Signaling via programmed death ligand-1 (PD-L1) and PD-L2 is crucial for maintaining peripheral tolerance. CD90+ myofibroblasts/fibroblasts (CMFs) are major programmed cell death-1 (PD-1) ligand–expressing cells in normal human colonic mucosa. CMFs suppress activated CD4+ T cell proliferation via PD-1 ligands. It is not known whether signaling through TLRs contribute to the regulation PD-1 ligands on CMFs upon colonic mucosal tolerance. In this study, we demonstrated that stimulation of TLR4 on human CMFs upregulates PD-L1, but not PD-L2, and reinforces CMF-mediated suppression of CD4+ T cell proliferation and IFN-γ production. TLR4-mediated upregulation of PD-L1 on CMFs involved NF-κB pathways and was JAK2 and MyD88 dependent. MyD88-dependent stimulation of TLR1/2 and TLR5 also upregulated PD-L1 expression on CMFs in culture. PD-L1 expression was drastically decreased in vivo in the colonic mucosa of mice devoid of MyD88. Induction of MyD88 deficiency in CMFs in fibroblast-specific MyD88 conditional knockout mice resulted in a strong increase in a mucosal IFN-γ expression concomitantly with the abrogation of PD-L1 expression in CMFs under homeostasis and epithelial injury induced by dextran sodium sulfate. Together, these data suggest that MyD88-dependent TLR stimulation of CMFs in the normal colonic mucosa may reinforce these cells’ anti-inflammatory capacity and thus contribute to the maintenance of mucosal tolerance.

The gastrointestinal (GI) mucosa plays a fundamental role in the immunoregulatory response to pathogens, sensitization/desensitization to dietary Ags, and tolerance to self-Ags and commensal flora in the GI system (1). The mammalian colon is highly populated by variety of different microorganisms. Using standard culture methods, it was estimated three decades ago that at least 400 species of bacteria are present in fecal samples (2), whereas current metagenomic methodology has suggested that those numbers may have been an underestimate, representing only 30–40% of the total microbial diversity in colon (35). Despite this high level of exposure to microbial Ags, the normal colonic mucosa is protected by innate and adaptive protolerogenic mechanisms that homeostatically control inflammatory responses. Disruption of mucosal tolerance in the colon is associated with several incurable and chronic diseases, such as inflammatory bowel disease (IBD). Therefore, understanding the underpinning mechanisms responsible for colonic mucosal tolerance has potential translational value for IBD therapy and management.

Despite significant progress in understanding GI mucosal tolerance, the cells and pathways that determine the establishment and maintenance of tolerance in the colon remain poorly understood. A prominent role in peripheral tolerance has been proposed for the inhibitory molecules programmed death ligand-1 (PD-L1) (also known as B7-H1 and CD274) and PD-L2 (also known as B7-DC and CD273). These B7 family molecules are mainly expressed on APCs and appear to have dual roles as both negative modulators of activated T cell responses and as costimulators of early T cell priming and differentiation (610). Signaling through these ligands limits the proliferation of activated effector T cells and regulates the balance among Th1, Th2, Th17, and CD4+ regulatory T responses (1013). The best-characterized tolerogenic receptor is programmed cell death-1 (PD-1; CD279), which negatively regulates TCR signals, controls the balance of T cell activation and immune-mediated tissue damage, and provides multiple tolerance checkpoints that prevent autoimmunity (7).

Despite extensive investigation of the B7 family of costimulators in various tissues, their involvement in the regulation of mucosal immune responses in the GI tract remains poorly understood. The crucial role of PD-L1–mediated signaling in mucosal tolerance in the small intestine has recently been highlighted by the report that loss of PD-L1/PD-1 signaling leads to a development of an autoimmune enteritis and destruction of the small intestinal epithelium (6). PD-L1−/− animals have elevated systemic IFN-γ production (14). PD-L1–mediated control of inflammatory IFN-γ–producing effector T cells in the GI mucosa has particular importance, as immune surveillance by these cells is critical for the host responses to tumors and infection and contributes to autoimmune responses (13). In particular, altered PD-L1–mediated signaling/expression has been reported to contribute to chronic inflammation during IBD (15, 16).

Limited data are available on cells expressing PD-L1 and PD-L2 in the GI tract. CD90+ stromal cells (also known as CD90+ myofibroblasts/fibroblasts [CMFs]) are located just beneath the epithelial basement membrane and represent a major cell population expressing both PD-L1 and PD-L2 in the normal colonic mucosa (17). We and others (18, 19) previously reported that CMFs represent a distinct class of local nonprofessional APCs and are negative for the markers of professional APCs such as macrophages and dendritic cells (DCs). CMFs express MHC class II molecules. Under homeostasis, CMFs do not express CD80 (18) (also known as B7-1), which is also capable of signaling via PD-L1 (20). CMFs limit activated CD4+ T cell responses via signaling initiated by PD-L1 and PD-L2. Taken together (1719, 21), these recently published data demonstrate that normal human colonic mucosa colonic stromal cells, although capable of serving as APCs, play an immunosuppressive role instead.

TLRs recognize conservative molecular motifs of commensal and pathogenic microorganisms. Activation of TLRs results in the transduction of innate pro- and anti- inflammatory gene expression programs through distinct intracellular signaling pathways, controlled by signaling adapter complexes. MyD88 is a shared cytoplasmic TLR adapter, which plays a critical role in innate signaling through the NF-κB and IFN-regulatory factor transcription factors (2224). Signaling through TLRs on innate and adaptive immune cells plays a critical role in the maintenance of mucosal homeostasis. These interactions contribute to the maintenance of tolerance to commensal microorganisms and the induction of the inflammatory responses against pathogens (23, 25, 26). Although the colonic epithelial layer is a dynamic structure undergoing constant renewal, CMFs located just beneath the epithelium and with a slower turnover rate are constantly exposed to a vast array of colonic microbiota products that might stimulate innate immune receptor-mediated signaling (19, 27). Under conditions of homeostasis, these interactions do not result in overt inflammatory responses, presumably due to mechanisms of mucosal tolerance. A critical contribution of PD-1 ligand–mediated signaling to peripheral tolerance in vivo has been demonstrated (6, 14). Recent studies showed that stimulation of TLRs induces PD-L1 expression on DCs, macrophages, and cancer cells (2830). However, it is not known whether signaling via TLRs contributes to the regulation of PD-1 ligands in the GI tract. Thus, the objective of the study presented in this paper was to investigate the role of MyD88-mediated TLR signaling in the modulation of the PD-L1 expression by CMFs.

In this study, we demonstrate that stimulation of TLR4 on CMFs derived from the human normal mucosa leads to the NF-κB–dependent upregulation of PD-L1, but not PD-L2, with reinforced CMF-mediated suppression of CD4+ effector T cell responses. Lack of intact MyD88 signaling specifically within the fibroblast compartment in vivo resulted in an increased expression of the inflammatory cytokine IFN-γ concomitantly with the decrease in PD-L1 expression by CMFs. Taken together with previous observations by ourselves and others (14, 17, 31), these new data suggest that upregulation of PD-L1 expression by CMFs via MyD88-dependent TLR signaling is among the critical processes contributing to the suppression of inflammation under mucosal tolerance, and disruption of this mechanism might be an important factor involved in acute and chronic colonic inflammatory conditions.

Fluorochrome-conjugated and unconjugated murine anti–α-smooth muscle actin (α-SMA; clone 1A4) monoclonal and anti-MyD88 goat polyclonal Abs were purchased from Sigma-Aldrich (St. Louis, MO). Fluorochrome-conjugated forms of IgG1κ, IgG2a, isotype controls, and mAbs directed against human CD90 (clone 5E10) were from BD Pharmingen and eBioscience (San Diego, CA). Fluorochrome-conjugated mAbs against human and murine CD4 (clone RPA-T4 and RM4-5, respectively), T-bet (clone eBIo4B10), isotype controls, as well as mAbs against human PD-L1 (clone M1H1), PD-L2 (clone M1H18), murine PD-L1 (clone M1H5), and murine and human Fc receptor blocker were from eBioscience. Alexa Fluor (AF) 488– and AF633-labeled donkey anti-mouse IgG2a and IgG1κ (respectively), Zenon Mouse IgG, and Apex Ab labeling kits were purchased from Invitrogen (Carlsbad, CA). Anti-total IκBα murine mAbs (clone L35A5) and anti-biotin HRP-conjugated Abs were from Cell Signaling Technology (Danvers, MA). Goat anti-human PD-L1 polyclonal biotinylated Abs were purchased from R&D Systems (Minneapolis, MN). Goat anti-murine IgG HRP-conjugated Abs were from Bio-Rad (Hercules, CA).

For CMF isolation, full-thickness fresh human mucosal samples were obtained from discarded surgical resection material of colons in compliance with protocols approved by the University of Texas Medical Branch Institutional Review Board. Areas of uninvolved colon tissue from patients undergoing colectomy for colon cancer were studied. Total mucosal cell preparation was done as described previously (18). CMFs were isolated according to the protocol of Mahida et al. (32), which is routinely used in our laboratory (18). The purity of isolated CD90+ CMFs (98 to 99%) was confirmed by flow cytometry, as previously described (18). Studies were performed with primary CMF isolates at passages 4–10. Cells were cultured as described previously (18).

