Liver Immunology

Abstract

The liver is the largest organ in the body and is generally regarded by non-immunologists as not having lymphoid function. However, such is far from accurate. This review highlights the importance of the liver as a lymphoid organ. Firstly, we discuss experimental data surrounding the role of liver as a lymphoid organ. The liver facilitates a tolerance rather than immunoreactivity, which protects the host from antigenic overload of dietary components and drugs derived from the gut and is also instrumental to fetal immune tolerance. Loss of liver tolerance leads to autoaggressive phenomena which if are not controlled by regulatory lymphoid populations may lead to the induction of autoimmune liver diseases. Liver-related lymphoid subpopulations also act as critical antigen-presenting cells. The study of the immunological properties of liver and delineation of the microenvironment of the intrahepatic milieu in normal and diseased livers provides a platform to understand the hierarchy of a series of detrimental events which lead to immune-mediated destruction of the liver and the rejection of liver allografts. The majority of emphasis within this review will be on the normal mononuclear cell composition of the liver. However, within this context, we will discus select, but not all, immune mediated liver disease and attempt to place these data in the context of human autoimmunity.

Introduction

The liver is the largest solid organ of the human body, accounting for almost 2% of adult body weight and weighing approximately 1.5 kg; it performs an amazing number of tasks that support the function of other organs and impacts all physiologic systems. An essential function of the liver is protein synthesis and metabolism, including the metabolism of amino acids, carbohydrates, lipids and vitamins. However, the liver is also responsible for the removal of pathogens and exogenous antigens from the systemic circulation. The key position of the liver (Figure 1) and its unique vasculature allow it to carry out the degradation of toxins and waste products.

Figure 1.

Anatomical location and external appearance of the liver. The falciform ligament, on the surface of the diaphragm, splits the liver into right and left lobe. The anatomical relationship of the liver with organs such as the gallbladder, stomach, duodenum, and pancreas is illustrated.

The role of the liver as the main metabolic organ increases the rate of exposure to newly produced neo-antigens and enhances the inherited risk of overactivation of components of the immune system with potentially harmful consequences for cell homeostasis. Thus, the immune system developed dedicated mechanisms to be able to “switch” from a tolerant to a responsive state at any given time¹. Early in the history of experimental transplantation, transplant surgeons were intrigued to note that while kidney, skin, pancreas and other allografts were rapidly rejected, allogeneic liver grafts were more tolerant. This prompted investigators to consider that the liver is predominantly an organ biased towards tolerance rather than a reactive state which would otherwise lead to rejections. The scientific basis for this tolerant state remained elusive for many years.

Historical Perspectives

The first human liver transplant was performed in 1963 by Starzl,² with significant short-term success in 1967 when a recipient survived for more than a year. Subsequently, the systematic administration of cyclosporin by Calne and colleagues dramatically improved the outcome of the patients receiving a liver allograft^{3, 4}. Further, it was noted in 1969 that liver allografts between unrelated pigs were not rejected in spite of MHC mismatch; the transplanted livers did not require high doses of immunosuppression to be sustained⁵. A seminal study, conducted two years before Calne’s report, from Cantor and Dumont demonstrated that administration of antigens to animals via the portal vein was tolerated better compared to systemic administration⁶. Subsequent studies confirmed the potential acceptance of MHC mismatched liver grafts in other species. Further, liver transplantation confers tolerance to heart and skin grafts from the same donors, while heart and skin grafts from other donors were immediately rejected. Interestingly, the rejection of other transplanted organs can be modulated by subsequent transplantation. Similarly, co-transplantation of human liver with another organ limits the likelihood of immediate rejection of the second organ and improves the survival of the allograft. The natural regenerative capacity of the liver parenchymal cells is significant; 25% of residual liver is sufficient for regeneration within a few weeks in rodents and a few months in humans. Because of its anatomical location, the liver is continuously exposed to an overload of antigenic stimuli which includes exogenous pathogens, dietary components and xenobiotics, including drugs and toxins.

Microanatomy of the Liver as an Immunological Organ

To achieve its multifaceted tasks, the liver is composed of a myriad of cell types, largely sub-divided in parenchymal and non-parenchymal cells (Table 1)⁷. Most of the liver volume is occupied by parenchymal cells (hepatocytes); these cells occupy approximately 78–80% of the total liver tissue, compared to just 5–6% of non-parenchymal cells^7–11 (Table 1). The remaining 14–17% of the total liver tissue corresponds to cellular components of the extracellular space (Figure 2)⁷. The non-parenchymal cells consist of a diverse set of cells, including 45% liver sinusoidal endothelial cells (LSECs), 33% Kupffer cells (KCs), and 22% hepatic stellate cells (HSCs)¹² (Table 1 and Figure 3). The liver can be considered to have two separate anatomic areas, the parenchyma and the portal tracts. Structurally, the liver can be further subdivided into five systems comprising the vascular system, the hepatic lobule, the hepatic sinusoidal system, the biliary system and the stroma. Each of these systems - directly or indirectly - plays an important role in the homeostasis of the innate and adaptive immune system.

Table 1.

Percentage of total volume of cellular and extracellular compartments in liver⁷.

	Mean %	±SE
Cells (84.1%)
Hepatocytes	77.8	1.15
Liver Sinusoidal Endothelial Cells	2.8	0.2
Kupffer Cells	2.1	0.3
Hepatic Stellate Cells	1.4	0.2
Extracellular Spaces (15.9%)
Sinusoidal lumen	10.6	0.45
Disse space	4.9	0.35
Biliary canaliculi	0.4	0.05
Total Sum (100%)	100%

Data are presented as mean% ± standard errors (SE) of the mean

Figure 2.

Cellular and extracellular composition of the liver

Figure 3.

The morphological appearance of cells within the liver.

Hepatic lobule

The simplest way to describe the cellular anatomy of the liver is by light microscopy. Thus, the hepatic lobule is not only the structural but also the functional unit of the liver¹³. These lobules are centered on central veins like spokes in wheel, and their periphery is demarcated by arbitrary lines joining each of the surrounding regions of portal tracts (Figure 4). Each portal tract consists of an intrahepatic bile duct and a collection of blood vessels including a branch of both the hepatic artery and portal vein. Such lining formulates a roughly hexagonal assembly of hepatocyte plates¹³, the extension of which forms the basis of the one-cell thick liver cell layers consisting of 15–25 cells each.

Figure 4.

The hepatic lobule is the structural unit of the liver. It consists of an hexagonal arrangement of hepatocyte plates with the central vein located in the center of the structure and the portal triads distributed at the vertices of the lobule. The portal triad consists of terminal branches of the portal vein and the hepatic artery and a bile duct.

The hepatic vasculature

The liver has a dual blood supply as it receives arterial blood from the right and left hepatic arteries and venous blood from the hepatic portal vein. The antigen-rich blood delivered through the portal vein accounts for more than 75–80% of the total blood. This blood originates from the stomach, alimentary tract, rectum and spleen, and contains large concentrations of antigens from dietary components and bacterial products from gut bacteria such as lipopolysaccharide endotoxin (LPS). This can be found at a concentration of up to 1 ng/ml¹⁴. Also, metastatic cells pass through the liver, which also has to deal with the load of detoxified byproducts with oncogenic potential. The remaining 20–25% is oxygenated blood delivered through the hepatic arteries which are branches of the celiac axis. The blood leaving the liver is drained into the inferior vena cava via the hepatic veins. The liver receives 1.5 L (30% of the total blood volume) each minute. Thus, the total volume of blood in the human body circulates through the liver approximately 360 times each day. The blood flows through the vascular sinusoids, and is drained into central veins or terminal hepatic venules, which are the branches of the hepatic veins.

Hepatic sinusoidal system

The hepatocytes form one cell thick plates and are separated from the bloodstream by a non-parenchymal, thin fenestrated (porous) barrier forming a labyrinth of specialized capillaries, termed liver sinusoids (Figure 5). These capillaries are primarily formed by liver sinusoidal endothelial cells (LSECs)¹⁵. The liver sinusoids are designed to allow easy transmission of molecules and cellular mediators between the sinusoids and the hepatocytes. To achieve such transfer, LSECs lack a basement membrane, making them permeable. Under normal conditions, the space of Disse formed between hepatocytes, LSECs and HSCs, allows exchange of cellular mediators without the need of a direct contact. A direct contact between hepatocytes and sinusoidal cells can be achieved from clusters of endothelial cells which contain small holes (or fenestrations)¹⁶ which perforate the cytoplasm and form ‘sieve plates’^17–19 (Figure 5). Fenestrae manage the exchange of solutes and particles between the sinusoidal blood and the space of Disse^{18, 19}. Blood also passes over a population of macrophages called Kupffer cells (KC). KCs account for approximately 80% of all macrophages in the body²⁰. Their strategic position enables them to clear the blood from endotoxins and antigens and to eliminate microorganisms by phagocytosis. HSCs (also known as Ito cells) are located in close proximity to the vascular sinusoids and in particular in the vicinity of the terminal hepatic venules. Because of the small diameter of sinusoids, even minimal increases in systemic venous pressure can create stasis. Such stasis encourages lymphocyte extravasation and extends direct contact of APCs and lymphocyte populations.

