Abstract
Intestinal dendritic cells (DCs) play key roles in mediating tolerance to commensal flora and inflammatory responses against mucosal pathogens. The mechanisms by which intestinal “conditioning” influences human DC responses to microbial stimuli remain poorly understood. Infections with viruses, such as HIV-1, that target mucosal tissue result in intestinal epithelial barrier breakdown and increased translocation of commensal bacteria into the lamina propria (LP). It is unclear whether innate LP DC responses to concurrent viral and bacterial stimuli influence mucosal HIV-1 pathogenesis. In this study, direct ex vivo phenotype and in vitro constitutive cytokine production of CD1c+ DCs in human intestinal LP were compared with those in peripheral blood (PB). To evaluate innate responses to viral and bacterial stimuli, intracellular cytokine production by LP and PB DCs following stimulation with ligands for TLRs 2, 4, 5, and 7/8 was evaluated. At steady state, LP CD1c+ DCs expressed higher levels of activation markers (CD40, CD83, CD86, HLA-DR, and CCR7) than did PB CD1c+ DCs, and higher frequencies of LP CD1c+ DCs constitutively produced IL-6 and -10 and TNF-α. LP DCs had blunted cytokine responses to TLR4 ligand and TLR5 ligand stimulation relative to PB DCs, yet similarly produced IL-10 in response to TLR2 ligand. Only synthetic TLR7/8 ligand, a mimic of viral ssRNA, induced IL-23 production by LP CD1c+ DCs, and this proinflammatory cytokine response was synergistically enhanced following combined TLR7/8 and TLR4 stimulation. These findings highlight a potential mechanism by which viruses like HIV-1 may subvert homeostatic mechanisms and induce inflammation in the intestinal mucosa.
The intestinal immune system maintains tolerance to commensal microbes and food Ags, yet it must mount protective responses against invading bacterial and viral pathogens. Dendritic cells (DCs) are APCs that bridge innate and adaptive immunity and serve as sentinels of the intestinal immune system (1, 2). An intricate maturation and activation process allows them to sense the presence of invading pathogens, migrate to local lymph nodes, and stimulate the activation and expansion of naive and Ag-specific T cells (3, 4). In the intestinal mucosa, DCs in the lamina propria (LP) were shown to play a crucial role in the induction of tolerogenic responses to commensal bacteria (1). We recently showed that commensal bacteria-reactive, effector Th1 and Th17 CD4+ T cells exist in normal human LP. Furthermore, we identified a subset of resident LP DCs expressing CD1c that mediated the expansion of bacteria-reactive effector T cell responses in vitro (5). These findings suggest a complex role for LP DCs in mediating homeostatic responses to commensal bacteria, as well as inflammatory responses to intestinal pathogens.
DCs typically detect the presence of pathogens through pattern-recognition receptors, especially those of the TLR family that recognize conserved microbial components (6). TLR ligand stimulation of DCs promotes their maturation and the induction of proinflammatory or regulatory responses, depending on the cytokine profiles induced (7). The manner in which intestinal DCs respond to recognition of TLR ligands expressed by pathogenic and commensal microbes within the intestinal environment is of crucial importance to understanding the immunore-gulatory nature of these cells.
In the steady state, resident intestinal DCs are likely condi-tioned by intestinal epithelial cell-derived factors to become tolerant to commensal microbes or to induce regulatory or Th2-type T cell responses (1, 8–10). It was postulated that DCs capable of responding to microbial stimuli in a proinflammatory manner in vivo consist of newly recruited DCs that have yet to be exposed to various conditioning factors or are resident DCs that are refractory to suppressive conditioning (1). Debate exists as to whether the conditioning effects on resident LP DCs are subset dependent, are permanent or require constant exposure to inhibitory factors, or whether suppressive effects of conditioning apply only to commensal bacteria but not pathogenic microbes. Even less is known about how intestinal DCs respond to viral stimuli and whether such interactions alter intestinal homeostasis. This is particularly relevant in the context of HIV-1 infection, in which HIV-1 replication, CD4+ T cell depletion, intestinal epithelial barrier dysfunction, and the translocation of microbial products from the lumen into LP and systemic circulation have been reported (11, 12). It was also recently shown that small intestinal DCs express HIV-1 coreceptors and are capable of transporting HIV-1 through the mucosa to transmit the virus to intestinal T cells (13).
HIV-1 ssRNA encodes for numerous ligands that bind and activate APCs, such as DCs, via TLR7/8 in vitro (14–17), and there is evidence that HIV-1 replicates at high levels in the gut, providing ample viral RNA to serve as these ligands (18). It is unknown whether innate sensing by LP DCs to the combination of replicating virus and translocating commensal bacteria contributes to intestinal inflammation and T cell depletion. Further, the linking of innate and adaptive immunity through TLR ligation makes TLR ligands ideal candidates as vaccine adjuvants (19, 20). Given the unique environment of human intestinal mucosa, it is imperative that these types of mucosal-specific innate immune responses are investigated in a near physiological environment.
In the current study, a systematic approach was taken to address the effect of gastrointestinal (GI) tract conditioning on human LP DCs by comparing maturation phenotype and innate function of “unconditioned” peripheral blood (PB) DCs with DCs isolated from the LP of normal small and large bowel. Cytokine responses of DCs to a range of bacterial-derived TLR ligands and to a synthetic viral TLR ligand were measured using multicolor flow cytometry. Additionally, we evaluated the response of LP DCs to the concurrent exposure of a viral and bacterial TLR ligand. These studies shed light on the mechanisms by which viruses, such as HIV-1, might induce intestinal inflammation and potentially alter the homeostatic response to commensal flora.
Materials and Methods
Study participants
Human intestinal biopsies (n = 7 jejunum, n = 10 colon) were obtained from patients undergoing elective abdominal surgery and represented otherwise discarded tissue from surgical anastomic junctions that was considered macroscopically normal. Patients with a history of inflammatory bowel disease or those receiving chemotherapy, radiation, or other immunosuppressive drugs were excluded from the study. This cohort consisted of 7 females and 10 males with a median age of 56 y (range, 21–72 y). PB samples were obtained from 22 healthy adults (12 females and 10 males) with a median age of 35 y (range, 22–58 y) who voluntarily gave written informed consent to participate. A statistical difference was noted in the median age of the two cohorts (p = 0.0004). Collection of PB samples was approved by the Colorado Multiple Institutional Review Board at the University of Colorado Denver.
Collection and preparation of human LP mononuclear cells
LP mononuclear cells (LPMCs) were isolated using techniques described in detail elsewhere (5). Briefly, tissue specimens were trimmed of fat and incubated in 1.67 mM DTT (Sigma-Aldrich, St. Louis, MO) in HBSS to remove additional mucus; the epithelial layer was subsequently removed with 1 mM EDTA solution supplemented with 1% BSA (Sigma-Aldrich). The remaining tissue was treated with collagenase D (1–2 mg/ml, Roche, Nutley, NJ) in RPMI 1640 (Invitrogen, Carlsbad, CA) + 1% penicillin/streptomycin/l-glutamine (Sigma-Aldrich) with or without 500 μg/ml piperacillin/tazobactam (Wyeth, Madison, NJ) (complete media [CM]) + 10% FBS (Sigma-Aldrich) or 0.1% BSA + 10 μg/ml DNase I (Sigma-Aldrich). All released LPMCs were cryopreserved in complete media + 10% FBS + 10% DMSO (Fisher Scientific, Pittsburgh, PA) and stored in liquid nitrogen. LPMCs were thawed in CM + 10% FBS + 10 μg/ml DNase I prior to use. The overall viability of thawed LPMC samples was 73.4% ± 4.0% (mean ± SEM), based on exclusion of a viability dye (live/dead fixable dead cell stain; see below).
