Overview

Cytokines are small, nonstructural proteins that are key regulators of the human response to noxious stimuli. Cytokines can be proinflammatory, antiinflammatory, or both depending on the timing and context in which they are secreted. Nearly all nucleated cells are capable of both synthesizing and responding to cytokines, but most cytokine genes are not expressed unless stimulated by a specific stressor. This chapter reviews the roles that specific stimuli—including endotoxin, microbial products, and cytokines—play in both the normal physiology and pathophysiology of liver, biliary, and pancreatic disease, with particular attention to surgical problems. Both a metabolic and immunologic organ, the liver has been the subject of extensive research, but cytokines undoubtedly play a role in pancreatic pathophysiology as well. The entire field of cytokine research has continued to grow exponentially since the last version of this text, with several key discoveries that are directly applicable to surgical practice. Exciting areas of therapeutics based on this new knowledge are also discussed.

Microbiologic Recognition and Toll-Like Receptors

In mounting an immune response, the challenge to the host is to identify the presence of an infectious process early and to coordinate its considerable resources to eradicate the pathogen, while minimizing tissue damage and chronic inflammation. The recognition of microbial pathogens and the ability to distinguish self from nonself relies heavily on the innate immunity and cell surface receptors present on antigen-presenting cells in the liver (Akira & Hemmi, 2003). Antigen-rich blood from the portal circulation has the potential to activate either the innate and adaptive immune responses against infections or to maintain immunologic tolerance of harmless antigens (Tacke et al, 2009). In addition, the infiltration of monocytes during liver injury is an important adaptive mechanism leading to chronic infammation (Karlmark et al, 2009). The immune response to a biologic insult is a highly organized process that involves pattern-recognition receptors (PRRs) that identify preserved structures of different pathogens. Toll-like receptors (TLRs) are an important family of at least 10 PRRs that are ubiquitously expressed in humans (Akira & Takeda, 2004). Current evidence demonstrates crucial roles for TLRs in alcoholic liver disease, nonalcoholic steatoheaptitis, hepatitis B, hepatitis C, hepatic fibrosis and cirrhosis, hepatocellular carcinoma, primary biliary cirrhosis, acetaminophen-induced hepatotoxicity, and ischemia-reperfusion (I/R) liver injury, all of which are potential therapeutic targets (Pimental-Nunes et al, 2010; Table 10.1).

Table 10.1 Toll-like Receptors and Their Potential Targets for Gastrointestinal Diseases

Endotoxin Structure and Biologic Activity

Lipopolysaccharides (LPSs) from gram-negative bacteria have a wide range of harmful biologic activities and are therefore also considered endotoxins (Dinarello, 2004). Research over the past decade has elucidated the mechanisms by which the host recognizes endotoxin and the innate immune response is initiated. Much more is known about the cell surface receptors and signaling pathways involved in the host response to endotoxin than is known about several other microbial products and TLRs. Although endotoxin has been implicated in the beneficial and injurious host responses to gram-negative infections, the exact role of endotoxin in the pathogenesis of human disease remains unclear. Similarities in host responses to various microbial products, including LPSs from gram-negative bacteria and the glycoconjugates of gram-positive bacterial and fungal infections, suggest that the host innate immune system has evolved to respond consistently to microbial invasion, regardless of its antigenic source (Fearon & Locksley, 1996).

Endotoxin is the major constituent of the outer cell wall of gram-negative bacteria and can comprise 65% by weight of the total bacterium. Endotoxin is shed spontaneously from the cell walls of living bacteria and is released in copious amounts upon cell death and cell lysis. Endotoxin comprises three major components: an inner lipid A, an intermediate R-core oligosaccharide, and an outer O-polysaccharide (Fig. 10.1). The inner lipid A portion and the R-core oligosaccharide generally are conserved, whereas the structure of the O-polysaccharide is unique for each strain of bacteria. The R-core oligosaccharide is exposed in rough forms of gram-negative bacteria. The internal lipid A moiety is poorly antigenic but seems to be a primary factor in inducing most of the biologic effects of LPS. Binding of endotoxin to serum proteins and its activation of host immunity through its cellular receptor are mediated predominantly via the lipid A moiety; the O-antigen is primarily immunogenic (Ulevitch & Tobias, 1999).

FIGURE 10.1 Schematic representation of the structure of endotoxin.

The host response to endotoxin is immediate (within minutes), is dose dependent, and affects all organ and tissue systems (Table 10.2). High-dose endotoxin exposure in rodents and primates reproduces the toxic effects of gram-negative bacteremia, characterized by hemodynamic collapse, shock, organ failure, and death (Beutler et al, 1985). Low-dose administration of endotoxin to human volunteers (Fong et al, 1990; Michie et al, 1988) produces a variety of constitutional symptoms consistent with a milder infectious process, including fever, myalgia, tachycardia, occasional hypotension, transient leukopenia followed by neutrophilia, and a hepatic acute-phase response (Table 10.3).

Table 10.2 Biologic Actions of Endotoxin

IL, interleukin; TNF, tumor necrosis factor; CSF, colony-stimulating factor; PAF, platelet activating factor

Table 10.3 Biologic Responses to Low-Dose Endotoxin in Human Volunteers

TNF-α, tumor necrosis factor α; IL, interleukin; s-TNFR, soluble TNF receptor

* ≤4 hours’ duration

† 6 to 24 hours’ duration

‡ 6 to 12 hours’ duration

Data from Fong et al, 1990: The acute splanchnic and peripheral tissue metabolic response to endotoxin in humans. J Clin Invest 85:1896-1904.

Although the host responses to endotoxin are likely to be dose dependent, there is now general acceptance that the clinical sequelae associated with systemic endotoxin administration are probably secondary to the host innate immune response rather than a result of direct interactions between endotoxin and cell membranes or secretory proteins (Fearon & Locksley, 1996; Ulevitch & Tobias, 1999). At high doses, endotoxin can produce direct endothelial injury, but most host responses to endotoxin are mediated by complement activation and the release of humoral factors, including proinflammatory cytokines, nitric oxide (NO), and prostaglandins.

