Dendritic cells limit fibroinflammatory injury in nonalcoholic steatohepatitis in mice


  • Potential conflict of interest: Nothing to report.

  • This work was supported, in part, by a Liver Scholar Award from the American Liver Foundation (to G.M.) and National Institutes of Health Awards 1UL1RR029893 (to J.R.H.), DK085278 (to G.M.), and CA155649 (to G.M.).

  • See Editorial on Page 494


Nonalcoholic steatohepatitis (NASH) is the most common etiology of chronic liver dysfunction in the United States and can progress to cirrhosis and liver failure. Inflammatory insult resulting from fatty infiltration of the liver is central to disease pathogenesis. Dendritic cells (DCs) are antigen-presenting cells with an emerging role in hepatic inflammation. We postulated that DCs are important in the progression of NASH. We found that intrahepatic DCs expand and mature in NASH liver and assume an activated immune phenotype. However, rather than mitigating the severity of NASH, DC depletion markedly exacerbated intrahepatic fibroinflammation. Our mechanistic studies support a regulatory role for DCs in NASH by limiting sterile inflammation through their role in the clearance of apoptotic cells and necrotic debris. We found that DCs limit CD8+ T-cell expansion and restrict Toll-like receptor expression and cytokine production in innate immune effector cells in NASH, including Kupffer cells, neutrophils, and inflammatory monocytes. Consistent with their regulatory role in NASH, during the recovery phase of disease, ablation of DC populations results in delayed resolution of intrahepatic inflammation and fibroplasia. Conclusion: Our findings support a role for DCs in modulating NASH. Targeting DC functional properties may hold promise for therapeutic intervention in NASH. (HEPATOLOGY 2013;58:589–602)


alanine aminotransferase


antigen-presenting cell


B-cell lymphoma 2


bone marrow


chronic liver disease


C-type lectin domain family 9 member A


dendritic cells


Fas ligand


granulocyte colony-stimulating factor


high-mobility group box 1








Kupffer cell




methionine-choline deficient


monocyte chemoattractant protein 1


macrophage inflammatory protein 1 alpha


matrix metalloproteinase 9




nonalcoholic fatty liver disease


nonalcoholic steatohepatitis


NASH with depletion of DCs


natural killer T cells


nonparenchymal cells




transforming growth factor beta


T-helper cell


tissue inhibitor of metalloproteinase 1


Toll-like receptor


tumor necrosis factor alpha


regulatory T cell

Nonalcoholic fatty liver disease (NAFLD) is the hepatic consequence of metabolic syndrome, which includes insulin resistance, hypertension, hyperlipidemia, and visceral adiposity. Obesity itself is an independent risk factor for NAFLD, which is currently recognized as the most common cause of liver dysfunction in the United States, representing 75% of all cases of chronic liver disease (CLD).[1] Moreover, future projections estimate that 50% of all Americans will have elements characteristic of NAFLD by 2030.[1] In most cases of NAFLD, liver steatosis is mild and reversible; however, 10%-20% of cases progress to nonalcoholic steatohepatitis (NASH), characterized by intense intrahepatic inflammation, exacerbated steatosis, hepatocellular injury, and incipient fibrosis.[2] Furthermore, NASH can progress to cirrhosis, liver failure, and hepatocellular carcinoma. Between 2000 and 2010, the percentage of orthotopic liver transplants performed for NASH in the United States increased from 1.2% to 7.4%.[3]

The precise cellular and biochemical pathogeneses of NASH are incompletely understood. However, a “two-hit” hypothesis has been gaining experimental traction. In general terms, hepatic lipid accumulation, the “first hit,” is thought to induce oxidative stress and hepatocyte damage, which subjects the liver to inflammatory cell infiltration—the “second hit”—leading to the cyclical development of further inflammatory injury and eventual fibrosis. A number of inflammatory mediators have been implicated. Kupffer cells (KCs) reside in liver sinusoids and contribute to hepatocyte cell death by Toll-like receptor (TLR)9-mediated production of interleukin (IL)-1β.[4] TNF-α production by activated KCs is essential for fibrosis development in NASH.[5] Moreover, NASH is mitigated in mice fed a methionine-choline–deficient (MCD) diet in the absence of KCs.[6] Neutrophils are also important mediators of hepatocellular damage in NASH. Neutrophils are activated by necrotic hepatocytes and perpetuate hepatitis through the release of proinflammatory cytokines and the secretion of myeloperoxidase (MPO), an abundant source of free radicals that contributes to disease progression by increasing oxidative hepatocyte damage.[7] An increased liver neutrophil/lymphocyte ratio has been shown to increase the likelihood of progression of steatosis to steatohepatitis and, ultimately, fibrosis in patients with NASH.[8]