Figure 5.

Illustration of the microanatomical localization of hepatocytes, liver sinusoidal endothelial cells, Kupffer cells, and hepatic stellate cells. The space of Disse separates hepatocytes from the liver sinusoids. The endothelium of liver sinusoids is discontinued (fenestrated) and is formed by a layer of liver sinusoidal endothelial cells¹⁶. These cells act as scavenger cells and form a physical filtering barrier between the sinusoidal blood and plasma^{69, 391}. Kupffer cells are resident macrophages, that are attached to the layer of liver sinusoidal endothelial cells. The hepatic stellate cells are located in the sub-endothelial space of Disse and play vital role in fibrogenesis.

The liver as a lymphoid organ

Organs such as the thymus, lymph nodes, and spleen are "classical" lymphoid organs. However, other organs such as the gut and liver consist of cells whose primary function may not be immunological but nonetheless still perform essential immune tasks. Within these respective organs there are resident cells of the innate and adaptive immune system. Within a normal liver, the lymphocyte population is largely resident in the portal tract but can be also scattered throughout the parenchyma. The composition and localization of these lymphocyte populations changes dramatically when the liver architecture is altered as a consequence of acute or chronic inflammatory conditions. It is therefore not surprising, that hepatocytes and BECs are sites of immune-mediated destruction induced by a variety of infectious xenobiotic and tumor-originated sources¹. The anatomic position of the liver and its distinctive vasculature accounts for its unique ability to continuously exchange immunological information. The antigen-rich blood passing through the liver sinusoids is "scanned" achieved by a complex network of conventional and non-conventional antigen presenting cells (APCs)^{21, 22}.

Composition of the Liver Immune System

The human liver has a lymphocyte population normally resident in the portal tract, but also scattered throughout the parenchyma^23–26. Approximately 500,000 to one million lymphocytes can be retrieved from 100 mg of normal human liver tissue, bringing the total number of these cells to approximately 0.75–1.5 10¹⁰ cells for a liver weighing 1.5 kg²³. The lymphoid repertoire includes significant numbers of T-cells, B cells, natural killer (NK) and natural killer T (NKT) cells (Figure 6). The relative contribution^{1, 27} of these cells in the liver is illustrated in Figure 7. Using CD3 as a pan-T cell marker, these lymphoid populations can be subdivided into CD3⁺ and CD3⁻. CD3⁺ T lymphocytes outnumber CD3⁻ lymphoid cells. The most widely used marker for NK cells is CD56; all three sub-populations CD3⁺CD56⁻ T cells, CD3-CD56⁺ NK cells and CD3⁺CD56⁺ NKT cells have cytotoxic activity. B lymphocytes comprise only 5% of the total lymphocytes^23–26. Many of the liver-resident lymphocytes differ phenotypically and functionally from circulating lymphocytes^{24, 25, 28–31}, for reasons poorly understood²⁷. The hepatic lymphocyte repertoire includes conventional and unconventional lymphocyte sub-populations of both the innate and the adaptive immune system.

Figure 6.

Cells comprising the liver including hepatocytes (HEP), liver sinusoidal ensothelial cells (LSEC), Kupffer cells (KC), hepatic stellate cells (HSC) and lymphoid cell sub-populations. NK, natural killer; NKT, natural killer T-cells; DC, dendritic cell; Treg, T-regulatory cell

Figure 7.

Distribution of cell sub-populations within intrahepatic lymphocytes

αβ- and γδ-T cells in liver immunity

The conventional T-cell population consists of CD8⁺ and CD4⁺ T cells exhibiting a diverse repertoire with αβ-chain T cell receptors (TCR). These receptors recognize short peptidyl sequences from antigens in the context of MHC class I and class II molecules for CD8 and CD4, respectively. More than 80% of the CD3⁺ T cells are αβ T-cells, with the remainder expressing the γδT-cell receptor in the liver¹. The mean prevalence rate of 15% for γδTCR+ cells is 5 times higher in the liver compared to the periphery (range 1–5%)^{32, 33}. Lymphocyte repertoires rich in γδTCR+ cells are also found in skin, the gut and the genitourinary tract. The role of γδTCR+ cells in the immune homeostasis of the liver remains elusive. Although a protective role of these cells in the concavalin A-induced animal model of acute liver failure has been reported³⁴, the protective role is initiated by Vγ4 and not Vγ1, γδT cells. IL-17A deficient γδT cells are unable to protect from liver destruction, suggesting that IL-17A is an important mediator of such protection. Vγ4 and not Vγ1, γδT T cells are the major subsets of peripheral lymphoid γδT T cells. A different set of experiments have shown that the γδT cells protect from liver failure by targeting NKT cells³⁴. An earlier report has suggested that IL-17A induced by γδT cells is expressed in the liver of mice infected with Listeria monocytogenes at an early stage of infection and plays an important role in the initiation of innate immune responses³⁵.

The role of γδT cells is not limited to control of infection but appears to also play distinct immunoregulatory roles in tumor immunity^{36, 37}. They have also been involved in the induction and maintenance of liver-specific autoimmunity, as they appear increased in patients with active autoimmune hepatitis and primary sclerosing cholangitis^38–41. Some T-cells do not express CD4 or CD8 and are known as ‘double negative’ T cells. They have been studied mainly in mice, and appear to develop extrathymically and expresses specific V TCR β genes^{42, 43}. They can be found in liver, expressing either the αβ or γδ TCR and are considered to be cells with primarily regulatory properties^{23, 44–46}, they may be critical to induction of autoimmunity⁴⁷.

The participation of conventional T-cells bearing αβ TCRs in the immune homeostasis of a normal liver is by far the best studied and is reviewed elsewhere^{1, 48, 49}; the role played by these cells in the induction of immune-mediated liver injury has been extensively studied^{1, 48, 49}.

NK and NKT cells in the liver

NK cells are bone-marrow derived large granular cells; their main task is to kill target cells. NK cells represent approximately 20–30% of the total number of liver-resident lymphocytes, a percentage unusually high compared to less than 5% seen in peripheral blood²⁵. Their over-representation in human liver probably relates to their primary role, which is surveillance for infection, killing of infected hepatocytes (Figure 8) and possibly malignant transformation. They are conventional constituents of the innate immune system, but emerging evidence, both in mice and humans, indicates that NK cells are ‘educated’ throughout the development in a way similar to that seen in lymphoid cells of the adaptive immune system⁵⁰. NK cells acquire antigen-specific receptors, go through clonal expansion during exposure to infectious agents and produce long-lived memory cells.

Figure 8.

Schematic illustration of NK cell receptors and killing of viral hepatitis infected cells. Under normal conditions, non-infected cells are not killed because inhibitory signals from HLA class I molecules prevail over activating signals. Virus-infected cells are characterized by altered expression of HLA class I molecules. This disrupts the inhibitory signals and allows activation of NK cells and subsequent lysis of the infected hepatocytes. NK-mediated killing of infected hepatocytes is not operated in viral hepatitides.

Liver lymphocytes are also enriched in NKT cells, accounting for 20–35% of mouse liver lymphocytes and 10–15% of rat and human liver lymphocytes (Gao et al., J Leukocyte Biology 2009, 86:513–528). NKT cells are a heterogeneous group of T lymphocytes that express both NK and T cell markers, and recognize the lipid antigens presented by the nonclassical MHC class I-like molecule CD1. The functions of NKT cells are mainly mediated via the production of a variety of cytokines (e.g. IFN-γ and IL-4), which play important roles in regulating innate and adaptive immunity. Accumulating evidence suggests that NKT cells play a diverse role in liver injury, inflammation, fibrosis, and regeneration (Gao et al., J Leukocyte Biology 2009, 86:513–528; Park et al., Hepatology 2009, 49:1683-94).

The hypothesis that lymphocyte sub-populations participate in routine immuno-surveillant functions in normal liver has largely been based on evidence demonstrating that they express mature/activated phenotypes^{24, 51}. Early evidence suggested that these populations may have arisen locally, a finding that has implications not only in the maintenance of immune homeostasis but also in the induction of immune-mediated liver injury. Phenotypic and functional characterization of these liver-related lymphocytes has led to the appreciation of a role in the immunopathogenesis of liver diseases, including autoimmune liver diseases, chronic viral hepatitides, liver-related tumor immunology, alcoholic hepatitis, drug-induced immune-mediated liver disease and allograft rejection^{1, 49}.

Liver-Related Antigen Presenting Cells in Immunity and Tolerance

The liver is unique in its ability to recruit distinct cell types to become APCs; LSECs, KCs and hepatic DCs may all be considered conventional or classical liver APCs. However, under pathophysiological circumstances and during persistent liver inflammation, hepatocytes and BECs can express MHC II antigens and act as non-conventional APCs. Hepatocytes and BECs as APCs play an important role in the initiation and maintenance of processes important for loss of immunological tolerance. All these cells come in continuous contact with naïve T cells recirculating through the blood, and under normal of pathological conditions may participate in their activation (Figure 8)⁵².

