Abstract
The liver has a number of important functions in innate and adaptive immunity. Contributions to the innate (nonspecific) immune system include production of acute phase proteins, nonspecific phagocytosis of particles, nonspecific pinocytosis of molecules, and nonspecific cell killing. Hepatic involvement in innate immunity contributes to the systemic response to local inflammation, clearance of particles and soluble molecules from the circulation, and killing of invading cells such as neoplastic cells. Liver involvement in the adaptive (specific) immune system includes deletion of activated T cells, induction of tolerance to ingested and self-antigens, extrathymic proliferation of T cells, and deletion of many of the signaling and effector molecules. Hepatic involvement in adaptive immunity allows clearance of activated T cells and signaling molecules following inflammatory reactions, and promotes immunologic tolerance toward potentially antigenic proteins that are absorbed from the intestinal tract. The liver is a major site of extrathymic T cell development, which assumes increasing significance with aging in mammals. Perturbations in hepatic structure or function can result in significant ramifications in both the innate and adaptive immune systems.
Keywords
Introduction
The major function of the immune system is resistance to various pathologic processes, with invading bacteria being the exemplary process. The immune system has innate (non-specific) immunity and adaptive (acquired or specific) subdivisions. Innate immunity is further subdivided into functional components: anatomic barriers, physiologic barriers, non-specific phagocytosis/pinocytosis, and nonspecific inflammation. Though not a textbook category of innate immunity, the critical role of nonspecific cell-killing suggests it should be considered a major component of innate immunity. Anatomic barriers are the first-line defense against bacterial invasion and are an obvious critical component of immunity. Physiologic barriers are less obvious than anatomic barriers, but nevertheless provide a substantial level of resistance to bacterial invasion and other pathologic processes. Physiologic components of innate immunity include temperature (body and local), pH, and chemical mediators such as lysozyme, interferon, and complement (Goldsby et al., 2000). Respiratory air movements, mucociliary movement, intestinal peristalsis, and urine flow should also be considered as components of innate immunity, since disruptions in these functions can result in disastrous immunologic alterations.
Adaptive immunity consists primarily of antigen-presenting cells, effector lymphoid cell populations, and a myriad of molecules that mediate cellular interactions. The adaptive immune system is essentially the macrophage-lymphocyte system that serves as a backup or fail-safe system in the event of failure in some component of the innate immune system. Adaptive immunologic processes exhibit the classic features of antigenic specificity, diversity, immunologic memory, and self-nonself-recognition (Goldsby et al., 2000).
Biologic processes do not necessarily adhere to orderly classification schemes, and this lack of order is painfully apparent in the immune system. There are numerous examples of interactions between the innate and adaptive immune systems, thus the distinction between these 2 man-made divisions must be considered blurred, at best. One clear-cut and important example of a cooperative innate/adaptive immunologic activity is antibody-dependent cell, particle, or molecule recognition followed by nonspecific phagocytosis, pinocytosis, or cell-killing based on nonspecific recognition of the Fc domain of the immunoglobulin molecule. The first step is dependent on specific antigen recognition, but the effector step is based on the nonspecific affinity of certain cells for the Fc domain of immunoglobulin molecules, regardless of the antigenic specificity of the immunoglobulin molecule.
The liver has a number of major functions in both innate and adaptive immunity, which can be outlined as follows:
Innate Immunity Involvement
Production of acute phase proteins
Nonspecific phagocytosis
Nonspecific cell killing
Disposal of waste molecules
Adaptive Immunity Involvement
Deletion of activated T cells
Induction of tolerance to ingested and self antigens
Extrathymic proliferation of T cells
Disposal of waste molecules
Following is a brief discussion of these major functions. As previously mentioned, there is considerable overlap between innate and adaptive immune functions and, in some cases, overlap between immune and nonimmune functions.
Production of Acute Phase Proteins
Acute inflammation in mammals is associated with a transient increase in a group of circulating proteins that are collectively known as acute phase proteins. The term “acute phase proteins” has assumed a broader definition in recent years, but in this discussion the term is confined to proteins produced by the liver in response to cytokine stimulation. The group includes serum amyloid A protein (SAA), fibrinogen, C-reactive protein (CRP), haptoglobin, complement factors C3 and C9, hemopexin, ceruloplasmin, α2-macroglobulin, CD14, α1-antichymotrypsin (ACT), α1-cysteine proteinase inhibitor (α1CPI), and α1-antitrypsin (AAT) (Fey et al., 1994). Circulating acute phase proteins are responsible for many of the systemic effects of inflammation, which are largely aimed at preparing the body for resistance to systemic invasion and facilitating local resistance to pathogens. The acute phase proteinase inhibitors (e.g., AAT, ACT, α1CPI, and α2M) reduce tissue damage due to proteinases that are released by dead or dying cells. Hemopexin and haptoglobin bind to heme and globin, respectively, which may be released by erythrolysis in inflammatory lesions. SAA and CRP have functions that suggest a scavenger role, but no single function has been identified that would necessitate the marked (up to 1000×) increase of these proteins in an acute phase reaction.
CRP is a member of the ancient, highly conserved pentraxin family of proteins and is arranged as a cyclic homopentamer. CRP acts as an opsonizing agent, activates complement, binds to IgG receptors on mammalian cells and phosphocholine in bacterial membranes and recognizes nuclear constituents in damaged cells. CRP has been shown to be a reliable indicator of inflammation associated with atherogenesis in humans (Dupuy et al., 2003). CRP is not a reliable indicator of inflammation in all species. Serum haptoglobin level has been shown to be a better indicator of systemic inflammation in swine and serum levels of α2-macroglobulin, haptoglobin or fibrinogen are more accurate than CRP levels as indicators of inflammation in laboratory rats (Chen et al., 2003; Giffen et al., 2003; Dasu et al., 2004).