Dendritic cells (DCs) are professional antigen-presenting cells (APCs) that initiate potent adaptive immune responses. DCs have also recently emerged as important mediators in noninfectious chronic fibroinflammatory conditions. For example, DCs modulate the severity of inflammation during exacerbations of asthma and are necessary for bleomycin-mediated pulmonary fibrosis.[9] Mucosal DCs in the small and large intestine are thought to be responsible for triggering deleterious T-cell responses to the endogenous microflora in inflammatory bowel disease.[10] We recently showed that, despite their activated phenotype, DCs can have a protective role in acute pancreatitis by limiting sterile inflammation.[11] The role of DCs in CLD is incompletely defined. We reported that DCs become highly proinflammatory in thioacetamide-induced chronic liver fibrosis.[12] However, the resolution of murine liver fibrosis was recently found to be accelerated by the recruitment of DCs.[13] In NASH liver, our initial investigations uncovered a robust recruitment of phenotypically activated DCs early in disease. Based on these data, we postulated that DCs augment the cycle of inflammation in NASH. However, our investigations, utilizing continuous in vivo depletion of DC populations, revealed a more-complex relationship, because DCs limit fibroinflammation in NASH by curtailing the destructive effects of KCs and neutrophils. Furthermore, during the recovery phase of disease, DC depletion delays the resolution of intrahepatic inflammation and fibroplasia. This work offers novel insight to the pathogenesis and resolution of NASH and has potential implications for targeting DCs in experimental therapeutics.

Materials and Methods

Animals and Model of NASH

Six-week-old male C57BL/6 (H-2kb), BALB/c (H-2kd), OT-I (B6.Cg-RAG2tm1Fwa-TgN), OT-II (B6.Cg-RAG2tm1Alt-TgN), CD45.1 (B6.SJL-Ptprca/BoyAiTac), and CD11c-DTR (B6.FVB-Tg[Itgax-DTR/EGFP]57Lan/J) mice were purchased from The Jackson Laboratory (Bar Harbor, ME). NASH was induced by administration of an MCD diet (MP Biomedicals, Solon, OH) for 6 weeks. Bone marrow (BM) chimeric mice were generated as previously described.[11] Briefly, C57BL/6 mice were anesthetized and irradiated (1,200 Rads), followed by intravenous transfer with 1 × 107 BM cells from CD11c.DTR mice or C57BL/6 controls. Chimeric mice were used in experiments 7 weeks later. DC depletion was achieved with serial intraperitoneal (IP) injections of diphtheria toxin (4 ng/g; Sigma-Aldrich, St. Louis, MO), beginning 1 day before initiation of the MCD diet. Serum alanine aminotransferase (ALT) was measured using the Olympus AU400 Chemistry Analyzer (Olympus, Tokyo, Japan). Control mice were aged matched, made chimeric using BM from wild-type mice, fed standard chow, and also received diphtheria toxin injections. In recovery experiments, mice were returned to standard chow and DC depletion was initiated at the time of reintroduction of a normal diet. In selected experiments, mice were treated with lipopolysaccharide (LPS) (300 μg, IP; InvivoGen, San Diego, CA) and sacrificed at 12 hours. All procedures were approved by the New York University School of Medicine Institutional Animal Care and Use Committee.

Cellular Isolation and Culture

Hepatic nonparenchymal cells (NPCs) were collected as previously described.[14] Briefly, the portal vein was cannulated and infused with 1% Collagenase IV (Sigma-Aldrich). The liver was then removed and minced. Hepatocytes were excluded with serial low-speed centrifugation (300 rpm), followed by high-speed centrifugation (1,500 rpm) to isolate the NPCs, which were then further enriched over a 40% OptiPrep gradient (Sigma-Aldrich). For DC isolation, CD11c+MHCII+ hepatic NPCs were selected by fluorescence-activated cell sorting. Splenocytes were isolated by mechanical disruption of the spleen, and splenic T cells were purified using immunomagnetic beads and positive selection columns (Miltenyi Biotec, Bergisch-Gladbach, Germany). NASH DC is defined as liver DCs harvested from mice at 6 weeks after initiation of an MCD diet. Cellular suspensions were cultured in complete media (RPMI 1640 with 10% heat-inactivated fetal bovine serum, 2 mM of L-glutamine, 100 U/mL of penicillin, 100 μg/mL of streptomycin, and 0.05 mM of 2-ME). In selected experiments, DCs were stimulated with TLR9 ligand CpG ODN1826 (5 uM; InvivoGen).