Collection and preparation of human PBMCs
PBMCs were isolated from heparinized blood by standard Ficoll-Hypaque (Amersham Biosciences, Picataway, NJ) density-gradient centrifugation, as described previously (5, 21). In some instances, PBMCs were treated with collagenase D, as detailed for the isolation of LPMCs. PBMCs were cryopreserved in 10% DMSO in CM containing 10% human AB serum (Gemini Bio-Products, Woodland, CA) and stored in liquid nitrogen.
Abs for flow cytometry
Various combinations of the following Abs were used: CD19 (PE-Cy5), HLA-DR (allophycocyanin-Cy7), CD83 (PE), CD86 (FITC) purified CCR7, biotinylated anti-mouse IgG, streptavidin PE-Texas Red, IL-12p40/p70 (PE), TNF-α (FITC), IL-10 (PE) (all from BD Biosciences, San Jose, CA); biotinylated TLR4 and IL-6 (FITC) (both from eBioscience, San Diego, CA); CD1c ([BDCA-1], allophycocyanin) and CD303 ([BDCA-2], FITC) (both from Miltenyi Biotec, Auburn, CA); TLR5 (PE; Imgenex, San Diego, CA); biotinylated CD40 (Ancell, Bayport, MN); and streptavidin-PE-Texas Red (Beckman Coulter, Fullerton, CA).
In all cases, the recommended, appropriate isotype-control Abs were used, and FcR-blocking reagent (Miltenyi Biotec) was included in all initial incubations to limit nonspecific Ab binding through FcRs.
Flow cytometry protocol for surface and intracellular Abs
Standard flow-cytometric staining protocols for surface markers are detailed elsewhere (5, 21). Eight-color flow cytometry was performed on LPMCs and PBMCs immediately after thawing (baseline) or after culture for assessment of intracellular cytokine production using a FACSAria or an LSRII flow cytometer (BD Biosciences). Cells were washed in cold Dulbecco’s PBS (Invitrogen) containing 1% BSA and supplemented with 2 mM EDTA (FACS buffer). Cells were incubated with appropriate surface Abs (detailed above) for 20 min at 4°C and washed in cold PBS. For CCR7, a three-step staining protocol was used with purified CCR7 + biotinylated anti-mouse IgG + streptavidin-PE-Texas Red, whereas CD40 staining was achieved with a two-step protocol using biotinylated CD40 + streptavidin-PE-Texas Red. TLR4 staining was performed using biotinylated TLR4 and streptavidin-PE-Texas Red. Cells were washed twice in FACS buffer before the final incubation with the streptavidin-labeled fluorochrome, after which cells were washed in cold PBS as for the single-Ab staining procedure detailed above. At the completion of surface Ab incubations, all cells were stained with a Live/Dead Fixable Dead Cell Stain (Aqua Fluorescent reactive dye; Invitrogen) for 30 min at 4°C. Cells were then washed in Dulbecco’s PBS, and, for surface-only staining, cells were resuspended in 4% paraformaldehyde (Sigma-Aldrich) for 15 min at room temperature, washed, and resuspended in FACS buffer prior to acquisition.
For intracellular cytokine staining, cells were washed in PBS at the completion of the surface-staining and live/dead cell-staining protocol, fixed and permeabilized in Cytofix/Cytoperm buffer (BD Biosciences) for 15 min, washed in Perm/Wash buffer (BD Biosciences), resuspended in Perm/Wash buffer containing the appropriate cytokine Abs, and incubated for 15 min at 4°C. After an additional wash in Perm/Wash buffer, cells were resuspended in 0.5% paraformaldehyde prior to acquisition.
In vitro stimulation of LPMCs and PBMCs
Thawed LPMC or PBMC samples were resuspended at 0.9–2 × 106 cells/ml in CM + 10% human AB serum. Cells were cultured at 37°C in a humidified 5% CO2 atmosphere without exogenous stimuli or were stimulated with 10 μg/ml peptidoglycan (PGN; TLR2 ligand [TLR2L]; Sigma-Aldrich), 10 μg/ml LPS from Salmonella minnesota (TLR4 ligand [TLR4L]; Sigma-Aldrich), 0.1 μg/ml S. typhimurium flagellin (TLR5 ligand [TLR5L]; InvivoGen, San Diego, CA), 5 μg/ml CLO97, a derivative of the imidazoquinoline compound R848 (TLR7/8 ligand [TLR7/8L]; InvivoGen), or a combination of TLR4L and TLR7/8L. For the assessment of intracellular cytokines, cells were cultured for 30 min, 1 μg/ml Golgi Plug (brefeldin A, BD Biosciences) was added, and cultures were incubated for an additional 17–24 h. For the assessment of secreted cytokines, culture supernatants were collected after 18–28 h of culture in the absence of Golgi Plug. Prior to evaluation of cytokine production, 10 μg/ml DNase I was added to all cultures for 5 min at 37°C to dissociate cell clumps.
In vitro stimulation of DC-depleted LPMC cultures
Total LPMCs were stained with fluorescent-labeled CD1c and CD19, as described above, and CD1c+CD19− myeloid DCs (mDCs) were depleted from total LPMCs by FACS using a FACS ARIA. Collected LPMCs were 99.99% depleted of mDCs (data not shown). Total LPMCs (also labeled with CD1c and CD19 Abs but not flow sorted) or DC-depleted LPMCs were stimulated with or without TLR7/8L, as described above, and culture supernatants were collected after 23–24 h of culture. Highly enriched populations of LP CD1c+CD19−mDCs (>85% purity, range 86–95%) were obtained from four donors by flow-sorting and stimulated with or without TLR7/8L for 24 h prior to collection of culture supernatants.
Detection of IL-23 and -12p70 within culture supernatants
IL-23 and -12p70 (both from eBioscience) ELISAs were run using culture supernatants from stimulated LPMC and PBMC cultures following the manufacturer’s recommended protocols. The lower detection limits were 15 pg/ml for IL-23 and 4 pg/ml for IL-12p70.
Data analysis
All flow-cytometry data analysis was performed using DIVA software (BD Biosciences). Analysis was performed if ≥100 DCs were acquired. For statistical analysis of surface marker and cytokine expression, Mann–Whitney analysis was performed comparing PBMCs with LPMCs. The Wilcoxon signed-rank test was used to compare PBMCs before and after collagenase treatment and to compare the difference in cytokine production before and after the addition of exogenous stimuli. The Friedman test was used for matched-paired comparisons across multiple groups. Statistical analysis was performed using GraphPad Prism Statistical Software, version 4.0 (GraphPad Software, San Diego, CA). No significant differences were observed between CD1c+ mDCs isolated from the jejunum or the colon for phenotypic or functional analyses (Mann-Whitney t test); therefore, results are shown using pooled samples, with the specific tissue location highlighted where possible.