Detection of the presence and effects of acute phase proteins has been a cornerstone of clinical practice, and serves as the basis for such time-honored diagnostic tests as the erythrocyte sedimentation rate. SAA levels are monitored as a marker for allograft transplant rejections in humans and α2-macroglobulin levels are used as a marker for adjuvant-induced chronic polyarthritis in rats (Lonberg-Holm et al., 1987).
Production of acute phase proteins has historically been considered as a function of hepatocytes, but recent evidence indicates that a similar spectrum of proteins is produced during involution of the mammary gland and uterus. Using oligonucleotide microarray analysis, RT-PCR, Western blotting, and/or immunohistochemistry, investigators have shown that involution of the mammary gland is associated with expression of acute-phase response genes in the mammary gland that resemble the acute-phase response in the liver (Stein et al., 2004).
Pharmacologic manipulation of the acute phase response in either the positive or negative direction has immediate clinical implications and has been the subject of intense research, much of which is beyond the scope of this overview. As a brief synopsis of the signaling pathways, the acute phase response is generated when focal injury at some extrahepatic location in the body prompts local macrophages to release a first wave of cytokines that includes IL-1, tumor necrosis factor alpha (TNFα), and a small amount of IL-6. Absorption of the first wave of cytokines into surrounding cells is followed by a second wave cytokine release, including a large amount of IL-6 that promotes massive production of acute phase proteins by hepatocytes (Fey et al., 1994).
IL-6 is the prototype signaling molecule in the induction of the acute phase response, but IL-22, which is produced by activated T cells, also has the ability to up-regulate production of acute phase proteins (Nagalakshmi et al., 2004). A high level of expression of IL-22 receptor-1 (IL-22R1) in colonic epithelial cells suggests IL-22 has an additional role in intestinal inflammation.
Nonspecific Phagocytosis
Hepatic sinusoids have historically been known to contain 4 resident cell populations: endothelial cells, macrophages known as Kupffer cells, liver-specific NK cells known as pit cells, and fat-storing cells (Kuiper et al., 1994). More recent observations suggest a fifth intrasinusoidal population of dendritic cells should be added to the list.
Nonspecific phagocytosis in the liver is mediated primarily by Kupffer cells. Kupffer cells are present throughout the liver, but there is variation in the population density, cytologic characteristics, and physiologic functions of Kupffer cells in different zones of the hepatic acinus/lobule. Kupffer cells are somewhat more numerous in the periportal region, and have been shown to be distributed through zone 1 (periportal), zone 2 (midzonal), and zone 3 (perivenous) of the rat liver acinus in a ratio of 4:3:2. Periportal Kupffer cells are larger and have higher lysosomal enzyme activities than Kupffer cells from the midzonal and perivenous regions. Periportal Kupffer cells also have greater phagocytic activity (Sleyster and Knook, 1982). These observations suggest a concentrated population of highly active Kupffer cells in the region of the acinus that is the first point of contact for incoming, potentially pathogen-laden blood.
Studies using peroxidase cytochemistry as a detector indicated that Kupffer cells constituted 31% of the sinusoidal cells. Liver tissue contained 14–20 × 106 Kupffer cells per gram of tissue. Studies involving the intravenous injection of latex particles revealed that the population of latex-labeled Kupffer cells did not change over a period of 3 months, suggesting a long lifespan for these resident macrophages. The impression of a long life span is supported by an observed low mitotic rate (0.06% after a 6-hour arrest by vinblastine) and by a low rate of labeling by 3Hthymidine (Bouwens et al., 1986).
Studies of Kupffer cells that are removed from the hepatic microenvironment confirm observations based on in situ studies. Studies on extracted hepatic cell populations reveal the macrophage population of the liver is heterogeneous in size as well as endocytic and lysosomal enzyme activity. Small-to medium-sized liver macrophages have the greatest activation and cytotoxic potential, while larger liver macrophages have a greater capacity for uptake of particulate matter such as liposomes(Daemen et al., 1989).
Rat liver macrophages of different sizes also have different patterns of secretion of various immunomodulatory molecules after stimulation. The highest levels of secretion of TNFα prostaglandin E and IL-1 were observed in the largest macrophages, while the highest level of nitric oxide secretion was observed in the smallest macrophages (Hoedemakers et al., 1995). When coupled with the observed size differences in macrophages in different areas of the hepatic lobule, it becomes apparent that different regions of the lobule have different profiles of signaling and effector molecules. This could be particularly significant in the case of locally active molecules such as nitric oxide.
As a general rule, Kupffer cells assimilate particulate material via phagocytosis and endothelial cells assimilate soluble materials via pinocytosis. There are variations to this general theme, particularly in the case of materials that have ill-defined physical characteristics. As an example, colloids are small particles that have characteristics of both soluble macromolecules and particles. Due to their small size, colloids tend to be assimilated into cells by pinocytosis rather than phagocytosis. It has been shown that colloidal gold of up to 100 nm diameter particle size is sequestered almost exclusively into sinusoidal endothelial cells. Colloidal carbon, by contrast, is assimilated primarily by phagocytosis into Kupffer cells. This apparent nonconformity in carbon assimilation is due to the fact that colloidal carbon binds to platelets, and the platelet-carbon complexes are subject to phagocytosis by Kupffer cells. The uptake of platelet-carbon complexes is so avid that transient thrombocytopenia is associated with intravenous injection of colloidal carbon (Salvidio and Crosby, 1960; Cohen et al., 1965).
