5th Floor it has a role in recruiting effector cells to areas of epithelial injury Institute for Biomedical Research, University of Birmingham, Birmingham B15 2TH, U.K. E-mail address: [email protected] (21). CCR10 expression has also been reported on nonlymphoid 3 Abbreviations used in this paper: Treg, regulatory T cell; LIL, liver-infiltrating malignant cells (22), and coexpression of CCR9 and CCR10 has lymphocyte; PSC, primary sclerosing cholangitis; PBC, primary biliary cirrhosis; been implicated in the formation of small bowel melanoma me- ALD, alcoholic liver disease; MAdCAM-1, mucosal addressin cell adhesion mole- ϩ cule-1; hpf, high-power field; BEC, biliary epithelial cell; CDCA, chenodeoxycholic tastases (23). CCR10 lymphocytes are positioned in the intraepi- ␣ ␤ acid. thelial compartment where coexpression of E 7 integrins (24)

Copyright © 2006 by The American Association of Immunologists, Inc. 0022-1767/06/$02.00 594 CCR10 EXPRESSION ON EPITHELIAL Tregs allows them to interact with E-cadherin expressed at epithelial ad- phocytes were separated using 33%/77% (v/v) Percoll (Amersham Bio- ␣ sciences) density gradient centrifugation at 2000 rpm (650 ϫ g) for 30 min. herens junctions (25). The expression of E integrins also defines ϩ a population of CD25 Tregs with enhanced suppressive proper- PBL isolation ␣ Ϫ ϩ ties compared with E CD25 Tregs (26). Recent studies have suggested that the function of ␣ ϩ Tregs is at least in part depen- Lymphocytes were isolated from peripheral venous blood and diluted 1/1 E with PBS before centrifuged over a Lymphoprep (Invitrogen Life Tech- dent on their ability to be recruited to inflamed tissue compart- nologies) gradient for 30 min at 2000 rpm (650 ϫ g). ments (27). Experimental animals that lack fucosultransferase required to synthesize E- and P-selectin ligands during inflammation Isolation and culture of BEC ␣ ϩ are unable to recruit E effector/memory Tregs to inflammatory BEC were isolated according to previously described methods (30). sites and as a consequence display a reduced ability to suppress Briefly, liver tissue was finely chopped and subjected to enzyme digestion ϩ␣ Ϫ (collagenase type IV) and density gradient centrifugation. Nonparenchymal inflammation (27). In contrast, naive Tregs (CD25 E ) express CCR7 and CD62L, allowing them to enter lymphoid tissues to cells were then removed and separated by immunomagnetic selection. Cells positive for mAb HEA-125 (Progen) were classified as BEC. Fol- control naive CD4 cell proliferation during the early phases of lowing isolation BEC were plated in collagen coated flasks (Corning). BEC inflammation. We hypothesized that the chemokine receptor were cultured in DMEM, Ham’s F12, containing 10% human serum, pen- CCR10 would define a population of Tregs associated with epi- icillin and streptomycin (100 IU/ml), glutamine (2 mM), epidermal growth thelial inflammation. We used chronic liver disease as a model of factor (10 ng/ml), hydrocortisone (2 mg/ml), choleratoxin (10 ng/ml), tri- iodo-thyronine (2 nM), insulin (0.124 U/ml), hepatocyte growth factor (5 persistent epithelial inflammation to test the hypothesis in humans. ng/ml), and vascular endothelial growth factor (10 ng/ml). Cells were cul- Inflammatory and autoimmune liver damage is often focused on tured to confluence and in all experiments used between passage 3 and 4. the biliary epithelium, which is contiguous with the mucosa of the Downloaded from gastrointestinal tract and in primary sclerosing cholangitis (PSC) In vitro BEC stimulation immune damage to bile ducts is mediated by mucosal T cells re- BEC from six livers (n ϭ 3 PSC, n ϭ 1 PBC, n ϭ 2 ALD) were grown to cruited as a consequence of aberrant expression of the gut homing confluence in 24-well plates and either left under basal conditions or stim- ␣ ␤ ␥ molecules CCL25 and mucosal addressin cell adhesion molecule-1 ulated with cytokines for 24 h rTNF- , IL-1 , IFN- (all 10 ng/ml from PeproTech), LPS (1 ␮g/ml), or 500 ␮M CDCA (chenodeoxycholic acid; (MAdCAM-1) (28, 29). Sigma-Aldrich). Following stimulation, mRNA as well as the supernatant

We report that CCL28 is expressed by biliary epithelial cells was isolated from each well and stored at Ϫ70°C until analyzed. All treat- http://www.jimmunol.org/ (BEC) in PSC and other forms of chronic liver disease (primary ments were performed in triplicate for each experiment. biliary cirrhosis (PBC), alcoholic liver disease (ALD)) with evi- Immunohistochemistry and dual color coimmunofluorescence dence of biliary inflammation. Expression of CCL28 by BEC is stimulated by LPS and IL-1␤ and recruits subsets of T lympho- Six cases each of PSC, normal liver, PBC, and ALD and three normal cytes expressing CCR10, which include CD4ϩCD25ϩFoxp3ϩ spleens were examined. Sections for immunohistochemistry were initially incubated with 20% normal goat serum for 1 h before the addition of a Tregs with potent suppressive properties. Thus CCL28 may be a mouse anti-human CCL28 mAb for 60 min. Control sections were incu- critical signal for positioning Tregs at inflamed epithelial surfaces bated without the primary Ab or mouse Igs. Subsequently, sections were and CCR10 expression defines a subset of Tregs that home to incubated sequentially for 20 min with biotinylated secondary Abs fol- epithelial sites. lowed by a HRP complex (StepABCcomplex Duet kit; DakoCytomation). by guest on September 26, 2021 Sections were incubated with diaminobenzidine as a peroxidase substrate for 5–10 min until a color reaction developed and counterstained with Materials and Methods hematoxylin (Mayer’s hemalum). Positive CCL28 staining was identified Tissues by the presence of a dark brown reaction product. All washes were performed with Tris-buffered saline (pH 7.6). Paraffin-fixed sections were ini- Diseased liver tissue and paired peripheral blood was obtained at the time tially dewaxed in xylene for 10 min and dehydrated in 100% ethanol for 10 of liver transplantation. Skin tissue was obtained from tissue removed dur- min. Sections were washed and microwave Ag retrieval was done for 10 ing reconstructive plastic surgery. All samples were collected with appro- min in 0.01 M trisodium citrate buffer (pH 6). Sections were subsequently priate patient consent and local research ethics committee approval. blocked, stained, and counterstained as per frozen sections described. Sections for dual immunofluorescence were incubated with 20% normal Abs and reagents goat/mouse serum for 30 min before the addition of primary Abs raised The following primary Abs were used: mouse anti-human CCL28 mAb against CCR10 or CCL28 for 1 h. Control sections were incubated without ␮ ␣ ␤ primary Ab. Subsequently, sections were incubated with either CD31- (62705, 5 g/ml; R&D Systems), mouse anti-human 4 7 (ACT-1, 15 ␮ Texas Red, HEA-labeled with Texas Red, or Foxp3-FITC and goat g/ml; a gift from M. Briskin (Millennium Pharmaceuticals, Cambridge, ␮ ␮ MA), CD31 mAb (JC70A, 1 ␮g/ml; DakoCytomation), cytokeratin-19 anti-mouse FITC (20 g/ml) or rabbit anti-goat Texas Red (20 g/ml) ␮ secondary Abs for 60 min in the dark. Nuclei were counterstained with mAb (BA17, 1 g/ml; DakoCytomation), MAdCAM-1 mAb (CA Ј Ј 102.2C1, 12.5 ␮g/ml; Serotec), mouse anti-human anti-␤1 mAb (3S3, 4 4 ,6 -diamidino-2-phenylindole. Dual immunofluorescence was assessed ␮g/ml; Abcam), mouse anti-human mAb HEA-125 (Progen), mouse anti- using AxioVision software. human ␤ -Cy5 mAb (FIB504, 1/20; BD Pharmingen), goat anti-human 7 Western blotting CCR10 (ab3904, 0.05 ␮g/ml; Abcam), CD8-FITC mAb (DK25; 1/50; Da- koCytomation), CD4-PE-Cy5 mAb (C7069; 1/50; DakoCytomation), CD3 Frozen tissue samples (n ϭ 6 PSC, n ϭ 6 PBC, n ϭ 6 normal liver, n ϭ R-PE-Cy7 mAb (UCHT1, 1/20; Beckman Coulter), CD25-FITC (F0801, 6 ALD, and n ϭ 3 spleen) were homogenized into loading sample buffer 1/50; DakoCytomation), CD56-PE mAb (MOC-1, 1/30; DakoCytomation), and normalized for total protein content using Coomassie blue staining of CD104-FITC mAb (Ber-ACT8, 1/25; DakoCytomation), CD49d-FITC gels or ␤-actin expression. Samples of equal protein content were loaded mAb (P4G9, 1/50; DakoCytomation), CCR7 mAb (fab197p, 1/20; R&D on 8% SDS-PAGE gels and following electrophoresis were transferred Systems), and CXCR3 mAb (fab160p, 1/20; R&D Systems). Relevant sec- onto Hybond membranes (Amersham Biosciences). The membranes were ondary Abs used were donkey anti-goat PE/FITC (1/100; Abcam), goat blocked with 10% skim milk and subsequently incubated with mouse anti- anti-mouse FITC (20 ␮g/ml; Southern Biotechnology Associates) and goat human CCL28 Abs (2.5 ␮g/ml) for 1 h. The blots were washed and incu- anti-mouse IgG1 or IgG2a Texas Red (20 ␮g/ml; Southern Biotechnology bated with a goat anti-mouse HRP-conjugated Ab for 45 min. Immunore- Associates). active bands were detected using the ECL detection system (Amersham Pharmacia). Bands were scanned on a Gel Doc (Bio-Rad) system and op- Liver-derived lymphocyte isolation tical densities for individual bands compared. 3 Liver tissues were reduced to 5-mm cubes and homogenized at 230 rpm for Real-time RT-PCR for CCL28 5 min in a Stomacher 400 circulator (Seward). For detection of Tregs, liver tissue was digested for another 5 min with collagenase type IV (Sigma- RNA was extracted from snap frozen tissue (n ϭ 6 PSC, n ϭ 6 PBC, n ϭ Aldrich). Homogenized tissues were filtered through a fine mesh and lym- 6 normal liver, n ϭ 6 ALD, n ϭ 3 spleen, n ϭ 3 skin) or stimulated BEC The Journal of Immunology 595