C57BL/6 mice were purchased from The Jackson Laboratory (Bar Harbor, ME). MyD88-deficient (MyD88−/−) mice on a C57BL/6 background were obtained from the University of California at San Diego. MyD88flox mice were obtained from The Jackson Laboratory and crossed with fibroblast-specific Cre mice (TgCol1α2-CreER) to generate fibroblast-specific tamoxifen-inducible conditional knockout (KO) mice under fibroblast-specific Col1α2 promoter selective for MyD88 (Myd88floxTgCol1α2-CreER or Fib-Myd88 KO). The fibroblast-specific Cre mice (TgCol1α2-CreER) mice were provided by Dr. Andrew Leask at University of Western Ontario (London, Ontario, Canada) and have been described previously (33). Deletion of myd88 was induced with the i.p. injection of tamoxifen (1 mg/mice for 4 d; total injection volume/animal is 100 μl) 1 wk prior to the experimental procedure. The presence of MyD88 floxed and Cre recombinase under Col1α2 promoter transgene was identified by using standard PCR protocols suggested by The Jackson Laboratory. For the detection of MyD88 floxed transgene, the following primer pair was used: oIMR9481fw 5′-GTT GTG TGT GTC CGA CCG T-3′ and oIMR9481rev 5′-GTC AGA AAC AAC CAC CAC CAT GC-3′. Cre recombinase transgene was detected using primers CreSfw 5′-AGG TTC GTT CAC TCA TGG A-3′ and CreASrev 5′-TCG ACC AGT TTA GTT ACC C-3′. Absence of MyD88 expression within the CMF compartment was controlled using immunostaining followed by confocal microscopy and flow cytometry. All mice (female, 6–12 wk of age) were housed under pathogen-free, but not germ-free, conditions. All experimental procedures were approved by the Institutional Animal Care and Use Committee of the University of California at San Diego and University of Texas Medical Branch at Galveston.

The dextran sodium sulfate (DSS) model of colitis was used to induce epithelial injury using the standard protocol (34). On day 1, animals received DSS solution (2% DSS in drinking water). Control animals received regular drinking water. On days 3 and 5, water supplies were refilled with fresh DSS and control solutions, respectively. On day 8, the DSS solution was replaced by regular water. Colitic changes and immune responses were analyzed at day 10. For this purpose, mice were anesthetized (80 μl ketamine at 80 mg/kg and xylazine 10 mg/kg, i.p.) and euthanized by cervical dislocation and samples taken for analysis enlisted below. Total mucosal cell preparation was done as described previously for human tissue (18).

Primary human CMFs with repressed expression of RelA, RelB, or MyD88 were generated in our laboratory using Stealth siRNA probes (Invitrogen, Carlsbad, CA). Negative small interfering RNA (siRNA) controls with appropriate GC content were included in each experiment. An optimal concentration of each siRNA (0.3 nM) was used for each transfection. Transfection of primary cells was performed using Nucleofector technology (Amaxa Biosystems, Gaithersburg, MD) according to the manufacturer’s instructions as previously described (10).

Assays were performed as previously described using allogeneic CMF/T cell cocultures (17). Briefly, human CD4+ CD45RA+ T cells were purified from PBMCs of healthy donors by negative selection using Naive CD4+ T Cell Isolation magnetic bead Kit II (Miltenyi Biotec, Auburn, CA). T cells were CFSE labeled as previously described (35) and preactivated with anti-human CD3 and CD28 microbeads (Miltenyi Biotec; per manufacturer’s instructions) for 1 h prior to coculture initiation. A ratio of 4:1 (T cells to CMFs) was used in these experiments. In some assays, CMFs were prestimulated with LPS (1 μg/ml) for 24 h. LPS was removed, and CMFs were washed three times with MEM containing polymyxin B (10 μg/ml) to remove residual LPS. Preactivated T cells were then added, and the cocultures were maintained in polymyxin-containing media. T cells were then recovered from cocultures on day 4, immunostained for the markers of interest, and analyzed by flow cytometry. Conditioned media from mono- and cocultures of T cells and CMFs were collected and analyzed for IFN-γ using single-plex analysis (Millipore) per the manufacturer’s instructions.

CMFs were detached from the culture flasks by treatment with cell dissociation buffer (Sigma-Aldrich) at 37°C for 15–30 min, followed by two washes with cold PBS. Single- and multicolor immunostaining were performed according to standard eBioscience surface and intracellular FACS staining protocols. Cells were analyzed by flow cytometry using LSRII cytometers (BD Biosciences) per the manufacturer’s procedure. Area, height, and width parameters for forward and side scatters were used to discriminate single live cells. An additional gate was set up based on the negativity for the fixable viability dye eFluor 780 (eBioscience), which was added during the surface marker staining to exclude dead cells from the analysis. Flow cytometry data were analyzed using FACSDiva 6.2 (BD Biosciences) and FlowJo (Tree Star) software.

Frozen murine colon tissue sections were fixed in 1% paraformaldehyde for 20 min at room temperature, blocked with normal rabbit serum (1:10 in PBS) for 15 min at room temperature, and then incubated with anti–PD-L1 (1 μg/ml) mouse mAbs (clone M1H5) overnight at 4°C. The sections were then stained with AF594-conjugated rabbit anti-rat IgG Abs (1:300) and AF488-conjugated anti–α-SMA mAbs for 30 min at room temperature. Each staining step was followed by six washes with PBS with Ca2+/Mg2+. Secondary mAbs and isotype controls were included in the analyses. The sections were then mounted in SlowFade Gold antifade reagent with DAPI (Invitrogen). Confocal microscopy was performed with a Zeiss LSM510 META laser-scanning confocal microscope (Carl Zeiss, Thornwood, NY) as previously described (17).

This analysis was performed on 10 μg protein as previously described (18).

Analysis was performed as previously described (17) according to the Applied Biosystems’s two-step RT real-time PCR protocol (Applied Biosystems, Foster City, CA). Briefly, all reagents were purchased from Applied Biosystems. The appropriate assays-on-demand gene expression assay mix (Applied Biosystems) for human 18S RNA and the gene of interest (a 20× mix of unlabeled PCR primers and TaqMan MGB probe, FAM dye-labeled) and 2 μl cDNA were added to the PCR reaction step. The reactions were carried out in a 20 μl final volume using a Bio-Rad Q5 real-time PCR machine (Bio-Rad) with the protocol 2 min at 50°C, 10 min at 95°C (1 cycle), 15 s at 95°C, and 1 min at 60°C (40 cycles).

Unless otherwise indicated, the results were expressed as the mean ± SE of data obtained from at least three independent experiments each performed in triplicate. Differences between means were evaluated by ANOVA using Student t test for multiple comparisons. The p values < 0.05 were considered statistically significant.

Because abnormal TLR4-mediated signaling is involved in chronic inflammation associated with IBD (3638), and TLR4 stimulation of the resident peritoneal macrophages upregulates PD-L1 (11), we first determined how LPS stimulation affected PD-L1 expression on CMFs in freshly digested normal colonic mucosal cell (MC) preparations. Stimulation of normal colonic MCs with LPS (1 μg/ml) led to strong upregulation of PD-L1 on CD90+ CMFs, which were the major population expressing PD-L1 in LPS-stimulated MC preparations (Fig. 1A). Minor increases in PD-L1 expression were also noted on nonstromal cells in the MC preparations. To discern whether this process is due to direct stimulation of TLR4 on CMFs, we analyzed pure primary cultures of human CMFs isolated from normal mucosa. By flow cytometry analysis, we confirmed that normal CMFs express surface TLR4 (Fig. 1B), as was previously reported (27). Stimulation of cultured CMFs with LPS also resulted in significant upregulation of PD-L1 surface protein (Fig. 1C) and mRNA (Fig. 1D). Analysis of the geometric size of CMFs (forward scatter) versus the fluorescence intensity of the anti–PD-L1 mAb-labeled cells confirmed that the observed LPS-mediated increase in PD-L1+ CMFs was due to surface expression of PD-L1 and not to the increase in the size of cells (data not shown). Further, the observed level of the LPS-mediated upregulation of PD-L1 was comparable to those reported earlier for macrophages and mucosal Langerhans cells (11, 28). The kinetics of PD-L1 mRNA expression by LPS increased significantly between 2 and 4 h of LPS exposure and returned to baseline levels after 24 h (Fig. 1D). Because PD-L2 and CD80 may serve as a ligand/receptor for PD-1 and PD-L1, respectively, we also analyzed whether TLR4 stimulation modulates their expression on normal CMFs. No detectable CD80 expression was observed upon TLR4 stimulation of the CMFs. There was a minor increase in surface PD-L2 expression by isolated normal CMF upon TLR4 stimulation. However, this increase did not reach statistical significance (Supplemental Fig. 1). These data suggest that in normal human colonic mucosa, CMFs respond to TLR4 stimulation with the upregulation of the T cell negative costimulator PD-L1.

FIGURE 1.

(A) Freshly digested normal colonic MC preparations were stimulated with 1 μg/ml of LPS for 16 h and stained with CD90, PD-L1, or isotype controls and analyzed by flow cytometry. One representative experiment and the collective results of five separate experiments are shown. (B) Primary normal CMF cultures express TLR4 as detected by flow cytometry. One representative experiment of 10 is shown. (C) Stimulation of cultured CMFs with 1 μg/ml of LPS (24 h) upregulates PD-L1 on the cell surface (flow cytometry analysis); values are expressed as percentage of positive cells ± SE of triplicate CMF cultures isolated (n = 9 donors). (D) Stimulation of cultured CMFs with 1 μg/ml of LPS upregulates PD-L1 mRNA levels (real time RT-PCR analysis). The means ± SE are shown as the results of duplicates of three representative experiments. *p < 0.05.