Immune Regulation and Response to Endotoxin (SEE CHAPTER 9)

The host defense against many gram-negative bacteria is dependent on the innate immune recognition of endotoxin, and this response must be highly sensitive and self-limited. The sensitivity of this system is crucial for prompt mobilization to combat infection, but a risk of severe immune-mediated pathology exists if this response not self-limited. Optimal sensitivity is achieved by the strictly ordered interactions of endotoxin with different extracellular and cell surface proteins, including lipopolysaccharide (LPS)-binding protein (LBP), cluster of differentiation (CD)14, MD-2, and Toll-like receptor 4 (TLR-4) (Gioannini & Weiss, 2007). The presence of endotoxin in the plasma and lymphatic system is initially recognized by serum proteins and lipoproteins, and this recognition initiates the activation of the innate immune response (Beutler et al, 2003). Redundancy is the hallmark of this activation process, as proteins, glycoproteins, lipids, and nucleic acids of prokaryotic origins are recognized by different TLRs and other components of innate immunity. For the most part, the innate immune system relies on cell surface receptors and hepatic secretory proteins, primarily opsonins, to recognize carbohydrate and lipid, protein, and DNA structures indicative of a microbial infection.

Different constituents of specific microbial products, including endotoxin, are recognized by different aspects of the innate immune system (Ulevitch & Tobias, 1999). The first step in the recognition of endotoxin by the innate immune system is its binding to a hepatic secretory protein, LBP, to form the LPS-LBP complex (Fig. 10.2). LBP is a member of a group of homologous lipid-binding proteins that function as lipid-transfer proteins (Tobias et al, 1986, 1989; Wright et al, 1989). Secreted by the liver and found in human blood in 2 to 20 mg/mL concentrations, LBP binds avidly to the lipid A moiety of endotoxin and mediates its transfer to the cell surface endotoxin receptor, CD14. Although CD14 functions as a ligand-binding protein for endotoxin-LBP complexes, CD14 is bound to the cell membrane through a glycosyl–phosphatidylinositol anchor, and it lacks an intracellular domain capable of transducing a signal. LPS-LBP complex binding to CD14 alone is insufficient to transduce a signal.

FIGURE 10.2 Endotoxin-binding proteins and signal transduction. Endotoxin in body fluids is rapidly recognized by endotoxin-binding proteins and lipoproteins. The balance between lipopolysaccharide binding protein (LBP) and bactericidal permeability–increasing protein (BPI) or high-density lipoprotein or both determines whether endotoxin in body fluids is detected by macrophages and other effector cells of the innate immune system. LBP-endotoxin complexes bind to CD14 and interact with Toll-like receptor-4 (TLR-4) and MD-2 membrane proteins, which activates intracellular myeloid differentiation response gene (88) (MyD88) dependent and independent pathways that ultimately lead to nuclear factor (NF)-κB activation and type I interferon production. IkB, inhibitor of NF-κB; IkK, inhibitor of NF-κB kinase; IRAK, interleukin-1 receptor–associated kinase; LPS, lipopolysaccharide; NIK, NF-κB–inducing kinase; NEMO, NF-κB essential modulator; TRAF, tumor necrosis factor receptor–associated factor; TRAM, Toll receptor–associated molecule; TRIF, Toll receptor–associated activator of interferon.

The signal-transducing component of the CD14 bipartite complex protein has been identified as a member of the family of TLRs, specifically TLR-4 (Beutler, 2002). Although TLR-4 is expressed in several cells of healthy liver, signs of inflammation are absent because of low expression levels of TLR-4 and modulation of TLR-4 signaling. Mounting evidence suggests that altered LPS–TLR-4 signaling plays a role in the pathogenesis of chronic liver disease (Soares, 2010). TLR-4 seems to interact with CD14, the LBP-endotoxin complex, and a third protein, MD-2 (Viriyakosol et al, 2001).

Defining the complex role of CD14 in endotoxin-induced activation of TLR-4 is a challenge, and whether this role is obligatory remains unclear (Beutler et al, 2006; Jiang et al, 2005). However, the activation of TLR-4 by endotoxin does appear to require the simultaneous binding of endotoxin and TLR-4 by MD-2 (Gioannini et al, 2004). TLR-4 signaling can occur via a Toll-like and interleukin (IL)-1 receptor (TIR) domain and the formation of a scaffold composed of members of myeloid differentiation factor 88 (MyD88) adaptor proteins. At the same time, TLR-4 signaling also occurs through a MyD88-independent pathway that involves the adaptor protein TIR domain–containing adaptor inducing interferon-β (TRIF). This latter pathway seems to be more essential for the expression of type I interferons (IFNs) and IFN-dependent proteins in response to endotoxin exposure (Yamamoto et al, 2003). Although it is presently unknown how CD14-dependent signaling is integrated from the cell surface via TLR-4, the confirmation of TLR-4 as the signaling complex is important, because it identifies a signal transduction pathway and potential therapeutic targets.

The presence of LBP in the plasma increases the sensitivity of cells bearing CD14 to endotoxin by at least 1000-fold. From a teleologic standpoint, this has distinct advantages, because endotoxin can be detected, and innate immunity activated early, in response to gram-negative infection. However, increasing the sensitivity of the innate immune response to endotoxin can have devastating pathologic consequences, as may occur during sepsis or when gastrointestinal integrity is lost. These pathologic consequences include loss of vascular integrity, shock, organ failure, and death. Hosts have evolved several mechanisms to reduce the likelihood of an exaggerated endotoxin-induced signal.