Results
CD1c+ mDCs, but not CD303+ plasmacytoid DCs, are readily identifiable within normal human LP
Previous investigations identified human intestinal mDCs by the expression of HLA-DR and CD11c in the absence of the expression of a range of lineage markers (CD3, CD14, CD16, CD19, CD20, and CD56) (22), a combination of Abs traditionally used to identify blood mDCs. The majority of blood mDCs can also be directly identified by the expression of CD1c and exclusion of CD19 (23), whereas plasmacytoid DCs (pDCs), a second subset of blood DCs, specifically express CD303 (23). Thus, we focused on identifying LP DCs based on the expression of CD1c and CD303, with comparisons made with profiles of DCs observed within PBMCs (Fig. 1A). Much like B cells within the blood (23), a subset of LP B cells also expressed CD1c and was excluded from the CD1c+ mDC population by the inclusion of CD19 in the Ab mixture. As in the blood, these LP CD1c+CD19− DCs were identified as a subset of the previously identified human CD11c+Lineage−HLA-DR+ LP DCs (22) (data not shown).
CD1c+ DCs are readily identifiable within the LP. A, A representative example is shown of the gating strategy used to identify DCs within PB (PBMC) or LP (LPMC). Viable cells are initially gated using a viable cell dye (not shown), and cellular debris is removed using a forward/side scatter gate. mDCs were identified based on the expression of CD1c in the absence of CD19 (CD1c+ DCs), whereas pDCs expressed CD303. B, A higher frequency of CD1c+CD19− DCs was measured in the LP (n = 9) compared with similarly identified DCs within the PB (n = 11), whereas fewer pDCs were identified in the LP compared with PB. Frequencies of CD1c+ DCs and CD303+ pDCs from jejunum (n = 5) and colon (n = 4) are shown and are expressed as the percentage of total viable cells. Lines represent median values. Statistical analysis comparing frequencies of CD1c+ DCs or CD303+ pDCs within the LP with those detected in PB was performed using the Mann–Whitney t test.
CD1c+ DCs are readily identifiable within the LP. A, A representative example is shown of the gating strategy used to identify DCs within PB (PBMC) or LP (LPMC). Viable cells are initially gated using a viable cell dye (not shown), and cellular debris is removed using a forward/side scatter gate. mDCs were identified based on the expression of CD1c in the absence of CD19 (CD1c+ DCs), whereas pDCs expressed CD303. B, A higher frequency of CD1c+CD19− DCs was measured in the LP (n = 9) compared with similarly identified DCs within the PB (n = 11), whereas fewer pDCs were identified in the LP compared with PB. Frequencies of CD1c+ DCs and CD303+ pDCs from jejunum (n = 5) and colon (n = 4) are shown and are expressed as the percentage of total viable cells. Lines represent median values. Statistical analysis comparing frequencies of CD1c+ DCs or CD303+ pDCs within the LP with those detected in PB was performed using the Mann–Whitney t test.
In comparing DCs from the blood (n = 11) with those from LP samples (n = 9) of nonautologous donors, a significantly higher frequency of CD1c+ mDCs was observed in LP preparations (Fig. 1B), with CD1c+ mDCs making up 0.76% (range, 0.40–1.13%) of viable LPMCs. Conversely, pDCs were virtually undetectable in normal LP (Fig. 1B). Thus, additional investigations focused on the CD1c+/CD19− mDC population in LP (hereafter referred to as CD1c+ DCs).
LP CD1c+ DCs display a more activated phenotype than PB CD1c+ DCs
The level of activation and the maturation state of mDCs are critical to their function as APCs (3), yet few studies have addressed the maturation status of human DCs exposed to local environmental factors present in the intestinal LP relative to PB DCs, which are unlikely to have been exposed to the same tissue-specific factors.
Expression of a panel of typical DC maturation markers CD40, CD83, CD86, HLA-DR, and CCR7 was compared between PB (n = 11) and LP (n = 7–9) CD1c+ DCs (Fig. 2, Table I). LP CD1c+ DCs expressed significantly higher levels of CD40, CD83, CD86, and HLA-DR (Table I) compared with PB CD1c+ DCs. Additionally, the lymph node-homing chemokine receptor CCR7 was expressed at significantly higher levels on LP CD1c+ DCs (Table I).
Representative example showing expression of various maturation/activation markers on CD1c+ DCs from PB (PBMC) and LP (LPMC) samples. Multiparameter flow-cytometry techniques were used to assess the expression of maturation/activation markers (CD86, CD83, CD40, HLA-DR and CCR7; shaded graph) on CD1c+ DCs, identified as described in Fig. 1A. Background staining was assessed using isotype controls (open graph).
Representative example showing expression of various maturation/activation markers on CD1c+ DCs from PB (PBMC) and LP (LPMC) samples. Multiparameter flow-cytometry techniques were used to assess the expression of maturation/activation markers (CD86, CD83, CD40, HLA-DR and CCR7; shaded graph) on CD1c+ DCs, identified as described in Fig. 1A. Background staining was assessed using isotype controls (open graph).
. | PB (Net MFI: Median [Range]) . | LP (net MFI: Median [Range]) . | Fold Increase LPMC/PBMCa . |
---|---|---|---|
CD86 | 6,813 (4,340–8,231) | 25,190 (13,154–41,594) | 3.70* |
CD83 | 11 (0–45) | 1,867 (268–4,123) | 169.7* |
CD40 | 4,783 (2,641–7,751) | 37,744 (28,263–46,623) | 7.89* |
HLA-DR | 16,008 (9,682–20,160) | 48,626 (30,873–59,697) | 3.04** |
CCR7 | 371 (240–965) | 1,746 (103–5,210) | 4.71*** |
. | PB (Net MFI: Median [Range]) . | LP (net MFI: Median [Range]) . | Fold Increase LPMC/PBMCa . |
---|---|---|---|
CD86 | 6,813 (4,340–8,231) | 25,190 (13,154–41,594) | 3.70* |
CD83 | 11 (0–45) | 1,867 (268–4,123) | 169.7* |
CD40 | 4,783 (2,641–7,751) | 37,744 (28,263–46,623) | 7.89* |
HLA-DR | 16,008 (9,682–20,160) | 48,626 (30,873–59,697) | 3.04** |
CCR7 | 371 (240–965) | 1,746 (103–5,210) | 4.71*** |
Expression of various maturation markers were determined on CD1c+ DCs (as defined in Fig. 1A) from PB (n = 10–11) and LP (n = 7–9). Values shown are net MFI, determined by removing background staining using appropriate isotype controls. Increased expression levels on LP CD1c+ DCs are shown as fold increase over PB CD1c+ DC expression levels.
Statistical analysis comparing expression levels on LP CD1c+ DCs with PB CD1c+ DCs was performed using the Mann-Whitney t test.