FIGURE 1.

(A) Freshly digested normal colonic MC preparations were stimulated with 1 μg/ml of LPS for 16 h and stained with CD90, PD-L1, or isotype controls and analyzed by flow cytometry. One representative experiment and the collective results of five separate experiments are shown. (B) Primary normal CMF cultures express TLR4 as detected by flow cytometry. One representative experiment of 10 is shown. (C) Stimulation of cultured CMFs with 1 μg/ml of LPS (24 h) upregulates PD-L1 on the cell surface (flow cytometry analysis); values are expressed as percentage of positive cells ± SE of triplicate CMF cultures isolated (n = 9 donors). (D) Stimulation of cultured CMFs with 1 μg/ml of LPS upregulates PD-L1 mRNA levels (real time RT-PCR analysis). The means ± SE are shown as the results of duplicates of three representative experiments. *p < 0.05.

Close modal

To assess the functional significance of the TLR4-mediated PD-L1 upregulation on CMFs, we analyzed their ability to suppress proliferation of activated CD4+ effector T cells. LPS prestimulation of CMFs led to an increase in the CMF-mediated suppression of activated CD4+ effector T cell proliferation. This process involved PD-L1 since its neutralization with anti–PD-L1 blocking mAbs, but not an isotype control, partially restored T cell proliferation (Fig. 2A, 2B). Upregulated PD-L1 expression is reported on activated T cells, and this may inhibit T cell proliferation in an endogenous manner (39, 40). In our experiments, addition of PD-L1 blocking mAbs did not alter proliferation of the activated T cell monocultures, suggesting that CMF-mediated PD-L1 signaling and not intrinsic T cell PD-L1 signaling is important in alteration of the T cell proliferation. Interestingly, we observed that the major reduction of CFSElow population of CD4+ T cells cocultured with CMFs was mostly in P10 (later) phase (Fig. 2A, 2B) and much less in earlier P5–P9 phases of proliferation. Because PD-1 ligand–mediated signaling may render PD-1+ effector T cells more susceptible to apoptosis (41, 42), our data may indicate that the apparent decline in T cell proliferation in the in vitro CMF/T cell cocultures may be due to increases in PD-L1–mediated cell death.

FIGURE 2.

LPS prestimulation of normal CMFs reinforces their capacity to suppress activity of the CD4+ T effector cells. CMFs were prestimulated with 1 μg/ml of LPS for 24 h, the remaining LPS was removed by extensive washes, and the CMFs cocultured with allogeneic CFSE-labeled, CD2/CD3/CD28-preactivated naive CD4+ T cells (Act T cells) at a ratio of 1:4 for 4 d in 24-well plates. T cell monocultures were included as experimental controls. In some experiments, cocultures were performed in the presence of anti–PD-L1 blocking mAbs (clone M1H1) or isotype controls. Proliferation of the CD4+ T cell was analyzed based on the dilution CFSE dye (A, B), and IFN-γ production in conditioned medium was analyzed using singleplex cytokine analysis (C). n = 5 allogeneic donor pairs, two experimental replicates each.

FIGURE 2.

LPS prestimulation of normal CMFs reinforces their capacity to suppress activity of the CD4+ T effector cells. CMFs were prestimulated with 1 μg/ml of LPS for 24 h, the remaining LPS was removed by extensive washes, and the CMFs cocultured with allogeneic CFSE-labeled, CD2/CD3/CD28-preactivated naive CD4+ T cells (Act T cells) at a ratio of 1:4 for 4 d in 24-well plates. T cell monocultures were included as experimental controls. In some experiments, cocultures were performed in the presence of anti–PD-L1 blocking mAbs (clone M1H1) or isotype controls. Proliferation of the CD4+ T cell was analyzed based on the dilution CFSE dye (A, B), and IFN-γ production in conditioned medium was analyzed using singleplex cytokine analysis (C). n = 5 allogeneic donor pairs, two experimental replicates each.

Close modal

Because PD-L1–mediated signaling is involved in downregulation of IFN-γ production by activated T cells, we then analyzed production of IFN-γ in these cocultures. LPS-prestimulated CMFs cocultured with activated CD4+ T cells were found to have an increased capacity to suppress T cell IFN-γ production in a PD-L1–dependent manner (Fig. 2C). Thus, our data suggest that upon exposure to the bacterial ligands that are recognized by TLR4 CMFs may contribute to mucosal tolerance via an increased capacity to suppress IFN-γ which is involved in proinflammatory CD4+ T effector cell responses.

Exposure to LPS induces activation of NF-κB in murine and human intestinal stromal cells (24, 37). We evaluated whether NF-κB is involved in LPS-mediated upregulation of PD-L1 on CMFs. Using Western blot analysis, we demonstrated that stimulation of CMFs with LPS resulted in a rapid degradation of IκBα, which was maximal at 1 h and returned to the basal level by 24 h post–LPS exposure (Fig. 3A). Moreover, LPS-induced upregulation of total (Fig. 3B) and surface PD-L1 (Fig. 4A, 4B) on CMFs was significantly inhibited by the presence of the NF-κB translocation inhibitor triptolide (20 ng/ml). Interestingly, we noted that basal expression of both PD-L1 and PD-L2 on CMFs was slightly decreased by treatment with triptolide (Fig. 4, Supplemental Fig. 1). Additional studies are currently in progress to understand the role of the NF-κB activity in the basal expression levels of B7 molecules. A previous report suggested that the PI3K/mammalian target of rapamycin (mTOR) pathway may also enhance PD-L1 expression (43). Thus, we tested the effect of the P70S6/mTOR kinase inhibitor rapamycin on CMFs, and this did not affect LPS-induced PD-L1 upregulation (data not shown). LPS-induced activation of the NF-κB–mediated response is reported to involve activation of protein tyrosine kinase JAK2 (44). We then analyzed whether inhibition of JAK2 activation decreased the LPS-induced PD-L1 expression by CMFs. AG490 specifically blocks JAK2 activation at the concentration range of 5–20 μM. Its application at 10 μM concentrations significant decreased the LPS-induced upregulation of PD-L1 (Fig. 4C). A previous study (4547) demonstrated that AG490 may have off-target effects (blocking of the constitutive activation of Stat3) when used at higher concentrations of 60–100 μM. This seems unlikely in our study because we used AG490 at a final concentration of 10 μM. Thus, activation of JAK2 is involved in LPS-mediated induction of PD-L1 on CMFs.

FIGURE 3.

The NF-κB pathway is involved in the TLR4-mediated upregulation of PD-L1. (A) Western blotting using Abs specific for total IκBα protein demonstrated that the total IκBα level rapidly decreases in LPS-treated (1 μg/ml) CMFs cultures compared with the untreated controls. (B) LPS-induced total cell PD-L1 upregulation was inhibited in the presence of triptolide (Trip, an NF-κB translocation inhibitor). Under denaturing conditions, the PD-L1 protein migrates as subunits of ∼50 kDa. Image J software available through the National Institutes of Health Web site was used to calculate relative density. Adjusted density values for tested samples were calculated by dividing the relative density of each sample lane obtained with anti–PD-L1 Abs by the relative density of the loading-control for the same lane (stain with anti–β-actin Abs). n = 3.

FIGURE 3.

The NF-κB pathway is involved in the TLR4-mediated upregulation of PD-L1. (A) Western blotting using Abs specific for total IκBα protein demonstrated that the total IκBα level rapidly decreases in LPS-treated (1 μg/ml) CMFs cultures compared with the untreated controls. (B) LPS-induced total cell PD-L1 upregulation was inhibited in the presence of triptolide (Trip, an NF-κB translocation inhibitor). Under denaturing conditions, the PD-L1 protein migrates as subunits of ∼50 kDa. Image J software available through the National Institutes of Health Web site was used to calculate relative density. Adjusted density values for tested samples were calculated by dividing the relative density of each sample lane obtained with anti–PD-L1 Abs by the relative density of the loading-control for the same lane (stain with anti–β-actin Abs). n = 3.

Close modal
FIGURE 4.

TLR4 stimulation of CMFs increases surface PD-L1 and depends on NF-κB and JAK2 activation. (A) Immunostaining followed by flow cytometry analysis demonstrated that LPS-induced PD-L1 upregulation was inhibited in the presence of the NF-κB translocation inhibitor triptolide (Trip). (B) Summary of PD-L1 surface expression modulation by LPS in presence/absence of Trip by CMF primary isolates. Values are expressed as a percentage of PD-L1+ cells ± SE of triplicate CMF cultures isolated from each donor; n = 10. (C) JAK2 is involved in the LPS-mediated upregulation of PD-L1 levels on CMFs. Real-time RT-PCR and flow cytometry were used to analyze expression of the PD-L1 on CMFs treated with LPS in presence/absence of JAK2 inhibitor AG490, 10 μM; n = 4). (D) CMFs isolated from normal colonic mucosa were transected with RelA- or RelB-specific siRNA or negative control siRNA, allowed to recover for 7 d, and stimulated with LPS (1 μg/ml) for 24 h. Immunostaining followed by multicolor flow cytometry demonstrated that silencing of either relA or relB gene expression abrogated LPS-induced surface upregulation of PD-L1 (n = 4 donors, two experimental replicates each). MFI, mean fluorescence intensity. p < 0.05.

FIGURE 4.