High-density lipoproteins can bind endotoxin in serum and may play a role in protecting against endotoxemia (Cue et al, 1994). An additional endogenous mechanism to suppress endotoxin responsiveness is through the release of an LBP-homologous protein, bactericidal/permeability-increasing protein (BPI; see Fig. 10.2). BPI is a 55- to 60-kD protein structurally similar to LBP, which binds endotoxin through its lipid A moiety. However, the resulting BPI-endotoxin complexes do not bind to CD14 or transduce a signal (Elsbach, 1998). BPI acts as an endogenous endotoxin inhibitor and has shown some modest beneficial effects in patients with meningococcal sepsis (Levin et al, 2000). In addition, BPI released from activated neutrophils can bind directly to endotoxin being expressed on the surface of gram-negative bacteria, and it is growth arresting and cytolytic for the bacteria (von der Mohlen et al, 1996).

BPI is stored in the granules of neutrophils and is released on neutrophil activation and degranulation. In a healthy adult, BPI levels in the serum are very low (15 to 50 ng/mL), but concentrations increase in response to an inflammatory or endotoxemic challenge (Calvano et al, 1994). In contrast, LBP concentrations are several logs higher in patients in the surgical ICU (2 to 20 mg/mL) and increase only modestly in response to inflammation. During infection, the ratio of LBP to BPI concentrations in the plasma approaches 1000:1, and this ratio favors endotoxin signaling. In closed-space infections infiltrated with neutrophils, however, BPI exceeds LBP concentrations, and BPI concentrations are directly proportional to neutrophil counts. These findings suggest that in the normal healthy adult and in patients with sepsis, the relationship between plasma LBP and BPI concentrations favors the recognition of endotoxin and activation of the innate immune response. In local infectious sites, where neutrophil and inflammatory cell recruitment has occurred, the increased BPI response is presumably compensatory and aimed at the continued activation of the immune system and reduction of bacterial growth.

Endotoxin signaling via the CD14 receptor and TLR-4 involves a novel IL-1 receptor-associated kinase, which activates a secondary kinase, NF-κB–inducing kinase (NIK) (Yang et al, 1999). Activation of this cascade rapidly induces the phosphorylation of additional kinases, which results in NF-κB translocation and the transcription of NF-κB–dependent genes (see Fig. 10.2). Genes containing NF-κB response elements are numerous and include most of the proinflammatory cytokines, including tumor necrosis factor (TNF)-α, IL-1, IL-6, IL-8, IL-12, IL-18, and IFN-γ (Blackwell & Christman, 1997).

Most of the systemic inflammatory responses to endotoxin are mediated by the release of proinflammatory cytokines and other humoral factors, predominantly TNF-α, IL-1, and IL-6 and, to a lesser extent, Fas ligand (FasL) (Suffredini et al, 1999). These proinflammatory cytokines not only globally regulate the inflammatory response to endotoxin, they also play a crucial role in reprogramming the metabolic and protein synthetic responses by the liver. As we will discuss in greater detail, TNF-α and IL-6 in particular play unique roles in regulating not only the hepatocyte acute-phase response to endotoxin but also hepatocyte proliferation versus apoptosis in liver regeneration (Michalopoulos & DeFrances, 1997; see Chapter 5), viral hepatitis (Hayashi & Mita, 1997; see Chapter 64), and toxic liver injury (Batey et al, 1999; Bradham et al, 1998; McClain et al, 1999). In contrast, FasL is a potent inducer of hepatocyte apoptosis and has been implicated in the pathogenesis of viral hepatitis (Kondo et al, 1997; Hayashi & Mita, 1997).

Tumor Necrosis Factor Superfamily

Since the discovery of TNF-α 35 years ago, the TNF superfamily has grown to comprise at least 20 related proteins that signal through greater than 30 receptors (Grewal, 2009). Members of the TNF superfamily are primarily homotrimeric proteins, with the exception of lymphotoxin, and they exist primarily in a membrane-associated form (Bazzoni & Beutler, 1996). As a general rule, members of the TNF family are primarily involved in the regulation of cell proliferation and apoptosis, although several of the members—including TNF-α, TNF-β, FasL, CD30L, and CD40L—also have proinflammatory properties.

TNF-α is synthesized as a bioactive cell-associated protein that is primarily involved in juxtacrine signaling of cytotoxicity. This 26-kD intermediate is enzymatically cleaved by a matrix metalloproteinase to a 17-kD secreted form that acts in a paracrine or endocrine fashion (Fig. 10.3). TNF-α originally was characterized as a factor that produced necrosis of solid tumors in vivo (Carswell et al, 1975). It subsequently was recognized that TNF-α modulates growth, differentiation, and metabolism in a variety of cell types; it can produce cachexia by stimulating lipolysis, inhibiting lipoprotein lipase activity in adipocytes, and stimulating hepatic lipogenesis; and it can initiate apoptosis in hepatocytes and lymphoid cells (Table 10.4; Ksontini et al, 1998).

FIGURE 10.3 Processing of tumor necrosis factor (TNF)-α, which is initially synthesized as a 26-kD membrane-associated protein that is processed by TNF-α converting enzyme (TACE) to a secreted 17-kD homotrimer. Cell-associated and secreted molecules can bind and transduce a signal through two TNF receptors (TNFRs; p55 [type I] and p75 [type II]). mTNF, membrane TNF; sTNF, soluble TNF.

Table 10.4 Biologic Activities of Tumor Necrosis Factor-α

Vascular Endothelial Cells

Fibroblasts

Adipocytes

Endocrine System

Central Nervous System

Cardiovascular

Gastrointestinal

Metabolic

Inflammatory

FSH, follicle-stimulating hormone; G-CSF, granulocyte colony-stimulating factor; GH, growth hormone; GM-CSF, granulocyte-macrophage colony-stimulating factor; ICAM, intercellular adhesion molecule; IL, interleukin; LIF, leukemia inhibitory factor; TSH, thyroid-stimulating hormone; VCAM, vascular cell adhesion molecule

TNF-α is a powerful inducer of the inflammatory response both directly and through stimulation of numerous downstream proinflammatory mediators. Secondary mediators that are known to be induced by systemically administered TNF-α include cytokines (IL-1, IL-2, IL-4, IL-6, IL-10, IL-12, IL-18, IL-23, type I and type II IFN, transforming growth factor-β, leukemia inhibitory factor [LIF], and macrophage migration inhibitory factor [MIF]), hormones (cortisol, epinephrine, glucagon, insulin, and norepinephrine), and assorted other molecules (acute-phase proteins, IL-1 receptor antagonist [IL-1Ra], leukotrienes, oxygen free radicals, NO, platelet-activating factor, and prostaglandins; Tracey & Cerami, 1994). Other principal biologic effects of TNF-α are listed in Table 10.4 (see Chapter 9).