One working hypothesis to explain the ability of the liver to induce systemic tolerance²¹ is based on the assumption that liver-resident DCs have distinct properties which promote tolerance rather than an immune response. Another plausible explanation is the intrinsic tolerogenic capacity of liver-related APCs such as the LSECs or the KCs. Donor cell chimerism has also been postulated to account for the observed tolerance⁵³. According to this scenario, donor-derived leukocytes including liver-resident APCs migrate to central lymphoid organs within 120 minutes post- transplantation, and persist for a long period, accounting for the hepatic tolerogenicity seen. Also, the systemic tolerance of the liver has been attributed to the induction of allospecific regulatory cells such as those with a CD4⁺, CD25⁺, FoxP3⁺ phenotype⁵⁴. These mechanisms may act in isolation or in combination.

Liver sinusoidal endothelial cells

Wisse was the first to demonstrate that LSEC is a distinct cell type with a characteristic open fenestration without basement membrane or diaphragm¹⁶. Aging, liver disease and various stimuli co-cultured with LSECs (such as vascular endothelial growth factor) alter the number, size and localization of the fenestrations^55–58. Wisse was also the first to suggest that because of the unusually high amounts of endocytic vehicles that LSECs contain, these cells are probably involved in the uptake of proteins circulating in sinusoidal blood¹⁶. This notion has been proven correct 15 years later when hyaluronic acid was identified as the first physiological macromolecule cleared from the blood by rat LSECs⁵⁹.

The role of LSECs in hepatic tolerance and the generation of immunity has been investigated and several surface markers have been found to be expressed by resting and activated LSECs (Table 2). The LSEC is an efficient APC with multiple functions^{21, 22}. The capacity of LSECs to possess several of the properties seen in DCs, such as the expression of MHC class II, various co-stimulatory molecules, and CD11c and their ability to activate naive T cells has been reported^60–68. LSECs can take up antigen using a multitude of receptors⁶⁹. The question as to whether a) LSECs are efficient APCs and b) can induce T-cell tolerance has also been addressed systematically. Lohse et al⁶⁰ were among the first to study the ability of LSECs to act as APC and demonstrated that LSECs are efficient APC, carrying functional B7-2 molecules⁶⁰. The ability of LSECs to present soluble antigens to CD4⁺ T cells is down-regulated by IL-10⁶⁰. Subsequent work⁶³ reflects that murine LSECs are efficient APCs⁶³. However, levels of MHC class II⁶³ are relatively low and stimulation of toll like receptors is unable to induce IL-12 expression⁶³. Also, treatment with endotoxin down-regulates the surface expression of constitutively expressed MHC class II, and CD80/CD86 co-stimulatory molecules⁶³. These data suggest that although endotoxin does not alter the ability of LSEC to remove gut-derived peptides from circulation, it affects antigen processing and expression of accessory molecules⁶³.

Table 2.

Phenotypic markers of liver sinusoidal endothelial cells as reported by various studies. Liver sinusoidal endothelial cells express CD54, vWF, CD31 in more than three studies

Molecule	References
CD4	62, 382, 383
CD14	382
CD16	382
CD31	67, 384–386
CD32	67, 382
CD34	384, 386, 387
CD54 (ICAM-1)	62, 382, 383, 385–388
CD106 (VCAM-1)	387
AcLDL	388, 389
vWF	67, 382, 384, 386, 390
VIII	388, 390
MHC-I	388
MHC-II	70, 382
CD11b	388
CD11c	388
CD40	70
CD80	62, 70
CD86	62, 70
CD105	70, 388
L-SIGN	388

The relative expression of these markers varies amongst studies. Methodology used for the assessment of their expression included immuno-histochemical analysis, PCR and flowcytometry using mouse, rat or human cultured LSECs (reviewed in ¹⁵); CD, cluster of differentiation; ICAM-1, intercellular adhesion molecule 1; VCAM-1, vascular cell adhesion molecule 1; AcLDL, acetylated low-density lipoprotein; vWF, von Willebrant factor; L-SIGN, liver/lymph node-specific ICAM-3-grabbing nonintegrin

Contrary to these results, LSECs isolated from primed mice can present antigen and induce antigen-specific CD8⁺ T cell tolerance⁶⁵. Also, adaptive transfer of LSECs isolated from ovalbumin-fed animals can induce antigen-specific T-cell tolerance to unfed animals, clearly indicating an involvement of LSECs in the induction of CD8 T cell tolerance against oral antigens⁶⁶. The ability of murine LSECs to tolerize T cells across MHC barriers has been studied⁷⁰; LSECs can regulate a polyclonal population of T cells with direct allospecificity. Data have demonstrated the Fas/Fas ligand pathway as important in the tolerizing capacity of LSECs towards alloreactive T cells⁷⁰. Though the molecular mechanisms responsible for LSEC-induced T cell anergy are poorly understood, reported data suggests that B7-H1 signaling on LSECs is a prerequisite for the induction of CD8⁺ T cell tolerance via programmed death (PD)-1 ligation⁷¹. Also, the contact of LSECs with DCs suppress neighboring antigen-presenting DCs to fully activate naive CD8 T cells⁷².

The capacity of LSECs as efficient APCs ex vivo and in vivo has been the focus of heated debate, as the results obtained have not been always supportive of their APC role²¹. Thus, in contrast to the findings presented above, Katz et al⁶⁷ challenged the dogma that LSECs have DC properties expressing MHC class II antigens, CD40, CD80, and CD86 co-stimulatory molecules, and CD11c and are able to effectively stimulate naïve T-cells. Using isolated LSECs with the phenotypic markers of endothelial cells (CD45⁻, CD31⁺, vWF⁺) (Table 2), it was reported that such cells lacked the expression of CD11c, the most widely used marker of murine DC. Additionally, they had minimal expression of MHC class II and undetectable levels of CD40, CD80, and CD86. Such LSECs were able to capture antigen but could not induce IFN-γ production or significant CD4⁺ or CD8⁺ T-cell proliferation. The possibility that the tolerogenic effect of the liver may be due to the continuous exposure of LSECs to gut bacterial products has been tested experimentally. Physiological concentrations of entotoxin contained within the blood drained from the portal veins to the liver can induce the secretion of interleukin (IL)-10 from LSECs⁷³; physiological concentrations of endotoxin appear to be able to down-regulate LSEC-mediated CD4⁺ T-cell activation via the modulatory effects they exert in the expression of MHC class II, CD80 and CD86⁶².

More recent studies have addressed the potential of diseased liver to provide a microenvironment which reverses the function of LSECs from tolerogenic to pro-inflammatory and highly immunogenic. LSECs from fibrotic livers induce T-cells to produce pro-inflammatory cytokines such as IFN-γ, IL-6, and TNF-α⁷⁴. Such data clearly implicate LSECs in intrahepatic immune-mediated inflammation seen in hepatic fibrosis and possibly to liver allograft tolerance^{70, 75, 76}.

Kupffer cells

The tolerogenic capacity of KCs has been demonstrated in the induction of tolerance to allergic and drug-induced reactions^{77, 78}. KCs have also been implicated in the induction of liver tolerance caused by preoperative donor-specific blood transfusion after liver transplantation⁷⁹. Their important role in tolerance has been suggested because gadolinium chloride-induced blockade of KCs prevents the induction of tolerance by the portal venous route in rat cardiac allograft transplantation^{80, 81}. KCs can also produce prostaglandin E2, which in turn can inhibit antigen (ovalbumin)-specific T-cell activation by DCs⁸².

The close interplay of KCs with LSECs or Tregs may be of relevance to the tolerogenic capacity of the liver. Early studies on human KCs challenged with LPS have suggested that these cells can also release IL-10⁷³. In vitro experiments have also demonstrated that exogenous IL-10 down-regulates the secretion of IL-6 and tumor necrosis factor (TNF)-α by LPS-stimulated human KCs⁷³. Kupffer cell-derived IL-10 may decrease the expression of both MHC class II and co-stimulatory molecules expressed by LSECs⁶⁴. Kupffer cell-produced prostaglandin E2 abrogates the capacity of LSEC to activate cloned, antigen-specific CD4⁺ T cells⁶⁴. Interaction of KCs with Tregs provoke the secretion of IL-10 by Tregs and facilitates the induction of systemic tolerance to hepatocyte-derived antigens, such as human α-1 antitrypsin⁸³. Earlier data have demonstrated in a concavalin A-induced model of liver injury the ability of both Tregs and KCs to provoke liver tolerance to concanavalin A through the induction of IL-10⁸⁴. KCs can freely transcytose LSECs and can secrete cytokines and chemokines to eliminate pathogens.