TLR4 stimulation of CMFs increases surface PD-L1 and depends on NF-κB and JAK2 activation. (A) Immunostaining followed by flow cytometry analysis demonstrated that LPS-induced PD-L1 upregulation was inhibited in the presence of the NF-κB translocation inhibitor triptolide (Trip). (B) Summary of PD-L1 surface expression modulation by LPS in presence/absence of Trip by CMF primary isolates. Values are expressed as a percentage of PD-L1+ cells ± SE of triplicate CMF cultures isolated from each donor; n = 10. (C) JAK2 is involved in the LPS-mediated upregulation of PD-L1 levels on CMFs. Real-time RT-PCR and flow cytometry were used to analyze expression of the PD-L1 on CMFs treated with LPS in presence/absence of JAK2 inhibitor AG490, 10 μM; n = 4). (D) CMFs isolated from normal colonic mucosa were transected with RelA- or RelB-specific siRNA or negative control siRNA, allowed to recover for 7 d, and stimulated with LPS (1 μg/ml) for 24 h. Immunostaining followed by multicolor flow cytometry demonstrated that silencing of either relA or relB gene expression abrogated LPS-induced surface upregulation of PD-L1 (n = 4 donors, two experimental replicates each). MFI, mean fluorescence intensity. p < 0.05.

Close modal

Because both canonical (RelA-dependent) and noncanonical (RelB-dependent) NF-κB pathways can be active in fibroblasts (48, 49), we determined which of these two NF-κB pathways is implicated in TLR4-mediated upregulation of PD-L1 expression on CMFs. We demonstrated that silencing the expression of RelA or RelB with specific siRNAs, but not control siRNA, abrogated LPS-induced PD-L1 upregulation on CMFs (Fig. 4D). These data suggest that both pathways contribute to the TLR4-mediated upregulation of PD-L1.

Because stimulation of TLR4 on APCs may activate both MyD88-dependent and -independent pathways (50), we analyzed whether the adaptor MyD88 is required for the TLR4-mediated PD-L1 upregulation on CMFs. Using an siRNA approach, we demonstrated that silencing MyD88 expression in CMFs abrogated the LPS-induced upregulation of PD-L1 (Fig. 5A).

FIGURE 5.

MyD88-dependent TLR signaling increases PD-L1 expression by human CMFs. (A) Silencing of the myd88 gene in CMFs abrogates TLR4-mediated increases in PD-L1 mRNA expression. Primary CMFs isolated from normal colonic mucosa were transfected with MyD88 siRNA or control siRNA 7 d prior to exposure to the TLR4 agonist LPS (100 ng/ml) for 4 h. PD-L1 expression was analyzed using real-time RT-PCR. Data are shown as means ± SE; n = 3. CMFs were exposed to TLR agonists and PD-L1 mRNA (B) and surface levels of PD-L1 (C) and analyzed using real-time RT-PCR (4-h treatment) and flow cytometry (24-h treatment), respectively. TLR agonists were used at following concentrations: TLR1/2 agonist (Pam3CSK4), 100 ng/ml; TLR4 agonist (Escherichia coli K12 LPS), 1 μg/ml; TLR5 agonist (S. typhimurium flagellin), 100 ng/ml; TLR7 agonist (imiquimod), 100 ng/ml; TLR8 agonist (ssRNA40), 100 ng/ml; and TLR9 agonist (ODN 2006), 0.5 mM. The results are shown as means ± SE; n = 4 donors. *p < 0.05.

FIGURE 5.

MyD88-dependent TLR signaling increases PD-L1 expression by human CMFs. (A) Silencing of the myd88 gene in CMFs abrogates TLR4-mediated increases in PD-L1 mRNA expression. Primary CMFs isolated from normal colonic mucosa were transfected with MyD88 siRNA or control siRNA 7 d prior to exposure to the TLR4 agonist LPS (100 ng/ml) for 4 h. PD-L1 expression was analyzed using real-time RT-PCR. Data are shown as means ± SE; n = 3. CMFs were exposed to TLR agonists and PD-L1 mRNA (B) and surface levels of PD-L1 (C) and analyzed using real-time RT-PCR (4-h treatment) and flow cytometry (24-h treatment), respectively. TLR agonists were used at following concentrations: TLR1/2 agonist (Pam3CSK4), 100 ng/ml; TLR4 agonist (Escherichia coli K12 LPS), 1 μg/ml; TLR5 agonist (S. typhimurium flagellin), 100 ng/ml; TLR7 agonist (imiquimod), 100 ng/ml; TLR8 agonist (ssRNA40), 100 ng/ml; and TLR9 agonist (ODN 2006), 0.5 mM. The results are shown as means ± SE; n = 4 donors. *p < 0.05.

Close modal

Signaling through other innate immunity receptors has been reported to activate PD-L1 expression on the professional immune cells (51, 52). We tested whether stimulation of other MyD88-dependent TLRs modulates PD-L1 expression by CMFs. Stimulation of CMFs with the bacterial agonists for TLR1/2 (Pam3CSK4, 100 ng/ml) and TLR5 agonist (Salmonella typhimurium flagellin, 100 ng/ml) also resulted in upregulation of PD-L1 (Fig. 5B, 5C). Similar to the TLR4-mediated PD-L1 upregulation, TLR1/2- and TLR5-mediated increase in PD-L1 expression by CMFs was MyD88 dependent. Silencing of the MyD88 mRNA expression using specific siRNA abrogated this PD-L1 increase (Supplemental Fig. 2). By contrast, tested agonists to TLR7 (imiquimod 100 ng/ml), TLR8 (ssRNA40, 100 ng/ml), and TLR9 (ODN 2006, 0.5 mM) did not increase PD-L1 expression by CMFs. Importantly, bacteria for which products are recognized by TLR1/2, -4, and -5 are highly represented within the normal colon (2, 4). Because tissue expression of PD-L1 has been demonstrated to be among the key factors mediating peripheral T cell tolerance (31), our data suggest that CMFs are important contributors to mucosal tolerance via upregulation of PD-L1 in response to stimulation by colonic bacterial products.

Our in vitro experiments demonstrated that MyD88-dependent TLRs stimulation results in the PD-L1 upregulation by CMFs. This observation raised the question whether MyD88 is required for the PD-L1 expression on CMFs in vivo. Using immunostaining followed by confocal microscopy, we demonstrated that CMFs (α-SMA+ subepithelial cells) in C57BL/6 wild-type mice express a strong basal level of PD-L1 in normal colonic mucosa (Fig. 6A). The level of total mucosal PD-L1 expression was vastly reduced in MyD88−/− mice on the same genetic background. We also noted that the number of the α-SMA+ CMFs was reduced in the colonic mucosa of these KO mice. This may be due to the reduced MyD88-dependent fibroblast migration and differentiation as has been noted in the small intestine and other organs (22, 53). However, the α-SMA+ cells (also known as activated CMFs) that remain were essentially devoid of PD-L1. Further, the number of α-SMA–negative cells expressing PD-L1 was also reduced in the colonic mucosa from MyD88−/− mice (Fig. 6B).

FIGURE 6.

Expression of PD-L1 is reduced in colonic mucosa from mice with impaired MyD88 signaling. Immunostaining of frozen mucosal colonic tissue cross-sections from wild-type C57B6 (A) and MyD88−/− mice (B) was performed, followed by confocal microscopy. DAPI was used to stain cell nuclei (blue); CMFs were detected by anti–α-SMA mAb (green; clone A4) and stained for PD-L1 with mAb (red; clone M1H5). A yellow-orange color on merged images indicates colocalization of α-SMA and PD-L1 and thus the expression of PD-L1 by CMFs (indicated by arrows). Cross-sections are representative of data from six wild-type and six MyD88−/− mice. Objective used: 20×.

FIGURE 6.

Expression of PD-L1 is reduced in colonic mucosa from mice with impaired MyD88 signaling. Immunostaining of frozen mucosal colonic tissue cross-sections from wild-type C57B6 (A) and MyD88−/− mice (B) was performed, followed by confocal microscopy. DAPI was used to stain cell nuclei (blue); CMFs were detected by anti–α-SMA mAb (green; clone A4) and stained for PD-L1 with mAb (red; clone M1H5). A yellow-orange color on merged images indicates colocalization of α-SMA and PD-L1 and thus the expression of PD-L1 by CMFs (indicated by arrows). Cross-sections are representative of data from six wild-type and six MyD88−/− mice. Objective used: 20×.

Close modal

We then tested how CMF-specific activation of MyD88 regulates the expression of PD-L1 at homeostasis and upon epithelial injury. For this purpose, we generated fibroblast-specific tamoxifen-inducible conditional KO mice selective for MyD88 (Fib-MyD88 KO). The specificity of MyD88 deletion within the CMF, as well as retained MyD88 expression in other relevant innate immunity cells (CD11c+ DC, F4/80+ macrophages, and epithelial cell adhesion molecule-positive cells) was confirmed by immunostaining followed by confocal microscopy analysis (Supplemental Fig. 3). Similar to the MyD88−/− mice, we observed a decrease in a PD-L1 expression within the colonic mucosa (real-time RT-PCR analysis, Supplemental Fig. 4A). Using confocal microscopy and flow cytometry, we demonstrated that CMF-intrinsic MyD88 deficiency resulted in the abrogation of PD-L1 expression by CMFs in Fib-MyD88 KO control group (Fig 7A, 7C, Supplemental Fig. 4B). Similar observations were made in the DSS acute inflammation model, in which the exposure of colonic mucosal lamina propria cells to commensal microbiota is increased due to the epithelial damage induced by DSS (Fig. 7B, 7C, Supplemental Fig. 4B).

FIGURE 7.