See Supporting Materials for a description of additional methods.


Hepatic DC Populations Expand in NASH

The number of CD45+ hepatic leukocytes increased by approximately 3-fold in NASH (Fig. 1A,B). Furthermore, the composition of hepatic NPC in NASH was markedly different from control liver (Fig. 1C and Supporting Fig. 1A). F4/80+ KCs expanded from a baseline of 20%-25% in control liver to 40%-50% in NASH. Gr1+ neutrophils and inflammatory monocytes expanded from ∼10% in controls to ∼25% in NASH, whereas both natural killer T (NKT) cells and B cells decreased as a fraction of total NPC (Fig. 1C). The fraction of hepatic CD3+ T cells remained fairly stable in NASH; however, we observed marked upward skewing of the CD8+/CD4+ ratio (Fig. 1D). Moreover, CD11c+MHCII+ DCs expanded from a baseline of ∼5% of liver leukocytes in control liver to 15%-18% in NASH (Fig. 1C,E). Expansion of CD11c+MHCII+ DCs began within days of initiating an MCD diet, plateaued by 2 weeks, and remained stably elevated for the duration of disease (Fig. 1F). By contrast, there was no change in splenocyte composition, splenomegaly, or evident expansion of splenic DCs in NASH, implying that the effects of NASH on DCs are specific to the liver (Supporting Fig. 1B,C).

Figure 1.

DCs expand in NASH liver. (A) Fraction and (B) total number of CD45+ leukocytes in control and NASH liver (C) as well as fraction of specific hepatic leukocyte subsets were determined by flow cytometry at 6 weeks after beginning an MCD diet in C57BL/6 mice. (D) Fraction of CD4+ and CD8+ T cells in control and NASH liver was determined by flow cytometry. (E) Coexpression of the DC markers, CD11c and MHCII, in control and NASH liver at 6 weeks and (f) time course of hepatic DC recruitment in mice fed an MCD diet were determined by flow cytometry. Experiments were repeated more than five times with similar results using 3-5 mice per data point (*P < 0.05; **P < 0.01; ***P < 0.001).

DCs Exhibit a Mature Phenotype in NASH Liver

Besides expanding in number, hepatic DCs underwent phenotypic maturation in NASH. MHCII and CD40, both essential for antigen presentation, were up-regulated on NASH DCs, as was the expression of costimulatory molecules CD54, CD80, and CD86 (Fig. 1E and Supporting Fig. 2A). CD1d, necessary for DC induction of NKT cells, was expressed at lower levels on NASH DCs (Supporting Fig. 2A), which correlates with the observed diminution in the fraction of NKT cells in NASH liver (Fig. 1C). The increased maturation of NASH DCs, compared to controls, was also evident after 24 hours of in vitro culture (Supporting Fig. 2B). Besides phenotypic maturation, the fractional subsets of liver DCs were markedly altered in NASH. The B220+ plasmacytoid DC population was decreased in NASH. Conversely, the CD11b+CD8 myeloid DC population expanded by approximately 20%-30%, whereas the fraction of CD11bCD8a+ lymphoid DC decreased proportionately (Supporting Fig. 2C). In contrast to liver DCs, spleen DC phenotype was unaltered in NASH (Supporting Fig. 2D).

NASH DCs Produce Elevated Immune-Modulatory Cytokines

Because secreted cytokines are critical in NASH pathogenesis and DCs can regulate inflammation through production of soluble inflammatory mediators, we tested cytokine production from DCs isolated from NASH liver. NASH DC produced increased levels of TNF-α, IL-6, monocyte chemoattractant protein 1 (MCP-1), and IL-10, compared to normal liver DCs (Fig. 2A,B). NASH DCs also exhibited increased cytokine responses to TLR9 ligation (Fig. 2C). Consistent with these observations, hepatic DCs increased their expression of TLRs in NASH (Fig. 2D).

Figure 2.

NASH DC are proinflammatory. DCs derived from control and NASH liver in C57BL/6 mice at 6 weeks after beginning an MCD diet were tested for production of (A) TNF-α, IL-6, MCP-1, and (B) IL-10 in cell-culture supernatant. (C) DC production of IL-6 and IFN-γ in cell-culture supernatant were analyzed after stimulation with TLR9 ligand CpG ODN1826. (D) DCs from NASH and control liver were analyzed for surface expression of TLR2 and TLR4 and intracellular expression of TLR7 and TLR9. MFIs are shown for each respective TLR. Experiments were repeated at least three times with similar results (*P < 0.05; **P < 0.01; ***P < 0.001).