cultures (n ϭ 3 PSC, n ϭ 1 PBC, n ϭ 2 ALD) using an RNeasy Mini kit Lymphocyte stimulation, IL-10 capture, and IL-10 detection by (Qiagen). mRNA were transcribed to cDNA and real-time PCR performed FACS on a PE7700 ABI Prism machine. Each reaction was performed in triplicate ϩ using QuantiTect Probe RT-PCR kit (Qiagen) according to the manufac- Enriched populations of CD4 lymphocytes were labeled with an IL-10 turer’s instructions. Reactions contained 400 nM CCL28-specific 5Ј-CA capture construct to retain secreted IL-10 (IL-10 secretion assay; Miltenyi GAGAGGACTCGCCATCGT and 3Ј-TGTGAAACCTCCGTGCAACA Biotec) and stimulated for4hat37°C in the presence of 50 ng/ml PHA or primers (AltaBioscience) and 200 nM CCL28-specific TaqMan probe 5Ј- 1000 IU IL-2 (proleukin). Lymphocytes were subsequently washed, la- FAM-CATGCCTCAGAAGCCATACTTCCCATTG-TAMRA-3Ј (Eurogen- beled with mouse anti-human IL-10 PE mAb (Miltenyi Biotec) and IL-10 tech). Data are presented as fold increases in gene expression (⌬CT) normal- production detected by flow cytometry. Unstimulated lymphocytes and iso- ized to the 18 S control and compared with normal skin tissue or untreated type-matched control Abs were used as controls for flow cytometry. BEC (normalized to 1). Samples with no cDNA were used to control for background fluorescent signals. Allogeneic lymphocyte proliferation assay Liver lymphocytes expressing CD4 were isolated by negative selection and RT-PCR for Foxp, CCR10, and IL-10 divided into CD25ϩCD4ϩ and CD25ϪCD4ϩ subsets by MACS using CD25 mAb and Dynabeads (Dynal Biotech). Beads were detached and RNA was extracted from snap frozen tissue lymphocytes using an RNeasy removed with DETACHaBEAD (Dynal Biotech). CCR10ϩ populations Mini kit (Qiagen). Foxp3 and CCR10 mRNA were reverse transcribed and were subsequently isolated using the described EasySep techniques. PCR performed using Qiagen One-Step RT-PCR kit according to manu- Matched naive PBL were labeled with CFSE (Sigma-Aldrich). facturer’s instructions. PCR products were normalized to ␤-actin mRNA CD25ϩCD4ϩ or CD25ϪCD4ϩ lymphocytes were cultured 1:1 with allo- expression and separated on a 1% agarose gel containing ethidium bromide genic myeloid dendritic cells (isolated as previously described) (32) and (Sigma-Aldrich). Positive bands were detected under UV light and quan- CFSE-labeled cells added in ratios 1:1 to 1:30. Samples were cultured for tified by densitometry (Gel Doc; Bio-Rad). Primers were designed from 7 days and cell division measured by flow cytometry. For some experi- GenBank sequences for CCR10 and Foxp3: CCR10 forward 5Ј-GAGGC ϩ ϩ ␣ Downloaded from ments the CD25 CD4 populations were further isolated based on E CACAGAGCAGGTTTC-3Ј, reverse 5Ј-CATCGGCCTTGTAGCAAAGC- integrin expression and repeated experiments as described. 3Ј; and Foxp3 forward 5Ј-ACACCACCCACCACCGCC ACT-3Ј, reverse 5Ј- Proliferation was also measured by incorporation of [3H]thymidine. TCGGATGATGCCACAGATGAAGC-3Ј. IL-10, ␤-actin, and GAPDH CD25ϩCD4ϩ or CD25ϪCD4ϩ lymphocytes were cultured with autologous primer pairs were obtained from R&D Systems. naive lymphocytes in ratios 1:1 to 1:30 with added 5 ␮l/ml anti-human CD3/CD28 beads (Dynal Biotech). After 24 h, 0.5 ␮Ci [3H]thymidine was CCL28 sandwich ELISA added to each well and the plate harvested on day 3. Results of radioactive