Fibroblast-intrinsic MyD88 signaling is required for the PD-L1 expression by CMFs and controls IFN-γ expression in colonic mucosa. PD-L1 expression in colonic mucosa upon homeostasis (A) and in DSS-colitis (B) (day 10) in wild-type (Wt) C57B6 and Fib-MyD88fl/fl mice was analyzed by confocal microscopy. DAPI was used to stain cell nuclei (blue); activated CMFs were detected by anti–α-SMA mAb (green; clone A4) and stained for PD-L1 with mAb (red; clone M1H5). A yellow-orange color on merged images indicates colocalization of α-SMA and PD-L1 and thus the expression of PD-L1 by CMFs (indicated by arrows). Cross-sections are representative of a two independent experiments including four to five animals per group. (C) Freshly digested murine colonic MC preparations were stained with mAbs against α-SMA and PD-L1 and analyzed by flow cytometry. Isotype controls were included for each staining. Five animals per group were analyzed. (D) IFN-γ mRNA level in colonic mucosa was determined using real time RT-PCR. The level of the IFN-γ mRNA was normalized to the level of murine β-actin mRNA. The results are shown as means ± SE from two independent experiments. Four and five animals per group per experiment were analyzed. Objective used: 63×.

FIGURE 7.

Fibroblast-intrinsic MyD88 signaling is required for the PD-L1 expression by CMFs and controls IFN-γ expression in colonic mucosa. PD-L1 expression in colonic mucosa upon homeostasis (A) and in DSS-colitis (B) (day 10) in wild-type (Wt) C57B6 and Fib-MyD88fl/fl mice was analyzed by confocal microscopy. DAPI was used to stain cell nuclei (blue); activated CMFs were detected by anti–α-SMA mAb (green; clone A4) and stained for PD-L1 with mAb (red; clone M1H5). A yellow-orange color on merged images indicates colocalization of α-SMA and PD-L1 and thus the expression of PD-L1 by CMFs (indicated by arrows). Cross-sections are representative of a two independent experiments including four to five animals per group. (C) Freshly digested murine colonic MC preparations were stained with mAbs against α-SMA and PD-L1 and analyzed by flow cytometry. Isotype controls were included for each staining. Five animals per group were analyzed. (D) IFN-γ mRNA level in colonic mucosa was determined using real time RT-PCR. The level of the IFN-γ mRNA was normalized to the level of murine β-actin mRNA. The results are shown as means ± SE from two independent experiments. Four and five animals per group per experiment were analyzed. Objective used: 63×.

Close modal

Using PD-L1 KO animals, Keir et al. (14) demonstrated that animals lacking intact PD-L1 signaling have elevated production of the inflammatory cytokine IFN-γ. Our in vitro experiments suggest that stimulation of MyD88-dependent TLRs potentiate CMF-mediated suppression of the IFN-γ production by preactivated T cells in a PD-L1–dependent manner. Thus, we analyzed IFN-γ levels in the mice lacking an intact MyD88–PD-L1 immunoregulatory axis. We observed that in the absence of MyD88–PD-L1 signaling (in Fib-MyD88 KO mice), there was a significant increase in mucosal IFN-γ expression at homeostasis (Fig. 7D). Similar observations were made in colonic mucosa from MyD88−/−: a decrease in mucosal IFN-γ was observed (Supplemental Fig. 4C). Further, production of an IFN-γ inflammatory response was even stronger in Fib-MyD88 KO mice upon intestinal injury induced by DSS (Fig. 7D).

Tolerance is defined as the immune system’s endogenous regulatory mechanisms that prevent potentially injurious inflammatory immune responses by producing a systemic state of Ag-specific nonresponsiveness to Ags previously encountered (54, 55). A similar but more localized process regulates responses to commensal bacteria in the intestinal mucosa and is known as mucosal tolerance (5658). However, physiological pathways, and cells responsible for the induction and maintenance of tolerance in the colonic mucosa are not well characterized. Taken together with our previous publications (17, 35), the current study provides additional evidence supporting the importance of human colonic CMFs in the maintenance of local mucosal tolerance.

Many mechanisms have been proposed to explain the defects in the activation of T cell–dependent immune responses that occur during tolerance, including clonal anergy and deletion as well active regulatory processes (3, 56). Immunosuppression due to the upregulation of the B7 negative costimulatory molecules, as well as induction of regulatory T cells, are important mechanisms contributing to the induction of the tolerogenic (homeostatic) state (30, 32). During mucosal tolerance, acute inflammatory effector T cell responses are regulated by their interactions with APCs. We and others recently reported that CMFs may act as nonprofessional APCs in normal colonic mucosa (18, 19) and so may be involved in the regulation of T cell responses. It has been suggested that, in contrast to parenchymal cells, a “tolerogenic” antiproliferative effect is a fundamental characteristic of all mesenchymal stromal cells (59). Our recently published data are consistent with this idea. CMFs isolated from normal human mucosa suppress proliferation of activated CD4+ T cells (17). The mechanisms responsible for CMF’s suppressive effect on activated T effector cell responses involve cell contact–mediated interactions and/or the production of soluble factors. We previously demonstrated that direct CMF-mediated suppression of activated T effector cell responses involves signaling through PD-L1 and PD-L2 (17). However, in colonic mucosal sites in vivo, these interactions occur within an extremely dynamic milieu containing abundant cytokines, chemokines, growth factors, and TLR ligands.

The continuous presence of physiological microflora in the colonic lumen provides a significant source for TLR ligands; thus, it is likely that TLR signaling plays a major role in orchestrating mucosal tolerogenic responses. Therefore, the immunosuppressive capacity of CMFs is likely to be modulated by signaling through TLRs. We have confirmed expression of TLR4 on normal CMFs and demonstrated that signaling through TLR4 upregulates PD-L1, but not PD-L2, on CMFs. TLR4-mediated upregulation of PD-L1 results in greater CMF-mediated suppression of CD4+ effector T cell proliferation. Although further investigation is required to understand the mechanism responsible for CMF-mediated suppression of proliferating T cells, our data suggest that this inhibitory effect may be caused by an increase in PD-L1–mediated T cell death. Such an APC-mediated PD-L1–dependent inhibitory action on T cell growth/survival was previously described in the other organs (41, 42)

CMF/T cell coculture experiments also demonstrated that signaling through TLR4 enhanced CMF-mediated suppression of IFN-γ production by activated CD4+ effector T cells, and this process was PD-L1 dependent. CMFs isolated from normal colonic tissue respond to exogenous IFN-γ by increased PD-L1 expression (60). This suggests that reduced production of IFN-γ by CMF-derived PD-L1 engagement of T cells may be an important physiological negative-feedback mechanism to control the inflammation in the GI track and mucosal homeostasis.

The intracellular mechanisms responsible for TLR-mediated upregulation of PD-L1 on APCs are not well characterized. IFN-γ– and LPS-induced PD-L1 upregulation on tumor cells was reported to be mediated by MEK and PI3K/AKT/mTOR, but not NF-κB pathways (30, 51). However, IFN-γ regulation of PD-L1 expression on dermal fibroblasts was demonstrated to be NF-κB mediated (61). NF-κB pathways are activated in the intestinal mucosal stromal cells upon stimulation with LPS (37), and we have made the novel observation that TLR4-mediated upregulation of PD-L1 on CMFs is NF-κB dependent, involving both the canonical and noncanonical NF-κB pathways. Further, our data suggest that activation of JAK2 is involved in the NF-κB–mediated upregulation of PD-L1 on CMFs in response to LPS. A similar mechanism was described for the LPS-induced production of NO in macrophages (44). The question that needs further investigation is whether activation of JAK2 is directly involved in the TLR4-mediated increase of PD-L1 or it is due to the LPS-mediated increase in the inflammatory cytokines and growth factors on CMFs. Recent data by Wölfle et al. (62) suggest that IL-6 and IL-10 via STAT3 activation may contribute to the increase in PD-L1 expression on professional APCs. LPS treatment may induce minor increases in IL-6, but not IL-10, production by CMFs (63). However, we did not observe significant upregulation of PD-L1 expression by CMFs after 24 h of treatment with IL-6 (10 ng/ml; data not shown). This suggests that JAK2 activation might be directly involved in the TLR4-mediated regulation of PD-L1 expression by CMFs, as reported previously in B cell lymphomas (64). However, we do not exclude the possibility that exogenous IL-6 and IL-10 produced by other cells in the normal colonic mucosa may contribute to the regulation of the PD-L1 expression in vivo. In contrast to previous findings with tumor-derived cells (30, 51), inhibition of mTOR with rapamycin did not affect TLR4-mediated upregulation of PD-L1 on CMFs. In summary, TLR-mediated upregulation of PD-L1 on different cell type may involve different intracellular pathways in health and disease and thus may have different pathophysiological effects.

TLR4-mediated signaling may activate two downstream pathways, MyD88-dependent and -independent, each able to directly activate NF-κB. The adaptor MyD88 is also required for the TLR1, -2, and -5–9–mediated signaling. This adaptor has been reported to be essential for the TLR4- and TLR7/8-medated upregulation of PD-L1 on plasma cells and DCs, respectively (51, 52). We observed that signaling through MyD88-dependent TLRs (-1/2, -4, and -5) upregulates PD-L1. In contrast to professional APCs, stimulation of CMFs via TLR6–9 agonists did not result in PD-L1 upregulation. This may be due either to limited availability of these receptors on CMFs or different regulatory functions of these TLRs on CMFs as contrasted to professional APCs, as has previously been reported for gingival fibroblasts (65).