*p ≤ 0.001; **p ≤ 0.0001; ***p ≤ 0.05.
CD103 is an integrin found on a subpopulation of murine DCs that are thought to play a crucial role in inducing gut-homing receptors on T cells and Foxp3+ T cell differentiation in vitro (24–26). It was recently identified on a subpopulation of human mesenteric lymph node DCs and shown to induce the gut-homing molecule CCR9 on allogeneic CD8+ T cells (27). Low levels of CD103 (net mean fluorescence intensity [MFI]: median, 153; range, 35–495; n = 9) were detected on a very small fraction of LP CD1c+ DCs (net percentage of CD103+ CD1c+ DCs: median, 3.8%; range, 0–18.5%), although LP DC expression of CD103 was higher than that expressed on PB CD1c+ DCs (net MFI: median, 4; range, 0–31; net percentage of CD103+ CD1c+ DCs: median, 0.04%; range, 0–0.74%; n = 11). It remains to be determined whether other human LP DC subsets express higher levels of CD103 than CD1c+ DCs.
To evaluate the possible effects of the enzymatic-digestion step required for the isolation of LPMCs on the DC phenotype, PBMCs were treated with a similar collagenase-digestion protocol and comparisons were made with untreated PBMCs from matched donors (n = 5–7). Exposure of PBMCs to collagenase induced increased expression of CD86 (2.67-fold increase over untreated DCs) and CD83 (32.9-fold increase over untreated DCs) on CD1c+ DCs. However, the expression levels of CD86 and CD83 on LP CD1c+ DCs (Table I) remained significantly higher than on collagenase-treated PB CD1c+ DCs (p = 0.008 and p = 0.005, respectively; data not shown). No significant increases in HLA-DR, CD40, or CCR7 expression were noted on PB CD1c+ DCs after collagenase exposure compared with untreated PB CD1c+ DCs. Thus, even after accounting for a potential effect of the LPMC-isolation procedure on the expression levels of CD86 and CD83, LP CD1c+ DCs remained consistently more mature than blood CD1c+ DCs.
Despite the greater expression of CD40, CD86, CD83, HLA-DR, and CCR7, a phenotype typical of DC maturation and activation, LP CD1c+ DCs were not in a fully mature state, as indicated by their ability to further upregulate the expression of maturation markers after overnight culture with or without specific stimulation (data not shown).
LP CD1c+ DCs constitutively produce more IL-6 and -10 and TNF-α than PB CD1c+ DCs
To assess the cytokines that LP CD1c+ DCs produce in vivo under homeostatic steady-state conditions, total LPMCs or PBMCs were cultured overnight in the absence of exogenous stimuli, and the constitutive production of IL-12p40/p70, -10 and -6 and TNF-α by LP or PB CD1c+ DCs was determined by an intracellular cytokine staining assay (Fig. 3A). IL-12p40/p70–producing CD1c+ DCs were virtually undetectable within PBMC or LPMC cultures without exogenous stimulation (Fig. 3B), and no significant difference between the groups was observed. However, significantly more IL-10+ and TNF-α+ and, to a lesser extent, IL-6+ CD1c+ DCs were identified within unstimulated LPMC compared with PBMC cultures (Fig. 3B). No significant differences in cytokine production were observed in CD1c+ DCs from PBMCs versus collagenase-treated PBMCs (n = 5; data not shown), suggesting that increased basal cytokine production in LP CD1c+ DCs did not result from effects of the tissue-digestion process.
Constitutive production of IL-6, -10, and -12p40/p70 and TNF-α by LP and PB CD1c+ DCs. LPMCs (LP) or PBMCs (PB) were cultured for 17–24 h in the absence of exogenous stimulation, and the frequencies of 12p40/p70+, IL-10+, TNF-α+, and IL-6+ CD1c+ DCs were determined using an intracellular cytokine staining assay. A, A representative example of intracellular cytokine levels within viable jejunal LP CD1c+ DCs with quadrants established using appropriate isotype controls for each cytokine. B, Frequencies of cytokine-producing CD1c+ DCs as the percentage of total CD1c+ DCs were compared between PB (n = 11) and LP (n = 9–12), with results from jejunum (n = 5–6) and colon (n = 4–6) delineated. Lines represent median values, and statistical analysis comparing LP CD1c+ DCs with PB CD1c+ DCs was performed using the Mann–Whitney t test.
Constitutive production of IL-6, -10, and -12p40/p70 and TNF-α by LP and PB CD1c+ DCs. LPMCs (LP) or PBMCs (PB) were cultured for 17–24 h in the absence of exogenous stimulation, and the frequencies of 12p40/p70+, IL-10+, TNF-α+, and IL-6+ CD1c+ DCs were determined using an intracellular cytokine staining assay. A, A representative example of intracellular cytokine levels within viable jejunal LP CD1c+ DCs with quadrants established using appropriate isotype controls for each cytokine. B, Frequencies of cytokine-producing CD1c+ DCs as the percentage of total CD1c+ DCs were compared between PB (n = 11) and LP (n = 9–12), with results from jejunum (n = 5–6) and colon (n = 4–6) delineated. Lines represent median values, and statistical analysis comparing LP CD1c+ DCs with PB CD1c+ DCs was performed using the Mann–Whitney t test.
LP CD1c+ DCs have a blunted cytokine response to TLR4L and TLR5L but a similar response to TLR2L and TLR7/8L relative to PB CD1c+ DCs
We next evaluated the cytokine profiles of CD1c+ DCs in response to viral and bacterial TLR ligands. Because HIV-1 ssRNA encodes for multiple TLR7/8Ls (14–17), we used a synthetic TLR7/8L to model these potential innate interactions of HIV-1 with LP CD1c+ DCs. Stimulation with TLR4L and TLR5L induced significant increases in the frequencies of TNF-α+ and IL-6+ PB CD1c+ DCs, but the same stimuli failed to significantly increase the frequencies of cytokine-producing LP CD1c+ DCs above constitutive levels (Table II). TLR2 stimulation of PBMCs and LPMCs resulted in increased frequencies of IL-10+, TNF-α+, and IL-6+ CD1c+ DCs but no change in IL-12p40/p70+ CD1c+ DC frequencies (Table II). In fact, TLR2L was the only stimulus that induced significant increases in the frequencies of IL-10–producing PB and LP CD1c+ DCs (Table II). In contrast to the other stimuli, only TLR7/8 stimulation significantly increased the frequency of IL-12p40/p70+ CD1c+ DCs in PB and LP, along with increased frequencies of TNF-α+ and IL-6+ CD1c+ DCs (Table II). Focused comparisons of TLR7/8L-induced IL-12p40/p70 responses showed similar frequencies of IL-12p40/p70+ CD1c+ DCs in PB (median, 8.6%; range 0–31.1%; n = 11) and LP (median, 13.2%, range, 5.0–28.1%; n = 11; p = 0.54) in response to TLR7/8 stimulation.