The liver has been shown to be the primary site for removal of experimentally administered antigen and immune complexes. Soluble IgG complexes are eliminated from the circulation mainly by the liver, predominantly by Kupffer cells and, to a lesser degree, endothelial cells. Kupffer cell recognition of the Fc domain of immunoglobulins results in nonspecific phagocytosis of immune complexes as well as antibody-coated particles such as microorganisms and eukaryotic cells. Presence of the Fc receptor allows Kupffer cells to have a significant role in control of inflammatory and immunologic processes (Ravetch, 1994).
Uptake of immunoglobulin complexes is mediated by various subtypes of the Fcγ receptor, principally Fcγ receptor IIB2 (Fcγ RIIB2) and Fcγ receptor III (Fcγ RIII). Both Kupffer cells and liver endothelial cells produce mRNA for Fcγ RIIB2 and Fcγ RIII, while the level of expression of Fcγ RIIB1 is negligible. Hepatocytes have no demonstrable production of Fcγ receptor of any type (Lovdal and Berg, 2001). It has been shown that there is a difference in the rate of catabolism of immunoglobulins that are assimilated via the Fcγ receptor, as compared to ligand molecules that are assimilated by the scavenger receptor (Lovdal et al., 2000).
Regulation of receptor expression and activity is an emerging field of research that has immediate implications for pharmacologic manipulation of biologic processes. Modulation of Fc receptor expression on Kupffer cells and hepatic sinusoidal endothelial cells would appear to have potential in the down-regulation of any process that involves production of circulating immunoglobulins. Details on receptor regulation are beyond the scope of this presentation. In general, it is apparent that receptor regulatory processes involve both extra-and intracellular signals, and that these processes vary between individual receptors. Studies in a number of systems have shown that the presence and level of ligand can influence the level of receptor expression. As a general scenario, presence of an increased level of ligand leads to increased receptor-ligand binding and internalization of the receptor-ligand complex, resulting in a deficient population of free receptors on the cell surface. Intracellular signaling pathways result in increased transcription/translation processes that replenish the deficient receptor population. In a second pathway in some receptor systems, the receptor is separated from the ligand in the early endosome and the receptor is recycled to the cell surface. The fate of Kupffer cell Fc receptors following internalization is unknown at present. There is conflicting data as to whether Fc receptors on Kupffer cells are recycled following binding to immune complexes, or are destroyed along with the immune complexes in lysosomes (Mellman and Plutner, 1984).
Additional systems of receptor regulation are continually identified. As an example, the cytoplasmic ubiquitin-proteasome system, which is recognized primarily for its role in marking, degradation, and recycling of proteins, has been identified as a regulator of the endocytosis of selected membrane proteins (Strous and Govers, 1999), suggesting the rate of degradation of engulfed molecules has some feedback that influences the rate of pinocytosis of additional molecules. This observation has significant implications if the rate of disposition of effete receptors by the ubiquitin-proteasome system is found to be a generalized rate-limiting step in regulation of receptor expression.
In addition to Fc receptors for binding and phagocytosis of erythrocytes coated with IgG, Kupffer cells also have complement receptors for binding and phagocytosis of erythrocytes coated with complement fragments (human C3b or mouse inactivated C3b) (Smedsrod et al., 1985b). The avid binding of immunoglobulin- or complement-coated erythrocytes allows Kupffer cells to have a major role in removal of erythrocytes from the circulation, resulting in the well-known Kupffer cell accumulation of iron-positive materials in disease processes that involve intravascular erythrolysis or erythrocyte sequestration.
Nonspecific Cell-Killing
Effector cells involved in nonspecific, intrahepatic cell-killing include natural killer (NK) cells and natural killer T (NKT) cells. An additional liver-specific population of NK cells, known as pit cells, has the general phenotypic and physiologic features of NK cells, plus some additional features that identify them as a specific subpopulation.
NK cells are bone marrow-derived mononuclear cells that have markers of both T lymphocytes and macrophages. A characteristic microscopic feature of NK cells is the presence of distinct azurophilic cytoplasmic granules, thus the classical name of “large granule leukocyte”. The azurophilic granules correspond to the osmiophilic granules seen via electron microscopy (Bouwens et al., 1987). The cytoplasmic granules have been shown to contain perforin and granzymes, which are involved in cell membrane attack and induction of apoptosis in target cells. As opposed to target recognition by cytotoxic T lymphocytes, recognition of target cells by NK cells is not restricted to major histocompatibility complex (MHC) antigen presentation. In fact, NK cells are known to selectively kill cells that have a deficient level of MHC surface presentation. No immunologic memory is involved in NK cell killing. NK cells exert antitumor effects by exocytosis of perforin/granzyme-containing granules, induction of apoptosis in target cells, and production of various cytokines that augment the functions of other immune cells (Nakatani et al., 2004).
With increasing age there are changes in the hepatic population of NK cells. Mouse liver contains high T-cell receptor (TCR(hi)) and intermediate TCR (TCR(int)) cell populations. The (TCR(int) population, which includes NK cells, is subdivided into NK1+ and NK1− subsets based on presence or absence of the NK surface marker. Natural killer activity in the liver is associated with the NK1+ TCR(int) population, not the NK1− TCR(int) or TCR(hi) populations. The NK1+ TCR(int) cell population in the liver increases until middle age, then declines, resulting in a reduction in this critical first-line defense against invading tumor cells at an age when tumor metastasis to the liver is most likely.