In the GI tract, MyD88 signaling is critical for host protection against DSS-induced colitis by limiting mucosal damage and acute inflammatory responses (58, 66). MyD88-dependent signaling has also been shown to limit the systemic enteropathogen burden in the model of Salmonella-induced colitis (22). At the same time, MyD88 signaling is critical in spontaneous and inflammation-induced intestinal tumorogenesis (67, 68). Because of the intrinsic complexity of MyD88 signaling in various type of cells (58, 68), the mechanisms underlying MyD88-protective versus pathological effects are unclear. PD-L1 signaling is known to play a dual role: it is critical for the maintenance of mucosal tolerance (6, 7, 10, 14), but its persistent, increased expression within a tumor promotes cancer-associated immunosuppression (10, 6971). The normal colonic mucosa is highly colonized by large variety of bacterial species that are a source of multiple ligands for MyD88-dependent TLRs (4, 72). However, the role of MyD88 signaling in the in vivo regulation of the PD-L1 in GI tract has not been investigated. Our in vivo data suggest that total lack of intact MyD88 signaling in colonic mucosa (Myd88−/− mice) drastically decreases PD-L1 expression especially by CMFs. We have demonstrated that fibroblast-restricted MyD88 signaling was required for the PD-L1 expression by CMFs in vivo. Using the DSS acute colitis-conditional MyD88 KO model, we demonstrated that despite intact MyD88 signaling on other immune cells that accumulate in the mucosa of this model, there does not appear to be any compensatory mechanism present to restore the expression of PD-L1 on CMFs during acute intestinal epithelial injury.

Our in vitro data in human cells and our in vivo observations in mice suggest that direct interactions between CMFs, stromal cells located just beneath the epithelial basement membrane, and colonic microbial products are critical for PD-L1 expression by CMFs. Further, our in vitro culture experiments suggest that activation of MyD88-dependent TLR4 signaling potentiates PD-L1–dependent CMF-mediated suppression of IFN-γ production by T cells. We observed also that impairment of MyD88/PD-L1 regulation in CMFs in vivo concomitantly increased the expression of IFN-γ in the colonic mucosa. Thus, taking into consideration the key role of PD-L1 in the negative regulation of the IFN-γ production in vivo (14), the shift in PD-L1 expression on CMFs even by the 10–20% observed by us in culture might be relevant to the TLR-mediated tuning of the immune balance in the colonic mucosa. This might be particularly relevant when the epithelial layer is disrupted and PD-L1 upregulation on CMFs might help to maintain the equilibrium between tolerance and immunity to protect the colonic mucosa against overt inflammatory responses toward innocuous microflora.

Data from these studies point to mechanisms that deserve further detailed investigation such as the contribution of MyD88 signaling to the basal levels of PD-L1 expression on CMFs in vivo and whether it affords a protective function in human IBD. Further in vivo investigation of the role of PD-L1 expression by CMFs in health and diseases is also needed. Finally, our data strongly suggest that modulation of PD-L1 expression by CMFs via MyD88-dependent TLR stimulation might be among the main processes contributing to the control of colonic inflammation during mucosal tolerance, and its disruption may be among the factors involved in several acute and chronic GI inflammatory diseases.

We thank Dr. Andrew Leask at University of Western Ontario (London, Ontario, Canada) for providing us with the TgCol1α2-CreER mice. We also thank Dr. David Konkel (University of Texas Medical Branch) for critical revision of the manuscript and our technical assistant Yu Lin for her excellent assistance in performing the animal work.

This work was supported by grants from the National Institute of Allergy and Infectious Diseases (1R56AI085474), the National Center for Advancing Translational Sciences (KL2TR000072 and TR000071), the American Gastroenterology Association, the American Cancer Society (RSG-10-159-01-LIB), the Clinical and Translational Science Award Multi-Institutional Pilot Fund, and the Crohn’s & Colitis Foundation of America.

The content is solely the responsibility of the authors and does not necessarily represent the views of the supporting agencies listed above.

The online version of this article contains supplemental material.

Abbreviations used in this article:

AF

Alexa Fluor

CMF

CD90+ myofibroblast/fibroblast

DC

dendritic cell

DSS

dextran sodium sulfate

GI

gastrointestinal

IBD

inflammatory bowel disease

KO

knockout

MC

mucosal cell

mTOR

mammalian target of rapamycin

MyD88−/−

MyD88-deficient

PD-1

programmed cell death-1

PD-L1

programmed death ligand-1

siRNA

small interfering RNA

α-SMA

α-smooth muscle actin.