. | PB CD1c+ DCs (%) Cytokine+ (Net) (Median [Range]) . | Trenda . | LP CD1c+ DCs (%) Cytokine+ (Net) (Median [Range]) . | Trenda . |
---|---|---|---|---|
TLR4L | ||||
IL-12p40/p70 | 0 (0–1.4) | ↔ | 0.15 (0–1.8) | ↔ |
TNF-α | 25.4 (13.2–41.2) | ↑* | 10.1 (−2.8–41.9) | ↔ |
IL-6 | 23.9 (12.1–36.9) | ↑* | 7.7 (−3.7–25.3) | ↔ |
IL-10 | 0.7 (−1.7–3.1) | ↔ | 0.9 (−1.3–8.5) | ↔ |
TLR5L | ||||
IL-12p40/p70 | 0 (0–1.4) | ↔ | 0.6 (0–5.2) | ↔ |
TNF-α | 3.1 (−1.7–11.1) | ↑** | 6.6 (−2.6–11.8) | ↔ |
IL-6 | 4.9 (2.4–13.9) | ↑*** | 0.6 (−0.6–10.7) | ↔ |
IL-10 | 0.3 (−0.7–3.4) | ↔ | 0.6 (−3.4–4.0) | ↔ |
TLR2L | ||||
IL-12p40/p70 | 0 (0–0.8) | ↔ | 0.2 (−1.9–1.6) | ↔ |
TNF-α | 31.8 (24.5–39.4) | ↑*** | 17.9 (−4.0–34.1) | ↑** |
IL-6 | 4.7 (−0.7–22.7) | ↑** | 10.8 (−1.9–21.1) | ↑** |
IL-10 | 1.7 (−0.2–9.1) | ↑** | 9.4 (2.2–13.0) | ↑*** |
TLR7/8L | ||||
IL-12p40/p70 | 8.6 (0–31.1) | ↑*** | 13.2 (5.0–28.1) | ↑* |
TNF-α | 48.8 (27.1–73.7) | ↑* | 34.5 (3.1–76.7) | ↑* |
IL-6 | 61.8 (21.3–73.9) | ↑* | 26.5 (23.9–44.0) | ↑*** |
IL-10 | 1.1 (−0.6–7.5) | ↔ | 4.2 (−5.8–16.1) | ↔ |
. | PB CD1c+ DCs (%) Cytokine+ (Net) (Median [Range]) . | Trenda . | LP CD1c+ DCs (%) Cytokine+ (Net) (Median [Range]) . | Trenda . |
---|---|---|---|---|
TLR4L | ||||
IL-12p40/p70 | 0 (0–1.4) | ↔ | 0.15 (0–1.8) | ↔ |
TNF-α | 25.4 (13.2–41.2) | ↑* | 10.1 (−2.8–41.9) | ↔ |
IL-6 | 23.9 (12.1–36.9) | ↑* | 7.7 (−3.7–25.3) | ↔ |
IL-10 | 0.7 (−1.7–3.1) | ↔ | 0.9 (−1.3–8.5) | ↔ |
TLR5L | ||||
IL-12p40/p70 | 0 (0–1.4) | ↔ | 0.6 (0–5.2) | ↔ |
TNF-α | 3.1 (−1.7–11.1) | ↑** | 6.6 (−2.6–11.8) | ↔ |
IL-6 | 4.9 (2.4–13.9) | ↑*** | 0.6 (−0.6–10.7) | ↔ |
IL-10 | 0.3 (−0.7–3.4) | ↔ | 0.6 (−3.4–4.0) | ↔ |
TLR2L | ||||
IL-12p40/p70 | 0 (0–0.8) | ↔ | 0.2 (−1.9–1.6) | ↔ |
TNF-α | 31.8 (24.5–39.4) | ↑*** | 17.9 (−4.0–34.1) | ↑** |
IL-6 | 4.7 (−0.7–22.7) | ↑** | 10.8 (−1.9–21.1) | ↑** |
IL-10 | 1.7 (−0.2–9.1) | ↑** | 9.4 (2.2–13.0) | ↑*** |
TLR7/8L | ||||
IL-12p40/p70 | 8.6 (0–31.1) | ↑*** | 13.2 (5.0–28.1) | ↑* |
TNF-α | 48.8 (27.1–73.7) | ↑* | 34.5 (3.1–76.7) | ↑* |
IL-6 | 61.8 (21.3–73.9) | ↑* | 26.5 (23.9–44.0) | ↑*** |
IL-10 | 1.1 (−0.6–7.5) | ↔ | 4.2 (−5.8–16.1) | ↔ |
PBMCs (PB; n = 8–11) or LPMCs (LP; n = 6-11) were stimulated with LPS (TLR4L; 10 μg/ml), S. typhimurium flagellin (TLR5L; 0.1 μg/ml), PGN (TLR2L; 10 μg/ml), or a derivative of the imidazoquinoline compound R848 (TLR7/8L; 5 μg/ml) for 17–24 h, and intracellular levels of IL-6, -10, and -12p40/p70 and TNF-α within PB CD1c+ DCs or LP CD1c+ DCs were determined (as shown in Fig. 3A). Values are expressed as the difference between frequencies of cytokine+ CD1c+ DCs in stimulated cultures and those in cultures without exogenous stimulation (net).
Overall change between cytokine+ CD1c+ DC frequencies in TLR-stimulated and unstimulated conditions; Wilcoxon signed-rank test. ↑, increase; ↔, no change.
*p ≤ 0.001; **p ≤ 0.05; ***p ≤ 0.01.
LP CD1c+ DCs express less TLR4 but more TLR5 than PB CD1c+ DCs
TLR expression levels may, in part, determine the response pattern of DCs to microbial products, and low TLR expression on gut DCs was postulated to play a role in the tolerance of commensal organisms (28–30). Given that we observed blunted cytokine responses by LP CD1c+ DCs to TLR4 and TLR5 stimulation, we compared the surface expression of TLR4 and TLR5 on CD1c+ DCs in LP (n = 6–7) versus PB (n = 9). Although TLR4 was expressed at very low levels on PB and LP DCs, a statistically lower frequency of CD1c+ DCs in LP expressed TLR4 compared with PB CD1c+ DCs (Fig. 4). Expression of TLR5 was low to undetectable on PB CD1c+ DCs, and a statistically greater frequency of LP CD1c+ DCs expressed TLR5 (Fig. 4). No differences were observed in TLR4 or TLR5 expression on CD1c+ DCs in comparisons of PBMCs with collagenase-treated PBMCs (n = 4; data not shown).
Expression of TLR4 and TLR5 by LP and PB CD1c+ DCs. Surface expression of TLR4 and TLR5 were determined within PB (n = 9) and LP [(n = 6–7); jejunum (n = 3–4); colon (n = 3)] CD1c+ DC populations using flow-cytometry techniques. Values are expressed as net MFI (top panels) or as net percentage positive (bottom panels) by removing background fluorescence based on isotype controls. Lines represent median values, and statistical analysis comparing levels of expression and frequencies of TLR-expressing CD1c+ DCs within the LP with those detected in PB was performed using the Mann–Whitney t test.