NK cell activity in the liver has a major role in defense of the liver against invading tumor cells. NK cell killing of target cells is mediated via 2 major pathways. The Fas/FasL pathway involves binding of the ligand, FasL, to the receptor Fas and subsequent activation of “death domain” signaling elements, resulting in activation of the caspase cascade and apoptosis. The perforin/granzyme pathway involves perforin-mediated introduction of pores in the cell membrane and introduction of granzymes into the cytosol. The perforin/granzyme pathway essentially constitutes a shortcut that bypasses the death domain-containing signaling molecules on the surface of target cells and proceeds directly to the downstream caspase cascade. Granzyme B activates caspase-3, which then removes the propeptide of caspase-7 and allows activation of the executioner caspase-7 by granzyme B (Yang et al., 1998). The perforin/granzyme pathway serves as a backup defense against cellular targets that are resistant to Fas/FasL-mediated apoptosis. Human and rat hepatic NK cells utilize the perforin/granzyme pathway exclusively in the induction of apoptosis in tumor cells that are resistant to killing by splenic or blood NK cells (Vermijlen et al., 2002)
Granzymes are the effector molecules in perforin/granzyme-mediated apoptosis, thus considerable attention has been given to details of granzyme activity and modulation of that activity. Available evidence suggests granzymes have multiple pro-apoptotic effects on the apoptosis cascade. Mitochrondria are thought to be involved in at least 1 pathway of granzyme B-associated apoptosis induction, as granzyme B and perforin cause the release of cytochromc c into the cytosol before the activation of apoptosis. Granzyme B-induced apoptosis is highly amplified by mitochrondria in a caspase-dependent manner, but granzyme B can also initiate caspase 3 processing and apoptosis in the absence of mitochrondria (MacDonald et al., 1999).
A number of signaling molecules are known to influence Fas/FasL- or perforin/granzyme-mediated apoptosis induction. Prominent among these is IL-18, which was first identified as an interferon-gamma (IFNγ)-inducing factor that is produced by activated Kupffer cells. IL-18 promotes Fas/FasL-mediated killing by NK cells (Tsutsui et al., 1996) and augments perforin/granzyme-mediated killing by NK-T cells (Dao et al., 1998). Kupffer cells also produce IL-12, which was first identified as NK cell-stimulating factor (Tsutsui et al., 1996). Identification of IL-18 (apoptosis enhancement) and IL-12 (NK cell stimulation) by Kupffer cells is further evidence of the synergistic cooperation between Kupffer cells and NK cells in cell-killing.
Pit cells are intrasinusoidal, liver-specific NK cells that are defined morphologically as large granular lymphocytes (LGLs) and functionally as liver-associated natural killer (NK) cells. Pit cells are located inside sinusoidal lumina, where they adhere to endothelial cells and Kupffer cells. As with NK cells, there is an apparent synergism in some actions of Kupffer cells and pit cells. Morphologic features of pit cells suggest they represent a more mature form of circulating NK cells (Nakatani et al., 2004).
Pit cells remain in the liver approximately 2 weeks, are dependent on Kupffer cells, and proliferate locally in the liver when stimulated by IL-2. Pit cells adhere directly to target tumor cells during killing, and act synergistically with Kupffer cells in killing tumor cells (Wisse et al., 1997). Pit cells exhibit a high level of endogenous cytotoxicity against a variety of tumor cells, similar to lymphokine-activated killer (LAK) cells. Pit cells have a higher level of cytotoxicity against certain tumor cells and have a higher level of expression of the cell adhesion molecule CD11a as compared with blood NK cells. Studies have shown that CD11a/CD18 (LFA-1) present on pit cells plays an important role in pit cell-mediated adhesion, lysis, and apoptosis of tumor cells (Luo et al., 1999). Given the need for pit cells to adhere to target cells in order to initiate apoptosis and the role of CD11a in NK cell adhesion, it would appear that CD11a has a central role in the overall effectiveness of pit cells in tumor cell killing.
Pit cells can be divided into high- and low-density sub-populations (HD and LD, respectively). Both HD and LD NK cells express the NK activation markers gp42, CD25, and ANK44 antigen. LD cells, like IL-2-activated NK cells, have a high expression of perforin, granzymes, INFγ, and TNFα. Pit cells can be considered as in vivo activated NK cells (Luo et al., 2001).
Not all features of the hepatic microenvironment are pro-apoptotic. Hepatic sinusoidal endothelial cells express serine protease inhibitors 6 and 9 (SPI-6 and SPI-9), which inhibit the perforin/granzyme pathway and thus alter the liver microenvironment to hinder pit cell-mediated killing of metastatic tumor cells (Vermijlen et al., 2002). Pharmacologic interference with SPI-6 and -9 would appear to be a possible mechanism for potentiating pit cell-mediated killing of tumor cells in the liver.
The NKT cell population is currently considered to be separate from NK and pit cell populations. NKT cells have many of the phenotypic and physiologic characteristics of NK cells, including surface expression of T cell receptor (TCR). The TCR on NKT cells interacts with CD1, as opposed to the MHC-1 or MHC-2 interaction with the TCR on T lymphocytes. Antigen processing into either MHC-1 or MHC-2 context involves time-consuming steps for antigen processing and molecule transposition to the cell surface. By contrast, constitutive expression of CD1 on cell surfaces allows NKT cells to interact with target cells without delay, thus constituting a rapid response team. NKT cells are abundant in the liver (Nakatani et al., 2004), and it has been shown that NKT cells can develop extrathymically from liver precursors (Shimamura et al., 1997). This liver-resident, locally regenerating pool of rapid response killing cells has a significant role in defending the liver from invading tumor cells.