1
Untersmayr
E.
,
Jensen-Jarolim
E.
.
2006
.
Mechanisms of type I food allergy.
Pharmacol. Ther.
112
:
787
798
.
2
Tannock
G. W.
1999
.
Probiotics: A Critical Review.
Wymondham
:
Horizon Scientific Ed
.
3
Novak
N.
,
Haberstok
J.
,
Bieber
T.
,
Allam
J. P.
.
2008
.
The immune privilege of the oral mucosa.
Trends Mol. Med.
14
:
191
198
.
4
Ricanek
P.
,
Lothe
S. M.
,
Frye
S. A.
,
Rydning
A.
,
Vatn
M. H.
,
Tønjum
T.
.
2012
.
Gut bacterial profile in patients newly diagnosed with treatment-naïve Crohn’s disease.
Clin. Exp. Gastroenterol.
5
:
173
186
.
5
Zhang
Z.
,
Geng
J.
,
Tang
X.
,
Fan
H.
,
Xu
J.
,
Wen
X.
,
Ma
Z. S.
,
Shi
P.
.
2014
.
Spatial heterogeneity and co-occurrence patterns of human mucosal-associated intestinal microbiota.
ISME J.
8
:
881
893
.
6
Reynoso
E. D.
,
Elpek
K. G.
,
Francisco
L.
,
Bronson
R.
,
Bellemare-Pelletier
A.
,
Sharpe
A. H.
,
Freeman
G. J.
,
Turley
S. J.
.
2009
.
Intestinal tolerance is converted to autoimmune enteritis upon PD-1 ligand blockade.
J. Immunol.
182
:
2102
2112
.
7
Francisco
L. M.
,
Sage
P. T.
,
Sharpe
A. H.
.
2010
.
The PD-1 pathway in tolerance and autoimmunity.
Immunol. Rev.
236
:
219
242
.
8
Jiang
X.
,
Zhou
J.
,
Giobbie-Hurder
A.
,
Wargo
J.
,
Hodi
F. S.
.
2013
.
The activation of MAPK in melanoma cells resistant to BRAF inhibition promotes PD-L1 expression that is reversible by MEK and PI3K inhibition.
Clin. Cancer Res.
19
:
598
609
.
9
Rozali
E. N.
,
Hato
S. V.
,
Robinson
B. W.
,
Lake
R. A.
,
Lesterhuis
W. J.
.
2012
.
Programmed death ligand 2 in cancer-induced immune suppression.
Clin. Dev. Immunol.
2012
:
656340
.
10
Podojil
J. R.
,
Miller
S. D.
.
2013
.
Targeting the B7 family of co-stimulatory molecules: successes and challenges.
BioDrugs
27
:
1
13
.
11
Loke
P.
,
Allison
J. P.
.
2003
.
PD-L1 and PD-L2 are differentially regulated by Th1 and Th2 cells.
Proc. Natl. Acad. Sci. USA
100
:
5336
5341
.
12
D’Addio
F.
,
Riella
L. V.
,
Mfarrej
B. G.
,
Chabtini
L.
,
Adams
L. T.
,
Yeung
M.
,
Yagita
H.
,
Azuma
M.
,
Sayegh
M. H.
,
Guleria
I.
.
2011
.
The link between the PDL1 costimulatory pathway and Th17 in fetomaternal tolerance.
J. Immunol.
187
:
4530
4541
.
13
Amarnath
S.
,
Mangus
C.W.
,
Wang
J.C.
,
Wei
F.
,
He
A.
,
Kapoor
V.
,
Foley
J.E.
,
Massey
P.R.
,
Felizardo
T.C.
,
Riley
J.L.
, et al
.
2011
.
The PDL1-PD1 axis converts human TH1 cells into regulatory T cells.
Sci. Transl. Med.
3
:
111ra20
.
14
Keir
M. E.
,
Liang
S. C.
,
Guleria
I.
,
Latchman
Y. E.
,
Qipo
A.
,
Albacker
L. A.
,
Koulmanda
M.
,
Freeman
G. J.
,
Sayegh
M. H.
,
Sharpe
A. H.
.
2006
.
Tissue expression of PD-L1 mediates peripheral T cell tolerance.
J. Exp. Med.
203
:
883
895
.
15
Nakazawa
A.
,
Dotan
I.
,
Brimnes
J.
,
Allez
M.
,
Shao
L.
,
Tsushima
F.
,
Azuma
M.
,
Mayer
L.
.
2004
.
The expression and function of costimulatory molecules B7H and B7-H1 on colonic epithelial cells.
Gastroenterology
126
:
1347
1357
.
16
Kanai
T.
,
Totsuka
T.
,
Uraushihara
K.
,
Makita
S.
,
Nakamura
T.
,
Koganei
K.
,
Fukushima
T.
,
Akiba
H.
,
Yagita
H.
,
Okumura
K.
, et al
.
2003
.
Blockade of B7-H1 suppresses the development of chronic intestinal inflammation.
J. Immunol.
171
:
4156
4163
.
17
Pinchuk
I. V.
,
Saada
J. I.
,
Beswick
E. J.
,
Boya
G.
,
Qiu
S. M.
,
Mifflin
R. C.
,
Raju
G. S.
,
Reyes
V. E.
,
Powell
D. W.
.
2008
.
PD-1 ligand expression by human colonic myofibroblasts/fibroblasts regulates CD4+ T-cell activity.
Gastroenterology
135
:
1228
1237
.
18
Saada
J. I.
,
Pinchuk
I. V.
,
Barrera
C. A.
,
Adegboyega
P. A.
,
Suarez
G.
,
Mifflin
R. C.
,
Di Mari
J. F.
,
Reyes
V. E.
,
Powell
D. W.
.
2006
.
Subepithelial myofibroblasts are novel nonprofessional APCs in the human colonic mucosa.
J. Immunol.
177
:
5968
5979
.
19
Owens
B. M.
,
Steevels
T. A.
,
Dudek
M.
,
Walcott
D.
,
Sun
M. Y.
,
Mayer
A.
,
Allan
P.
,
Simmons
A.
.
2013
.
CD90(+) Stromal Cells are Non-Professional Innate Immune Effectors of the Human Colonic Mucosa.
Front. Immunol.
4
:
307
.
20
Park
J. J.
,
Omiya
R.
,
Matsumura
Y.
,
Sakoda
Y.
,
Kuramasu
A.
,
Augustine
M. M.
,
Yao
S.
,
Tsushima
F.
,
Narazaki
H.
,
Anand
S.
, et al
.
2010
.
B7-H1/CD80 interaction is required for the induction and maintenance of peripheral T-cell tolerance.
Blood
116
:
1291
1298
.
21
Powell
D. W.
,
Pinchuk
I. V.
,
Saada
J. I.
,
Chen
X.
,
Mifflin
R. C.
.
2011
.
Mesenchymal cells of the intestinal lamina propria.
Annu. Rev. Physiol.
73
:
213
237
.
22
Månsson
L. E.
,
Montero
M.
,
Zarepour
M.
,
Bergstrom
K. S.
,
Ma
C.
,
Huang
T.
,
Man
C.
,
Grassl
G. A.
,
Vallance
B. A.
.
2012
.
MyD88 signaling promotes both mucosal homeostatic and fibrotic responses during Salmonella-induced colitis.
Am. J. Physiol. Gastrointest. Liver Physiol.
303
:
G311
G323
.
23
Fukata
M.
,
Arditi
M.
.
2013
.
The role of pattern recognition receptors in intestinal inflammation.
Mucosal Immunol.
6
:
451
463
.
24
Burke
J. P.
,
Cunningham
M. F.
,
Watson
R. W.
,
Docherty
N. G.
,
Coffey
J. C.
,
O’Connell
P. R.
.
2010
.
Bacterial lipopolysaccharide promotes profibrotic activation of intestinal fibroblasts.
Br. J. Surg.
97
:
1126
1134
.
25
Kamdar
K.
,
Nguyen
V.
,
DePaolo
R. W.
.
2013
.
Toll-like receptor signaling and regulation of intestinal immunity.
Virulence
4
:
207
212
.
26
Maynard
C. L.
,
Elson
C. O.
,
Hatton
R. D.
,
Weaver
C. T.
.
2012
.
Reciprocal interactions of the intestinal microbiota and immune system.
Nature
489
:
231
241
.
27
Otte
J. M.
,
Rosenberg
I. M.
,
Podolsky
D. K.
.
2003
.
Intestinal myofibroblasts in innate immune responses of the intestine.
Gastroenterology
124
:
1866
1878
.
28
Allam
J. P.
,
Peng
W. M.
,
Appel
T.
,
Wenghoefer
M.
,
Niederhagen
B.
,
Bieber
T.
,
Bergé
S.
,
Novak
N.
.
2008
.
Toll-like receptor 4 ligation enforces tolerogenic properties of oral mucosal Langerhans cells.
J. Allergy Clin. Immunol.
121
:
368
374, e1
.
29
Bao
H.
,
Lu
P.
,
Li
Y.
,
Wang
L.
,
Li
H.
,
He
D.
,
Yang
Y.
,
Zhao
Y.
,
Yang
L.
,
Wang
M.
, et al
.
2011
.
Triggering of toll-like receptor-4 in human multiple myeloma cells promotes proliferation and alters cell responses to immune and chemotherapy drug attack.
Cancer Biol. Ther.
11
:
58
67
.
30
Berthon
C.
,
Driss
V.
,
Liu
J.
,
Kuranda
K.
,
Leleu
X.
,
Jouy
N.
,
Hetuin
D.
,
Quesnel
B.
.
2010
.
In acute myeloid leukemia, B7-H1 (PD-L1) protection of blasts from cytotoxic T cells is induced by TLR ligands and interferon-gamma and can be reversed using MEK inhibitors.
Cancer Immunol. Immunother.
59
:
1839
1849
.
31
Owens
B. M.
,
Simmons
A.
.
2013
.
Intestinal stromal cells in mucosal immunity and homeostasis.
Mucosal Immunol.
6
:
224
234
.
32
Mahida
Y. R.
,
Beltinger
J.
,
Makh
S.
,
Göke
M.
,
Gray
T.
,
Podolsky
D. K.
,
Hawkey
C. J.
.
1997
.
Adult human colonic subepithelial myofibroblasts express extracellular matrix proteins and cyclooxygenase-1 and -2.
Am. J. Physiol.
273
:
G1341
G1348
.
33
Zheng
B.
,
Zhang
Z.
,
Black
C. M.
,
de Crombrugghe
B.
,
Denton
C. P.
.
2002
.
Ligand-dependent genetic recombination in fibroblasts : a potentially powerful technique for investigating gene function in fibrosis.
Am. J. Pathol.
160
:
1609
1617
.
34
Wirtz
S.
,
Neufert
C.
,
Weigmann
B.
,
Neurath
M. F.
.
2007
.
Chemically induced mouse models of intestinal inflammation.
Nat. Protoc.
2
:
541
546
.
35
Pinchuk
I. V.
,
Beswick
E. J.
,
Saada
J. I.
,
Boya
G.
,
Schmitt
D.
,
Raju
G. S.
,
Brenmoehl
J.
,
Rogler
G.
,
Reyes
V. E.
,
Powell
D. W.
.
2011
.
Human colonic myofibroblasts promote expansion of CD4+ CD25high Foxp3+ regulatory T cells.
Gastroenterology
140
:
2019
2030
.
36
Abreu
M. T.
,
Fukata
M.
,
Arditi
M.
.
2005
.
TLR signaling in the gut in health and disease.
J. Immunol.
174
:
4453
4460
.
37
Walton
K. L.
,
Holt
L.
,
Sartor
R. B.
.
2009
.
Lipopolysaccharide activates innate immune responses in murine intestinal myofibroblasts through multiple signaling pathways.
Am. J. Physiol. Gastrointest. Liver Physiol.
296
:
G601
G611
.
38
Seki
E.
,
Schnabl
B.
.
2012
.
Role of innate immunity and the microbiota in liver fibrosis: crosstalk between the liver and gut.
J. Physiol.
590
:
447
458
.
39
Chen
L.
,
Flies
D. B.
.
2013
.
Molecular mechanisms of T cell co-stimulation and co-inhibition.
Nat. Rev. Immunol.
13
:
227
242
.
40
Latchman
Y. E.
,
Liang
S. C.
,
Wu
Y.
,
Chernova
T.
,
Sobel
R. A.
,
Klemm
M.
,
Kuchroo
V. K.
,
Freeman
G. J.
,
Sharpe
A. H.
.
2004
.
PD-L1-deficient mice show that PD-L1 on T cells, antigen-presenting cells, and host tissues negatively regulates T cells.
Proc. Natl. Acad. Sci. USA
101
:
10691
10696
.
41
Dong
H.
,
Strome
S. E.
,
Salomao
D. R.
,
Tamura
H.
,
Hirano
F.
,
Flies
D. B.
,
Roche
P. C.
,
Lu
J.
,
Zhu
G.
,
Tamada
K.
, et al
.
2002
.
Tumor-associated B7-H1 promotes T-cell apoptosis: a potential mechanism of immune evasion.
Nat. Med.
8
:
793
800
.
42
Venkatachari
N. J.
,
Buchanan
W. G.
,
Ayyavoo
V.
.
2008
.
Human immunodeficiency virus (HIV-1) infection selectively downregulates PD-1 expression in infected cells and protects the cells from early apoptosis in vitro and in vivo.
Virology
376
:
140
153
.
43
Wei
S.
,
Shreiner
A. B.
,
Takeshita
N.
,
Chen
L.
,
Zou
W.
,
Chang
A. E.
.
2008
.
Tumor-induced immune suppression of in vivo effector T-cell priming is mediated by the B7-H1/PD-1 axis and transforming growth factor beta.
Cancer Res.
68
:
5432
5438
.
44
Jones
E.
,
Adcock
I. M.
,
Ahmed
B. Y.
,
Punchard
N. A.
.
2007
.
Modulation of LPS stimulated NF-kappaB mediated Nitric Oxide production by PKCepsilon and JAK2 in RAW macrophages.
J. Inflamm. (Lond.)
4
:
23
.
45
Burdelya
L.
,
Catlett-Falcone
R.
,
Levitzki
A.
,
Cheng
F.
,
Mora
L. B.
,
Sotomayor
E.
,
Coppola
D.
,
Sun
J.
,
Sebti
S.
,
Dalton
W. S.
, et al
.
2002
.
Combination therapy with AG-490 and interleukin 12 achieves greater antitumor effects than either agent alone.
Mol. Cancer Ther.
1
:
893
899
.
46
Meydan
N.
,
Grunberger
T.
,
Dadi
H.
,
Shahar
M.
,
Arpaia
E.
,
Lapidot
Z.
,
Leeder
J. S.
,
Freedman
M.
,
Cohen
A.
,
Gazit
A.
, et al
.
1996
.
Inhibition of acute lymphoblastic leukaemia by a Jak-2 inhibitor.
Nature
379
:
645
648
.
47
Abe
M.
,
Funakoshi-Tago
M.
,
Tago
K.
,
Kamishimoto
J.
,
Aizu-Yokota
E.
,
Sonoda
Y.
,
Kasahara
T.
.
2009
.
The polycythemia vera-associated Jak2 V617F mutant induces tumorigenesis in nude mice.
Int. Immunopharmacol.
9
:
870
877
.
48
Marienfeld
R.
,
May
M. J.
,
Berberich
I.
,
Serfling
E.
,
Ghosh
S.
,
Neumann
M.
.
2003
.
RelB forms transcriptionally inactive complexes with RelA/p65.
J. Biol. Chem.
278
:
19852
19860
.
49
Jacque
E.
,
Tchenio
T.
,
Piton
G.
,
Romeo
P. H.
,
Baud
V.
.
2005
.
RelA repression of RelB activity induces selective gene activation downstream of TNF receptors.
Proc. Natl. Acad. Sci. USA
102
:
14635
14640
.
50
Mukherjee
S.
,
Chen
L. Y.
,
Papadimos
T. J.
,
Huang
S.
,
Zuraw
B. L.
,
Pan
Z. K.
.
2009
.
Lipopolysaccharide-driven Th2 cytokine production in macrophages is regulated by both MyD88 and TRAM.
J. Biol. Chem.
284
:
29391
29398
.
51
Liu
J.
,
Hamrouni
A.
,
Wolowiec
D.
,
Coiteux
V.
,
Kuliczkowski
K.
,
Hetuin
D.
,
Saudemont
A.
,
Quesnel
B.
.
2007
.
Plasma cells from multiple myeloma patients express B7-H1 (PD-L1) and increase expression after stimulation with IFN-gamma and TLR ligands via a MyD88-, TRAF6-, and MEK-dependent pathway.
Blood
110
:
296
304
.
52
Meier
A.
,
Bagchi
A.
,
Sidhu
H. K.
,
Alter
G.
,
Suscovich
T. J.
,
Kavanagh
D. G.
,
Streeck
H.
,
Brockman
M. A.
,
LeGall
S.
,
Hellman
J.
,
Altfeld
M.
.
2008
.
Upregulation of PD-L1 on monocytes and dendritic cells by HIV-1 derived TLR ligands.
AIDS
22
:
655
658
.
53
Li
G.
,
Liu
D.
,
Zhang
Y.
,
Qian
Y.
,
Zhang
H.
,
Guo
S.
,
Sunagawa
M.
,
Hisamitsu
T.
,
Liu
Y.
.
2013
.
Celastrol inhibits lipopolysaccharide-stimulated rheumatoid fibroblast-like synoviocyte invasion through suppression of TLR4/NF-κB-mediated matrix metalloproteinase-9 expression.
PLoS ONE
8
:
e68905
.
54
Rothstein
D. M.
2006
.
Immunosuppression and regulation: cast in new light?
J. Am. Soc. Nephrol.
17
:
2644
2646
.
55
Halloran
P. F.
,
Bromberg
J.
,
Kaplan
B.
,
Vincenti
F.
.
2008
.
Tolerance versus immunosuppression: a perspective.
Am. J. Transplant.
8
:
1365
1366
.
56
Pabst
O.
,
Mowat
A. M.
.
2012
.
Oral tolerance to food protein.
Mucosal Immunol.
5
:
232
239
.
57
Kuhn
K. A.
,
Stappenbeck
T. S.
.
2013
.
Peripheral education of the immune system by the colonic microbiota.
Semin. Immunol.
25
:
364
369
.
58
Kirkland
D.
,
Benson
A.
,
Mirpuri
J.
,
Pifer
R.
,
Hou
B.
,
DeFranco
A. L.
,
Yarovinsky
F.
.
2012
.
B cell-intrinsic MyD88 signaling prevents the lethal dissemination of commensal bacteria during colonic damage.
Immunity
36
:
228
238
.
59
Jones
S.
,
Horwood
N.
,
Cope
A.
,
Dazzi
F.
.
2007
.
The antiproliferative effect of mesenchymal stem cells is a fundamental property shared by all stromal cells.
J. Immunol.
179
:
2824
2831
.
60
Saada
J. I.
,
Barrera
C. A.
,
Reyes
V. E.
,
Adegboyega
P. A.
,
Suarez
G.
,
Tamerisa
R. A.
,
Pang
K. F.
,
Bland
D. A.
,
Mifflin
R. C.
,
DI Mari
J. F.
,
Powell
D. W.
.
2004
.
Intestinal myofibroblasts and immune tolerance.
Ann. N. Y. Acad. Sci.
1029
:
379
381
.
61
Lee
S. K.
,
Seo
S. H.
,
Kim
B. S.
,
Kim
C. D.
,
Lee
J. H.
,
Kang
J. S.
,
Maeng
P. J.
,
Lim
J. S.
.
2005
.
IFN-gamma regulates the expression of B7-H1 in dermal fibroblast cells.
J. Dermatol. Sci.
40
:
95
103
.
62
Wölfle
S. J.
,
Strebovsky
J.
,
Bartz
H.
,
Sähr
A.
,
Arnold
C.
,
Kaiser
C.
,
Dalpke
A. H.
,
Heeg
K.
.
2011
.
PD-L1 expression on tolerogenic APCs is controlled by STAT-3.
Eur. J. Immunol.
41
:
413
424
.
63
Rogler
G.
,
Gelbmann
C. M.
,
Vogl
D.
,
Brunner
M.
,
Schölmerich
J.
,
Falk
W.
,
Andus
T.
,
Brand
K.
.
2001
.
Differential activation of cytokine secretion in primary human colonic fibroblast/myofibroblast cultures.
Scand. J. Gastroenterol.
36
:
389
398
.
64
Green
M. R.
,
Monti
S.
,
Rodig
S. J.
,
Juszczynski
P.
,
Currie
T.
,
O’Donnell
E.
,
Chapuy
B.
,
Takeyama
K.
,
Neuberg
D.
,
Golub
T. R.
, et al
.
2010
.
Integrative analysis reveals selective 9p24.1 amplification, increased PD-1 ligand expression, and further induction via JAK2 in nodular sclerosing Hodgkin lymphoma and primary mediastinal large B-cell lymphoma.
Blood
116
:
3268
3277
.
65
Mahanonda
R.
,
Sa-Ard-Iam
N.
,
Montreekachon
P.
,
Pimkhaokham
A.
,
Yongvanichit
K.
,
Fukuda
M. M.
,
Pichyangkul
S.
.
2007
.
IL-8 and IDO expression by human gingival fibroblasts via TLRs.
J. Immunol.
178
:
1151
1157
.
66
Rakoff-Nahoum
S.
,
Paglino
J.
,
Eslami-Varzaneh
F.
,
Edberg
S.
,
Medzhitov
R.
.
2004
.
Recognition of commensal microflora by toll-like receptors is required for intestinal homeostasis.
Cell
118
:
229
241
.
67
Rakoff-Nahoum
S.
,
Medzhitov
R.
.
2007
.
Regulation of spontaneous intestinal tumorigenesis through the adaptor protein MyD88.
Science
317
:
124
127
.
68
Schiechl
G.
,
Bauer
B.
,
Fuss
I.
,
Lang
S. A.
,
Moser
C.
,
Ruemmele
P.
,
Rose-John
S.
,
Neurath
M. F.
,
Geissler
E. K.
,
Schlitt
H. J.
, et al
.
2011
.
Tumor development in murine ulcerative colitis depends on MyD88 signaling of colonic F4/80+CD11b(high)Gr1(low) macrophages.
J. Clin. Invest.
121
:
1692
1708
.
69
Brahmer
J. R.
,
Tykodi
S. S.
,
Chow
L. Q.
,
Hwu
W. J.
,
Topalian
S. L.
,
Hwu
P.
,
Drake
C. G.
,
Camacho
L. H.
,
Kauh
J.
,
Odunsi
K.
, et al
.
2012
.
Safety and activity of anti-PD-L1 antibody in patients with advanced cancer.
N. Engl. J. Med.
366
:
2455
2465
.
70
Brahmer
J. R.
2012
.
PD-1-targeted immunotherapy: recent clinical findings.
Clin. Adv. Hematol. Oncol.
10
:
674
675
.
71
Brahmer
J. R.
,
Drake
C. G.
,
Wollner
I.
,
Powderly
J. D.
,
Picus
J.
,
Sharfman
W. H.
,
Stankevich
E.
,
Pons
A.
,
Salay
T. M.
,
McMiller
T. L.
, et al
.
2010
.
Phase I study of single-agent anti-programmed death-1 (MDX-1106) in refractory solid tumors: safety, clinical activity, pharmacodynamics, and immunologic correlates.
J. Clin. Oncol.
28
:
3167
3175
.
72
Lozupone
C. A.
,
Stombaugh
J. I.
,
Gordon
J. I.
,
Jansson
J. K.
,
Knight
R.
.
2012
.
Diversity, stability and resilience of the human gut microbiota.
Nature
489
:
220
230
.

The authors have no financial conflicts of interest.

Supplementary data