NASH DCs Differentially Activate CD4+ T Cells

Liver DCs have the capacity to induce either immunogenic responses or tolerance, depending on physiologic circumstance.[15] NASH liver DCs exhibited an increased ability to induce allogeneic T-cell stimulation (Supporting Fig. 3A). Similarly, liver DC capacity to induce antigen-restricted CD4+ T-cell proliferation (Supporting Fig. 3B), as well as CD4+ T-cell production of T-helper cell (Th)1, Th2, and Th17 cytokines, was increased in NASH (Supporting Fig. 3C). NASH DCs also down-regulated expression of the CD25+FoxP3+ regulatory T-cell (Treg) phenotype in DC/T-cell coculture experiments to a greater extent than control DCs (Supporting Fig. 3D). However, DC activation of antigen-restricted CD8+ T cells was unchanged in NASH. In particular, peptide-pulsed control and NASH DCs induced comparable antigen-restricted CD8+ T-cell proliferation (Supporting Fig. 3E) and cytokine production (Supporting Fig. 3F). Similarly, the antigen-specific lytic capacity of hepatic CD8+ T cells against Ova-expressing targets was equivalent after in vivo adoptive transfer immunization using Ova-pulsed control or NASH DCs (Supporting Fig. 3G). Taken together, these data suggest that, in NASH, hepatic DCs gain enhanced capacity to activate CD4+ T cells, but not CD8+ T cells.

DCs Modulate Hepatitis and Fibrosis in NASH

Because DC expand, mature, and gain enhanced capacity to produce inflammatory mediators in NASH, we postulated that DCs may contribute to exacerbation of disease. To test this, we employed BM chimeric CD11c.DTR mice in which continuous DC depletion could be accomplished (Fig. 3A and Supporting Fig. 4). Control mice were made chimeric using BM from WT mice. Surprisingly, ablation of DC populations—rather than mitigating hepatic insult—worsened disease. In particular, NASH(-DC) (NASH with depletion of DCs) mice experienced more precipitous weight loss, compared with NASH mice with intact DC populations (Supporting Fig. 5A). Furthermore, DC depletion in NASH resulted in a larger intrahepatic inflammatory cell infiltrate, compared to controls (Fig. 3B). In addition, analysis of cytokines produced by liver NPC revealed that DC depletion resulted in increased NPC production of numerous cytokines linked to hepatic injury in NASH, including TNF-α, IL-6, and IL-1β (Fig. 3C), as well as chemokines critical for hepatic leukocyte recruitment, including macrophage inflammatory protein 1 alpha (MIP-1α) and granulocyte colony-stimulating factor (G-CSF) (Fig. 3D). Conversely, IL-10, a regulatory cytokine, had decreased expression in NASH liver in the context of DC depletion (Fig. 3E). ALT levels were similarly elevated in NASH and NASH(-DC) liver (Supporting Fig. 5B). DC depletion did not alter hepatic NPC composition (Supporting Fig. 6a-e) or production of inflammatory mediators (Supporting Fig. 6F) in mice on a control diet. DC depletion similarly had no effect on NPC composition in LPS-treated mice on a normal diet (Supporting Fig. 7).

Figure 3.

DC depletion exacerbates inflammation in NASH liver. (A) Splenocytes from NASH and NASH(-DC) mice at 6 weeks after beginning an MCD diet were tested for expression of CD11c. (B) Paraffin-embedded hepatic sections were stained using a monoclonal antibody against CD45. The number of CD45+ cells per HPF was calculated. NPC production of (C) TNF-α, IL-6, IL-1β, (D) MIP-1α, G-CSF, and (E) IL-10 in cell-culture supernatant was determined in control, NASH, and NASH(-DC) liver. Experiments were repeated at least three times using 3-5 mice per group (*P < 0.05; **P < 0.01; ***P < 0.001).

Intrahepatic inflammation has a reciprocal pathogenic relationship with cellular apoptosis in NASH liver.[16] Consistent with elevated intrahepatic inflammation, NASH(-DC) liver exhibited the increased presence of apoptotic bodies (Fig. 4A). Accordingly, expression of PAR4, a marker of apoptosis, was increased in NASH liver in the context of DC depletion (Fig. 4B). Cleaved caspase-3 was also more prevalent in NASH(-DC) liver, compared to controls (Fig. 4C). Furthermore, p53, Fas ligand (FasL), and B-cell lymphoma 2 (Bcl-2), well-described mediators of apoptosis in NASH,[17, 18] exhibited markedly elevated expression in NASH(-DC) liver (Fig. 4D).

Figure 4.