thymidine are reported in cpm. http://www.jimmunol.org/ The concentration of CCL28 in BEC supernatant samples (n ϭ 6) were determined by sandwich ELISA (31) using human CCL28 DuoSet kits Transwell chemotaxis of lymphocytes (DY717; R&D Systems). Briefly, relevant capture Abs were prepared in The migration of lymphocytes from inflamed livers (n ϭ 4) and peripheral 0.1 M carbonate/bicarbonate buffer (pH 9.6), added to 96-well immunosorp blood (n ϭ 4) were assessed using fibronectin-covered (Sigma-Aldrich) plates (Nunc) and allowed to coat overnight at 4°C. Protein standards of 6.5-mm diameter, 5-␮m pore Transwell inserts (Corning). Responses to known concentration or test samples were then added to relevant wells and CXCL12 were positive controls because large numbers of liver-infiltrating incubated at room temperature for 1 h before washing and further 1 h lymphocytes (LIL) express CXCR4. The 100 ng/ml recombinant human incubation with relevant biotinylated detection Abs. The plates were then CCL28 or 100 ng/ml recombinant human CXCL12 was placed in the bot- washed thoroughly before incubation with peroxidase-conjugated strepta- tom of the well, and 5 ϫ 105 lymphocytes were added to the upper cham- vidin (1/5000; Zymed Laboratories). The ELISA was developed using ber. Cells were collected from the top and bottom chambers after2hand TMB substrate (3,5,3,5-tetramethylbenzidine, 100 ␮g/ml) in 0.11 M so- by guest on September 26, 2021 measured by fixed volume counting and phenotyped for CCR10 expression dium acetate (pH 5.5), and 0.0003% H O (both from BDH) and the en- 2 2 by flow cytometry. Assays were conducted in duplicate and compared with zymatic reaction was stopped using 2.5 M H SO . Colorimetric analysis 2 4 control wells that only contained medium with BSA alone. Results of mi- was performed by measuring absorbance values at 450 nm. All measure- grated cells are reported as a percentage of input cells. ments were performed using duplicate samples for each experiment. Static adhesion assay Four-color flow cytometry Isolated LIL from inflamed liver (n ϭ 6) were phenotyped by flow cytom- Lymphocytes were initially incubated with 3 mg/ml mouse Igs before in- etry for CCR10 expression and subsequently used in the static adhesion cubation with a CCR10 Ab (0.05 ␮g/ml) for 30 min. Cells were washed, assay. Briefly, 18-well Teflon-coated slides (Erie Scientific) were incu- centrifuged for 10 min at 2000 rpm (650 ϫ g), and labeled with a donkey bated with recombinant human VCAM-1 (10 ␮g/ml), recombinant MAd- anti-goat FITC/R-PE secondary Ab. Cells were subsequently washed and CAM-1 (10 ␮g/ml) or BSA (1 ␮g/ml) at 4°C for 16 h in a humidified labeled with fluorochrome-labeled primary mAbs. Control samples were chamber. Slides were blocked with 10% FCS for 60 min before the addi- labeled with matched isotope control Ig. Samples were run on a Coulter tion of lymphocytes. Binding of lymphocytes to MAdCAM-1 or VCAM-1 ␣ ␤ ␤ Epics XL flow cytometer. Results were analyzed using Summit software was blocked by adding an anti-human 4 7 Ab or anti- 1 Ab, respectively, (DakoCytomation). for 30 min at 37°C before lymphocytes were added. A total of 8 ϫ 104 cells was added per well and allowed to bind for 10 min at 37°C. Some lym- Lymphocyte enrichment and culture phocytes were preincubated with pertussis toxins (100 ng/ml). Adhesion was triggered by the addition of CCL28 (10 ng/ml; PeproTech) or MnCl2 Lymphocytes were positively enriched to 95% purity for CCR10 with Ea- (100 nM/ml). After incubation nonadherent cells were washed away using sySep (StemCell Technologies). Briefly, lymphocytes were suspended in cold PBS. Slides were analyzed by manual counting of adherent lympho- cold PBS with 1 mM EDTA and 2% FCS. Fc receptors were blocked (100 cytes in three representative high-power fields (hpf) per well. ␮ l/ml EasySep human Fc blocker) for 10 min and followed by incubation Statistical analysis with CCR10 R-PE primary Ab for 15 min. Lymphocytes were centrifuged at 10 min at 2000 rpm (650 ϫ g) and subsequently incubated with EasySep Paired or independent t tests were used to assess data normally distributed, FITC-PE selection mix, followed by magnetic nanoparticles. Magnetic whereas nonparametric data were compared using Wilcoxon signed rank positive selection of labeled cells was then performed and the purity of test (for related samples) or Mann-Whitney U test (for unrelated samples). selected populations was confirmed by flow cytometry. Statistical analyses were performed with Axum 7 software. Tregs were isolated using the CD4ϩCD25ϩ Treg kit (Dynal Biotech) according to the manufacturer’s instructions. A total of 1 ϫ 108 lympho- Results cytes was suspended in 1 ml of PBS (pH 7.4) with 0.1% BSA and negative ϩ Increased expression of CCL28 in chronic liver disease isolation of CD4 cells performed by magnetic depletion with Abs to CD14, CD56, CD19, CD8, and CD235a. CD25ϩ regulatory cells were The chronic biliary diseases, PSC and PBC, are characterized by a subsequently positively isolated by CD25 Abs conjugated with Dynabeads. chronic, predominantly T cell infiltrate in the portal tracts that is Beads were removed from the cells by DETACHaBEAD (Dynal Biotech). Subsequent enrichment of Tregs based on CCR10 expression was done as centered on inflamed bile ducts (Fig. 1a). We investigated whether described. Purity of isolated populations was confirmed by flow cytometry the epithelial chemokine CCL28 might be involved in this recruit- and Treg status confirmed by Foxp3 RT-PCR. ment by staining frozen liver tissue from patients with chronic 596 CCR10 EXPRESSION ON EPITHELIAL Tregs Downloaded from http://www.jimmunol.org/

FIGURE 1. Livers from patients with chronic inflammatory liver diseases, including PSC, PBC, and ALD are heavily infiltrated by CD3ϩ lymphocytes. a, T cells are centered on inflamed bile ducts (arrow). Immunohistochemistry revealed intense CCL28 staining (detected as a brown pigment) on inflamed bile ducts (B) and to a lesser extent on portal vein (PV) and hepatic artery (A) endothelium. b, Little staining was seen on normal liver or spleen. Control by guest on September 26, 2021 sections had no detectable staining. We confirmed our observations with Western blotting (c), which demonstrated minimal CCL28 protein in normal liver and spleen but enhanced expression in chronic liver disease. Sample loading was normalized for ␤-actin and band intensity quantified using a Gel Doc system. Real-time RT-PCR of total liver mRNA samples confirmed increased CCL28 mRNA in diseased liver (p Ͻ 0.001, Student’s t test) but also normal liver (p ϭ 0.006, Student t test) compared with skin after correction for 18S mRNA levels. d, Samples consisted of tissue samples from six donors each for PSC, PBC, normal liver, and ALD and three donors of spleen and skin tissue samples. Representative immunohistochemistry sections and Western blot .p Ͻ 0.001 ,ءء p ϭ 0.006 ,ء .results are shown. Results are expressed as mean Ϯ SEM

inflammatory liver disease with Abs against CCL28 (Fig. 1b). Nor- CCL28 and CD31 on portal vascular endothelium and with HEA- mal liver and splenic tissue revealed little or no detectable CCL28, 125 on biliary epithelium. Lower levels of CCL28 were detectable whereas CCL28 was readily observed in the portal tracts of pa- on sinusoidal endothelium. The distribution of CCL28 on BEC tients with PSC, PBC, and ALD. CCL28 expression was particu- was consistent with membranous expression (Fig. 2). larly intense on the cell membranes of injured bile ducts but was also detected on portal endothelium and on reactive bile ductules. Western blot analysis from whole liver protein lysates confirmed IL-1 and LPS stimulate CCL28 release by BEC little CCL28 expression in normal liver and spleen tissue but a To determine the factors that stimulate CCL28 secretion by epi- significant increase in CCL28 in inflamed liver tissue (Fig. 1c). thelial cells, primary human BEC (cholangiocytes) were isolated Total protein loading of samples was normalized by ␤-actin stain- from liver tissue removed at transplantation using established tech- ing. PSC and PBC consistently expressed the highest levels of niques and grown to confluence (Fig. 3a). Phenotype was con- CCL28. mRNA from the same liver samples was analyzed by real- firmed by cytokeratin-19 expression and all experiments were time RT-PCR using CCL28-specific primers and TaqMan probes done with cells at passage 4 or less. Cholangiocytes were grown in (Fig. 1d). Skin was used as a negative control and results normal- 24-well plates and either left under basal conditions or stimulated ized to 18S mRNA expression. Normal liver tissue contained with the following factors for 24 h: recombinant TNF-␣, IFN-␥, 4-fold higher levels of CCL28 mRNA when compared with skin TNF-␣ with IFN-␥, and IL-1␤ (all 10 ng/ml), LPS (1 ␮g/ml), or ( p ϭ 0.006) and chronically inflamed liver tissue showed a 15- to 500 ␮M of the bile acid CDCA. CCL28 protein was measured 25-fold increase in CCL28 mRNA with the highest levels again using ELISA (Fig. 3b) and mRNA with real-time PCR (Fig. 3c). seen in PSC and PBC ( p Ͻ 0.001). Selected tissues (PSC sample Stimulation with TNF-␣, IFN-␥ alone, or a combination of both shown) were stained by dual immunofluorescence using the anti- did not induce expression of CCL28, whereas both IL-1␤ and LPS CD28 mAb and Abs against either an endothelial marker (CD31) induced high levels of CCL28 mRNA and secretion of protein or an epithelial Ag (HEA-125). The results confirmed the previous ( p Ͻ 0.002). CDCA induced CCL28 expression to a lesser but still cellular distribution of staining with intense coexpression of significant extent ( p Ͻ 0.01). The Journal of Immunology 597 Downloaded from http://www.jimmunol.org/

FIGURE 2. Coimmunoﬂuorescence was used to conﬁrm the cellular distribution of CCL28. Abs to CCL28 labeled with FITC green colocalized with anti-CD31 mAb labeled with Texas Red (TxRED) on portal endothelium (upper; yellow) product. There was little CCL28 expression on sinusoidal endothelium (middle). The strongest staining of CCL28 was detected on bile ducts and where anti-CCL28 labeled with FITC colocalized with an Ab against the epithelial Ag HEA labeled with Texas Red