Expression of TLR4 and TLR5 by LP and PB CD1c+ DCs. Surface expression of TLR4 and TLR5 were determined within PB (n = 9) and LP [(n = 6–7); jejunum (n = 3–4); colon (n = 3)] CD1c+ DC populations using flow-cytometry techniques. Values are expressed as net MFI (top panels) or as net percentage positive (bottom panels) by removing background fluorescence based on isotype controls. Lines represent median values, and statistical analysis comparing levels of expression and frequencies of TLR-expressing CD1c+ DCs within the LP with those detected in PB was performed using the Mann–Whitney t test.
LP CD1c+ DCs produce IL-23 but not IL-12p70 in response to TLR7/8 stimulation
Given the finding that only TLR7/8 stimulation induced significant increases in the frequency of LP CD1c+ DCs producing IL-12p40/p70 and that the p40 subunit of IL-12 also forms part of the IL-23 complex (31), we next evaluated whether the TLR7/8L-induced IL-12p40/p70 responses reflected IL-12p70 or -23 production. Because of the paucity of IL-23–specific Abs appropriate for intracellular detection, IL-23 released into culture supernatants following TLR stimulation was assessed by ELISA. In accord with the increase in IL-12p40/p70+ CD1c+ DCs observed in the intracellular cytokine staining flow cytometry assay, TLR7/8 stimulation induced significant amounts of IL-23 from LPMCs, whereas stimulation of the same LPMCs with bacterial TLRs resulted in limited IL-23 production (Fig. 5A). Minimal IL-12p70 production was detected for any stimulation condition (Fig. 5A). To evaluate the contribution of LP CD1c+ DCs to the total IL-23 production observed by ELISA within LPMC cultures, CD1c+ DCs were first depleted from LPMCs by flow sorting, and CD1c+ DC-depleted and nondepleted LPMCs were cultured with or without TLR7/8L. Levels of measurable IL-23 were reduced by 92.7% (range: 41.4–100%) in TLR7/8L-stimulated cultures following depletion of CD1c+ DCs, indicating that the majority of IL-23 measured in LPMC cultures following TLR7/8 stimulation was produced by or dependent upon CD1c+ DCs (Fig. 5B).
IL-23 production by LP CD1c+ DCs. A, To detect IL-23 and -12p70 within LPMC cultures (n = 6–13), total LPMCs were stimulated with LPS (TLR4L; 10 μg/ml), PGN (TLR2L; 10 μg/ml), S. typhimurium flagellin (TLR5L; 0.1 μg/ml), or TLR7/8L (5 μg/ml) for 17–28 h, and levels of IL-23 and -12p70 in culture supernatants were evaluated by ELISA. Values represent the net amount of cytokine (picogram) per 1 × 106 total LPMCs within TLR-stimulated cultures after subtraction of cytokine amounts (per 1 × 106 total LPMCs) detected in cultures without exogenous stimuli (line = median). Statistical analysis was performed on paired samples with and without each specific TLR stimulation using the Wilcoxon signed-rank test. B, To determine the contribution of LP CD1c+ DCs to IL-23 production observed in total LPMC cultures with TLR7/8 stimulation, LPMCs from a subset of samples (n = 3) were depleted of CD1c+ DCs by flow-sorting techniques, cultured with or without TLR7/8L (5 μg/ml) for 23–24 h, and levels of IL-23 were determined by ELISA. Comparisons of total IL-23 detected within CD1c+ DC-depleted cultures were made to parallel cultures of matched total LPMCs similarly stimulated with or without TLR7/8L. The values (median, range) represent the net amount of cytokine (picogram) per 1 × 106 LPMCs determined by subtracting the amount of cytokine detected in cultures without exogenous stimuli from that within TLR7/8-stimulated cultures. C, To detect IL-23 and -12p70 within LPMC cultures (n = 7) in response to combined viral and bacterial TLR ligand stimulation, total LPMCs were stimulated with TLR4L (10 μg/ml), TLR7/8L (5 μg/ml), or a combination of both TLR ligands for 24–28 h, and IL-23 and -12p70 in culture supernatants were evaluated by ELISA. Values represent the net amount of cytokine (picogram) per 1 × 106 total LPMCs within TLR-stimulated cultures after subtraction of cytokine amounts (per 1 × 106 total LPMC) detected in cultures without exogenous stimuli (line = median). Statistical analysis was performed using the Friedman test. D, Levels of IL-23 and -12p70 in PBMC (PB; n = 6) or LPMC (LP; n = 7) culture supernatants were evaluated by ELISA 24–28 h after stimulation with a combination of TLR4L (10 μg/ml) and TLR7/8L (5 μg/ml). Values represent the net amount of cytokine (picogram) per 1 × 106 total LPMCs or PBMCs within stimulated cultures after subtraction of cytokine amounts (per 1 × 106 total LPMCs or PBMCs) detected in cultures without exogenous stimuli (line = median). Statistical analysis was performed on paired samples with and without stimulation using the Wilcoxon signed-rank test.
IL-23 production by LP CD1c+ DCs. A, To detect IL-23 and -12p70 within LPMC cultures (n = 6–13), total LPMCs were stimulated with LPS (TLR4L; 10 μg/ml), PGN (TLR2L; 10 μg/ml), S. typhimurium flagellin (TLR5L; 0.1 μg/ml), or TLR7/8L (5 μg/ml) for 17–28 h, and levels of IL-23 and -12p70 in culture supernatants were evaluated by ELISA. Values represent the net amount of cytokine (picogram) per 1 × 106 total LPMCs within TLR-stimulated cultures after subtraction of cytokine amounts (per 1 × 106 total LPMCs) detected in cultures without exogenous stimuli (line = median). Statistical analysis was performed on paired samples with and without each specific TLR stimulation using the Wilcoxon signed-rank test. B, To determine the contribution of LP CD1c+ DCs to IL-23 production observed in total LPMC cultures with TLR7/8 stimulation, LPMCs from a subset of samples (n = 3) were depleted of CD1c+ DCs by flow-sorting techniques, cultured with or without TLR7/8L (5 μg/ml) for 23–24 h, and levels of IL-23 were determined by ELISA. Comparisons of total IL-23 detected within CD1c+ DC-depleted cultures were made to parallel cultures of matched total LPMCs similarly stimulated with or without TLR7/8L. The values (median, range) represent the net amount of cytokine (picogram) per 1 × 106 LPMCs determined by subtracting the amount of cytokine detected in cultures without exogenous stimuli from that within TLR7/8-stimulated cultures. C, To detect IL-23 and -12p70 within LPMC cultures (n = 7) in response to combined viral and bacterial TLR ligand stimulation, total LPMCs were stimulated with TLR4L (10 μg/ml), TLR7/8L (5 μg/ml), or a combination of both TLR ligands for 24–28 h, and IL-23 and -12p70 in culture supernatants were evaluated by ELISA. Values represent the net amount of cytokine (picogram) per 1 × 106 total LPMCs within TLR-stimulated cultures after subtraction of cytokine amounts (per 1 × 106 total LPMC) detected in cultures without exogenous stimuli (line = median). Statistical analysis was performed using the Friedman test. D, Levels of IL-23 and -12p70 in PBMC (PB; n = 6) or LPMC (LP; n = 7) culture supernatants were evaluated by ELISA 24–28 h after stimulation with a combination of TLR4L (10 μg/ml) and TLR7/8L (5 μg/ml). Values represent the net amount of cytokine (picogram) per 1 × 106 total LPMCs or PBMCs within stimulated cultures after subtraction of cytokine amounts (per 1 × 106 total LPMCs or PBMCs) detected in cultures without exogenous stimuli (line = median). Statistical analysis was performed on paired samples with and without stimulation using the Wilcoxon signed-rank test.