Presence of a pro-inflammatory cytokine profile in the normal healthy liver suggests the normal liver is maintained in a constant state of inflammation, and high hepatic levels of IL-12 help maintain this pro-inflammatory status. In addition to the effects on NK cells described above, IL-12 promotes the maturation of CD8+cells, double positive CD4+ CD8+T cells, and NKT cells, all of which have cytotoxic activities (O’Farrelly, 2004).
With these effective, redundant systems for nonspecific cell killing in the liver, metastasis of neoplastic cells to the liver should be uncommon. However, the high mortality rate of liver cancer in humans is largely due to liver metastasis, suggesting a failure in liver defense mechanisms. It has been shown that pit cell-mediated killing of rat colon carcinoma cells occurs through a Ca++-dependent perforin/granzyme apoptotic pathway rather than a Ca++-independent Fas apoptotic pathway (Vermijlen et al., 1999), suggesting that 1 of the 2 major nonspecific tumor cell-killing mechanisms is inoperative in the case of metastatic colonic carcinoma in the liver.
In addition to the colon carcinoma-associated alterations in nonspecific cell-killing, colon carcinoma cells are known to influence adaptive immune processes in the liver. These effects are mediated primarily through the Fas/FasL signaling pathways. Binding of FasL to Fas initiates the death domain-mediated steps that activate the apoptosis cascade in many cells, resulting in apoptosis. The interaction between Fas and FasL is employed for apoptosis induction in numerous sites and circumstances, and has been implicated in the maintenance of immunologically priviledged sites such as the eye. FasL is not produced by normal colonic epithelial cells but is expressed by a large percentage of metastatic colonic carcinoma cells. Expression of FasL by metastatic colonic carcinoma cells permits the apoptotic deletion of activated T cells in the liver through a “Fas counterattack,” thus creating an immunologically privileged site for colon carcinoma cells in the liver (O’Connell et al., 2000). FasL also induces apop-tosis of Fas-expressing hepatocytes, the death of which allows ready expansion of colon carcinoma metastases (Shiraki et al., 1997; Yoong et al., 1999). In addition, rat colon carcinoma cells have been shown to induce apoptosis in hepatic endothelial cells by the Fas/FasL pathway (Vekemans et al., 2003).
Disposal of Waste Molecules
Hepatic sinusoidal endothelial cells (SEC) have a unique cell marker phenotype that suggests myeloid lineage, though available evidence indicates SEC are derived from hepatocyte precursors (O’Farrelly, 2004). SEC resemble immature dendritic cells more than endothelial cells of other organs, and may represent a population of organ-specific antigen-presenting cells.
SECs have a voracious appetite for circulating molecules, to the degree that SEC are known as professional pinocytes. SEC receptor-mediated endocytosis occurs primarily via 4 categories of surface receptors: collagen receptor, mannose receptor, scavenger/hyaluronan receptor, and Fc receptor.
Collagens are dynamic molecules that constitute a major component of the mammalian body. Formation and turnover of collagen results in the release of large quantities of collagen components into the circulation, and these must be cleared in order to prevent untoward immunologic and other consequences. SEC of rat liver express a type of receptor that specifically recognizes and mediates the endocytosis of collagen alpha 1 monomers and denatured collagen (gelatin) (Smedsrod et al., 1985a). However, clearance of all collagen products is not necessarily mediated via the collagen receptor on SEC. Circulating C-terminal propeptide of type I procollagen is cleared mainly via the SEC mannose receptor (Smedsrod et al., 1990a) and clearance of NH2-terminal propeptides of types I and III procollagen is a function of the SEC scavenger receptor (Melkko et al., 1994).
Carbohydrates function as labels that mark circulatory glycoproteins for rapid clearance. The mannose receptor on SEC, which recognizes terminal mannose residues on macromolecules, appears to represent an essential element in the regulation of serum glycoprotein homeostasis (Lee et al., 2002). Proteomic analysis of mice that were genetically deficient in mannose receptor had elevated levels of 8 different lysosomal hydrolases as well as 4 additional proteins that are up-regulated during inflammation and wound healing, indicating that functional hepatic mannose receptor is important in the control and resolution of inflammation.
Endocytosis via the mannose receptor is extremely rapid. The endocytotic rate constant for mannose receptor-mediated internalization of ovalbumin by isolated liver endothelial cells was 4.12 min−1, corresponding to a half-life of approximately 10 seconds for surface receptor-ligand complexes. This is one of the fastest known rates of internalization of a receptor-ligand complex (Magnusson and Berg, 1989).
SECs contain a high level of many lysosomal enzymes, and some specific enzyme levels are actually higher in SEC than in professional phagocytes such as Kupffer cells (Knook and Sleyster, 1980). The high lysosomal enzyme activity of SEC is explained in part by the fact that SECs sequester the enzymes from the circulation via attachment of the mannose receptor to a terminal mannose on the enzymes (Smedsrod and Tollersrud, 1995). In contrast to other engulfed macromolecules, lysosomal enzymes are preserved and remain physiologically active within SECs.