(lower). The staining shown on tissue sections from a patient with PSC. by guest on September 26, 2021 The findings are representative of the patterns of immunofluorescence staining seen in tissue sections from six donors each for PSC, PBC, and FIGURE 3. To determine factors that induce CCL28 expression, we ALD. Sections stained with control Abs demonstrated minimal background isolated and cultured primary BEC from human livers (a) and stimulated tissue fluorescence. confluent highly pure cultures for 24 h with combinations of TNF-␣, IFN-␥, TNF-␣, IL-1␤ (all 10 ng/ml), LPS (1 ␮g/ml), or 500 ␮M of the bile acid CDCA. CCL28 protein was measured by sandwich ELISA (b) and mRNA by real-time PCR (c). LPS and IL-1␤ were the most potent inducers ϩ p Ͻ 0.002). Treatment with CDCA also induced ,ء) CCR10 CD4 T cells are enriched in inflamed liver tissue of CCL28 expression ␤ Ͻ ءء Given the identification of CCL28 in inflamed liver, we hypothe- CCL28 secretion ( , p 0.01), but at a lower level than LPS and IL-1 treatment. Experiments reflect BEC isolated from six donors (n ϭ 3 PSC, sized that a fraction of the lymphocytes within the inflammatory n ϭ 1 PBC, n ϭ 2 ALD) and assays conducted in triplicate. Results are p Ͻ 0.002 by Student t test ,ء p Ͻ 0.01 ,ءء .lesion might have been recruited on the basis of CCR10 expres- expressed as mean Ϯ SEM sion. Lymphocytes were therefore isolated from the livers of pa- compared with unstimulated cultures. tients with chronic liver disease (samples, n ϭ 6 PSC, n ϭ 6 PBC, and n ϭ 6 ALD) and their phenotype compared with that of PBL using flow cytometry (Fig. 4a). Gut-derived CD19ϩ B cells were used as a positive control. This analysis revealed that CCR10 was ϩ ϩ normal liver CD4 T cells compared with T cells isolated from expressed by 16.7 Ϯ 4.1% of CD3 T cells derived from diseased ϩ inflamed liver (Fig. 4c); 2.2 Ϯ 1.8% of CD4 T cells in blood liver tissue. The majority of CCR10 expression on intrahepatic T expressed CCR10 compared with 6.3 Ϯ 3.5% of CD4 T cells iso- cells was confined to the CD4 compartment with only low levels ϩ ϩ lated from noninflamed liver ( p Ͻ 0.04 compared with blood). of CCR10 detectable on CD8 T cells and CD56 NK cells (33). ϩ ϩ The patterns of expression of CCR10 obtained by flow cytometry CCR10 CD4 T cells were further enriched in inflamed liver tis- Ϯ ϩ Ͻ were confirmed using RT-PCR to assess levels of CCR10 mRNA sue with 11.4 3.0% CD4 T cells expressing CCR10 ( p 0.04 Ͻ in subpopulations of lymphocytes from the liver. Consistent with compared with normal liver or p 0.002 compared with PBL). the flow cytometry data, the CCR10 message was largely restricted Because integrins act in conjunction with chemokines to regu- to the CD3ϩ and CD4ϩ subsets (Fig. 4b). Trace amounts of late the homing of T cells, we analyzed integrin expression on the ϩ ϩ CCR10 mRNA within the CD56ϩ and CD8ϩ populations may liver-infiltrating CCR10 CD4 T cells (Fig. 4d). Virtually all ϩ ϩ ϩ ␣ ␤ Ϯ reflect contamination with CD4 cells as magnetic separation only CCR10 CD4 T cells expressed 4 1 integrin (78.6 5.1%), ␣ ␤ Ϯ ␣ ␤ selects to 95% purity. whereas subpopulations expressed 4 7 (11.1 4.4%) and E 7 We extended detection of CCR10 on CD4ϩ T cells to include (3.5 Ϯ 2.0%) integrins. Results are expressed as the mean Ϯ SEM lymphocytes derived from peripheral blood and from normal liver and were obtained following subtraction of control samples that tissue. CCR10 was expressed at lower frequencies on PBL and were stained with isotype-matched control Abs. 598 CCR10 EXPRESSION ON EPITHELIAL Tregs Downloaded from http://www.jimmunol.org/ by guest on September 26, 2021

FIGURE 4. CCR10 expression on LIL. a, CCR10 expression by CD3ϩ T cells derived from diseased liver tissue were assessed by flow cytometry and compared with isotype-matched negative control samples. Gut CD19ϩ B cells were used as a positive control. CCR10 was expressed by 11.4 Ϯ 3.0% (n ϭ 18) of the LIL with most staining seen on CD4ϩ T cells and little staining of CD56ϩ NK cells and CD8ϩ T cells. As shown, CCR10 is expressed by 7.6% of CD4ϩ T cells. Parentheses show value for the mean channel fluorescence. b, CCR10 expression by the CD4ϩ subset was confirmed using RT-PCR with primers specific for human CCR10 mRNA on subsets of T cells purified by immunomagnetic selection as described in Materials and Methods.A representative agarose gel is shown; overall samples from 10 donors were assayed in duplicate. c, There were significant differences in the expression of CCR10 on CD4 T cells derived from different tissues. A representative experiment of CCR10 expression on gated CD4 T cells (left). Isotype-matched control (open histogram) and representative samples (grayed overlay histograms) are shown for peripheral blood (0.4%), normal liver (6.2%), and inflamed liver (15.6%). Overall 2.2 Ϯ 1.8% (n ϭ 10) of PBL CD4ϩ T cells were CCR10ϩ, 6.3 Ϯ 3.5% (n ϭ 6) of CD4ϩ T cells in normal liver expressed CCR10 p Ͻ 0.04 compared ,ء) p Ͻ 0.04 compared with PBL), and 11.4 Ϯ 3.0% (n ϭ 18) of liver-infiltrating CD4 T cells in chronic inflammatory liver disease ,ء) ϭ ϩ ϩ ␣ ␤ Ϯ ءء with normal liver or , p 0.002 compared with PBL). d, The majority of CCR10 CD4 cells expressed 4 1 integrins (78.6 5.1%), whereas ␣ ␤ Ϯ ␣ ␤ Ϯ subpopulations expressed 4 7 (11.1 4.4%) and E 7 (3.5 2.0%) integrins. Representative flow cytometry histograms are shown from 10 PBL, six normal liver, and 18 inflamed liver (n ϭ 6 PSC, n ϭ 6 PBC, and n ϭ 6 ALD) donors. Results are expressed as the mean Ϯ SEM. Values for p were determined by Student’s t test.

␤ ␤ CCL28 stimulates CCR10-dependent migration and 1 and 7 mediated adhesion of LIL to immobilized MAdCAM-1 and integrin activation on liver-derived lymphocytes VCAM-1, both of which are present on portal endothelium in chron- ␣ ␤ ␤ ϩ Our ﬁnding of 4, 1, and 7 integrins on CCR10 lymphocytes in ically inﬂamed liver. Exposure of LIL to 10 ng/ml CCL28 activated the liver led us to determine whether CCL28 can trigger integrin- integrin-mediated adhesion to both recombinant MAd-CAM-1 The Journal of Immunology 599

(78 Ϯ 2.8 cells/hpf) and recombinant VCAM-1 (240 Ϯ 13.4 cells/hpf) (both p Ͻ 0.02 compared with BSA controls). Adhesion was inhibited by preincubation of lymphocytes with pertussis toxin, suggesting G ␣ ␤ protein-coupled receptors were required, and blocking Abs to 4 7 ␤ reduced binding to MAdCAM-1, whereas anti- 1 Abs inhibited adhesion to VCAM-1. Manganese was used as a positive control as a nonselective activator of integrin adhesion cells without CCL28 to determine basal binding and BSA as a control substrate (Fig. 5a). ␤ ␤ Thus CCL28 binding to CCR10 can activate both 1 and 7 integrins. Transwell migration assays were used to demonstrate that CCL28 can promote chemotaxis of CCR10ϩ lymphocytes isolated from inflamed human livers. BSA was used as a negative control and 100 ng/ml CXCL12 as a positive control (Fig. 5b). PBL (11.1 Ϯ 3.7%; p Ͻ 0.03) and LIL (16.0 Ϯ 3.9%; p Ͻ 0.004) demonstrated significant migration to CCL28 compared with BSA controls. Phenotyping by flow cytometry of the lymphocytes that migrated to CCL28 revealed that the vast majority were T cells expressing CCR10 (45 Ϯ 3.1%) with a smaller population of B ϩ cells (33.4 Ϯ 2.2%). The majority of migrated T cells were CD4 Downloaded from and included both CD25ϩ and CD25Ϫ subsets. Thus, CCR10 expressed on liver infiltrating CD4ϩ lymphocytes support migration to CCL28.