Hepatic dendritic cells

Hepatic DCs appear in high numbers throughout the portal triad, surrounding the central vein⁸⁵. A smaller number of DCs can also be found scattered throughout the parenchyma⁸⁶. These DCs are largely sub-divided into five sub-sets on the basis of their phenotypic characteristics, including classical myeloid DCs (mDCs) and plasmacytoid DCs (pDCs). They also consist of a unique mixture of mDCs and pDCs, a group of lymphoid DCs and a group of NK DCs (or DC-like NKs). A simplified phenotypic subdivision based on specific surface markers for these sub-populations is as follows: lymphoid (CD8α⁺, B220⁻, CD11b⁻); plasmacytoid (CD8α⁻, B220⁺); myeloid (CD8α⁻, B220⁻, CD11b⁺); myeloid & plasmacytoid mixture (B220⁻, CD11b⁻); and NK DCs (B220⁻,CD11c_int, CD69⁺, 2B4⁺⁺, DX5⁺)^{21, 87, 88}. Several other markers have been used in the recent past to better identify and investigate these subsets of dendritic cells in mouse and human liver. Myeloid and lymphoid DCs are the main sub-populations^{22, 89, 90}. Hepatic DCs are less immunogenic compared to their counterparts in spleen or other tissues⁸⁹. The low immunogenicity of hepatic DCs has been attributed to differences in subtype composition between the liver and spleen, which reflects the lack of expression of constitutive costimulatory molecules⁸⁹. However, hepatic DCs produce significantly higher amounts of cytokines and have a greater phagocytic capacity than the lymphoid organ counterparts^{91, 92}. The relative contribution of liver dendritic cell subsets in mouse liver⁹¹ is illustrated in Figure 5. Comparative analysis and morphometrics estimation of DCs in normal mice demonstrates that the normal liver contains more interstitial DCs compared to any other parenchymal organ⁹³.

It has been suggested - but not yet been proven - that the large number of DCs within the liver may result from the large number of pathogen-associated molecular patterns (PAMPs) contained in portal blood²². PAMPs are responsible for the activation of pattern recognition patterns (PRPs) which activates hepatic scavenger cells and induces the production of IL-6 or other cytokines, subsequently inducing the over-expression of complement C and C-reactive protein. This sequence of events is important for the induction of innate immunity, as acute phase proteins bind to pathogens and accelerate phagocytosis but abrogate the production of tumor necrosis factor (TNF)-α by Kupffer cells (KCs)⁹⁴. The constitutive expression of PPRs initiates the expression of chemokines and adhesion molecules and promotes immune recruitment to liver. This recruitment drives a series of processes which control the fine balance between recruited effector and regulatory cells and their cytokine milieu, subsequently leading to either hepatic tolerance or immunity²⁰. The cytokine milieu can cause hepatic DCs to become tolerogenic; these cytokines include - but are not limited to - IL-10, TGF-β and are induced by the complex interplay of KCs, LSECs, HSCs and other cell composites^{22, 95–100}.

Hepatic stellate cells

HSCs have a dual role. Under normal conditions they control blood flow through the sinusoidal system, while in pathological conditions and upon exposure to various inflammatory stimuli appear to differentiate into myofibroblasts. They then secrete inhibitors of tissue matrix metalloproteinases, deposit collagen and generate fibrous tissue leading to liver fibrosis. Experimental data in support of the capacity of HSCs to act as APCs are limited^101–103, in comparison to that of HSECs. Their tolerogenic capability has been indicated by studies demonstrating that transplanted HSCs effectively protect islet allografts from rejection in an islet transplantation mouse model¹⁰⁴. HSC-exerted immunomodulation is regulated by the inducible expression of B7-H1, an inhibitory molecule of B7 family¹⁰⁴; depletion of activated HSCs with gliotoxin decreases transplanted hepatocyte engraftment in rats¹⁰⁵.

HSCs present lipids to CD4⁺, CD8⁺ T-cells and NKT cells. Cultured HSCs can perform fluid-phase and receptor-mediated endocytosis and retain their ability for phagocytosis¹⁰¹. They express MHC class I and II molecules and lipid-presenting CD1b and CD1c molecules, as well as CD86, CD40 and other co-stimulatory molecules^{101, 102}. CD86 is also over-expressed when HSCs have been activated in vivo in a state of extended fibrosis and cirrhosis¹⁰¹. Exposure of HSCs to proinflammatory cytokines such as IFNγ markedly up-regulates CD80¹⁰¹ while CD40 activation leads to a 10-fold increase of the secretion of IL-8 by HSCs and a 2-fold increase of the monocyte chemoattractant protein-1 by the same cells¹⁰². Their potential tolerogenic capacity has been supported by their ability to produce vitamin A-derived retinoic acid and TGFβ. Activated HSCs express the negative co-stimulator programmed death (PD)-L1 and can inhibit T-cell responses via B7-H1-mediated apoptosis¹⁰⁶.

Hepatocytes and biliary epithelial cells as APCs

A wealth of experimental data reflect the ability of hepatocytes to serve as APCs. Murine studies have shown that the fenestrations of LSECs allow direct hepatocyte-T lymphocytes interaction¹⁰⁷. Hepatocytes constitutively express intercellular adhesion molecule-1 and addition of cytokines such as IFN-γ leads to moderate expression of HLA class I molecules. This cytokine leads to significant enhancement of HLA class II only if used at high doses within cultures¹⁰⁸. Hepatocytes prime naïve CD8⁺ T cell, even when they express low levels of MHC class I molecules¹⁰⁹. These hepatocyte-primed naive T-cells can expand, but in the absence of co-stimulatory signals they undergo apoptosis, leading to intrahepatic tolerance by clonal deletion^109–111. Also, when allogeneic hepatocytes are exposed to T-cells, T-cells activate and apoptose¹¹². In vivo experiments have also provided convincing evidence in support of the APC capacity of hepatocytes^{52, 113, 114}.

The study of the immunogenicity of human BECs or human BEC lines has been studied^115–120. BECs from normal human livers express HLA class I at a low frequency. HLA class II molecules are not expressed¹²¹. Infection with hepatotropic or hepatotrophic viruses enhances HLA class I expression¹¹⁵; cytomegalovirus infection leads to overexpression of HLA class leaves unaltered HLA class II¹¹⁵. Also, the same virus appears to reduce the rate of IFN-γ induced de novo expression of HLA class II, leaving unaffected HLA class I¹¹⁵. Data have shown that cytokine-induced expression of adhesion molecules such as intercellular adhesion molecule-1 (ICAM-1), vascular cell adhesion molecule-1 (VCAM-1), and lymphocyte function-associated antigen-3 (LFA-3) of human BECs is comparable to that seen in Epstein-Barr virus-transformed B-cell lines, a known antigen presenting cell¹¹⁷. The expression of these adhesion molecules is a prerequisite for effector cells to exert their cytolytic action¹¹⁸. In hepatocytes, the expression of HLA class II is increased in cultures performed in the presence of IFN-γ or when other pro-inflammatory cytokines are added^117–120. The lack of costimulatory CD28 ligands renders cytokine-stimulated human intraepithelial BECs unable to induce effective T-cell activation¹¹⁹. In pathological conditions such as that of PBC, destructed BECs overexpress HLA class II, as well as CD80 and CD86 co-stimulatory molecules¹¹⁶. While antigen presentation by CD80/CD86-positive APCs induces T-cell activation, antigen presentation in the absence of sufficient CD80/CD86 costimulation may induce tolerance. However, the capacity of BECs to efficiently present antigens and to activate effector cells also depends on whether there is efficient control by regulatory cells.

Efficient antigen presentation and T-cell activation of cells not undergoing apoptosis is a complex process, the fate of which depends on the action exercised by regulatory T-cells in the microenvironment.

Immune Mediated Liver Injury

Viral or bacterial antigens that pass through the liver sinusoids generally induce an immune response; in most cases, an efficient immune response will remove the microorganism. However, this cannot always be achieved, and chronic infectivity may be the final outcome. Interestingly, neither hepatitis B nor hepatitis C are considered cytopathic. In fact, the destruction of the hepatocytes is the outcome of the host’s immune response to clear the virus. This scenario implies that the liver becomes the victim of a friendly fire targeting the virus rather than the hepatocyte. Similarly, autoimmune liver diseases are models for investigation of mechanisms which lead to loss of immunological tolerance.

Autoimmune liver diseases

Autoimmune diseases of the liver affect the hepatocytes in the case of AIH and biliary epithelial cells in the case of PBC. Although select individuals can have features of both diseases, the clinical phenotypes and the immunological characteristics of these two diseases differ significantly^{1, 122, 123}.

AIH has a strong female preponderance, affects all ages, sexes and races and responds well to immunosuppressive treatment. Immunosuppression has no benefit in patients with PBC, a disease which is rarely seen in men and children and is currently treated with ursodeoxycholic acid, a bile constituent¹²².

Genetic susceptibility associations with HLA and non-HLA genes have been noted in AIH but large genome wide association studies are still lacking. Contrary to AIH, large genome wide studies have been conducted for PBC; a large number of loci have been found to be associated with disease, including HLA and non-HLA genes with immunological significance (such as that of IL12A, IL12RB, STAT4 and CTLA4)^124–128. STAT4 has been identified in several previous studies, and is of interest given that it is closely involved with IL12 signalling and participates in the immunopathogenesis of autoimmune diseases.

Despite enormous efforts, the pathogenesis of both diseases remains poorly understood. Some progress has been made and major experimental findings in support of the involvement of the innate and the adaptive arm of immunity are presented^{123, 129–131}. Translational research on the immunopathogenesis of AIH focused on the role played by humoral responses against specific antigens, the involvement of conventional and unconventional T-cells, and the immunomodulatory role of Tregs. Attempts have also been made to identify environmental triggers of the disease and to develop animal models resembling the human condition. In PBC, credible antigen-specific animals models of the disease have been developed^132–144.

Primary biliary cirrhosis

PBC is a chronic cholestatic liver disease characterized by an inflammatory destruction of the medium-size and small intrahepatic bile ducts (Figure 5), which eventually leads to fibrosis, cirrhosis and liver failure¹²².

PBC was first recognized by Addison and Gull in 1851 but its nature became better understood in 1958 when Mackay reported a case with high titers of complement-fixing antibodies to tissue homogenates¹⁴⁵; this autoantibody could be absorbed out using a mitochondrial fraction of rat liver¹⁴⁶, and was present in the majority of patients¹⁴⁷. The target antigens of these anti-mitochondrial antibodies (AMA) localized to the inner membrane of the mitochondria¹⁴⁸. A major step forward in the study of PBC was the cloning of the major AMA target, the E2 subunit of the pyruvate dehydrogenase complex^{149, 150}. Subsequent studies identified as AMA antigens other E2 subunits of the 2-oxo-acid dehydrogenase multienzyme family such as the branched-chain 2-oxo acid dehydrogenase complex (BCOADC) and 2-oxoglutarate dehydrogenase complex (OGDC)^151–156. Each of these three multifunctional complexes is involved in set of chain reactions and holds a key position in energy metabolism as PDC links glycolysis to the Krebs cycle, OGDC is essential to the Krebs cycle itself, and BCOADC is involved in the regulation of the oxidation of the branched-chain amino acids^{157, 158}.

Anti-PDC-E2 antibodies belong primarily to the IgG3 subclass, but IgM and IgA autoantibodies targeting this antigen can also be found. Autoantibodies to the 2-OADC enzymes are also detected in bile, saliva, and urine of patients with PBC^159–161. Notably, high-titer AMAs in PBC sera can block in vitro the catalytic function of the 2-OAD multienzyme complexes and the AMA bound to PDC-E2 with the greatest affinity are effective in inhibiting PDC-E2 enzymatic activity¹⁶². The fine specificity of AMA within their respective antigens has been delineated^{150, 163–169} and the core epitopic region on PDC-E2 has been mapped within the lipoyl domain of PDC-E2, though, at a 100-fold higher concentration, AMA also react with the outer domain¹⁷⁰. AMA specific for BCOADC-E2 and OGDC-E2 are directed against conformational epitopes that include the inner lipoyl domain¹⁷¹.

Adaptive Immunity in PBC

There is evidence of antigen-specific T-cell responses in PBC^{172, 173}. CD4 and CD8-T cell mapping studies have provided intriguing findings, as there is a major CD4 and CD8 epitope, which overlaps with the core B-cell epitope^174–180. This overlap of the autoepitopic regions of B- and T-cells has promoted investigators to suggest that this bears a pathogenetic significance for the induction of PBC. This was especially the case after reports indicating that soluble PDC-E2 complexed with a PDC-E2-specific human monoclonal antibody promotes the generation of PDC-E2-specific cytotoxic cells, at a 100-fold lower concentration than otherwise required in the presence of the soluble antigen alone¹⁷⁷.

Apotopes and PBC

The basis for the selective autoimmune attack of BECs lining the medium and small-size intrahepatic bile ducts remains elusive. However, it is interesting that specific destruction of small intrahepatic bile ducts is a consequence of the unique characteristics of human intrahepatic BECs during apoptosis and can be explained by exposure to the immune system of intact immunoreactive PDC-E2 within apoptotic blebs^{181, 182}. After apoptosis, human intrahepatic biliary epithelial cells (HiBECs), but not other epithelial cells, translocate PDC-E2 immunologically intact into apoptotic bodies, forming apotopes¹⁸². This observation suggests that PDC-E2 is accessible to the immune system during apoptosis. Subsequent experiments demonstrated an intense inflammatory cytokine production in the presence of the unique triad of BEC apotopes, mature monocyte-derived macrophages macrophages from patients with PBC, and AMAs¹⁸¹; other PBC-specific autoantigens also appear to be contained intact into apoptotic bodies appear to contain¹⁸³.

The role of innate immunity in PBC

The role of innate immunity in the immunopathogenesis of PBC has been supported by a plethora of experimental data demonstrating the intrinsic ability of cholangiocytes to express a variety TLRs, the cellular activators of innate immunity, or other PPRs^184–188. Peroxisome proliferator-activated receptor γ (PPARγ) is constitutively expressed in BECs isolated from small intrahepatic bile ducts. They appear to be downregulated in PBC bile ducts¹⁸⁵. PBC is also characterized by an upregulation of TLR4 and TLR9 in cholangiocytes and of TLR3 and type I interferon gamma signaling pathways in portal tracts and parenchymal cells^{92, 185}.

A significant role for periductal Langerhans cells and biliary epithelial cell-derived macrophage inflammatory protein-3alpha¹⁸⁴. Langerin-positive cells are predominantly within or scattered around biliary epithelial layers of bile ducts on liver biopsy specimens from PBC patients and may act as APC¹⁸⁴. The close interaction of innate immunity cells with lymphoid cells with immunoregulatory importance has been strengthened by data demonstrating the induction and perpetuation of Th17 cells in the periductal area in cases of PBC and the differentiation into Th17 cells in periductal dendritic cells and macrophages¹⁸⁹. It appears that in PBC, there is a characteristic periductal accumulation of IL-17-positive mononuclear cells¹⁸⁹. This is of interest given the wealth of experimental data from murine models of the disease based on the regulatory T-cell defects^{190, 191}.

Autoimmune hepatitis

Early data in patients with AIH^192–195 suggested that the disease is characterized by disease-specific circulating lymphocytes ‘sensitized’ to liver constituents of the so called liver soluble protein which were able to kill target cells in vitro. Subsequent experiments using separated T- and non-T- cell subsets from the peripheral blood of AIH patients and xenogenic target cells implies that cytotoxic cells were present in the non-T-cell subpopulation. The participation of antibody dependent cell-mediated cytotoxicity has been considered the most likely mechanism to explain the damage. Subsequent work has shown that IgG bound to hepatocytes are prone to injury by lymphocytes from healthy individuals¹⁹⁶. However, clonal analysis studies clearly demonstrated that cytotoxicity against specific antigens of the liver is due to T-cells¹⁹⁷. Further, CD4 T-cells have the capacity to initiate autoantibody production by autologous B lymphocytes¹⁹⁷. Neutralization with antibodies specific for HLA-DR, CD4 and IL-2R abolishes the ability to produce autoantibodies¹⁹⁷.

Although there are serotypes of the disease characterized by autoimmune responses against poorly defined nuclear and smooth muscle autoantigens^198–200, there are others (albeit less frequent) where the immune responses target liver-specific antigens^{194, 201–206}. Such an example is the immune response against cytochrome P450IID6 (CYP2D6), a member of the cytochrome P450 family of enzymes²⁰⁷. This is a family of exquisitely hepatic detoxifying enzymes, the most prevalent of those within the liver being the CYP3 and CYP2C families. It remains unclear as to why members of the CYP3 family are not target autoantigens in immune-mediated liver injuries. Also, peculiar to AIH is the fact that despite the high degree of homology between CYP2D6, CYP1A2 and CYP2A6 (Figure 12), antibodies against CYP2D6 do not cross-react with the two other cytochromes^{202, 208, 209}. The overall homology of the three antigens is not restricted to amino acid similarity but is also present at the 3D level²⁰² (Figure 12). The apparent lack of cross-recognition may be due to the fact that the epitopic regions recognized by individual anti-CYP antibodies are significantly dissimilar and can generate individual non cross-reactive humoral immune responses. This plausible explanation has been addressed experimentally for a limited number of peptides and proved to be valid²¹⁰. The epitope specificity of antibodies against CYP2D6 has been delineated and the major linear epitopic regions appear to be exposed on the surface of the molecule (Figure 13).

Figure 12.

Cell-Cell interaction of biliary epithelial cells (BECs) as antigen presenting cells, effector and regulatory T-cells

Figure 13.

CYP2D6, CYP1A2, and CYP2D6 amino acid homology.

The three cytochromes appear highly conserved but autoantibody responses against the one does not invoke cross-reactive immunity targeting the other. Amino acid analysis has been performed using the T-coffee software. Highlights of red, yellow and green correspond to areas of good, average and bad degree of homology; cons, conservation of amino acids (* indicate identical amino acids, : indicate conserved and . semi-conservative substitutions).

Recent studies have identified the CD4 and CD8-T cell epitopes on CYP2D6 and have demonstrated a partial overlap of major B-and T-cell epitopic regions, a finding that has been described also before in studies for PBC²¹¹. T-cell clones generated from liver tissue and peripheral blood recognising a major B- and T-cell epitope (aa 262–285) appear to acquire a Th1 CD4 characterized by their ability to induce IFNγ production phenotype²¹¹. Recent work has demonstrated the presence of HLA class I-restricted, IFN-γ-producing CYP2D6-specific CD8 T-cells²¹².

This has followed earlier studies demonstrating the presence of CD8 T-cell responses specifically recognizing the asialoglycoprotein receptor (ASGPR), a hepatic lectin the antigen recognition of which appears to be associated more with AIH than any other autoimmune or non-autoimmune liver disease.

Early publications provided hints in support of an impairment of suppressory lymphocyte populations in AIH^{213, 214} These findings have been recapitulated by new evidence supporting a numerical and functional impairment of Tregs which characterizes patients with AIH^215–217. This evidence warranties external validation. Nevertheless, functionally enhanced Tregs can be expanded and generated de novo in patients with AIH²¹⁷. This has been achieved using a polyclonal T-cell stimulation approach which is based on the use of exogenous IL-2 and the engages the T-cell receptor using CD3 and the co-stimulatory molecule CD28²¹⁷.

Alcoholic hepatitis (AH)

AH is a syndrome characterized by infiltration of the liver by inflammatory cells, including neutrophils, and hepatocellular injury. In addition to the alcohol-induced hepatotoxicity and oxidative stress, activation of innate immunity involving TLR4 and complement also plays an important role in initiating AH, but the role of adaptive immunity in the pathogenesis of AH remains largely unknown (Gao and Bataller, Gastroenterology, 2011,141:1572-85; Gao et al. Am J Physiol Gastrointest Liver Physiol. 2011;300:G516-25). The studies from last decade suggest that activation of a TLR4-mediated MyD88-independent (TRIF/IRF-3-dependent) signaling pathway and complement system in the liver play an important role in the initiation and progression of AH (Gao et al. Am J Physiol Gastrointest Liver Physiol. 2011;300:G516-25). In contrast to activation of TLR4 and complement, chronic alcohol consumption can attenuate NK cells, another important innate immunity component, and subsequently abolish the anti-viral and anti-fibrotic effect of NK cells, contributing to alcohol-mediated acceleration of viral infection and liver fibrosis in patients with chronic viral hepatitis (Jeong et al., Gastroenterology. 2008;134:248-58).

Pathogen-induced Liver Immunity

To better understand the mechanisms responsible for the initiation of an efficient immune response against bacteria, we need to be able to separate anti-bacterial defense in the sinusoidal compartment versus that noted in the parenchymal compartment. Under normal conditions, pathogens circulating in the sinusoidal blood will be eliminated by resident immune cells and will prevent them from accessing or entering hepatocytes.

Bacterial infection targeting the liver

Early sensing of pathogens is important for the initiation of efficient innate immunity and the successful elimination of pathogens^218–220. The adaptive arm of immunity also appears to play an important role in the induction of early anti-bacterial immunity^{207, 208}. Thus, KCs, NKT, NK and T cells collaborate in the eradication of pathogens in the sinusoidal blood^219–222. Under normal conditions, KCs phagocytose bacteria leading to the rapid clearance of these bacteria from liver. KCs also present bacterial glycolipids on CD1 molecules to immune cells such as NKT²¹⁸. Recent findings have demonstrated that KCs ingest Borrelia burgdorferi and induce chemokine receptor CXCR3-dependent clustering of invariant NKT cells²¹⁸. KCs produce CXCL9 and other CXCR3 ligands in response to B. burgdorferi. The antigen-presenting molecule C1d allows for a stable contact of KCs and invariant NKTs in addition to activating NKTs. The attraction of NKT by KCs is achieved in a CXC-chemokine receptor 3-dependent manner²¹⁸. Depletion of KCs prevents the formation of invariant NKT clusters. Also, the lack of KCS promotes a further increase of B. burgdorferi in liver parenchyma²¹⁸. No internalization of spirochetes by LSECs has been noted, and the adherence spirochetes to LSECs is augmented when these cells are exposed to the pathogen and KCs are depleted²¹⁸.

All together, these findings suggest that B. burgdorferi uses LSECs as a tool to escape resistance against pathogens provided by KCs in an attempt to gain access to the extravascular space²¹⁸. Those Borrelia burgdorferi escaping KCs enter hepatocytes and survive despite initiation of immune responses by HSCs²¹⁸. It seems that the cross-talk of KCs and NKT cells initiates an anti-bacterial immune response which prevents the host from persistent bacterial infection²¹⁸. They also support the notion supported by data on Plasmodium yoelii, that KCs do not act as a portal for the entry of spirochetes into liver parenchyma²²³.

Granulomatous infections and liver immunity

Detection of granulomas in histological liver specimens is a relatively frequent finding, found in up to 4% of undivided samples analyzed²²⁴. In one of the largest studies conducted so far, evidence for granulomas was found in 442 (3.6%) of 12,161 liver biopsies analyzed. The presence of infectious organisms was document in only 15 samples (3.4%), with M. tuberculosis being detected in three of the 15 (20%). Other mycobacteria, Yersinia, Toxoplasma gondii, Bartonella henselae and Listeria monocytogenes have also been considered responsible for the formation hepatic granulomas^224–226. The exact mechanisms by which these infectious agents induce the development of giant cells and subsequent hepatic granulomas remain unclear²²⁷. Granulomas mainly consist of focal accumulations of macrophages and the prevailing theory is that infection of macrophages plays an important role in the their formation²²⁷; the formation of granulomas is an attempt of the innate immune to control invasion of pathogens, especially when the adaptive arm of immunity is impaired. Listeria monocytogenes infection can induce the development of formatted by dendritic cells over-expressing indoleamine 2,3-dioxygenase^{227, 228}. The study of granulomas induced by mycobacterial infection has also revealed an important role for matrix metalloproteinase 9 expression in their formation²²⁹. Expression of this matrix metalloproteinase is required for the recruitment of macrophages and the formation of granulomas²²⁹.

Parasitic infections of the liver

Malarial transmission to humans is achieved by sporozoite infection of hepatocytes²³⁰. Infection with malarial parasites is an ideal model to study host–parasite interactions and the mechanisms that allow a pathogen to avoid elimination from immune cells, resident in the sinusoids, and subsequent hepatocyte infection^230–234. Work on animal models has provided a better understanding of the mechanisms responsible for parasite invasion. Sporozoites injected by mosquito bites enter the blood circulation, flow through liver sinusoids and subsequently infect hepatocytes,^235–237 using poorly understood immune-mediated mechanisms.

Plasmodium spp. sporozoites appear to first target KCs^{238, 239}. Pradel and Frevert²³⁹ were the first to provide conclusive data suggesting that sporozoites selectively target and actively invade KCs to avoid the sinusoidal barrier. However, experiments on KC-depleted rats have demonstrated that such a depletion does not abrogate the ability of sporozoites to infect the hepatocytes²³⁴. Contrary to the these findings, subsequent experiments conducted by Baer et al²²³ have shown that KC-deficient mice are resistant to sporozoite infection, probably because of the reduced number of portals to the liver parenchyma which prevents from direct access to hepatocytes. Recent data suggests that circumsporozoite protein, the sporozoite's major surface protein, has an adhesive confirmation in which the C-terminal cell-adhesive domain is exposed and a non-adhesive conformation in which the N terminus covers this domain, the former being important for hepatocyte invasion²⁴⁰. The state of hepatocytes also appears to play an important role for sporozoite infectivity; hepatocytes over-expressing the CD81, a tetraspanin which has been regarded as receptor for hepatitis C virus entry to hepatocytes, are susceptible to infection by Plasmodium falciparum and Plasmodium yoelli^{241, 242}. Subsequent findings have indicated that CD81 may not interact directly with the ligand of Plasmodium yoelli during Plasmodium infection, but through an unidentified protein which is regulated by CD81²⁴¹.

In endemic areas, repeated exposure to sporozoites during infection does not always lead to the clearance of the parasite from the circulation. This is probably due to the lack of protective immunity^{243, 244} manifested as antigen-specific antibody and CD8⁺ T cell responses targeting the circumsporozoite protein, which is one of the most immunodominant targets of malaria-specific immunity²⁴⁵. The number of Plasmodium-specific memory CD8 T cells required for immunity greatly exceeds the number required for resistance to other pathogens^{244, 246}. Also, IL-4 secreting CD4⁺ T cells appear important for the development of CD8⁺ T-cell responses against malaria liver stages²⁴⁷. If CD4⁺ T cell are absent, CD8⁺ T cells responses are reduced by 90% compared to that seen in the presence of CD4⁺ T cells and the cell-cell interaction largely depends on the presence of IL-4²⁴⁷.

Viral Hepatitides

Hepatitis B and C viruses (HBV and HCV) are the most common infectious causes of chronic liver disease worldwide. More than 500 million people are infected with HBV and HCV worldwide, and a significant proportion of those develop chronic liver disease, with mortality predominantly from the complication of cirrhosis and from hepatocellular carcinoma^{248, 249}. HBV and HCV are members of the hepadnaviridae and flaviridae families, respectively. The half-life of HBV is 2–3 days, while that of HCV is approximately 3 hours. An effective vaccine for the prevention of HBV has been available for years, but such a vaccine does not currently exist for HCV. Hepatitis A and E are food-borne RNA viruses, which infect the liver when they reach the blood through transversing gut epithelial cells^{250, 251}. They are transmitted from person-to-person by ingestion of contaminated water or food or water or through direct contact with an infected individual. Their incidence is high in developing countries, and both can cause acute hepatitis.

Virus-specific immunity has been considered responsible for the clearance of the virus and protective immunity but also for the immune-mediated destruction of hepatocytes and the development of cirrhosis^252–254. The study of the immunobiology of these viruses has been hampered by the lack of well-defined small animal models that accurately resemble the human disease^255–258. Virus-host interactions have been studied in the chimpanzee model (the only animal that can be infected with both viruses) and in translational research projects using biological material from infected individuals^{252, 253, 259–264}. From these studies it has become apparent that HBV and HCV share in common various immunopathologic features, but are highly variable in their virologic properties and kinetics during acute and chronic infection. They also differ in their viral- immune evasion and survival tactics^{219, 253}.

While perinatal HBV infection normally leads to chronic hepatitis, 90% or more of the acutely infected adults resolve from symptoms, develop virus-specific antibodies and preserve lifelong protective immunity. T-cell mediated liver destruction manifested as an increase of serum alanine aminotransferase is seen 10–15 weeks after infection. Humans and chimpanzees infected with HCV develop adaptive cellular and humoral responses only after at least 1 month and 2 months, respectively, after the infection^{265, 266}. An increase of transaminases levels is noted 8–12 weeks after infection with HCV.

Innate immune responses in viral hepatitides

A wealth of experimental data in infected chimpanzees has provided a better understanding of the role of innate immunity in the early phases of the infection^{252, 267–270}. The study of the chimpanzee model has shown that HBV DNA is cleared from the serum and the livers of the infected animals long before the development of significant anti-viral adaptive immunity²⁵². Work on transgenic mice has shown that IFN-γ produced by NKT cells can significantly suppress HBV replication^{271, 272}. Also, type I IFNs produced by KCs and other innate immune cells play a major role in the control of viral replication and spreading to neighbouring hepatocytes^273–275; such IFNs also prevent the induction of severe immune-mediated hepatocyte destruction^273–275. These data have prompted investigators to consider an important role for innate immunity in controlling virus replication^{271, 272}. Genetic microarray analyses in sequential liver samples from HBV infected chimpanzees have failed to show remarkable changes in the expression of type I IFN²⁶⁸ leading to the assumption that the inability to provoke strong innate immune responses in the first weeks of infection is an evasion strategy used by the virus in order to persist. KCs appear to control, to some extent, the magnitude of innate responses, both in HBV and HCV. Capsid proteins of these viruses trigger the expression of TLR-1, TLR-2 and TLR-6 expression on KCs and the subsequent release of pro-inflammatory and regulatory cytokines^276–278, involving various PPRs in cellular activation by viral proteins.

Despite the lack of strong innate responses, the recovery rate following HBV infection remains paradoxically very high. On the other hand, HCV infection provokes early changes in the expression of type I IFN and other genes²⁶⁹. The fact that HCV infected patients, who are homozygous for certain killer cell immunoglobulin-like receptor haplotypes are more likely to clear the virus and recover from HCV infection adds strength to the notion innate immunity is required for an efficient control of the virus²⁷⁹.

These data demonstrate that HBV and HCV are highly sensitive to type I IFN^280–282 and this has been the basis for the treatment of chronically infected patients, which is based on the administration of IFN-α, alone or in combination with anti-viral agents. Specific genetic polymorphisms in the IL28B locus predict a favourable outcome and response to IFN-α^283–288. It also appears that the induction of strong immune responses at early stages of the infection predicts a more favourable outcome and indicates subsequent clearance of HCV²⁸⁹. Against this background, it has become apparent that initiation of strong innate immune responses per se does not immediately preclude the clearance of HCV, as viral chronicity is established in more 50% of cases.

Hepatocyte receptors of hepatotropic viruses and innate immunity

The attachment of HBV and HCV to hepatocytes requires cross-linking with heparin sulphate proteoglycans. Clear demonstration of a molecule which can serve as a receptor for HBV has not been provided. None of the molecules reported so far as receptors of HCV are hepatocyte-specific. The tetraspin CD81 was amongst the first molecules to be identified as hepatocyte receptors of HCV²⁹⁰.

Other molecules which can bind to HCV and seem important for the internalization of HCV into hepatocytes include the low density lipoprotein receptor²⁹¹ and the scavenger receptor B1²⁹². Also, the envelope E2 protein of HCV can bind to the liver and lymph-node specific ICAM3 grabbing non-integrin (L-SIGN) which is also expressed by LSECs and the DC-specific ICAM3 grabbing non-integrin (D-SIGN) which is also expressed by KCs^293–297. Claudin-1²⁹⁸, occludin²⁹⁹, the epidermal growth factor receptor and ephrin type A receptor 2³⁰⁰ are also co-receptors of HCV and play a role in the late steps of viral entry. It appears, that when HCV reaches the liver attached to heparin sulphate proteoglycans in liver sinusoids, it binds to low density lipoprotein, CD81 or scavenger receptor B1 and before endocytosis attaches to claudin and occludin. The receptor tyrosine kinases epidermal growth factor receptor and ephrin type A receptor mediate HCV entry by regulating CD81-claudin-1 co-receptor interactions.

The role of the HCV receptors is not limited to the entry of the virus into hepatocytes. They appear to alter host-virus innate immunity responses with mechanisms poorly understood. Thus, binding of HCV envelope 2 protein with NKs expressing CD81 inhibits the production of IFN-γ and other pro-inflammatory cytokines and suppresses their cytotoxic effects innate immune cells expressing CD81 can alter host-virus innate immunity^{301302, 303} (Figure 14). Collectively, these data demonstrate that the cytotoxic potential of NK cells from healthy individuals is impaired in the presence of high concentrations of HCV envelope 2 protein, and that NKs from patients infected with HCV are unable to produce pro-inflammatory cytokines and to activate DCs^{301302, 303}. Such findings may not be limited to CD81-mediated HCV/NK interactions but may include other innate immune cells.

Figure 14.

3D-prediction model of the three major linear epitopic regions of human cytochrome P450IID6 (CYP2D6).

The B-cell epitopes of anti-CYP2D6 antibodies (also known as anti-liver kidney microsomal type 1 antibodies- anti-LKM1)³⁹² has been studied and the three main epitopic regions recognized span CYP2D6_254–271, CYP2D6_193–212 and CYP2D6_321–351 sequences, being targeted by more than 55% of the patients with CYP2D6 autoantibodies^393–396. The antigenicity of this area may in part been explained by the exposure of these sequences to the surface of the molecule as it is illustrated in Figure 7.

Aminoacids of the autoepitopic regions are presented in the form of space fill in different colours and the remaining in a wire worm backbone (grey); in red and yellow are the dominant CYP2D6_254–271 and CYP2D6_193–212 epitopes. Prediction analysis anticipates that the epitopes are exposed on the surface of the molecule. The structure was analyzed with the Cn3D visualization tool.

A mucin-like class I integral membrane glycoprotein has been considered the attachment receptor of HAV^{304, 305}. Anti-HAV specific antibodies of the IgA class can carry the virus. Asialoglycoprotein receptor is a C-type lectin which clears IgA from the circulation^{306, 307}. Anti-HAV specific IgA antibodies use asialoglycoprotein receptor to infect hepatocytes³⁰⁸. Dotzauer et al³⁰⁸ used a mouse hepatocyte model which, because it does not permit infection with HAV, can be used as a model to study carrier-mediated HAV entry into host cells without any interference by the HAV receptors. These authors have shown that HAV is taken up by hepatocytes in the form of HAV-specific IgA immunocomplexes and are endocytosed by ASGPR. This has led Protzer et al to hypothesize that KCs expressing the Fcα receptor³⁰⁹ may bind the IgA-HAV complexes²¹⁹. They may then transfer these complexes to hepatocytes, where the virus can be internalized via the asialoglycoprotein receptor²¹⁹.

Adaptive immune responses in viral hepatitis B and C

Neither HBV nor HCV clearance can be sustained through the development of strong innate immunity. It appears that the elimination of these viruses can only be achieved in the presence of strong antigen-specific CD4⁺ and CD8⁺ T cell responses, a finding that clearly supports the thesis that the adaptive arm of immunity is the most critical for the immune-mediated clearance of the virus^{281, 310, 311}.

Humoral Immunity

The isotype and antigen-specificity of antibody responses in acute and chronic HBV indicates the stage of the disease. IgM antibodies against the hepatitis B core antigen are early markers of infection. Acutely infected patients who recover develop neutralizing antibodies against hepatitis surface antigen (and antigen-specific CD4⁺ and CD8⁺ T-cell responses against the core and other antigens of the DNA virus) that confer lifelong protection. The great majority of virus-free individuals who are vaccinated with the surface antigen develop high-titer neutralizing antibodies and are protected from exposure to HBV. Recent data has demonstrated reactivation of the disease following B-cell depletion, indicating the importance of these cells in controlling the virus³¹².

Chronic HBV infection is roughly subdivided in the immunotolerant, immunoreactive, low replicative and high replicative phases. The duration of these stages varies amongst infected individuals. Antibodies against hepatitis e and surface antigens appear late and are signs of a favorable course while the presence of antibodies against the core and the surface antigens are present forever after the resolution of clinical symptoms.

The influence of anti-HCV antibodies over the course of the disease is far from clear. Infection with HCV does not initiate early humoral responses and the highest concentrations of anti-HCV antibody responses are typically present in patients with well-established chronic HCV infection. In contrast, patients who recovered from HCV do not have detectable anti-HCV antibodies, a finding that has prompted investigators to consider the emergence of HCV escape mutants. How HCV constantly evades neutralizing antibodies during chronic infection remains unclear, though it is most likely due to its ability to ‘creep’ from one hepatocyte to another (cell-to-cell transmission) and to generate a selection of escape mutants are the most likely scenarios.^{313, 314}.

Cellular immunity against hepatitis B and C

The study of the immunobiology of HBV and HCV has led to the appreciation that persistent infection is largely due to the inability of the immune system to clear the virus. Depletion and exhaustion of cytotoxic T lymphocytes specific for viral antigens is the main cause for the inability to control the virus. Antigen-specific cytotoxic lymphocytes are prone to apoptosis³¹⁵. Also, HCV-specific CD8⁺ T cells undergo substantial apoptosis in the periphery during acute HCV infection and in the liver during chronic infection³¹⁶. BCL-2-Interacting Mediator, a key pro-apoptotic mediator which is known as BIM, plays an important role in the apoptosis of virus-specific CD8⁺ T cells^{315, 317}. Consistent exposure to antigenic overload during infection leads to the ‘exhaustion’ of cytotoxic T lymphocytes^{318, 319}. This phenomenon is largely mediated through the programmed death PD-1/PD-L1 T cell co-inhibitory pathway. PD-1 is a co-inhibitory molecule that controls T-cell reactivity. PD-L1 and PD-L2 binding to the PD-1 receptor negatively regulates T cells, leading to decreased proliferation and down-regulation of IL-2 and IFN-γ. This ‘exhaustion’ operates at several levels and allows the virus to escape immune recognition and to establish persistent infection³²⁰. Blocking PD-1 restores intrahepatic HBV-specific T-cell responses^{321, 322}. This may explain why HBV patients have over-expressed PD-1 compared to healthy individuals³²². Also, HCV-specific cytotoxic T lymphocytes that are phenotypically exhausted are characterized by PD-1 over-expression^{323, 324}. Patients with chronic HBV and HCV infection over-express co-inhibitory molecules such as PD-1, CTL4, T cell immunoglobulin domain protein 3 and 2B4³²⁵. An imbalance on the operation of co-inhibitory pathways in viral hepatitides B and C appears not to be limited at the lymphocyte level, as it includes KCs LSECs and HSCs. These cells appear to express (at least) PD-L1, and their induction has been demonstrated during viral infection^{71, 106, 326–331}. The role of CD4⁺ T cells has been extensively studied in patients and chimpanzees infected with HBV or HCV. These studies have shown that induction of vigorous, antigen-specific CD4 T cell responses positively correlates with the recovery from the infection³³². Epitope mapping studies have delineated the disease-specific autopepitopic regions recognized by acutely and chronically infected patients at various phases of the diseases²⁵⁴. Despite the low frequency of CD4⁺ T cells in the periphery, flow cytometric analyses have shown that in the early phase of the infection with HCV CD4⁺ T cell become detectable at the time CD8⁺ T cell recover²⁶⁶. The close interplay between CD4⁺ and CD8⁺ T cells has been shown in studies indicating that CD4 depletion from a chimpanzee recovered from HCV infection leads to the loss of protective immunity upon re-exposure to the virus³¹¹. These studies have provided data to suggest that CD4⁺ T cells confer protective immunity but are also participating the induction of immune-mediated hepatocyte destruction.

Extrahepatic immune-mediated manifestations are frequently seen in chronic HBV and HCV. In the great majority of the cases are manifested as low to medium titer autoantibodies against nuclear and smooth muscle antibodies³³³. Whether these autoantibodies are induced as a result of polyclonal B-expansion during hepatocyte destruction or they are the outcome of antigen-driven mechanisms such as molecular mimicry and immunological cross-reactivity remains elusive^{202, 208, 333–338}. Immune complex-mediated rheumatic manifestations, mixed cryoglobulinemia, and non-Hodgkin lymphomas are more frequently seen in chronic HCV infected patients than HBV³³⁹. Viral-induced autoimmune liver diseases are rarely noted, and a link between the virus and the autoimmune condition is difficult to make^340–342.

The immunopathology of hepatocellular carcinoma

It has become apparent that while a vigorous antigen-specific immune response against HBV and HCV leads to viral clearance, the depletion and exhaustion of cytotoxic T lymphocytes specific for viral antigens induces chronic hepatitis and subsequent cirrhosis. In an attempt to respond to the process of necroinflammation, the liver regenerates, and in this process activates macrophages which release free radical with carcinogenic potential. The concert action of mitogenic and mutagenic stimuli causing cellular and DNA damage and the dysregulation of cellular growth leads to the development of hepatocellular carcinoma^{343, 344}. HBV vaccination programs prevented the development of newly infected cases and indirectly decreased the number of new hepatocellular carcinoma cases³⁴⁵. In this respect, the hepatitis B vaccine can be considered one of the very first successful immunotherapy cancer vaccines.

Tumor markers such as alpha fetoprotein are increased during the course of hepatocarcinogenesis and T-cell responses specific for this antigen are present in the peripheral blood of patients with cancer^346–349. Several other antigens have been identified as potential tumor markers and target autoantigens but none of them is highly disease-specific. Nevertheless, strong tumor-specific CD8⁺ T cell responses control tumor progression and prevent from the recurrence of hepatocellular carcinoma^{350, 351}. Several studies have shown that tumors have adopted numerous immune escape mechanisms, that include the induction of immunosuppressory cells. These include regulatory T cells and myeloid-derived suppressor cells. Such cells have the potential to mask tumor-specific immune responses in patients with hepatocellular carcinoma.

The frequency of circulating CD4(+)CD25(+)FoxP3(+) Treg is increased significantly in patients with hepatocellular carcinoma compared to controls and correlated with disease progression^{352, 353}. Clustering of Tregs and reduction rates of infiltration CD8⁺ T cells are characteristically found in tumor regions compared with nontumor regions and these Tregs suppress the anti-CD3/CD28 induced cytolytic activity of CD8⁺ T cells³⁵². Current attempts concentrate in targeting regulatory T cells as this will potentially enhance tumor-specific CD8⁺ T cells^{354, 355}.

Conclusions

Historically, the immune system was divided into mucosal versus systemic immunity. While this division is still accurate, it is equally important to note specific immunological contributions of specific organs; this is illustrated not only by the liver as discussed herein, but also with respect to skin, lung and other tissues. We have not attempted to discuss loss of tolerance in the detail deserved in a paper devoted to liver and the immune response and we refer to a number of recent publications which deal specifically with genetics, environment and immunity^356–375. However, it is noteworthy that the liver is anatomically unique and its function is essential for fetal tolerance and host protection from gut flora and the enormous repertoire of materials that pass through the portal circulation. It is indeed ironic that the liver, which is so critical for immune tolerance, can itself become a victim in diseases such as autoimmune hepatitis and primary biliary cirrhosis. We have not discussed primary sclerosing cholangitis herein because the scope is far beyond that of the thesis herein. However, it also brings to mind great voids that exist with respect to our knowledge on chemokines and their cognate receptors with respect to lymphoid homing and we refer to a number of seminal publications by Adams and colleagues^376–380. Finally, there is also the hope that the liver as a facilitator of tolerance can be used as a tolerizing vehicle to restore immune homeostasis in other examples of human autoimmune disease³⁸¹.

Figure 9.

Cell-cell interaction which can lead to activation of naïve T lymphocytes within the liver include contact with Kupffer cells (KC), liver sinusoidal endothelial cells (LSEC) or hepatocytes (indicated by the respective arrows).

Figure 10.

Flow cytometric analysis of peripheral blood reflects the relative proportion of bright and low NK (CD36⁺) cells, NKT (CD3⁺CD56⁺) and T-cells (CD3⁺CD56⁻) in a representative donor. Bright and low NK cells can be seen.

Figure 11.

Hepatic dendritic subsets in mouse

Figure 15.

The tetraspin CD81 is used by hepatitis C virus to entry the hepatocyte. Expression of CD81 by NK suppresses the induction of pro-inflammatory cytokines such as interferon-γ and inhibits the cytotoxic capability of these cells allowing for the persistence of the virus. Other HCV co-receptors (for further details, see main text), expressed by cells of the innate immune system resident within the sinusoids, may participate in a similar fashion facilitating the inability of the host to clear the virus.

Abbreviations

AIH

Autoimmune hepatitis

AMA

Antimitochondrial antibody

ANA

Antinuclear antibody

APC

Antigen presenting cell

DC

Dendritic cell

HSC

Hepatic stellate cell

IFN

Interferon

IL

Interleukin

MHC

Major Histocompatibility Complex

KC

Kupffer cell

LPS

Lipopolysaccharide endotoxin

LSEC

Liver sinusoidal endothelial cell

mDC

myeloid DC

NK

Natural Killer

NKT

Natural Killer T-cell

OADC

oxo-acid dehydrogenase complex

OGDC

2-oxoglutarate dehydrogenase complex

PAMP

pathogen-associated molecular pattern

PBC

primary biliary cirrhosis

PD

programmed death

pDC

plasmacytoid DC

PD-L1

programmed death ligand 1

PPR

Pattern recognition patterns

TLR

Toll-like receptor