Not only is TNF-α involved in tissue inflammation, but there is a growing recognition that it is also a prominent ligand for the activation of programmed cell death. Apoptosis occurs naturally during growth and development, but it may also result from certain pathologic conditions in which local and systemic production of TNF-α is increased. Hepatocytes are particularly sensitive to TNF-α–induced apoptosis, especially during simultaneous transcriptional inhibition (Leist et al, 1994).

TNF-α has been administered systemically in low doses to human volunteers and regionally in higher doses to sarcomas and melanomas. The physiologic responses to low-dose (50 mg/m²) TNF-α administration were remarkably similar to the responses seen with low-dose endotoxin administration: fever and constitutional symptoms of pain, headache, myalgia, and nausea. Hematologically, the volunteers developed a rapid leukopenia, followed by neutrophilia with sustained lymphopenia and monocytopenia (van der Poll et al, 1992). Plasma IL-6 levels increased 40-fold, and the volunteers developed an acute-phase protein response. Prostaglandin (6-keto prostaglandin [PG] F_1a) production also was markedly increased. Metabolically, the patients exhibited increased lipolysis and glucose turnover (van der Poll et al, 1991b). There also were significant effects on the vascular endothelium. Within 1 hour, TNF-α administration induced activation of the fibrinolytic pathway, followed by activation of the coagulation cascade (van der Poll et al, 1991a).

At higher doses of TNF-α, as achieved with regional perfusion of limbs, the systemic release of large quantities of protein occasionally occurs, and the effects on the vascular endothelium are profound. Under these conditions, activation of the endothelium is evident by increased release of soluble selectins and integrins, and hemodynamic instability is common (Aderka et al, 1998; Zwaveling et al, 1996); therefore continuous monitoring for systemic leakage is required.

Interleukin-1 Family

The IL-1 family of ligands and cytokines are similar to the TNF superfamily in that they are also primarily associated with inflammation. IL-1 possesses several biologic properties that result in increased expression of downstream proinflammatory genes. The most salient and relevant is the ability of IL-1 to initiate and sustain the expression of cyclooxygenase type 2 and inducible NO synthase (Dinarello, 1996). This property accounts for the large amount of PGE₂ and NO produced by cells exposed to IL-1 or in subjects injected with IL-1. Other important proinflammatory properties of IL-1 are its ability to increase IL-8 synthesis and to express adhesion molecules on endothelial cell surfaces, which accounts for the infiltration of inflammatory cells into the extravascular space. IL-1 also potentiates the biologic activities of TNF-α, and modest doses of the two in combination can be lethal in experimental animals.

The original IL-1 superfamily consisted of two members: IL-1α and IL-1β. Currently, the IL-1 superfamily also contains IL-1Ra, IL-16, IL-17, IL-18, IL-33, and at least six other homologues in the IL-1 family (Schmitz et al, 2005). IL-1α and IL-1β are agonists, and IL-1Ra is a specific receptor antagonist for the IL-1 receptor family. The naturally occurring IL-1Ra seems to be unique in cytokine biology (Arend, 1993). The intron–exon organization of three primary IL-1 genes suggests the duplication of a common gene some 350 million years ago. Processing of IL-1α or IL-1β to “mature” forms requires specific cellular proteases. In contrast, IL-1Ra evolved with a signal peptide and is readily transported out of the cell and is termed secreted IL-1Ra.

Even under conditions of cell stimulation, human blood monocytes do not process or readily secrete mature IL-1α, which is probably only released by dying cells, as it is a key factor released by necrotic cells that promotes inflammation (Chen, 2007). The IL-1α precursor proIL-1α is synthesized in association with cytoskeletal structures (microtubules), which is in contrast to most proteins translated in the endoplasmic reticulum. ProIL-1α is fully active as a precursor and remains intracellular. The opposite is the case with the IL-1β precursor proIL-1β, which is not fully active, and a considerable amount is secreted after cleavage by a specific, intracellular cysteine protease, IL-1β–converting enzyme, or caspase-1. Caspase-1 plays a central role in the biology of IL-1β, IL-18, and IL-33 and has become a therapeutic target for modulating IL-1–based diseases (Martinon & Tschopp, 2004). IL-33 is an activator of Th2 cells and has recently been shown to play a protective role in the development of atherosclerosis (Miller et al, 2008; Sanada et al, 2007) and is thought to be released by cell death (Carriere et al, 2007).

Nearly all microbes and microbial products induce production of the three IL-1 family proteins, but stimulants of nonmicrobial origin also can elicit transcription and synthesis. Stimulants such as the complement component C5a, hypoxia, adherence to surfaces, and clotting of blood induce the synthesis of large amounts of IL-1 mRNA in monocytic cells without significant translation into the IL-1 protein.

Although animal experiments revealed that IL-1 was proinflammatory, a great deal of information has been gathered from studies in which humans have been injected with recombinant IL-1α or IL-1β. Humans have received IL-1α infusions as part of clinical trials following bone marrow transplantation because of its hematopoietic properties. Although the duration of leukopenia and thrombocytopenia were reduced, patients had serious signs of systemic inflammation that included fever, hypotension, and flulike symptoms, similar to endotoxemia (Smith et al, 1993). Interestingly, the exquisite sensitivity of humans to IL-1 given systemically was not appreciated from animal studies.