The scavenger receptor of SEC cells assists in the clearance of numerous physiological and foreign waste macromolecules from the blood, including polysaccharides and proteins released during turnover of the extracellular matrix, intracellular macromolecules, modified serum proteins, and bacterial and fungal proteins (Smedsrod et al., 1990b). Bony fishes have a similar system of nonmacrophagic scavenger endothelial cells in either the kidney or heart, but not in liver (Sorensen et al., 1998, 2001). Studies have shown that species from all 7 vertebrate classes have a population of nonmacrophagic scavenger endothelial cells that eliminate soluble waste macromolecules from the blood, constituting an important part of the innate immune system (Seternes et al., 2002).
Hyaluronan is a widely distributed component of connective tissue. Blood levels of hyaluronan are increased in immune-mediated and liver diseases as well as certain malignancies. The major route of clearance of hyaluronan from the blood is via the liver, where it is taken up predominantly by SEC. An increased level of circulating hyaluronan may result from increased connective tissue synthesis/destruction or impaired hepatic capacity for waste molecule removal, as occurs in hepatic cirrhosis. The hyaluronan receptor shares functional properties with the scavenger receptor family, a group of proteins shown to play a key role in the uptake of atherogenic lipids and other waste molecules (McCourt et al., 1999).
The Fc receptor on SEC has specificity similar to the Fc receptor on Kupffer cells. Both cell types are capable of removal of waste immunoglobulins from the circulation, though Kupffer cells have the major activity. In addition to binding the Fc region of IgG, the Fc receptor on SEC is also capable of binding IgA and IgA complexes (Kuiper et al., 1994).
Liver Involvement in Adaptive Immunity
In addition to serving as an arena for a number of immune-mediated pathologic processes, the normal liver has continual direct involvement in adaptive (specific) immunobiology. Major facets of this involvement consist of (1) deletion of activated T cells that originate from inflammatory reactions at any site in the body, (b) induction of tolerance to ingested and self-antigens, (c) extrathymic proliferation of T lymphocytes, and (d) disposal of waste molecules that result from immunologically mediated events. The major cellular elements involved in these processes include dendritic cells, macrophages, and various subpopulations of T lymphocytes.
Dendritic Cells
Dendritic cells (DCs) are professional antigen-presenting cells (APC) that are located in numerous tissues. Based on location and other features, DCs are classified as interstitial or interdigitating DC. DCs in specific locations have specific names, e.g., Langerhans cells in the skin and circulating veiled cells.
Normal rat liver contains resident dendritic cells that mature as they migrate from the portal vein to the central vein. Hepatic sinusoids serve to select and concentrate circulating dendritic cells into the hepatic regional lymph nodes. Mature DCs in the central region of the hepatic acinus traverse the space to Disse and enter the hepatic lymph system (Sato et al., 1998). Some unspecified interaction between Kupffer cells and DCs is involved in the translocation of DC from blood to hepatic lymph (Kudo et al., 1997). After intravenous injection of particulate matter to rats, particle-laden dendritic cells (DCs) were present in hepatic lymph and migrated to thymus-dependent areas of regional lymph nodes (Matsuno et al., 1996).
Lymphocytes (T Cells and CTL)
T cells, which express the CD3 surface marker, are sub-categorized as CD4+ helper (Th) or CD8+ cytotoxic (CTL) populations. CD4+ helper T cells are further classified as Th1 or Th2 subsets based on the cytokine profile of the cells. Interferon gamma (IFNγ) is a commonly used marker for the Th1 subset, while IL-4 is a common marker for the Th2 subset.
Resident hepatic T cells are phenotypically different from T cells in blood, lymph nodes, or spleen. The CD4:CD8 ratio of hepatic T cells is reversed (1:3.5 for liver versus 2:1 for blood lymphocytes). The liver has a higher percentage of CD3+CD4+CD8+ (double positive) and CD3+CD4−CD8− (double negative) populations. More than 15% of CD3+ lymphocytes from human liver express TCRγδ rather than TCRα β, compared to the 2.7% TCRγδ population that is seen in blood. The CD8α chain with no CD8β chain was present in 15.4% of hepatic CD3+ cells, while that phenotype was not observed in blood CD3+ cells. These observations suggest local control of function and/or differentiation of hepatic lymphocyte populations (Norris et al., 1998).
The lytic activity of cytotoxic T lymphocytes (CTL) can occur by at least 2 pathways. In the perforin/granzyme-mediated pathway the pore-forming agent perforin, probably in conjunction with granzymes, induces apoptosis in target cells. In the Fas-mediated pathway, engagement of Fas and FasL trigger apoptosis of the CTL-bound target cell by a death domain-initiated caspase cascade. Regardless of the initiating pathway, the downstream events that lead to apoptosis appear to be similar (Berke, 1995).
In the classic model of perforin-mediated CTL killing, which is similar to NK cell-killing, perforin contained in cytoplasmic granules of CTL introduces pores in the plasma membrane of target cells and granzymes, also contained in CTL granules, penetrates through the membrane pores to initiate the apoptosis cascade in target cells. These processes require physical binding of the CTL to the target cell, thus the process is classified as juxtacrine rather than paracrine.
Recent evidence suggests a more complex mechanism of apopotosis induction by CTL. Perforin and granzymes exist as multimeric complexes with the proteoglycan ser-glycin within CTL granules. Monomeric perforin alone, as well as perforin-serglycin complexes, have been shown to mediate cytosolic delivery of macromolecular complexes of granzyme-serglycin without production of detectable plasma membrane pores, suggesting the traditional model of “punch hole and deliver granzymes” is somewhat oversimplified. Target cells apparently receive granule contents as a trimeric complex consisting of serglycin, perforin, and granzymes that are, respectively, the scaffold, translocator, and targeting/informational components of the modular delivery system (Metkar et al., 2002).