DC depletion exacerbates apoptosis and fibrosis in NASH liver. (A) Terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) staining was performed in control, NASH, and NASH(-DC) liver from mice sacrificed 6 weeks after beginning an MCD diet. The number of apoptotic bodies per high-power field was quantified. (B) Lysates from control, NASH, and NASH(-DC) liver were probed for expression of PAR4. (c) Cleaved caspase-3 staining was performed on paraffin sections from control, NASH, and NASH(-DC) liver and results were quantified. (D) Messenger RNA from each group was tested for p53, FasL, and Bcl-2 by polymerase chain reaction. (E) Picric acid sirius red (PASR) staining was performed in control, NASH, and NASH(-DC) liver. Fibrosis was quantified by examining 10 high-power fields per liver. (F) TGF-β, Collagen Iα1, and MMP9 expression were determined by polymerase chain reaction. Experiments were repeated at least three times with similar results (*P < 0.05; **P < 0.01; ***P < 0.001).

Besides augmented intrahepatic inflammation and an increased prevalence of apoptosis, NASH(-DC) mice exhibited accelerated hepatic fibrosis (Fig. 4e). Accordingly, transforming growth factor beta (TGF-β) and Collagen Iα1 (Figure 4f) as well as tissue inhibitor of metalloproteinase 1 (TIMP-1) (not shown) were more highly expressed in NASH(-DC) liver, compared to controls. Matrix metallopeptidase 9 (MMP9), which is associated with extracellular matrix remodeling, was similarly increased in NASH(-DC) liver (Fig. 4F). Taken together, these data imply that the absence of DCs in NASH leads to exacerbated intrahepatic fibroinflammation.

Ablation of DC Populations Augments Effector Cell Expansion and Activation in NASH

To better understand the mechanism for exacerbated hepatitis in NASH(-DC) liver, we investigated whether ablation of DC populations was associated with a compensatory expansion or activation of specific effector cell subsets linked to disease pathogenesis. We found that there was a large fractional increase in neutrophils, inflammatory monocytes, and KCs upon DC depletion in NASH (Fig. 5A). Immunohistochemical (IHC) staining confirmed an increase in total number of neutrophils (Fig. 5B) and KCs (Fig. 5C) in NASH(-DC) liver. Conversely, the fractional decrease in NK1.1+ cells in NASH was unchanged upon DC depletion (Fig. 5A). CD8+ T cells have also been implicated in intrahepatic inflammation, whereas the expansion of FoxP3+ Tregs has been associated with mitigation of hepatic injury.[19, 20] We found that DC depletion resulted in markedly greater skewing of the intrahepatic CD8/CD4 ratio and diminished accumulation of Tregs in NASH (Fig. 5a). Similar observations were made when examining the total numbers of leukocyte subsets in NASH(-DC), compared to NASH liver (Supporting Fig. 8). Taken together, these data imply that DCs may limit hepatic injury in NASH by regulating the expansion of innate and adaptive immune cellular subsets. Consistent with these observations, we further found that there was a decrease in Annexin V+ apoptotic KCs, neutrophils, and monocytes in NASH(-DC) liver (Fig. 5d-f), suggesting that DCs may limit effector cell expansion in NASH by inducing apoptosis of innate effector cells, as we have previously described in acute liver injury.[21] DC depletion in CD11c.DTR chimeric mice did not appreciably alter splenocyte composition in NASH or in inflammation induced by LPS, suggesting the effects are specific to the role of DC in NASH liver (Supporting Fig. 9A,B).

Figure 5.

DC depletion increases KC and granulocyte proliferation in NASH liver. (A) Gr1, F4/80, and NK1.1 expression in bulk liver NPC, coexpression of CD4 and CD8 on hepatic CD3+ T cells, and expression of FoxP3 on CD4+CD25+ T cells were analyzed in control, NASH, and NASH(-DC) liver from mice sacrificed 6 weeks after initiating an MCD diet. (B) MPO and (C) CD68 IHC were performed on paraffin-embedded liver sections. The number of positive cells per high-power field was quantified. (D-F) Fraction of Annexin V+ apoptotic (D) KCs, (E) neutrophils, and (F) inflammatory monocytes in control, NASH, and NASH(-DC) liver were determined by flow cytometry. Experiments were repeated three times with similar results (*P < 0.05, **P < 0.01, ***P < 0.001).