Interleukin-6

IL-6 is another pleiotropic cytokine produced by a wide variety of cells, including T cells, B cells, endothelial cells, fibroblasts, monocytes, and macrophages. The biologic activities of IL-6 include immune regulation, hematopoiesis, inflammation, and oncogenesis (see Chapter 9). The IL-6 gene has been mapped to chromosome 7, and its product varies from 21 to 28 KD, depending on posttranslational modifications. IL-6 belongs to a much larger superfamily of related cytokines, including leukemia inhibitory factor, oncostatin M, ciliary neurotrophic factor, cardiotrophin-1, and IL-11. All members of this superfamily share a modest degree of structural homology, but more importantly, all these related cytokines use a common signal transduction pathway through heterocomplexes composed of gp130. Each ligand has its own receptor, which combines in a duplex to transduce its signal through the JAK/Stat signaling cascade.

Although microbial products are potent inducers of IL-6, the synthesis of IL-6 seems to depend on the release of more proximal cytokines, primarily TNF-α and IL-1. The biologic activities of IL-6 are numerous and can be divided into those involved in the regulation of the hematopoietic system and those involved in activation of the innate immunity. Recent studies have elucidated a crucial role for IL-6 in regulating the balance between T-regulatory cells and a new subset of T-helper cells that produce IL-17 (Kimura & Kishimoto, 2010). In the liver, IL-6 is secreted by Kupffer cells and is an important regulator of innate immune response and a stimulator of hepatocyte proliferation. Although IL-6 shares several biologic responses with TNF-α and IL-1, IL-6 is the predominant regulator of the hepatic acute-phase response and induces hepatocyte proliferation during liver regeneration (Table 10.5).

Fas Ligand

FasL is another member of the TNF superfamily. Similar to TNF-α, FasL is synthesized and expressed first as a 40-kD membrane–associated protein and is then processed further by matrix metalloproteinases to a homotrimeric 26-kD secreted form (Nagata & Golstein, 1995). Historically, FasL was presumed to have predominantly proapoptotic properties, especially for hepatocytes, and these were associated primarily with its membrane-associated form. Although it was noted that secreted FasL formed trimeric structures and retained its ability to ligate the Fas receptor, it was unclear whether soluble FasL could induce apoptosis (Schneider et al, 1998). A model of Fas activation has recently been proposed that distinguishes five separate stages (Lee et al, 2006). FasL occurs in a membrane-bound found and in a soluble form that is the result of splicing of the membrane-bound form. Studies in mice in particular suggested that the soluble forms of FasL may be receptor antagonists of Fas, which inhibit cell-associated FasL-mediated apoptosis.

The hypothesis that soluble FasL is strictly an inducer of apoptosis has been dispelled by several groups, who reported that soluble FasL also has proinflammatory properties. FasL can induce NF-κB through NF-κB–inducing kinase (NIK) signaling (Malanin et al, 1997). Ottonello and colleagues (1999) showed that soluble FasL in humans is endowed with potent chemotactic properties for neutrophils at concentrations incapable of producing apoptosis. FasL does not seem to activate neutrophils, only to recruit them; neutrophil calcium flux, superoxide production, and degranulation are not affected. Other investigators have come to similar conclusions (Chen et al, 1998; Seino et al, 1998). In addition, Park and colleagues (2003) suggested that FasL can activate blood monocytes. These authors showed that these inflammatory properties are separate and distinct from the proapoptotic properties of FasL. This observation is significant in part because it suggests that FasL, similar to TNF-α and other members of the TNF superfamily, may contribute to neutrophil and monocyte infiltration and recruitment as part of the host inflammatory response. This observation is particularly relevant in viral hepatitis, in which increased FasL expression and neutrophil infiltration have been observed.

Although FasL was originally thought to be expressed only by cells of the lymphoid or myeloid lineage—including predominantly T and B cells, phagocytes, and natural killer cells—it is now recognized that FasL frequently is expressed by nonlymphoid cells (Kiener et al, 1997; Liles et al, 1996). The Fas receptor, also called CD95 or Apo1, is widely expressed on a variety of cell types, including hepatocytes. The Fas receptor shares structural homology with the TNF receptor I (Fig. 10.4), and Fas signaling can also lead to proinflammatory events through the activation of NF-κB–dependent pathways, which may be responsible for some its chemotactic properties.

FIGURE 10.4 Tumor necrosis factor (TNF) receptor I and Fas signaling of apoptosis and nuclear factor (NF)-κB activation. TNF type I and Fas signaling share a common signal transduction pathway, leading to apoptosis through caspase-8 and caspase-3. TNF type I receptor signaling leads directly to NF-κB activation through a divergent tumor necrosis factor receptor–associated factor (TRAF2)-dependent pathway. Fas signaling also can induce NF-κB through NF-κB–inducing kinase (NIK) activation, but under in vivo conditions, Fas signaling is a weaker inducer of NF-κB than TNF type I receptor. FADD/Mort1, Fas-associated death domain; FLICE, FADD-like interleukin-1β–converting enzyme; mACh, muscarinic acetylcholine; RIP, receptor interacting protein; TNFR, TNF receptor; TRADD, TNFR type 1–associated death domain.

Transforming Growth Factor-β

TGF-β is another ubiquitously expressed cytokine that plays a role in several cellular functions, including apoptosis, cell cycle regulation, immune regulation, and inflammation (Massague, 2008; Bierie & Moses, 2006). TGF-β is produced in an inactive form that requires activation by proteases such as matrix metalloproteinase (MMP)2 and MMP9. Interestingly, TGF-β and its proteases are present at high levels in the periphery of certain tumors (Reiss & Barcellos-Hoff, 1997), and TGF-β also seems to be a potent inducer of fibrosis (see Chapter 6) and eventually of carcinogensis of the liver (see Chapter 8C). TGF-β signaling is via phosphorylation of Smad3, but the location of phosphorylation determines whether the effect will be oncogenic or tumor suppressive. Phosphorylation at the C-terminal region of Smad3 transmits a tumor-suppressor signal, and phosphorylation at the linker region initiates oncogenic sequelae. Therefore TGF-β signaling appears to play a role in chronic liver disease progressing to cancer. In addition, mutations in the genes encoding the receptors for TGF-β and decreased expression of downstream Smad proteins have been reported in pancreatic cancer (Derynck et al, 2001).