To determine whether LP CD1c+ DCs produced IL-23 in response to direct stimulation by TLR7/8L, experiments were performed using sort-purified LP CD1c+ DCs. Using the bulk of total LPMCs available from intestinal samples from four donors, LP CD1c+ DCs with >85% purity (range, 86–95%) were obtained by flow sorting. Following overnight stimulation of sorted LP CD1c+ DCs (6,400–36,500 CD1c+ DCs per condition) with TLR7/8L versus media alone, low levels of IL-23 were measured in stimulated culture supernatant from three of four samples (1.39–7.47 pg IL-23 per 1 × 104 CD1c+ DCs) but not in unstimulated cultures, confirming that direct stimulation of LP CD1c+ DCs induced IL-23 production. Given that the levels of IL-23 induced were low, likely due in part to the small numbers of CD1c+ DCs obtained by sorting, which necessitated plating low cell concentrations, it is difficult to conclude that direct stimulation of LP CD1c+ DCs by TLR7/8L accounted for the majority of IL-23 production observed in TLR7/8-stimulated LPMCs. Thus, DC-dependent production of IL-23 may also occur through indirect mechanisms.
Combined TLR7/8 and TLR4 stimulation synergistically increases IL-23 production in LPMCs
Because altered epithelial integrity during mucosal viral infections may expose LP DCs to translocated bacterial products, as well as viral products, we evaluated the impact of combined TLR4 and TLR7/8 stimulation on IL-23 and -12p70 production in a subset of LPMC samples (n = 7; Fig. 5C). Despite the absence of significant IL-23 production following TLR4 stimulation alone, combined stimulation with TLR4L and TLR7/8L resulted in a synergistic enhancement of IL-23 production in LPMCs (Fig. 5C). Previous studies showed that combined TLR stimulation is required for IL-12p70 production by blood mDCs (32, 33); however, a significant induction of IL-12p70 was not observed following combined TLR4L and TLR7/8L stimulation within LPMC cultures (p = 0.13; Fig. 5C).
Next, the relative levels of IL-23 and -12p70 induced in response to combinatorial TLR stimulation in PBMCs relative to LPMC cultures was evaluated. Similar levels of IL-23 were observed in PBMC and LPMC cultures following stimulation (p = 0.84; Fig. 5D). In agreement with previous reports (32, 33), a significant increase in IL-12p70 production in response to combined TLR stimulation was observed in PBMC cultures (Fig. 5D). The observation that combined TLR4 and TLR7/8 stimulation failed to induce significant IL-12p70 production within LPMC cultures (Fig. 5C, 5D) suggests that LP CD1c+ DCs selectively produce IL-23, even in the presence of stimulatory conditions that induced IL-12p70 production from PBMCs.
Discussion
To evaluate the effects of intestinal conditioning on the innate function of DCs obtained from the human GI tract, we directly compared unconditioned CD1c+ DCs in PB with CD1c+ DCs obtained from the LP of small and large bowel. LP CD1c+ DCs expressed higher levels of markers typical of DC activation than did blood CD1c+ DCs, even after accounting for possible effects of the tissue-digestion process. Importantly, LP CD1c+ DCs still had the potential to upregulate these activation markers further, an observation in keeping with the upregulation of maturation markers observed in LP CD11c+ DCs that had migrated out of human colonic biopsies (22). Thus, resident LP CD1c+ DCs that have been exposed to the intestinal environment are more activated or mature than their counterparts in PB. This activation of LP DCs may result from their exposure to intestinal microbes, from soluble factors in the mucosal environment, or from contact with other cell populations in the intestinal mucosa. Interestingly, it was reported that CD3+ TCR γ/δ T cells, a subset of T cells resident in the intestinal mucosa, could induce DC maturation in a CD1c-restricted manner (34). The presence of DCs residing in the LP in a state of partial activation may allow for a more rapid immune response in the event of microbial invasion.
Higher frequencies of LP CD1c+ DCs constitutively producing IL-6 and -10 and TNF-α also likely reflect the more activated state of these cells compared with blood CD1c+ DCs. Spontaneous production of IL-10 and a concurrent lack of IL-12p40/p70 production were observed in human colonic CD11c+ DCs (35) However, in contrast to our study, Hart et al. (35) did not observe IL-6–producing CD11c+ DCs in the normal intestine. Constitutive production of IL-10 and -6 by CD1c+ DCs may reflect activation by commensal bacterial signals and aid in maintaining an anti-inflammatory environment within the LP through the induction of regulatory T cells, Th2 cells (36), and the induction of IgA-secreting B cells (37). Additionally, these cytokines were shown to play a role in maintaining epithelial barrier integrity (38, 39). The constitutive production of TNF-α is more intriguing given that it has typically been defined as a proinflammatory cytokine. However, a number of studies demonstrated the potential of TNF-α to be anti-inflammatory through specific regulation of IL-12 and -23 production by blood-derived macrophages and DCs (40, 41). It is possible that the combined production of these cytokines, in addition to other local factors, contributes to resident CD1c+ DCs existing in a state of inflammatory anergy that is necessary for GI tract homeostasis (42).
The effect of GI tract conditioning on innate responses by CD1c+ DCs was investigated using bacterial and viral TLR ligands. TLR4L and TLR5L stimulation of blood CD1c+ DCs led to increased production of IL-6 and TNF-α, despite the low expression of TLR4 and minimal expression of TLR5. Blood CD1c+ DCs were shown to express TLR5 using RT-PCR (43). Thus, it is possible that the flow cytometry techniques and specific Abs used in the present study lacked the required sensitivity to detect very low levels of surface protein adequate to allow stimulation with a TLR5L. Because these experiments were conducted using total PBMCs, it is also possible that DC-specific TLRs are upregulated in a bystander fashion during TLR stimulation in vitro, which may more accurately reflect the in vivo situation.
In contrast, TLR4L and TLR5L stimulation failed to significantly increase the frequency of cytokine-producing LP CD1c+ DCs over steady-state levels, even though LP CD1c+ DCs expressed significantly higher levels of TLR5 compared with their blood counterparts. Lower frequencies of TLR4-expressing LP CD1c+ DCs compared with blood CD1c+ DCs is in agreement with earlier observations of TLR expression on intestinal CD11c+ DCs (35). Notably, the actual levels of expression of TLR4 were very low on blood and LP CD1c+ DCs. Although a lack of IL-12p40 production by LP DCs in response to TLR4L contrasts with the results from one study of human CD11c+ intestinal DCs (44), the observation is consistent with other studies showing that murine LP CD11c+ DCs and rat lymph DCs were hyporesponsive to LPS, likely related to lower TLR4 expression (28–30, 45, 46).