To investigate whether DCs regulate effector cell activation—in addition to expansion—in NASH, we harvested KCs, neutrophils, and inflammatory monocytes from NASH(-DC) mice and controls and measured their expression of intracellular cytokines implicated in disease pathogenesis.[4, 5] We found that the absence of DCs resulted in markedly higher production of TNF-α and IL-1β by KCs, neutrophils, and inflammatory monocytes in NASH liver (Fig. 6A-C). IL-6 was also up-regulated by these cellular subsets in NASH(-DC) liver (not shown). Furthermore, because the pathogenesis and severity of NASH has been linked to TLR4 and TLR9 activation of KCs,[23, 24] we tested whether ablation of DC populations results in up-regulation of KC expression of TLRs. We found that KCs from NASH(-DC) liver exhibited markedly elevated TLR9 expression (Fig. 6D). IHC staining confirmed increased TLR9 expression in NASH(-DC) liver (Fig. 6E). TLR4 was similarly up-regulated on KCs and liver tissues in NASH(-DC) mice (Fig. 6F). Taken together, these data imply that DC depletion results in activation of innate immune cells in NASH.

Figure 6.

DC depletion increases activation of KCs, neutrophils, and inflammatory monocytes in NASH liver. (A-C) Intracellular expression of TNF-α and pro-IL-1β in freshly isolated (A) F4/80+ KCs, (B) Gr1HiCD11b+ neutrophils, and (C) Gr1IntCD11b+ inflammatory monocytes from control, NASH, and NASH(-DC) liver is shown. (D) Expression of TLR9 in KCs was determined by flow cytometry. Mean Fluorescence intensity is indicated, and the fraction TLR9+ KCs was quantified for each group. (E) Paraffin-embedded hepatic sections were stained using a monoclonal antibody directed against TLR9. The number of positive cells per high-power field was quantified. (F) F4/80+ KC expression of TLR4 on was determined by flow cytometry and whole liver tissue expression level of TLR4 was determined by polymerase chain reaction in control, NASH, and NASH(-DC) groups. In all experiments, mice were sacrificed 6 weeks after initiating an MCD diet. Experiments were repeated at least three times with similar results (**P < 0.01, ***P < 0.001).

DCs Limit Sterile Inflammation in NASH

Because DCs have recently been implicated in the clearance of dead cells in other contexts,[11, 22] and a pathogenic role for sterile inflammation is emerging in NASH,[23] we postulated that—in the absence of DCs—delayed the clearance of apoptotic cells and necrotic debris results in augmentation of sterile inflammation within the liver, precipitating effector cell proliferation and activation. Augmented sterile inflammation in the hepatic microenvironment is supported by our observation of increased apoptotic bodies and mediators of apoptosis in NASH(-DC) liver (Fig. 4A-D). Additionally, levels of high-mobility group box 1 (HMGB1), a marker of sterile inflammation, were elevated in NASH(-DC) liver, compared to controls (Supporting Fig. 10A). We also found that—compared with other hepatic APCs—liver DCs express high levels of C-type lectin domain family 9 member A (CLEC9A) (Supporting Fig. 10B), a type II membrane protein with an extracellular C-type lectin domain, which is essential for DC recognition and clearance of necrotic cells.[26, 27] To directly test whether hepatic DCs are vital to the clearance of necrotic debris in NASH liver, we compared in vivo uptake of exogenously administered 7-amino-actinomycin-positive necrotic cells by CD11c+MHCII+ liver DCs, compared with other MHCII+ APC subsets. We found that DCs achieved greater capture of necrotic elements in vivo (Supporting Fig. 10C). Consistent with these observations, DCs from NASH liver also captured necrotic debris in vitro at a higher rate than other APC subsets (Supporting Fig. 10D). Furthermore, in NASH, DCs acquired greater capacity for necrotic cellular clearance, compared to DCs from control liver (Supporting Fig. 10E). We also tested DC capacity to clear apoptotic bodies in NASH. We found that NASH DCs captured Annexin V+ apoptotic cells in vivo at higher rates, compared with other MHCII+ APC subsets (Supporting Fig. 10F). Furthermore, NASH DCs captured apoptotic bodies at modestly higher rates than DCs from control liver (Supporting Fig. 10F). Taken together, these data suggest that DCs may limit sterile inflammation in NASH by their clearance of necrotic cellular debris and apoptotic bodies, whereas absence of DCs leaves the diseased liver with APCs less equipped for this task. Notably, uptake of apoptotic or necrotic cellular debris by KCs was unchanged in NASH(-DC) liver, compared with NASH (Supporting Fig. 10G).

DC Depletion Delays Recovery From NASH

NASH is reversible in its early stages. However, the role of DCs in the recovery phase of disease is unknown. To investigate this, WT chimeric or CD11c-DTR chimeric mice fed an MCD diet for 6 weeks were abruptly transitioned to normal chow. In selected cohorts, DCs were depleted beginning at the time of cessation of the MCD diet. Consistent with our data implicating a protective role for DCs in NASH, DC depletion delayed the resolution of NASH (Fig. 7A). In particular, absence of DCs on day 3 of normal diet resumption markedly delayed clearance of the intrahepatic CD45+ leukocytic infiltrate (Fig. 7B), neutrophilic infiltrate (Fig. 7C), and apoptotic bodies (Fig. 7D). Residual fibroplasia was also conspicuously more pronounced in mice depleted of DCs during the recovery period (Fig. 7E). Furthermore, DC depletion delayed resolution of the elevated cytokine and chemokine secretion by NASH NPC (Fig. 7F). Similar differences between control and DC-depleted mice were noted on day 7 of NASH recovery. However, by 14 days, there was complete resolution of NASH, even in mice depleted of DCs (not shown). Taken together, these data imply that DCs facilitate recovery from NASH.

Figure 7.

DC depletion delays resolution of fibroinflammation during recovery from NASH. (A) Representative hematoxylin and eosin (H&E) staining is shown in livers of mice sacrificed 3 days after of cessation of an MCD diet (Recovery) and in mice where DCs were depleted coincident with MCD diet cessation (Recovery(-DC)). (B) CD45, (C) MPO, and (D) cleaved caspase-3 expression were tested by IHC in Recovery and Recovery(-DC) liver. (E) Picric acid sirius red (PASR) staining was performed in Recovery and Recovery(-DC) liver. Data were quantified by examining 10 high-power fields per mouse. (f) NPC production of IL-6, TNF-α, and MCP-1 in cell-culture supernatant were measured in each group (n = 3-5 mice/group: *P < 0.05; **P < 0.01; ***P < 0.001).


This is the first investigation to report a significant role for hepatic DCs in NASH. We demonstrated that DCs are recruited to the liver soon after MCD diet initiation, plateau at 3-4 times normal levels by 2 weeks, and remain at an elevated level, unless there is disease resolution. NASH DCs exhibit an activated surface phenotype and increase their production of proinflammatory cytokines. Consistent with their mature phenotype, our in vitro experimentation shows that NASH DCs potently induce proliferation of both allogeneic T cells and antigen-restricted CD4+ T cells while reducing CD4+ T-cell expression of the CD25+FoxP3+ Treg phenotype. The finding of intrahepatic DC activation after hepatic insult is consistent with our previous reports showing immunogenic transformation of liver DCs in thioacetamide-induced liver fibrosis and acute hepatic injury induced by acetaminophen overdose.[12, 21] However, despite their phenotypic and functional activation, ablation of DC in NASH results in increased hepatic inflammation, diminished numbers of Tregs, expansion of CD8+ T cells, enhanced viability and production of proinflammatory cytokines by immune effector cells, increased hepatocyte apoptosis, and, ultimately, accelerated liver fibrosis. These ostensibly paradoxical findings are not entirely unprecedented. Recent studies have shown that, despite adopting a proinflammatory phenotype, hepatic DCs can accelerate the regression of hepatic fibrosis and ameliorate hepatic ischemia-reperfusion (I/R) injury.[13, 28] For example, exogenous expansion of hepatic DC populations by Flt3 ligand administration accelerates the regression of CCl4-induced liver fibrosis, despite phenotypic activation of DCs.[13]

Our findings show that effects of DC contrast sharply with the role of KCs, whose expansion have been strongly linked to worsening intrahepatic fibroinflammation in NASH.[5] Our investigations suggest multiple parallel mechanisms by which DC may regulate hepatitis. Importantly, we found that DCs in NASH liver are differentially capable of activating CD4+ T cells, in comparison with CD8+ T cells. Furthermore, upon DC depletion, the CD8/CD4 T-cell ratio is skewed markedly upward with associated diminution of Tregs. The protective role of Tregs in CLD is well established.[29, 30] Furthermore, relative suppression of CD8+ T-cell expansion may be protective, because CD8+ T cells have recently been shown to drive adipose tissue inflammation and have an emerging role in NASH pathogenesis.[31, 32]

Additionally, the exacerbated hepatic insult associated with ablation of DC populations may be mechanistically related to the DC's role in limiting sterile inflammation through clearance of apoptotic bodies and necrotic debris. Sterile inflammation in the liver increases recruitment, viability, and activation of innate immune cells.[33] We show that liver DCs express high CLEC9A, which recognizes and binds death signals on necrotic cells and is primary in DC capacity to clear necrotic products.[26, 27] Accordingly, we found that NASH liver DCs have remarkable capacity to capture necrotic cellular debris and apoptotic targets, when compared to other hepatic APC subsets and DCs from control liver. Furthermore, we found that DC depletion leads to an accentuation of sterile inflammation within the liver, because NASH(-DC) liver contains modestly higher HMGB1 and elevated markers of apoptosis, including p53, which has been demonstrated to play a pivotal role as a mediator of apoptosis in experimental NASH.[17] This also results in augmented production of proinflammatory cytokines—including IL-1β, TNF-α, and IL-6—and enhanced viability and expression of TLR4 and TLR9 in innate effector cells. Miura et al. demonstrated that signaling through TLR9 leads to progression of NASH by KC production of IL-1β.[4] TLR4 signaling in KCs has also been linked to severity of steatohepatitis.[6]

DC production of IL-10 may also have an important role in limiting hepatic damage in NASH. Bamboat et al. recently showed that DNA released from apoptotic hepatocytes stimulates liver DC to secrete IL-10 in a TLR9-dependent manner.[28] Furthermore, IL-10 derived from hepatic DCs can ameliorate liver injury through suppression of inflammatory monocyte function.[28] Additional studies in contexts such as allergen-induced asthma and cisplatin-induced nephrotoxicity have shown that DCs attenuate sterile inflammation through release of IL-10.[34, 35] We found that NASH DCs exhibited markedly elevated IL-10 production, compared to normal liver DCs. Moreover, IL-10 production by NASH NPC was reduced by 40%-60% in the absence of DCs, suggesting that DC production of IL-10 may have an important regulatory role in NASH.

Interestingly, despite showing varied evidence of increased inflammation, fibrosis, hepatocyte apoptosis, and delayed recovery from NASH upon DC depletion, we did not find significant elevations in serum ALT in NASH(-DC), compared with NASH mice with intact DC populations. However, this is consistent with previous reports showing that the severity of NASH may not correlate with serum ALT levels.[36, 37] Furthermore, clinically severe NASH can exist without overt elevations in serum ALT.[38] These studies suggest that ALT alone cannot be used as a “hard endpoint” in NASH.

Numerous studies have used the CD11c.DTR model to investigate the role of DCs in diverse inflammatory conditions within the liver, including I/R injury and acute acetaminophen hepatotoxicity.[21, 28] Similarly, the CD11c.DTR model has been useful in determining the role of DCs in many extrahepatic diseases, including allergic asthma, acute lung injury, pancreatitis, and renal I/R injury.[11, 28, 39] However, a sobering report by Tittel et al. recently showed that DC depletion in CD11c.DTR mice is associated with an early nonspecific neutrophilia in multiple organs, including a modest neutrophilia within the liver, implying that conclusions drawn using the CD11c.DTR model may be confounded by nonspecific effects.[42] The mechanism for the reported neutrophilia in CD11c.DTR mice depleted of DCs remains uncertain. However, we did not observe unintended changes in leukocyte composition in BM chimeric CD11c.DTR mice upon DC depletion that were independent of NASH. Possible explanations for our disparate results may be that the BM chimeric CD11c.DTR model is not affected by the neutrophilia associated with the endogenous model. Endogenous CD11c.DTR mice are distinct from the chimeric model in that repeated administration of diphtheria toxin is lethal. Furthermore, chronic DC depletion as in NASH may not cause the neutrophilia associated with acute single-dose depletion. Nevertheless, the CD11c.DTR model, though it is the best available tool to study the role of DCs in vivo in mice, is not the perfect model because the effects of DC depletion may not necessarily faithfully mimic the role of DC in situ. Thus, additional insight on the role of DCs in NASH and other inflammatory diseases may be forthcoming pending the advent of additional experimental tools to study DC effects in vivo.

In summary, our data suggest that DCs have complex influences on both the pathogenesis and resolution of steatohepatitis, which may have implications to human disease. However, a limitation of our study is that there is no perfect murine model of NASH mimicking human disease. Additionally, direct comparison of the current data, even to other murine studies of NASH employing an MCD diet, may be confounded by alternate durations of treatment between studies. That is, because the development of NASH, as well as its resolution, is a dynamic process, examining the intrahepatic phenotype and immune milieu after varied durations of feeding mice an MCD diet may yield inconsistent findings. Furthermore, interrupting the pathogenesis of NASH by targeting DCs in experimental therapeutics may prove challenging, given the technical limitations in modulating human DC function in vivo. Thus, additional investigations are needed to evaluate the clinical utility of these findings in treating patients with NASH or preventing disease onset.