Cytokines in the Liver and Pancreas

LPS, lipopolysaccharide; IL, interleukin.

Modified from Gabay C, Kushner I, 1999: Acute-phase proteins and other systemic responses to inflammation. N Engl J Med 340:448-454.

Cytokines and the Pancreas

Many pathways that regulate pancreatic development have already been identified and include TGF-β, Notch, Hedgehog, fibroblast growth factor, and epidermal growth factor (EGF) (Rane, 2006; see Chapter 1A). Dysregulation of these pathways appears to be important in many pancreatic disease states.

Much of the current data pertaining to the role of cytokines in the pancreas comes from the diabetes, pancreatitis, and adenocarcinoma literature. It is known that islet cells produce a variety of cytokines in response to both physiologic and pathologic exposure to nutrients (Donath et al, 2010). The inflammatory mediators of cytokine production are produced by α, β, endothelial, ductal, and immune cells. A chronic state of metabolic stress is believed to induce an inflammatory state that leads to islet cell dysfunction and insulin secretion failure. A significant role for TGF-β in the progression of type 1 diabetes has been demonstrated in mice (Baxter et al, 1995).

TGF-β proteins are regulators of pancreatic cell function and appear to have key roles in the development of nondiabetic pancreatic disease as well. The first model of pancreatic adenocarcinoma was generated with overexpression of TGF-β (Greten et al, 2001). These mice developed fibrosis and then pancreatic adenocarcinoma. Additionally, TGF-β2 subset levels appear to be associated with advanced tumor stage (Friess et al, 1993). Patients with long-standing pancreatitis are thought to be at increased risk of developing adenocarcinoma. TGF-β levels are elevated in patients with chronic pancreatitis, and this appears to correlate with both fibrosis and hyperglycemia (Fogar et al, 1998).

Cytokines and Apoptosis

TNF-α seems to be a prominent ligand for the activation of cellular apoptosis, but how TNF-α induces apoptosis in hepatocytes remains unclear. Although hepatocytes are particularly sensitive to TNF-α–mediated apoptosis, the cells first must be exposed to transcriptional inhibition. Incubating cultured hepatocytes with TNF-α or administering TNF-α does not produce hepatocyte apoptosis unless D-galactosamine or actinomycin D is coadministered. The requirement for transcriptional inhibition is likely explained by TNF-α simultaneously activating signaling pathways in hepatocytes that both induce and antagonize apoptosis, suggesting a delicate balance among TNF-α signaling pathways (Fig. 10.5; Beg & Baltimore, 1996; Van Antwerp et al, 1996). Induction of the proinflammatory properties of TNF-α, induced in part through activation of NF-κB–dependent “survival genes,” seems to inhibit apoptosis directly (Baker & Reddy, 1998). NF-κB is a ubiquitous transcription factor that plays a crucial role in the cellular response to signal transduction by TNF-α and IL-6. All members of this family (NF-κB) share a conserved region of approximately 300 amino acids important for their dimerization, nuclear translocation, and DNA binding (Baeuerle, 1998). These NF-κB proteins form homodimers and heterodimers, and their activity is modulated by interactions with inhibitory proteins of the I-κB family.

FIGURE 10.5 Hepatocyte fate is determined by the balance between tumor necrosis factor-α (TNF-α)–mediated nuclear factor (NF)-κB activation and apoptosis. TNF-α signaling simultaneously activates pathways that induce and antagonize capase-dependent apoptosis. Drugs or agents that inhibit NF-κB activation promote apoptosis, whereas cytokines that induce NF-κB protect against TNF-α–mediated apoptosis. IL, interleukin.

These findings have several clinical implications. The role of NF-κB in the development of type 1 diabetes and its effect on β cells is still unclear, but both antiapoptotic and proapoptotic effects have been described (Donath et al, 2008). Recent studies suggest that enhanced activation of the NF-κB pathway via NIK overexpression may exist in pancreatic adenocarcinoma (Nishina et al, 2009). Robin and colleagues (2005) showed that alcohol abuse increases the sensitivity of hepatocytes to TNF-α–mediated apoptosis and hepatotoxicity. This increased sensitivity seems to be secondary to the failure of TNF-α to induce NF-κB activation in the livers of animals after alcohol abuse. Similarly, radiation to hepatocytes prevents induction of survival genes and increases the apoptotic response to TNF-α (Christiansen et al, 2004). In contrast, ischemic preconditioning seems to protect livers from subsequent apoptotic cell death secondary to induction of NF-κB and IL-6 (Iwasaki et al, 2002). In contrast, activation of TNF-α signaling of caspase-8 and caspase-3 via tumor necrosis factor receptor type 1–associated death domin/Fas-associated death domain (TRADD/FADD) leads to apoptosis through pathways shared by FasL signaling. The balance between these two pathways ultimately determines whether a cell responds to TNF-α by proliferating or undergoing apoptosis (Beg & Baltimore, 1996; Van Antwerp et al, 1996).

The role that specific cytokines play in apoptotic liver injury secondary to endotoxin administration has been well characterized. Similar to high-dose TNF-α in the absence of transcriptional inhibitors, high-dose endotoxin administration results in significant liver injury secondary to hepatocyte necrosis, although hepatocyte apoptosis is minimal (Bohlinger et al, 1996). Simultaneous inhibition of transcription, either by depleting hepatocyte nucleotides with D-galactosamine or by blocking transcription with actinomycin D, results in a 1000-fold increased sensitivity to endotoxin-induced lethality, and the hepatic injury is characterized by massive liver apoptosis (Solorzano et al, 1997). The liver injury seems to depend primarily on TNF-α expression, although FasL, interferon-γ, IL-12, IL-18, and IL-23 probably also contribute. Mice lacking either TNF-α or the TNF type I receptor are completely protected from endotoxin and D-galactosamine–induced hepatocyte apoptosis and mortality, whereas they are still susceptible to high-dose endotoxin-induced lethality. Fas agonists differ from TNF-α in that they apparently can induce massive hepatocyte apoptosis in the absence of transcriptional inhibition. The differential response to apoptosis mediated by TNF-α and FasL seems to be related to the observation that FasL is a significantly weaker inducer of NF-κB than is TNF-α. IL-1 and NO protect hepatocytes from TNF-α–mediated apoptosis, presumably by stimulating NF-κB activation and inhibiting caspase activities (Bohlinger et al, 1996). Caspases play central roles as the “suicide effectors” of hepatocyte apoptosis, demonstrated by the treatment of mice with broad-acting caspase inhibitors to prevent apoptotic liver injury and mortality (Kunstle et al, 1997; Mignon et al, 1999).

Nitric Oxide

Oxidative stress has been implicated in the pathogenesis of apoptosis, cancer, diabetes, hypertension, and inflammation. This implies that it may be a final common pathway by which numerous diseases exert their harmful effects. The contribution of NO to liver injury and apoptosis is complex. NO is a lipophilic, short-lived molecule (with a half-life of seconds) that physiologically seems to have roles in vasodilation, immune response, apoptosis, and protection of cells from oxidative killing (Wink et al, 1999). NO derived from NO synthase originating in endothelial cells and other nonparenchymal liver cells seems to be hepatoprotective in most cases (Ou et al, 1997). NO from inducible NO synthase (iNOS) in hepatic parenchymal cells can either damage or protect hepatocytes. Upregulation of iNOS occurs after stimulation by endotoxin and by cytokines such as IL-1b (Kitade et al, 1996; Zamora et al, 2000). Depending on the level of redox stress, iNOS can cause detrimental effects, such as direct cytotoxicity and suppression of hepatic protein synthesis, in a manner similar to the septic state (Taylor et al, 1998; Chen et al, 2003). After resuscitation from profound hemorrhagic shock, iNOS can also mediate activation of NF-κB and increases in IL-6 expression, which leads to acute liver injury followed by recovery (Hierholzer et al, 1998).

Alternatively, with lower redox states, NO plays a cytoprotective role by suppressing apoptosis in hepatocytes and reducing mitochondrial dysfunction to limit liver injury (Brookes et al, 2000; Wang et al, 2002; Chen et al, 2003). NO exerts its cytoprotective effect during acute inflammation by suppressing caspase activation, inducing heat shock proteins (Kim et al, 1997), upregulating hemoxygenase-1 (Zamora et al, 2000), and activating the cyclic guanosine monophosphate signal pathway to interrupt apoptosis (Kim et al, 1997; Wang et al, 2002; Chen et al, 2003). NO can act as an electron acceptor for S-nitrosylation, which can alter the catalytic site of caspases and render them inactive (Li et al, 1997). NO also can induce the expression of heat shock proteins 70 and 32 (Kim et al, 1995; Li et al, 2000). These proteins may function by refolding damaged proteins in the liver, modulating caspase activation, and regulating expression of hemoxygenase-1, an inducible enzyme that catalyzes the degradation of heme into biliverdin, releasing carbon monoxide as a by product, which is thought to confer local antiinflammatory effects (Otterbein et al, 2000). NO also can bind soluble guanylyl cyclase, which can increase intracellular cyclic guanosine monophosphate, suppressing apoptosis and caspase activity (Lucas et al, 2000). Which particular mechanism predominates likely depends on the specific circumstances (Chen et al, 2003). To delineate further the effect of NO on hepatocytes, Wang and others (2002) are continuing to evaluate genetic changes using DNA microarray and large-scale proteomic analysis.

Liver Regeneration

The liver’s ability to regenerate is a unique quality that is essential for its functions of controlling metabolism and detoxification (see Chapter 5). This capability to fully regenerate after insult differs significantly from other organs, which heal with scar (Fausto et al, 2006; Michalopoulos et al, 2007). However, in certain chronic liver injury states, the liver’s regenerative capacity can be insufficient, leading to the replacement of functional epithelial tissue with connective tissue and to development of cirrhosis. A thorough understanding of the mechanisms and factors by which liver regeneration takes place is important in developing therapies to enhance regenerative capacity and prevent cirrhosis.

It has been shown that humoral factors are required, because hepatectomy in a rodent joined in parabiotic circulation results in proliferation and regeneration of the intact liver from the other member of the pair (Moolten & Bucher, 1967). This work provides convincing evidence that some mitogenic factor or factors are released into the peritoneal or systemic circulation during liver regeneration. Studies with recombinant proteins, cytokine inhibitors, and transgenic mice have convincingly shown that at least three cytokines—TNF-α, IL-6, and hepatocyte growth factor—play crucial roles in this regenerative process (Michalopoulos & DeFrances, 1997). The regenerative process of the liver can be divided into three phases: 1) initiation, 2) expansion, and 3) termination (Zimmermann et al, 2004) and is discussed elsewhere (Chapter 5). However, cytokines and growth factors play a significant role. The initiation phase is regulated by cytokines such as IL-6 and TNF-α, the expansion stage is regulated by mitogenic factors such as hepatocyte growth factor, and the termination phase is where mitosis stops, and sometimes apoptosis begins to control overgrowth. Additional cytokines, TGF-β and EGF, also may have roles. The TGF-β superfamily of cytokines appears to play a significant role in maintaining embryonic stem cell identity, and the differentiation of pancreas and liver progenitor cells is restricted by this pathway (Wandzioch & Zaret, 2009).

Several converging lines of evidence suggest that TNF-α, and particularly its signaling through the type I receptor, is essential for liver regeneration. Treatment of animals with neutralizing antibodies against TNF-α delays regeneration and decreases DNA synthesis (Yamada et al, 1997). DNA synthesis is also impaired after partial hepatectomy in TNF-receptor type I knockout mice. Whether these defects are due directly to the effects of TNF-α or are mediated through TNF-α–dependent expression of other mediators, including IL-6, is not fully understood, because treating TNF-receptor type I knockout mice with exogenous IL-6 corrects this delay in regeneration.

The role of IL-6 in hepatic regeneration also has been established (Drucker et al, 2010). Plasma IL-6 concentrations increase after partial hepatectomy, peaking within 24 to 48 hours (Matsunami et al, 1992; Rai et al, 1996). Although it is controversial whether IL-6 is directly mitogenic for hepatocytes in culture, it is clear that IL-6 is mitogenic for biliary epithelial cells (Matsumoto et al, 1994), and the absence of IL-6 delays regeneration in vivo (Cressman et al, 1996). Hepatocyte DNA synthesis after a partial hepatectomy is suppressed in IL-6 knockout mice. A recent randomized trial suggested a beneficial effect of pentoxifylline, a TNF-α inhibitor, on liver regeneration that seems to be IL-6 mediated (Petrowsky et al, 2010). IL-6 expression required for a normal regenerative response seems to depend on NF-κB translocation. NF-κB–dependent pathways are required to protect hepatocytes from TNF-α–mediated apoptosis. Studies from several groups have shown that inhibition of TNF-α–mediated NF-κB activation exaggerates apoptotic pathways initiated by TNF-α and FasL. Xu and colleagues (1998) noted that blocking NF-κB in RALA hepatocytes resulted in TNF-α–mediated apoptosis. Actinomycin D pretreatment of the same cells also promoted TNF-α apoptosis but did not prevent NF-κB translocation. Actinomycin D prevented the transcription of NF-κB–induced genes in response to TNF-α. In addition, inhibition of NF-κB using an inhibitory adenoviral vector in the livers of mice after hepatic resection produces massive apoptosis and liver failure (Iimuro et al, 1998). Taken together, these findings suggest that blockade of NF-κB–dependent transcription of certain genes in hepatocytes during liver regeneration results in TNF-α–mediated apoptosis.

Pathogenesis of Endotoxin and Cytokines in Liver, Biliary, and Pancreatic Disease

Future Directions

Over the past two decades, cytokines have been the subject of intense research, particularly from pharmaceutical companies. As a result of both academic and industrial interest, several agents are currently available that either stimulate cytokine activity or block their effects to treat cancer, infectious disease, inflammatory disorders, and autoimmunity.

Although it is clearly premature to celebrate the widespread use of cytokine or anticytokine therapies for patients with liver, biliary, or pancreatic disease, research is certainly heading in that direction. Several treatment modalities that offer the opportunity to interrupt endotoxin and the proinflammatory cytokine responses in liver disease remain of considerable interest. Antiendotoxin therapies are of particular interest in obstructive jaundice and acute alcoholic hepatitis. In chronic viral hepatitis, in which increased TNF-α–mediated and FasL-mediated apoptosis are presumed to occur, enthusiasm remains for the opportunity to intervene with either immunoadhesins or antibodies. The clinical use of antiinflammatory cytokines, such as IL-10, is likely to occur at some point in the near future, although the risk of added immunosuppression remains a concern. The pegylated form of IFN-α, which has greater stability and in vivo activity than IFN-α alone, has substantially improved viral response rates in hepatitis C patients (see Chapter 64). As a result, it is now considered the standard of care for treatment of hepatitis C and may also have some efficacy in hepatitis B patients.

The greatest excitement, however, has been focused on the availability of antiendotoxin and cytokine therapies currently being used for other disease states that may eventually be utilized for diseases of the liver, biliary system, or pancreas.

Antiendotoxin antibodies have been evaluated in patients with sepsis, generally with disappointing results. Although concerns have been raised about the neutralizing capacity and specificity of some of these antibodies, the lack of efficacy could be explained in part by the inadequate selection of patients and the timing of administration. Only a fraction of the patients receiving antiendotoxin therapies were actually endotoxemic, and the administration of the antibody usually occurred 12 to 24 hours after the onset of symptoms. Recently, focus has been placed on the development of antiendotoxin vaccines, which likely will be tested in Phase I human trials soon (Cross, 2010). An alternative approach is recombinant BPI. In human volunteers treated with low-dose endotoxin, recombinant BPI administration reduced the proinflammatory cytokine responses and activation of the fibrinolytic and coagulation pathways (von der Mohlen et al, 1995). Recombinant BPI continues to be under U.S. Food and Drug Administration (FDA) investigation for its potential use in children with meningococcal sepsis (Levin et al, 2000). Although the results of this trial indicated a trend toward a survival benefit, this study was underpowered to detect significant differences in mortality. In rats subjected to a partial hepatectomy, recombinant BPI reduced liver injury and the magnitude of the inflammatory response (Kimmings et al, 1999).

Another group of cytokine modulators being utilized to treat other diseases are TNF-α blockers. Three TNF-α blocking agents—etanercept (Enbrel), adalimumab (Humira), and infliximab (Remicade)—have received FDA approval for rheumatoid arthritis. TNF-α blockers are also currently the cornerstone of therapy for inflammatory bowel disorders, particularly infliximab and adalimumab. Recombinant human IL-1Ra, anakinra (Kineret), a potent inhibitor of IL-1, is currently approved for the treatment of rheumatoid arthritis.

Although most of the cytokine and anticytokine therapies currently in use are not specifically indicated for liver, biliary, or pancreatic disease, much information about their effects will be learned in the coming years, and some of these therapies may well emerge as effective therapies in patients with inflammatory, neoplastic, or other hepatopancreatobiliary disorders. A recent phase I/II trial of intratumoral injections of TLR-2 and TLR-6 antagonists for R2 resected pancreatic adenocarcinoma shows a possible increased survival, but larger trials will be needed (Schmidt et al, 2007). Preclinical studies indicate that TLR-3 antagonists may delay the development of pancreatic cancer (Schwartz et al, 2009). The now common use of TNF-α and pegylated IFN demonstrate the potential application of cytokine therapy to liver disease. The liver’s complex interplay of metabolism, immunology, cytokine productivity, and regenerative capacity make it ideally suited for the development of cytokine-directed therapies.