LIL contain a population of Foxp3ϩ Tregs that express CCR10

Given the role of Tregs in maintaining chronic inflammation in http://www.jimmunol.org/ animal models (34), we investigated whether lymphocytes in inflamed human livers contain a population of Tregs that express CCR10. Using flow cytometry and negative CD4ϩ T cell enrichment, we were able to identify a population of CD4ϩCD25ϩ T cells (10.2 Ϯ 4.8%) within the total CD4 population of the inflamed liver. Gating on CCR10ϩ expression demonstrated that 42.6% of CD25ϩCD4ϩ T cells expressed CCR10. To confirm that the CCR10-expressing lymphocytes in inflamed liver include a population of Tregs, we used RT-PCR to analyze the Treg-specific by guest on September 26, 2021 transcription factor Foxp3. mRNA for Foxp3 was clearly identified within lymphocytes isolated from the inflamed liver. Moreover, when we used immunomagnetic isolation to enrich or deplete for CCR10, Foxp3 mRNA was preferentially located in the CCR10- enriched fraction. This finding implies that the T cell infiltrate in inflamed liver tissue includes a population of Foxp3ϩ Tregs that express CCR10 (Fig. 6a). To determine the tissue anatomical distribution of Foxp3 cells in the liver we conducted immunohistochemical analysis of liver tissues. In the normal liver Tregs occurred infrequently and were almost exclusively found in the portal tracts close to a bile duct (Fig. 6c). There was a marked increase in the number of Foxp3ϩ cells in chronic inflammation with the majority detected in ex- panded and heavily inflamed portal tracts. Dual immunofluorescence confirmed coexpression (Fig. 6c, lower panel, yellow) of FIGURE 5. LIL express functional CCR10. a, The ability of CCL28 to ϩ activate lymphocyte binding to MAdCAM-1 and VCAM-1 was studied Foxp3 in a subpopulation of the CCR10 lymphocytes. ϭ using total T lymphocyte populations from diseased livers (n 6). Treat- ϩ ϩ ment of LIL with either 10 ng or 50 ng/ml CCL28 significantly increased CCR10 Foxp3 intrahepatic Tregs are functional suppressor adhesion to both VCAM-1 and MAdCAM-1. Adhesion was inhibited by cells preincubation of lymphocytes with pertussis toxin (ptx) or blocking mAbs We used allogeneic stimulation/suppression assays to confirm that to either ␤ or ␣ ␤ integrins. Manganese (Mn) was used as a positive ϩ ϩ ϩ 1 4 7 CCR10 CD25 CD4 lymphocytes are fully functional Tregs control as a nonsignaling activator of integrin adhesion. b, Migration of ϩ ϩ (Fig. 7, a and b). Freshly isolated and unstimulated CD25 CD4 LIL to CCL28 was assessed using fibronectin-coated Transwell migration chambers. CXCL12 and BSA were use as controls. A total of 5 ϫ 105 cells were isolated from inflamed livers and enriched for CCR10 lymphocytes isolated from inflamed livers (n ϭ 4) and peripheral blood (n ϭ 4) were loaded per well, and experiments were done in duplicate. Liver lymphocytes showed significant migration to CCL28 compared with p Ͻ 0.004). Phenotyping by flow cytometry of the lymphocytes CD4ϩ and included both CD25ϩ and CD25Ϫ subsets. Representative ,ء) BSA that migrated to CCL28 revealed that the migrated lymphocytes were pre- FACS plots with percentage expression are shown. Results are expressed ϩ ϩ p Ͻ 0.03 by Student’s t test and ,ءء p Ͻ 0.004 and ,ء .dominantly CCR10 and include both CCR10 T cells (45 Ϯ 3.1%) and as mean Ϯ SEM ϩ CCR10 B cells (33.4 Ϯ 2.2%). The majority of migrated T cells were Wilcoxon signed rank test where appropriate. 600 CCR10 EXPRESSION ON EPITHELIAL Tregs

by positive immunomagnetic selection. Isolated lymphocyte frac- tions were cultured 1:1 with allogeneic myeloid dendritic cells and CFSE-labeled lymphocytes added in ratios 1:1 to 1:30. CD25ϪCD4ϩCCR10ϩ lymphocytes did not inhibit proliferation of stimulated CFSE-labeled naive T cells, whereas CD25ϩCD4ϩ CCR10ϩ lymphocytes mediated potent, dose-dependent inhibition of proliferation ( p Ͻ 0.003 compared with CD25Ϫ samples). Fur- ␣ ther enrichment for E expression increased the suppressor activity, but this was not statistically significant. We confirmed the suppressive effects of CD25ϩCD4ϩCCR10ϩ T cells in a standard [3H]thymidine proliferation assay. Ratios of 1:5 to 1:1 with autologous naive T cells activated by anti-CD3/CD28 beads significantly reduced proliferation of naive lymphocytes (Fig. 7c).

IL-10 production by CCR10ϩCD4ϩCD25ϩ intrahepatic T cells A feature of Tregs is their ability to secrete the anti-inflammatory cytokine IL-10. To assess whether CD25ϩCD4ϩCCR10ϩ lymphocytes are able to produce IL-10, cell surface labeling was used to capture and detect IL-10 secreted by the different subpopula- Downloaded from tions of lymphocytes. CD4ϩCD25ϩ and CD4ϩCD25Ϫ subsets as well as those enriched for CCR10 expression were cultured with IL-2 or PHA and IL-10 detected by flow cytometry after 4 h. CD4ϩCD25ϩ and CD4ϩCD25ϩCCR10ϩ populations secreted IL-10 and this result was confirmed using RT-PCR and IL-10-

speciﬁc primers with GAPDH used as a loading control (Fig. 7, d http://www.jimmunol.org/ and e).

CCR10ϩ Foxp3ϩ intrahepatic Tregs are CCR7low and express high levels of the inflammatory chemokine receptor CXCR3 CCR10ϩ Tregs from inflamed livers expressed high levels of CXCR3 and relatively low levels of CCR7 consistent with an activated, tissue infiltrating phenotype. This finding is in contrast to CCR10Ϫ Tregs in blood that expressed low levels of CXCR3 and high levels of CCR7 consistent with peripheral naive-like Tregs by guest on September 26, 2021 and a proclivity to migrate to secondary lymphoid tissue. Given the lack of a reliable surface marker to identify Tregs within the CD4ϩCD25ϩ T cell population, we used RT-PCR to confirm Foxp3 mRNA expression in the CXCR3ϩ and CCR7ϩCD25ϩ CD4ϩCCR10ϩ populations (Fig. 8).

Discussion Natural Tregs suppress tissue damage that occurs as a consequence of inflammatory or antimicrobial immune responses (15). Tregs are positively selected in the thymus as a consequence of their high affinity for self Ags expressed on thymic epithelium and can be detected in blood, lymphoid tissues, and inflamed peripheral tissues. Their function in vivo relies on their ability to produce IL-10 and surface-bound TGF-␤ but it remains unclear exactly where Tregs operate. A subset of “naive” circulating Tregs has recently FIGURE 6. Characterization of CCR10 expression by Tregs in human been reported in the peripheral circulation, which express the liver. a, CD4ϩ T cells were enriched from inflamed livers and flow cy- lymph node homing receptors CCR7 and CD62L, suggesting that tometry was used to identify the CD25ϩCD4ϩ population. These cells they home to secondary lymphoid tissue where they may suppress comprised 10.2 Ϯ 4.8% of the total CD4ϩ T cell infiltrate. Gating on CCR10ϩ expression revealed that 42.6% of the CCR10ϩ cells were CD25ϩCD4ϩ T cells. b, To confirm that the CCR10ϩ T cell population in inflamed liver includes a population of bona fide Tregs, rather than acti- percentage expression and agarose gels are shown (n ϭ 6). c, Immunohis- vated CD25ϩ T cells, we performed RT-PCR for the Treg-specific tran- tochemistry was used to investigate the tissue location of Foxp3ϩ Tregs. In scription factor Foxp3. mRNA for Foxp3 was clearly identified within total the normal liver Foxp3ϩ Tregs were infrequent but were consistently lo- lymphocyte isolates from the inflamed liver. The expression of Foxp3 on cated in close proximity to bile ducts (arrow). Foxp3ϩ Treg numbers were CCR10ϩ liver T cells was determined by magnetic sorting of lymphocytes increased in chronic inflammatory liver disease predominantly in ex- isolated from inflamed liver tissue to obtain CCR10ϩ and CCR10Ϫ pop- panded, inflamed portal tracts. Dual immunofluorescence (bottom) using ulations followed by RT-PCR with specific human Foxp3 primers. Intense Abs to Foxp3-Texas Red and CCR10 FITC-green revealed coexpression expression of Foxp3 was detected in the CCR10-enriched population and (yellow) of CCR10 on these portal Foxp3ϩ Tregs. Typical tissue sections confirmed the presence of Tregs within the CCR10ϩ population. Re- are shown that were representative of the staining seen in tissue sections sults are expressed as mean Ϯ SEM. Representative FACS plots with from six donors each for PSC, PBC, normal liver, and ALD. The Journal of Immunology 601 Downloaded from http://www.jimmunol.org/