To our knowledge, this is the first study to show that human LP CD1c+ DCs express TLR5, yet they fail to respond to stimulation with TLR5L. Tolerance in human monocytes induced by prior exposure to flagellin did not alter TLR5 expression, but it altered downstream signaling proteins (47). It is possible that a similar alteration of TLR5 signaling occurs in human intestinal CD1c+ DCs as a result of the specific microenvironment to limit responses against commensal bacteria and that additional signaling is required from virulence factors to induce a proinflammatory response (48). The inability of TLR5L stimulation to induce cytokine frequencies above constitutive levels differs with some reports in murine models, in which a subset of LP DCs expressing TLR5 mRNA produced IL-12p70 in response to stimulation with flagellin (46, 49). Differences in the experimental approach and the specific DC subset evaluated, whether isolated DCs or DCs in whole LPMCs were evaluated, and species differences may account for these observed differences in TLR responses between studies. Importantly, we showed that in contrast to blood CD1c+ DCs, intestinal-conditioned human LP CD1c+ DCs did not significantly alter cytokine production profiles in response to TLR4 or TLR5 stimulation, indicating a form of anergy or tolerance to these two specific microbial products in vitro.
Stimulation with TLR2L induced a distinct cytokine profile in DCs from blood and LP, characterized by the production of IL-10 and -6 and TNF-α and a lack of IL-12p40/p70 upregulation. The observation that blood and LP CD1c+ DCs produced IL-10 in response to TLR2L may indicate a strong propensity of this TLR signaling pathway to lead to the generation of regulatory responses, independent of intestinal conditioning. Indeed, TLR2L stimulation of systemic DCs was reported to promote T regulatory responses through regulation of the MAPK pathway (50, 51).
The finding that TLR7/8 activation of LP CD1c+ DCs resulted in the potent induction of IL-12p40/p70 to frequencies comparable to those observed with similarly stimulated blood CD1c+ DCs suggest that this response was not significantly altered by the intestinal environment. This TLR7/8L-induced increase in IL-12p40/p70 by LP CD1c+ mDCs was associated with CD1c+ DC-dependent production of IL-23 rather than IL-12p70. Furthermore, the synergistic enhancement of IL-23 production, but not of IL-12p70, following combined TLR4 and TLR7/8 stimulation suggests a greater propensity toward IL-23 production within the intestinal tract (52). The mechanism behind synergistic IL-23 production by LP CD1c+ DCs in response to bacterial and viral TLR ligand costimulation is under investigation, but it may relate to regulation of TLR expression (53).
The novel finding that human LP CD1c+ DCs produced IL-23, a proinflammatory cytokine involved in the expansion of Th17 and Th1 cells (54), only in response to a viral TLR ligand adds a crucial component to our understanding of the role that DCs may play in the intestinal innate immune response and their contribution to intestinal homeostasis, host defense, and inflammation. Given that we observed anergic-type responses by this same population of DCs to TLR4 and TLR5 stimulation, and a typical anti-inflammatory response was generated in response to TLR2 stimulation, the ability of this DC subset to respond in a proinflammatory manner is likely to be TLR ligand specific. This has significant implications for the development of effective mucosal vaccines, with the recent focus on the use of TLR ligands as vaccine adjuvants (19, 20).
Importantly, because resident LP CD1c+ DCs in our in vitro model were presumably conditioned in the gut in vivo, these results indicate that a population of DCs with the potential to respond in a proinflammatory manner exists in normal human LP. Constitutive production of IL-23 by murine terminal ileum DCs was reported (55), and Denning et al. (56) showed that a subset of murine intestinal DCs promoted Th17 production, suggesting that DC subsets with a nonregulatory function also exist within the GI tract of rodents. A CD14+ population of macrophages, recently identified in human LP with a macrophage/DC phenotype, were capable of producing greater amounts of proinflammatory cytokines, such as IL-23, compared with CD14− cells, an observation particularly noted in patients with Crohn’s disease (57). Interestingly, this population of cells did not express CD1c, suggesting the existence of multiple populations of cells within the LP that are capable of producing proinflammatory responses, with the potential for exacerbation of these responses when disruption of the GI tract occurs.
We previously demonstrated the existence of commensal bacteria-reactive Th1 and Th17 effector CD4+ T cells in normal human LP (5). In the context of increased IL-23 production by LP CD1c+ DCs in response to ssRNA viruses and concomitant exposure to bacteria, these bacteria-reactive T cells would likely be induced to expand in vivo. This scenario may be particularly relevant in the setting of HIV-1 infection, in which a compromised epithelial barrier could lead to increased contact of HIV-1–exposed LP CD1c+ DCs to bacteria, potentially resulting in enhanced IL-23 production, activation and infection of bacteria-reactive Th1 and Th17 cells, and persistent HIV-1 replication. This proinflammatory response would create a vicious cycle, leading to further disruption of intestinal homeostasis, local immune activation, T cell depletion, and increased microbial translocation.
There are several limitations to this study. First, a statistical difference in donor age was noted between the unmatched PBMC and LPMC donors. Age-related changes in blood DC function were reported, i.e., increased spontaneous cytokine production and decreased responses to TLR stimulation (58). No significant associations between age and CD1c+ DC frequency, phenotype, or cytokine responses to TLR stimulation were observed in our study subjects. However, a statistically significant correlation between age and spontaneous TNF-α production by PB CD1c+ DCs was noted (Spearman’s test: r = 0.60, p = 0.05). Therefore, it possible that the age differences between cohorts may account for some of the differences in DC function that were observed in blood versus intestinal samples. An additional limitation of this study may be the wide range of DC cytokine responses to stimulation observed between the different donors. This type of variability is inherent in human-based research; in a larger cohort of healthy donors, Lombardi et al. (53) noted that at least one in three donors could be classified as a high responder with respect to blood DC cytokine production in response to combined TLR stimulation. With this in mind, we only concluded that differences existed between LP and PB CD1c+ DCs in either phenotype or function when confirmed by appropriate statistical analysis.
In summary, we describe a novel mechanism by which ssRNA viruses, through innate stimulation of CD1c+ DCs via TLR7/8, may subvert the normal homeostatic mechanisms in place in the intestinal mucosa to induce inflammation. As such, these findings have important implications for understanding viral pathogenesis in the intestinal mucosa. A better understanding of the pro- and anti-inflammatory responses of intestinal DCs to different microbial stimuli may also facilitate the development of more effective mucosal vaccines.
Acknowledgements
We thank all of the subjects for their generous participation in our study. We thank Dr. Ricardo Gonzalez and all of the members of the surgical teams of the Department of Surgery, University of Colorado Hospital, Denver, CO, for assistance with the collection of tissue samples. We acknowledge the Colorado Center for AIDS Research Immunology Core for assistance with flow cytometry and the Clinical Investigation Core for assistance with the recruiting of subjects.
Disclosures The authors have no financial conflicts of interest.
Footnotes
This work was supported by National Institutes of Health Grants R01 AI065275 and K24 AI07434 (to C.C.W) and was facilitated by the infrastructure and resources provided by the Colorado Center for AIDS Research (AI054907).
Abbreviations used in this paper: