Hepatocytes as Immunological Agents

Hepatocytes are the most abundant cell type in the liver, accounting for perhaps 90% of the biomass. When these cells are infected, whether by a hepatotropic virus, a parasite, or an intracellular bacterium, a complex local response involves diverse immunologically active cells that are found in the nonparenchymal fraction: liver sinusoidal endothelial cells, Kupffer cells, and hepatic stellate cells, as well as trafficking monocytes, dendritic cells (DCs), NK cells, NKT cells, and diverse varieties of CD4⁺ and CD8⁺ T cells. Much recent research clarified the roles of each of these cell types in antiviral, antimalarial, and antibacterial immunity, but hepatocytes are not passive recipients of immune signals from these surrounding cells. The purpose of this review is to shift the immunological focus to hepatocytes and emphasize their active role in both innate and adaptive immunity to many hepatic insults. Hepatocytes are not simply target cells but are immunological agents in their own right.

Hepatocytes are relatively tractable experimentally. In mice, they can be manipulated using a transgenic approach, because the albumin promoter results in highly cell type–specific transgene expression. Hepatocyte-like immortalized cell lines reproduce many features of primary hepatocyte activity. In addition, well-defined protocols exist for the isolation of hepatocytes from mouse liver, through perfusion of the organ with collagenase, followed by selective depletion of the nonparenchymal fraction. In the human this task is more daunting, but the logistical hurdles can be overcome, and primary human hepatocytes are widely available from several sources. Therefore, our understanding of the immunology of hepatocytes comes from in vivo experiments in mice and ex vivo experiments with murine and human cell lines and primary cells.

Hepatocytes are central in the systemic innate immune response to sepsis, which is extensively conserved in vertebrates from mammals to bony fish (1). In the acute-phase response, either generalized or localized infection increases the circulating levels of several key cytokines, including IL-1α, TNF-α, and IL-6 (Fig. 1). These mediators result in coordinated changes in the transcriptional activity of hepatocytes, with the secretion of molecules that limit tissue injury and molecules that participate in host defense (2). Among the cytokines that drive the acute-phase response, IL-6 acts directly on hepatocytes. In human hepatocytes, rIL-6 induced a classic pattern of acute-phase proteins, including serum amyloid A, C-reactive protein, haptoglobin, α1-antichymotrypsin, and fibrinogen. Conversely, the normal hepatocyte secretion products albumin, transferrin, and fibronectin were reduced (3). At least some of the IL-6–induced effects appear to be hepatoprotective, because elimination of the gp130 component of the IL-6R in transgenic mice attenuated the acute-phase response and led to an increase in liver injury in hepatitis induced by bacterial endotoxin (4). Many of the acute-phase proteins are counterregulated by hepatocyte growth factor (HGF), which plays a key role in liver growth and regeneration; however, in primary human hepatocytes, HGF also suppressed α1-antichymotrypsin and haptoglobin while increasing the synthesis of albumin, transferrin, and fibronectin (5). The coordinated regulation of acute-phase protein genes is regulated, in part, by microRNAs that interact with the conserved 3′ untranslated regions of those genes that are differentially expressed during the acute-phase response (6).

New members of the IL-6–induced class of host defense proteins continue to be identified. Recently, hepatocytes were identified to be the important source of the antibacterial protein lipocalin-2 (7). Although exogenous IL-6 can drive the acute-phase response, hepatocytes also synthesize IL-6 in response to diverse stimuli, including HGF (8), giving this molecule a role in activating and in suppressing the acute phase. The acute-phase response is not monolithic; variants of this response have been identified during ischemia-reperfusion injury (9, 10), in liver regeneration following partial hepatectomy (11), during abdominal surgery (12), and in noninfected bone fractures (13).

This brief introduction to a complex topic makes two key points: that the liver makes a major contribution to systemic innate immunity and that hepatocytes are both sensors of such injury and the principal source of the protective proteins thus induced.

Hepatocytes express a wide variety of pattern recognition receptors, including cell surface receptors (e.g., TLR4), endosomal receptors (e.g., TLR3), and cytoplasmic receptors (e.g., stimulator of IFN genes [STING], retinoic acid inducible gene-1 (RIG-1), and nucleotide-binding oligomerization domain (NOD) family members) (14). Isolated hepatocytes respond directly to microbial products (15) and to endogenous signals, such as heat shock protein 72, via their TLR2 and TLR4 receptors (16). The expression of these receptors may be differentially regulated. For example, diverse stimuli upregulate the expression of TLR2, but not TLR4, on hepatocytes (17). TLR2 may mediate liver-protective signals, for example in parasite infection (18). Conversely, in liver injury mediated by a TLR9 ligand, IRF7 and type 1 IFN signaling via the IFN-αR ameliorate liver injury through the action of IL-1ra, an antagonist of IL-1R signaling (19).

The innate immune response of hepatocytes to a major human pathogen, hepatitis C virus (HCV), is relatively well understood. Viral RNA engages the RIG-1 system, resulting in signaling via MAVS that activates the transcription factor IRF3 and its target, type 1 IFN. In parallel, dsRNA engages TLR3, signaling via the TRIF adapter protein to activate IRF3. It is clear that both of these signaling pathways are important in host defense because the virus has evolved a protease, NS3/4a, which can cleave both TRIF and MAVS, disabling IRF3 activation (20, 21). Therefore, it came as a surprise that primary human hepatocytes infected with HCV strain JFH1 were able to synthesize multiple cytokines and chemokines, whereas hepatocyte-like Huh7 cells secreted very few, principally IL-28 (IFN-λ). However, efforts to remove the trace numbers of nonparenchymal cells from the primary hepatocyte culture reduced the secretion of many of these chemokines and cytokines (22). Therefore, in the presence of viral immune-subversion mechanisms, the cell-autonomous self-defense capacity of hepatocytes is curtailed; however, some pathways remain active, whereas others are the task of nonparenchymal cells.

Hepatocytes also harbor the capacity to mount cell-autonomous defense against hepatitis B virus (HBV), but just as HCV has evolved ways to disable cytoplasmic RNA sensing, HBV acts against cytoplasmic DNA sensing. Specifically, the viral polymerase interacts directly with the STING DNA sensor, inhibiting its function (23). Conversely, direct stimulation of STING using a synthetic agonist resulted in cytokines dominated by type 1 IFN and an antiviral effect that suppressed HBV replication (24). Thus, understanding hepatocyte innate immune pathways may lead to novel approaches to antiviral therapy.

Cells may respond to insult by undergoing programmed cell death, of which the principal recognized forms are apoptosis (initiated by caspases), necroptosis (initiated by RIP kinases), and pyroptosis (initiated by activation of the inflammasome). Hepatocyte-resident innate immune pathways can contribute to cell death in immunopathology that occurs in the absence of a hepatocellular pathogen. Thus, ethanol-associated liver injury depends on TLR4, IRF3, and STING (25, 26) to trigger the mitochondrial pathway of apoptosis via Bax. Hepatocytes may undergo different varieties of programmed cell death in response to diverse insults. Thus, ethanol-induced liver injury also involves necroptosis mediated via RIPK3 (27). Similarly, inhibition of RIPK1 protects mice from acetaminophen-induced injury (28), but acetaminophen also acts via TLR9 and the NALP3 inflammasome, implicating pyroptosis. This led to the striking result that aspirin, an inhibitor of inflammasome activation, suppressed acetaminophen toxicity (29).

Hepatocellular cancer (HCC) frequently arises in the context of chronic liver inflammation, but the links between the inflammation and the cancer are complex. Innate immune responses of hepatocytes modify the response to carcinogenesis. In a mouse model of HCC, TLR2 expressed on hepatocytes suppressed the expression of IL-18, whereas IL-18 was responsible for the accumulation of myeloid-derived suppressor cells (MDSCs). Thus, in the absence of this signaling pathway, HCC growth was enhanced; when a TLR2 agonist was administered, HCC was suppressed (30). Both TLR3 and TLR7 are underexpressed in hepatoma cells, and there exists a hepatocyte-specific regulatory region in the TLR-7 promoter (31). Further, IL-7 expression in hepatoma cells was suppressed by IFN-γ. The TLR2–IL-18 and IFN-γ–TLR7 pathways establish links between liver inflammation and HCC.

Therefore, hepatocytes are active participants in cell-autonomous innate immune responses that may suppress pathogens but also result in immunopathology, including the activation of cell death pathways. In HCC, some aspects of innate immunity seem to be harmful (IFN-γ, IL-18), but other pathways enhance anticancer immunity (TLR2, and perhaps TLR7).

The major circulatory input to the liver is the portal vein; therefore, innate immune signals from the intestinal microbiota impinge strongly on hepatic biology. One clear example is the impact of LPS endotoxin from Gram-negative bacteria, which supports the capacity of the liver to sequester activated and apoptotic T cells from the circulation (32). In addition, the intestinal microbiota has been implicated in liver disease. Specifically, changes in the microbiota impact diverse liver pathologies, including diet-induced liver inflammation (33), autoimmune hepatitis (34), antiviral immunity (35), fibrosis (36), and even liver cancer (37). Given the wide array of innate immune receptors expressed on hepatocytes, it is likely that they play a central role in these responses; however, information is limited, and this is an important topic for future research.

In addition to cell-autonomous responses and the systemic acute-phase response, hepatocytes interact locally with innate immune cells. Thus, hepatocytes express the NK cell targets MIC-A and MIC-B; this is biologically significant in host defense, because HBsAg modulates their expression through the induction of cellular microRNAs (38). Similarly, the canonical NKT cell restriction element, CD1d, is expressed on hepatocytes and upregulated in chronic HCV infection (39). In mice, CD1d promotes liver injury and fibrosis, arguing for an important role for the hepatocyte–NKT cell axis in immunopathology (40). Finally, hepatocytes may modulate innate immune cells through the acute-phase response. Specifically, the gp130–STAT3 signaling axis promotes recruitment of immature myeloid cells (MDSCs) that regulate innate immune inflammation (41). In a further link between the acute-phase response and innate immunity, MDSCs can be induced by HGF (42).

In addition to innate signals from the intestinal microbiota, the liver encounters a wide range of Ags among the products of digestion, and it creates more through hepatocellular metabolism. Therefore, it is clear that a bias toward immune tolerance to hepatocellular Ags serves a valuable purpose in preventing futile immunity to nonhazardous molecules.

Immune tolerance associated with Ags present in the liver may be due, in part, to the ambiguous relationship of hepatocytes to T cells. The structure of the liver sinusoidal endothelium features small openings (fenestrae) grouped into sieve plates, and these openings permit direct interaction between hepatocytes and the blood space (43). Because T cells do not need to extravasate to make contact with hepatocytes, the opportunities for hepatocytes to talk to naive T cells are enhanced. The outcome of such a conversation is usually to the T cell’s disadvantage. Ex vivo, Ag-specific CD8⁺ T cells directly confronted with antigenic hepatocytes undergo initial activation (44, 45), followed by loss of function and premature cell death. This might be due to the lack of either costimulatory signals or CD4⁺ T cell help, because it can be partially reversed with exogenous IL-2 (46). More dramatically, the hepatocytes may endocytose and kill CD8⁺ T cells that recognize them, a process known as suicidal emperipolesis (47). Because hepatocytes express MHC class I, and not MHC class II, CD4⁺ T cells are not impacted directly by these hepatocyte-based tolerance mechanisms.

These mechanisms may act primarily to abort the activation of naive CD8⁺ T cells. In vivo, introduction of naive Ag-specific CD8⁺ T cells results in T cell proliferation and hepatitis; however, in general, the Ag persists, accompanied by T cell clonal exhaustion (48, 49). The level of Ag expression modulates this, and experiments that restrict Ag to rare hepatocytes may result in strong immunity if high-affinity CD8⁺ T cells are present (50); this cannot be a general conclusion because malaria parasites, which infect very few hepatocytes, normally induce little liver-stage–specific immunity. However, the introduction of Ag-specific CD8⁺ T cells into a transgenic mouse that expresses hepatocyte Ags readily results in inflammation accompanied by hepatocyte death. This is most clearly seen in mice expressing HBV as a hepatocellular transgene (51). Therefore, however immune failure arises, it does not necessarily protect the liver from injury.

The direct access of naive CD8⁺ T cells to hepatocellular Ags via fenestrae raises the question of how such CD8⁺ T cells might interact with CD4⁺ T cell help. A major pathway by which CD4⁺ and CD8⁺ T cells interact is the licensing mechanism whereby CD4⁺ and CD8⁺ T cells both interact with DCs. The CD4⁺ T cell, acting via CD40, induces changes in the DCs, including increased expression of costimulatory molecules, such as CD80, CD86, and CD70, all of which facilitate optimum CD8⁺ T cell activation (52). However, if CD8⁺ T cells receive their primary activation signals from hepatocytes, they are engaging a cell that lacks MHC class II expression and so cannot be licensed. It follows that, for the delivery of efficient CD4⁺ T cell help to a CD8⁺ T cell for a hepatocellular Ag, that Ag must be cross-presented by a different cell type (53).

Hepatocytes themselves may impose a barrier to such cross-presentation. The process is mediated by secreted Ags, Ag-carrying exosomes, and apoptotic fragments derived from dead and dying cells. Such fragments are readily taken up by neighboring hepatocytes (e.g., via the asialoglycoprotein receptor) (54). This may explain why, in a radiation bone marrow chimera, there was no evidence of cross-presentation by bone marrow–derived cells of hepatocellular Ags encoded by an adeno-associated virus vector (55). Ags acquired from dying hepatocytes and cross-presented by their neighbors would be subject to the same constraints as Ags in their hepatocyte of origin: CD8⁺ T cell engagement by an APC that cannot receive licensing signals.

Hepatocytes are active participants in innate immune responses. They are central drivers of the systemic acute-phase response and can respond to diverse insults, including stress and infection, resulting in either host defense or programmed cell death. In CD8⁺ T cell immunity, they mostly promote tolerance, but this is not a passive process; instead, hepatocytes abortively activate and may actively kill CD8⁺ T cells that recognize them, and their inherent capacity to take up hepatocyte debris effectively subverts the mechanisms of cross-presentation that are essential for the delivery of CD4⁺ T cell help.

I.N.C. received a research grant from Janssen Sciences to study liver immunology in human cells, including hepatocytes. This has had no influence on the content of this Brief Review.