The serine protease granzyme B is crucial for the rapid induction of target cell apoptosis by cytotoxic T cells (CTL). In addition to target cell delivery via the perforin-mediated pathway, granzyme B has been shown to enter target cells via a perforin-independent pathway. This suggests the existence of a surface receptor for granzyme B. Recent studies have shown the granzyme receptor is the same as the cation-independent mannose 6-phosphate/insulin-like growth factor receptor. Absence of this receptor on tumor cells would render the tumor cells less sensitive to granzyme-mediated apoptosis induction, thus is a potential mechanism for tumor cell evasion of the immune system (Motyka et al., 2000).
Deletion of Activated T Cells
After an inflammatory reaction subsides, there remains a population of immunologically active cells and molecules that require neutralization or disposal. The liver has a major role in disposal of circulating macromolecules, as discussed before, and has a specific role in removal of T cells that were activated by inflammation at sites distant from the liver.
A basic understanding of leukocyte emigration from blood vessels is required for discussion of activated T cell clearance by the liver. Leukocyte emigration from postcapillary venules into inflamed tissues is a 2-step process that involves an initial low-avidity, selectin-mediated rolling process followed by a more tenacious integrin-mediated adhesion of leukocytes to the luminal surface of endothelial cells. The initial rolling step serves to slow the movement of leukocytes to a rate where they have time for contact with integrin molecules expressed on activated endothelial cells. Integrin molecules involved in the final adhesion are up-regulated and expressed on the luminal surface of endothelial cells following the release of cytokines by various cell populations in the local inflammatory reaction. The combination of selectin- and integrin-mediated adhesion serves to localize leukocyte emigration to the site of inflammation.
Leukocyte emigration in the liver occurs in sinusoids rather than postcapillary venules, and does not require the initial selectin-mediated rolling step (Wong et al., 1997). Selectin-mediated rolling is not required because hemodynamic factors, coupled with Kupffer cell migration and leukocyte interactions with vessel walls, serve to slow the rate of blood flow through liver sinusoids. These alterations in rate of blood flow vary in different regions of the hepatic acinus, and vary between species (MacPhee et al., 1995). The end result is that leukocytes in hepatic sinusoids have extensive, slow-motion exposure to integrin-type adhesion molecules without the need for a preliminary rolling step.
Endothelial cells in vessels at sites of peripheral inflammation have increased expression of adhesion molecules such as ICAM-1 and VCAM-1, which are up-regulated by cytokines produced locally at sites of inflammation. Local expression of ICAM-1 and VCAM-1 serve to localize emigration of leukocytes to the site of inflammation. SECs exhibit constitutively high expression of ICAM-1 and VCAM, which facilitates integrin-mediated adhesions in the absence of a local inflammatory reaction. Intrahepatic accumulation of murine CD8+T cells was reduced significantly when either ICAM-1 or VCAM-1 was blocked by specific antibody, suggesting these adhesion molecules are responsible for the majority of trapping of activated CD8+T cells in the mouse liver (John and Crispe, 2004).
Perfusion of the liver with lymphocyte mixtures revealed selective hepatic retention of activated CD8+T cells, but not resting T cells or T cells that were actively involved in apoptosis. Following trapping there was a temporary expansion of the population of activated T cells in the liver, but the massive proliferation of T cells seen at sites of inflammation does not occur in the liver. Activated CD8+T cells underwent an approximately 8-fold expansion in the first 48 hours after injection into the portal vein of mice (Kuniyasu et al., 2004). The expansion was followed by apoptosis and a reduction in the intrahepatic population of CD8+T cells over the next four days.
Failure of continued expansion of the trapped CD8+T cell population may be explained by the cytokine microenvironment of the liver, which is different from the microenvironment of a macrophage-and helper T cell-populated inflammatory reaction. Activated T cells must receive a costimulatory signal, often in the form of B7 from macrophages, otherwise the activated T cells undergo apoptosis. Input from helper T cell populations is also critical for survival and expansion of activated T cell populations. Absence of costimulatory and helper molecules in the liver microenvironment results in apoptosis of sequestered activated T cells (“death-by-neglect”). As a result of these processes the liver is regarded as a “sink” for activated T cells (Mehal et al., 1999).
In addition to the death-by-neglect mechanism presented previously, there is evidence for direct apoptosis induction in trapped CD8+T cells. Galectin-1 is a beta-galactoside binding protein that is produced by endothelial cells and induces apoptosis in bound activated T cells (Lotan et al., 1994; Perillo et al., 1995; Rabinovich et al., 1998). As a result of this activity, activated T cells that are bound to sinusoidal endothelial cells may be actively killed by galectin-1 in addition to dying of neglect due to lack of a costimulatory signal.
Intrasinusoidal binding of activated T cells appears to be a general clearance mechanism, and may explain the accumulation of T cells in the liver of mice with defects in apoptosis mechanisms, such as the lpr/lpr mutant (Huang et al., 1994). In this situation T cells are bound in sinusoids but lack the necessary intracellular machinery to complete the apoptotic process (Cohen and Eisenberg, 1991).
Despite the fact that deletion of activated T cells is a major function of the liver, CD8+T cells that have been recently activated by immunization and circulate through the liver can survive long enough to transiently modulate the phenotype of hepatocytes, allowing hepatocytes to activate naïve CD8+T cells that have unrelated antigenic specificity (Dikopoulos et al., 2004). This observation raises the possibility that any systemic immune response, including immunization, may increase the chance of unrelated immunologic response to hepatocyte-presented antigens, which may include ingested or self-antigens.
Induction of Tolerance to Ingested and Self-Antigens
During induction of cell-mediated immunity, professional, marrow-derived antigen-presenting cells (APC) typically present antigens to CD8+ T cells in association with MHC molecules, thereby inducing protective immunity against intracellular microorganisms. Formation of fully functional effector cells requires the activity of helper T cells. Liver sinusoidal lining cells can take up antigen, process the antigen and present it to T cells but, probably due to the lack of input from helper T cells, the end result is tolerance rather than immunity (Rubinstein et al., 1986, 1987; Limmer et al., 2000). This major function of sinusoidal endothelial cells prevents immunologic reaction to the wide spectrum of potentially antigenic molecules that are assimilated from the gastrointestinal tract.
In addition to a myriad of other functions, hepatocytes also function as antigen-presenting cells. Extension of hepatocellular microvilli through intercellular junctions between SEC allows direct contact between hepatocytes and naïve CD8+T cells. However, T cell activation by hepatocytes leads to premature T cell death or tolerance rather than the formation of fully competent CTL. T cells activated by hepatocellular antigen presentation are phenotypically different from T cells that are activated by professional APC in the spleen or lymph nodes. T cells activated by hepatocytes express a lower level of the bcl-x(L) survival gene products and have less IL-2 mRNA than T cells that are activated by splenic antigen-presenting cells. Apoptosis of hepatocyte-activated T cells is suspected to be an example of death by neglect resulting from absence of an effective costimulatory signal. Cross-linking of CD28 on hepatocyte-stimulated T cells, which simulates costimulation by B7 molecules from macrophages, abrogated the early apoptosis of T cells, caused an increase in expression of bcl-x(L) and IL-2 in hepatocyte-activated T cells and resulted in sustained T cell proliferation and cytotoxic activity (Bertolino et al., 1995, 1998, 1999).
The mechanisms of induction and maintenance of tolerance in self-reactive T cells in the periphery are poorly understood. Current models assume that successful T cell activation occurs only if ligation of the T cell receptor (signal 1) is accompanied by a costimulatory signal (signal 2), and that signal 1 in the absence of signal 2 is either ignored or induces tolerance. In addition to this widely accepted mechanism, there is accumulating evidence for the existence of a suppressor macrophage population (Mφ) that actively promotes tolerance in T cells (Attwood and Munn, 1999).
Extrathymic T Cell Proliferation
T cell development and differentiation occur primarily in the thymus, but not exclusively so. Extrathymic pathways to T cell differentiation exist in the intestine and liver (Sato et al., 1995), where T cell populations may arise from their own preexisting precursor cells rather than thymus (Sugahara et al., 1999). Recombinase-activated genes 1 and 2 (RAG1 and RAG2), which are involved in gene rearrangements in lymphocytes, are present in high levels in lymphocytes extracted from adult human liver, suggesting lymphocyte differentiation occurs in adult liver (Collins et al., 1996).
Extrathymic T cell differentiation is of minor importance in normal mice, but becomes predominant in mice with autoimmune diseases, athymic mice, and aged mice. Extrathymically derived T cells in mice have a number of phenotypic alterations, including an intermediate level of expression of TCR (TCR(int)). TCR(int) cells are generated in situ in the liver parenchyma and subsequently migrate to the sinusoidal lumen. This is accompanied by a reverse migration of thymus-derived (‘normal’) T cells from sinusoids into the hepatic parenchyma (Yamamoto et al., 1999).
The liver may be involved in proliferation of leukocytes other than T lymphocytes. C-kit+ cells are essential for hematopoiesis, especially for the self-renewal of hemopoietic progenitor cells (Ogawa et al., 1991). Livers of adult mice are known to contain c-kit+ stem cells that can reconstitute thymocytes, multiple lineage cells and bone marrow stem cells (Taniguchi et al., 1996; Watanabe et al., 1996).
Multiple observations suggest a gender difference in immunologic responses in the liver. There is a female preponderance in the incidence of autoimmune hepatitis and primary biliary cirrhosis in humans, and several studies have suggested that gender mismatching has a major impact on the outcome of orthotopic liver transplantation (McFarlane and Heneghan, 2004). There is a synchronous expansion of TCR(int) cells in the liver and uterus during pregnancy (Kimura et al., 1995). Estrogen administration activates extrathymic T cells in mice. TCR(int) cells are more common in the liver and other organs of female mice than male mice (Kimura et al., 1994). Ovariectomy of female mice results in a reduction in the population of TCR(int) cells in the liver, and the effect is reversed by administration of physiologic doses of estrogen (Yahata et al., 1996). Injection of estrogen into male mice has been shown to increase the population of hepatic TCR(int) cells and a reciprocal reduction in thymic T cell differentiation as hepatic T cell differentiation is enhanced. It is interesting to note that estrogen replenishment resulted in an increase in the number of mononuclear cell aggregates in liver. Due to less stringent negative selection processes in the liver, extrathymic T cells have pronounced autoreactive characteristics. This combination of observations suggests estrogen has a role in the known predominance of autoimmune diseases in females of several mammalian species (Okuyama et al., 1992).
Disposal of Waste Molecules from Immune-Mediated Reactions
In addition to Fc receptor-mediated assimilation and disposal of immunoglobulin molecules, as discussed before under innate immunologic functions, there is accumulating evidence that the liver is a major site of assimilation of cytokines. The liver has a large number of cells with cytokine receptors and a high density of cytokine receptors per cell, thus is regarded as a cytokine sink (Fey et al., 1994).