FIGURE 7. CCR10ϩ Tregs in human liver can suppress allogeneic proliferation in vitro. a and b, CD4ϩ liver lymphocytes were isolated from diseased liver tissue by negative selection and positively enriched for CD25 and CCR10 expression. Where indicated, CCR10ϩCD25ϩCD4ϩ Tregs were further ␣ ϩ ϩ ϩ ϩ Ϫ ϩ isolated based on E expression. Matched naive PBL were labeled with CFSE. CCR10 CD25 CD4 or CCR10 CD25 CD4 lymphocytes were cultured with allogeneic myeloid dendritic cells and with CFSE-labeled cells added in ratios 1:1 to 1:30. Samples were cultured for 7 days and cell division measured by flow cytometry. The mean number of cell divisions was calculated based on the number of CFSEϩ cells per cell division and divided by the total viable ϩ ϩ ϩ Ϫ p Ͻ 0.003; n ϭ 4) at ratios of 1:1 to 1:10. by guest on September 26, 2021 ,ء) cell count. CCR10 CD25 CD4 Tregs inhibited proliferation compared with CD25 subsets ϩ ϩ ϩ ␣ ϩ ϭ CCR10 CD25 CD4 E populations appeared more potent at inhibiting proliferation, although the difference did not reach statistical significance (p 0.09). Representative histograms from one of the flow cytometric analyses are shown. c, To further illustrate the suppressive effects of CCR10ϩCD25ϩCD4ϩ T cells, we used [3H]thymidine proliferation assays to confirm suppression of proliferation at 1:5 to 1:1 ratios with naive lymphocytes activated by anti-CD3/CD28 beads. Values for p were calculated by paired Student’s t test. d and e, IL-10 secretion by CD4ϩCD25ϩ Tregs isolated from liver tissue was demonstrated using an extracellular capture assay. CD4ϩCD25ϩ cells, further separated into CCR10Ϫ and CCR10ϩ subsets, were stimulated for 4 h with 1000 U/ml recombinant human IL-2, and expression of IL-10 was measured by extracellular capture detected by flow cytometry. Consistent with a regulatory phenotype, stimulation of CCR10ϩCD4ϩCD25ϩ cells in culture with PHA or by high dose IL-2 resulted in detectable IL-10 in 47 Ϯ 4.4% and 42 Ϯ 3.0% of the CCR10ϩCD4ϩCD25ϩ cells, respectively. Transcription of IL-10 by CD25ϩCD4ϩ and CD25ϩCD4ϩCCR10ϩ populations was confirmed by RT-PCR with GAPDH expression used as a comparator for loading (mean Ϯ SEM; n ϭ 4). the proliferation of T cells during early stages of immune activa- ligands are expressed on endothelium and in some cases epithe- tion (27, 35). In contrast a subset of highly differentiated Tregs, lium of inflamed tissues and thus might aid Treg entry but are not ␣ which express the E integrin is found in inflamed peripheral tissue tissue specific (39). Our finding of CCR10 expression on a popu- where they exert anti-inflammatory properties (27). Inflammation lation of liver infiltrating Tregs suggests that CCL28 might provide is markedly increased in the absence of such cells and it is likely a mucosal signal to selectively recruit Tregs to epithelial surfaces. that they act to maintain suppression of immune responses and The coexpression of CCR10 with CXCR3 we describe would aug- particularly to limit the tissue damage resulting from attempts to ment recruitment of Tregs at inflamed epithelial sites such as the control infectious agents. Recent studies suggest that chronic in- liver where CXCR3 ligands are strongly expressed (39). We de- flammation is the result of a balance between pro- and anti- tected CCR10ϩ Tregs in the chronically inflamed liver in close inflammatory pathways in which Tregs promote stable chronic in- proximity to intrahepatic bile ducts the epithelial cells of which flammation while preventing fulminant destructive inflammation showed strong expression of CCL28. These intrahepatic Tregs ex- (34). pressed high levels of CXCR3, which is known to be critical for The factors responsible for the recruitment of Tregs to periph- the recruitment of effector T cells to the inflamed liver (39), but eral tissues are not well understood. CCR4, CCR8, and CXCR3 low levels of the lymph node homing chemokine receptor CCR7. ϩ ␣ have been reported on Tregs and could provide entry to dermal A subset of the CCR10 Tregs expressed high levels of the E sites (CCR4, CCR8) (36) and areas of general inflammation integrin, which defines Tregs with increased anti-inflammatory (CXCR3) (37). In addition, CCR4 and CCR8 ligands may also properties, and confers on them the ability to be retained at mu- contribute to Treg-dendritic cell interactions after tissue entry (38). cosal surfaces by binding to E-cadherin on epithelial cells. A However, specific signals that promote trafficking of Tregs to ep- higher proportion of intrahepatic T cells were Tregs in the diseased ithelial sites such as the gut, lung, and liver are not known. CXCR3 livers, suggesting that these cells are recruited as a consequence of 602 CCR10 EXPRESSION ON EPITHELIAL Tregs

dendritic cells to facilitate cross-homing between the gut and the skin (45). Tregs are likely to play a critical role in the pathogenesis of chronic liver disease. Hepatitis C infection is characterized by increased numbers of circulating CD4ϩ Tregs, depletion of which enhances Ag-specific CD8 effector responses (46). Increased intrahepatic CD4ϩ Tregs have also been reported in the liver of patients with hepatocellular carcinoma (47). In both cases Tregs are implicated in the disease pathogenesis by suppressing poten- tially beneficial T cell responses against virus or tumor, respectively. Regulation of liver inflammation is not only restricted to CD4ϩ Tregs because an intrahepatic population of CD8ϩ T cells with regulatory activity has been reported in chronic hepatitis C infection (48). In chronic inflammatory disease, the role of Tregs may be more complex. We studied two diseases targeted to bile ducts, PBC and PSC. Autoimmune and infectious etiologies have been proposed for both conditions but no definitive pathogenic mechanism has been reported for either disease (49, 50). Both

diseases are characterized by chronic inflammation focused on the Downloaded from FIGURE 8. CCR10ϩ Tregs have a tissue infiltrating phenotype that is Ϫ biliary epithelium and both show a slow progression to fibrosis and distinct from CCR10 naive Tregs. a, Flow cytometry revealed that CCR10ϩ Tregs from inflamed livers expressed high levels of CXCR3 and cirrhosis. It is possible that Tregs are initially recruited to limit the relatively low levels of CCR7 whereas CCR10Ϫ Tregs from blood ex- tissue damage associated with bile duct inflammation and subse- pressed low levels of CXCR3 and high levels of CCR7. b, Given the lack quently maintain the stability of chronic inflammation, which char- of a reliable surface markers for Tregs we measured expression of FoxP3 acterizes both diseases. Support for this model is provided by ev- ϩ in CCR7 and CXCR3 subsets using RT-PCR with primers specific for idence that Tregs play such a role in animal models of chronic T http://www.jimmunol.org/ human FoxP3 and controlled sample loading based on ␤-actin expression. cell-mediated inflammation in the lung and inflammatory bowel ϩ Ϫ RT-PCR confirmed Foxp3 mRNA expression in the CCR10 or CCR10 disease (34, 51). Chronic Ag exposure in the lung mucosa pro- ϩ Treg populations that have been further enriched based on CXCR3 and motes the recruitment of Tregs and the establishment of a local ϩ ϭ CCR7 expression (n 4). Representative FACS plots and agarose gels microenvironment in which secretion of the proinflammatory cy- are shown. Results are expressed as mean Ϯ SEM. tokines IL-2 and IFN-␥ is reduced and levels of IL-10 and IL-5 increased. The intrahepatic T cells we describe secrete IL-10 and may not only suppress inflammatory damage but also promote the chronic inflammation to limit tissue damage. However, ϳ10% of development of a fibrogenic microenvironment and progression to intrahepatic T cells in noninflamed livers from organ donors were fibrosis and cirrhosis. by guest on September 26, 2021 Tregs, suggesting that these cells may play a role in suppressing In summary we propose that expression of CCL28 by epithelial responses in the liver under physiological conditions. This out- cells in response to microbial products or IL-1 provides a signal to come is consistent with a recent murine study suggesting that localize CCR10ϩ Tregs at mucosal surfaces. This report is the first Tregs play a critical role in regulating the transition from hepatic linking a specific chemokine receptor with recruitment of Tregs to immune tolerance to hepatic inflammation (40). epithelial surfaces and suggests that CCR10 expression defines a We found very few CCR10ϩ Tregs in the peripheral circulation. population of Tregs with the capacity to traffic to mucosal sites to This finding is in contrast to CCR7ϩ Tregs, which can be detected limit autoimmunity and inflammation. at relatively high frequencies in blood and secondary lymphoid tissue (35). This implies different pathways of homing for these Acknowledgments ϩ two Treg subsets. It has been suggested that CCR7 naive Tregs We thank Gary Reynolds for expert help with immunohistochemistry as operate in secondary lymphoid tissues to suppress early events well as Janine Youster and Jean Shaw for contribution in the collection and during T cell activation including the expression of chemokine processing of the tissue samples. We acknowledge the advice of Dr. Vin- receptors required for tissue infiltration (41). CCR4, CCR5, and cent Lai in isolating dendritic cell subsets. CCR8 have been implicated in the recruitment of Tregs to lymphoid tissues (42, 43) but as yet very little is known about the Disclosures signals responsible for recruiting Tregs to peripheral tissues. Our The authors have no financial conflict of interest. data suggest that CXCR3 and CCR10 might be critical in this process and it is possible that activation of naive Tregs in lymph References nodes leads to the differentiation of Tregs with a tissue-infiltrating 1. Sakaguchi, S. 2004. Naturally arising CD4ϩ regulatory T cells for immunologic phenotype characterized by expression of CCR10, CXCR3, and in self-tolerance and negative control of immune responses. Annu. Rev. Immunol. ␣ 22: 531–562. some cells high levels of E integrin. Thus naive-type and tissue- 2. Walker, L. S., A. Chodos, M. Eggena, H. Dooms, and A. K. Abbas. 2003. An- infiltrating Tregs may operate at different sites and at different tigen-dependent proliferation of CD4ϩCD25ϩ regulatory T cells in vivo. J. Exp. stages of the immune response to suppress inflammation. The fact Med. 198: 249–258. 3. Fisson, S., G. Darrasse-Je`ze, E. Litvinova, F. Septier, D. Klatzmann, R. Liblau, that a subset of skin-homing T cells express CCR10 and respond and B. L. Salomon. 2003. Continuous activation of autoreactive CD4ϩCD25ϩ to the other CCR10 ligand, CCL27 in the skin suggests that regulatory T cells in the steady state. J. Exp. Med. 198: 737–746. 4. Kronenberg, M., and A. Rudensky. 2005. Regulation of immunity by self-reac- CCR10 expression on Tregs would allow them to enter both der- tive T cells. Nature 435: 598–604. mal and mucosal sites depending on their coexpression of addi- 5. Hsieh, C. S., Y. Liang, A. J. Tyznik, S. G. Self, D. Liggitt, and A. Y. Rudensky. ϩ ϩ tional skin- or gut-specific homing molecules (19, 44). Expression 2004. Recognition of the peripheral self by naturally arising CD25 CD4 T cell receptors. Immunity 21: 267–277. of gut and skin adhesion molecules are, however, subject to sig- 6. Hori, S., T. Nomura, and S. Sakaguchi. 2003. Control of regulatory T cell de- nificant plasticity and could be altered by interactions with local velopment by the transcription factor Foxp3. Science 299: 1057–1061. The Journal of Immunology 603

7. Khattri, R., T. Cox, S. A. Yasayko, and F. Ramsdell. 2003. An essential role for lymphocyte adhesion to hepatic endothelium (MAdCAM-1 in chronic inflamma- Scurfin in CD4ϩCD25ϩ T regulatory cells. Nat. Immunol. 4: 337–342. tory liver disease). Hepatology 33: 1065–1072. 8. Tivol, E. A., F. Borriello, A. N. Schweitzer, W. P. Lynch, J. A. Bluestone, and 30. Joplin, R., A. J. Strain, and J. M. Neuberger. 1989. Immuno-isolation and culture A. H. Sharpe. 1995. Loss of CTLA-4 leads to massive lymphoproliferation and of biliary epithelial-cells from normal human-liver. In Vitro Cell. Dev. Biol. 25: fatal multiorgan tissue destruction, revealing a critical negative regulatory role of 1189–1192. CTLA-4. Immunity 3: 541–547. 31. Burdick, M. D., S. L. Kunkel, P. M. Lincoln, C. A. Wilke, and R. M. Strieter. 9. Raimondi, G., W. J. Shufesky, D. Tokita, A. E. Morelli, and A. W. Thomson. 1993. Specific ELISAs for the detection of human macrophage inflammatory 2006. Regulated compartmentalization of programmed cell death-1 discriminates protein-1␣ and ␤. Immunol. Invest. 22: 441–449. ϩ ϩ CD4 CD25 resting regulatory T cells from activated T cells. J. Immunol. 176: 32. Goddard, S., J. Youster, E. Morgan, and D. H. Adams. 2004. Interleukin-10 2808–2816. secretion differentiates dendritic cells from human liver and skin. Am. J. Pathol. 10. Nakamura, K., A. Kitani, and W. Strober. 2001. Cell contact-dependent immu- 164: 511–519. ϩ ϩ nosuppression by CD4 CD25 regulatory T cells is mediated by cell surface- 33. Klugewitz, K., D. H. Adams, M. Emoto, K. Eulenburg, and A. Hamann. 2004. bound transforming growth factor ␤. J. Exp. Med. 194: 629–644. The composition of intrahepatic lymphocytes: shaped by selective recruitment? 11. Asseman, C., S. Mauze, M. W. Leach, R. L. Coffman, and F. Powrie. 1999. An Trends Immunol. 25: 590–594. essential role for interleukin 10 in the function of regulatory T cells that inhibit 34. Westendorf, A. M., M. Templin, R. Geffers, S. Deppenmeier, A. D. Gruber, intestinal inflammation. J. Exp. Med. 190: 995–1004. M. Probst-Kepper, W. Hansen, R. S. Liblau, F. Gunzer, D. Bruder, and J. Buer. 12. Mottet, C., H. H. Uhlig, and F. Powrie. 2003. Cutting edge: cure of colitis by ϩ ϩ ϩ 2005. CD4 T cell mediated intestinal immunity: chronic inflammation versus CD4 CD25 regulatory T cells. J. Immunol. 170: 3939–3943. immune regulation. Gut 54: 60–69. 13. Eksteen, B., L. S. Walker, and D. H. Adams. 2005. Immune regulation and 35. Valmori, D., A. Merlo, N. E. Souleimanian, C. S. Hesdorffer, and M. Ayyoub. colitis: suppression of acute inflammation allows the development of chronic 2005. A peripheral circulating compartment of natural naive CD4 Tregs. J. Clin. inflammatory bowel disease. Gut 54: 4–6. Invest. 115: 1953–1962. 14. MacDonald, T. T., and G. Monteleone. 2005. Immunity, inflammation, and al- 36. Lee, I., L. Wang, A. D. Wells, M. E. Dorf, E. Ozkaynak, and W. W. Hancock. lergy in the gut. Science 307: 1920–1925. 2005. Recruitment of Foxp3ϩ T regulatory cells mediating allograft tolerance 15. Belkaid, Y., and B. T. Rouse. 2005. Natural regulatory T cells in infectious depends on the CCR4 chemokine receptor. J. Exp. Med. 201: 1037–1044. disease. Nat. Immunol. 6: 353–360.

37. Debes, G. F., M. E. Dahl, A. J. Mahiny, K. Bonhagen, D. J. Campbell, Downloaded from 16. Rodriguez, M. W., A. C. Paquet, Y. H. Yang, and D. J. Erle. 2004. Differential ϩ Ϫ K. Siegmund, K. J. Erb, D. B. Lewis, T. Kamradt, and A. Hamann. 2006. Che- ␤ ␤ ϩ gene expression by integrin 7 and 7 memory T helper cells. BMC Immunol. motactic responses of IL-4-, IL-10-, and IFN-␥-producing CD4 T cells depend 5: 13. on tissue origin and microbial stimulus. J. Immunol. 176: 557–566. 17. Kunkel, E. J., C. H. Kim, N. H. Lazarus, M. A. Vierra, D. Soler, E. P. Bowman, 38. D’Ambrosio, D., F. Sinigaglia, and L. Adorini. 2003. Special attractions for sup- and E. C. Butcher. 2003. CCR10 expression is a common feature of circulating pressor T cells. Trends Immunol. 24: 122–126. and mucosal epithelial tissue IgA Ab-secreting cells. J. Clin. Invest. 111: 39. Curbishley, S. M., B. Eksteen, R. P. Gladue, P. Lalor, and D. H. Adams. 2005. 1001–1010. CXCR3 activation promotes lymphocyte transendothelial migration across hu- 18. Wang, W., H. Soto, E. R. Oldham, M. E. Buchanan, B. Homey, D. Catron, man hepatic endothelium under ﬂuid ﬂow. Am. J. Pathol. 167: 887–899.

N. Jenkins, N. G. Copeland, D. J. Gilbert, N. Nguyen, et al. 2000. Identification http://www.jimmunol.org/ 40. Wiegard, C., C. Frenzel, J. Herkel, K. J. Kallen, E. Schmitt, and A. W. Lohse. of a novel chemokine (CCL28), which binds CCR10 (GPR2). J. Biol. Chem. 275: 2005. Murine liver antigen presenting cells control suppressor activity of 22313–22323. ϩ ϩ CD4 CD25 regulatory T cells. Hepatology 42: 193–199. 19. Soler, D., T. L. Humphreys, S. M. Spinola, and J. J. Campbell. 2003. CCR4 41. Sarween, N., A. Chodos, C. Raykundalia, M. Khan, A. K. Abbas, and versus CCR10 in human cutaneous Th lymphocyte trafficking. Blood 101: L. S. Walker. 2004. CD4ϩCD25ϩ cells controlling a pathogenic CD4 response 1677–1682. inhibit cytokine differentiation, CXCR-3 expression, and tissue invasion. J. Im- 20. Ogawa, H., M. Iimura, L. Eckmann, and M. F. Kagnoff. 2004. Regulated pro- munol. 173: 2942–2951. duction of the chemokine CCL28 in human colon epithelium. Am. J. Physiol. 287: G1062–G1069. 42. Bystry, R. S., V. Aluvihare, K. A. Welch, M. Kallikourdis, and A. G. Betz. 2001. 21. Hieshima, K., H. Ohtani, M. Shibano, D. Izawa, T. Nakayama, Y. Kawasaki, B cells and professional APCs recruit regulatory T cells via CCL4. Nat. Immunol. F. Shiba, M. Shiota, F. Katou, T. Saito, and O. Yoshie. 2003. CCL28 has dual 2: 1126–1132. roles in mucosal immunity as a chemokine with broad-spectrum antimicrobial 43. Iellem, A., M. Mariani, R. Lang, H. Recalde, P. Panina-Bordignon, F. Sinigaglia, and D. D’Ambrosio. 2001. Unique chemotactic response profile and specific activity. J. Immunol. 170: 1452–1461. ϩ ϩ by guest on September 26, 2021 22. Mu¨ller, A., B. Homey, H. Soto, N. Ge, D. Catron, M. E. Buchanan, expression of chemokine receptors CCR4 and CCR8 by CD4 CD25 regulatory T. McClanahan, E. Murphy, W. Yuan, S. N. Wagner, et al. 2001. Involvement of T cells. J. Exp. Med. 194: 847–853. chemokine receptors in breast cancer metastasis. Nature 410: 50–56. 44. Homey, B., H. Alenius, A. Mu¨ller, H. Soto, E. P. Bowman, W. Yuan, L. McEvoy, 23. Hwang, S. T. 2004. Chemokine receptors in melanoma: CCR9 has a potential role A. I. Lauerma, T. Assmann, E. Bu¨nemann, et al. 2002. CCL27-CCR10 interac- in metastasis to the small bowel. J. Invest. Dermatol. 122: xiv–xxv. tions regulate T cell-mediated skin inflammation. Nat. Med. 8: 157–165. 24. Lazarus, N. H., E. J. Kunkel, B. Johnston, E. Wilson, K. R. Youngman, and 45. Mora, J. R., G. Cheng, D. Picarella, M. Briskin, N. Buchanan, and E. C. Butcher. 2003. A common mucosal chemokine (mucosae-associated epi- U. H. von Andrian. 2005. Reciprocal and dynamic control of CD8 T cell homing thelial chemokine/CCL28) selectively attracts IgA plasmablasts. J. Immunol. by dendritic cells from skin- and gut-associated lymphoid tissues. J. Exp. Med. 170: 3799–3805. 201: 303–316. 25. Shiraishi, K., K. Tsuzaka, K. Yoshimoto, C. Kumazawa, K. Nozaki, T. Abe, 46. Sugimoto, K., F. Ikeda, J. Stadanlick, F. A. Nunes, H. J. Alter, and K. M. Chang. K. Tsubota, and T. Takeuchi. 2005. Critical role of the fifth domain of E-cadherin 2003. Suppression of HCV-specific T cells without differential hierarchy dem- ␣ ␤ onstrated ex vivo in persistent HCV infection. Hepatology 38: 1437–1448. for heterophilic adhesion with E 7, but not for homophilic adhesion. J. Immu- nol. 175: 1014–1021. 47. Unitt, E., S. M. Rushbrook, A. Marshall, S. Davies, P. Gibbs, L. S. Morris, 26. Huehn, J., K. Siegmund, J. C. Lehmann, C. Siewert, U. Haubold, M. Feuerer, N. Coleman, and G. J. Alexander. 2005. Compromised lymphocytes infiltrate G. F. Debes, J. Lauber, O. Frey, G. K. Przybylski, et al. 2004. Developmental hepatocellular carcinoma: the role of T-regulatory cells. Hepatology 41: 722–730. stage, phenotype, and migration distinguish naive- and effector/memory-like 48. Accapezzato, D., V. Francavilla, M. Paroli, M. Casciaro, L. V. Chircu, CD4ϩ regulatory T cells. J. Exp. Med. 199: 303–313. A. Cividini, S. Abrignani, M. U. Mondelli, and V. Barnaba. 2004. Hepatic ex- ϩ 27. Siegmund, K., M. Feuerer, C. Siewert, S. Ghani, U. Haubold, A. Dankof, pansion of a virus-specific regulatory CD8 T cell population in chronic hepatitis V. Krenn, M. P. Scho¨n, A. Scheffold, J. B. Lowe, et al. 2005. Migration matters: C virus infection. J. Clin. Invest. 113: 963–972. regulatory T-cell compartmentalization determines suppressive activity in vivo. 49. Aoki, C. A., C. L. Bowlus, and M. E. Gershwin. 2005. The immunobiology of Blood 106: 3097–3104. primary sclerosing cholangitis. Autoimmun. Rev. 4: 137–143. 28. Eksteen, B., A. J. Grant, A. Miles, S. M. Curbishley, P. F. Lalor, S. G. Hubscher, 50. Nishio, A., E. B. Keeffe, and M. E. Gershwin. 2002. Immunopathogenesis of M. Briskin, M. Salmon, and D. H. Adams. 2004. Hepatic endothelial CCL25 primary biliary cirrhosis. Semin. Liver Dis. 22: 291–302. mediates the recruitment of CCR9ϩ gut-homing lymphocytes to the liver in pri- 51. Bruder, D., A. M. Westendorf, R. Geffers, A. D. Gruber, M. Gereke, R. I. Enelow, mary sclerosing cholangitis. J. Exp. Med. 200: 1511–1517. and J. Buer. 2004. CD4 T Lymphocyte-mediated lung disease: steady state be- 29. Grant, A. J., P. F. Lalor, S. G. Hubscher, M. Briskin, and D. H. Adams. 2001. tween pathological and tolerogenic immune reactions. Am. J. Respir. Crit. Care MAdCAM-1 expressed in chronic inflammatory liver disease supports mucosal Med. 170: 1145–1152.