|Home | About | Journals | Submit | Contact Us | Français|
The hepatitis B virus (HBV) and hepatitis C virus (HCV) are noncytopathic, hepatotropic members of the hepadnavirus (HBV) and flavivirus (HCV) families that cause acute and chronic necroinflammatory liver disease and hepatocellular carcinoma (HCC) (29, 38, 79). On a worldwide basis, over 500 million people are persistently infected by these viruses and are at great risk of dying prematurely from HCC (29, 64, 79, 111). It is widely believed that the outcome of both infections and the pathogenesis of the associated liver diseases are determined by host-virus interactions mediated by the immune response. It has been difficult to elucidate the viral and host factors at play in these infections, however, largely because the host range of HBV and HCV is limited to humans and chimpanzees (3, 19) and because cell culture systems and small animal models that are susceptible to HBV and HCV infection do not exist. Thus, the current state of our understanding of the biology and pathogenesis of these infections reflects what has been learned about their natural history (47, 64) and immunobiology (29, 137) in humans and chimpanzees, by the virological and immunological analysis of related hepadnavirus (131) and flavivirus (18) infections in their natural hosts, and by biochemical, molecular, virological, and immunological analysis of cell lines (5, 15, 67, 73, 76, 88, 89, 100, 109, 114, 135, 174) and mouse models that express individual viral genes or reproduce the viral life cycles to various degrees (30, 33, 56, 63, 74, 75, 77, 83, 90, 101, 102, 107, 110, 161). Thanks to these efforts, in recent years we have gained important new insight into the viral and host factors that determine pathogenesis and outcome of HBV and HCV infection. As we will describe in this review, it now appears that HBV is a stealth virus that establishes itself very efficiently without alerting the innate immune system to its presence, although it is readily controlled when the adaptive immune response is induced. In contrast, HCV strongly induces yet cunningly evades the innate immune response and also defeats the adaptive immune response by mutation and functional inactivation.
Although host factors contribute importantly to the outcome of HBV and HCV infection, viral factors are also critically involved. Perhaps the strongest evidence that viral factors play a role in the outcome of HBV and HCV infection is that more than 95% of adult-onset HBV infections are self limited, while more than 70% of adult-onset HCV infections persist, and the outcomes for humans and chimpanzees are similar (19, 81, 145, 149, 151). Moreover, as illustrated in Fig. Fig.1,1, we have recently shown that chimpanzees that have previously cleared HBV (58) (Fig. (Fig.1,1, upper panels) become persistently infected when subsequently inoculated with HCV (Fig. (Fig.1,1, lower panels) (149). These results suggest that the different outcomes of infection could not be due to host genetic differences in these animals. Figure Figure11 also demonstrates an interesting and underappreciated difference between HBV and HCV, i.e., that the logarithmic phase of HCV amplification occurs much earlier than for HBV, even when the HBV inoculum is several orders of magnitude greater than that of HCV (58, 149, 152). This and other viral factors that could contribute to the usually different outcomes of HBV and HCV infection are listed in Table Table11 and will now be briefly discussed.
Following infection, the 3.2-kb partially double-stranded DNA HBV genome is delivered to the nucleus and converted into a covalently closed circular double-stranded HBV DNA (cccDNA) molecule that serves as a transcriptional template for the host RNA polymerase II machinery, which produces four capped and polyadenylated mRNAs that encode the viral core and envelope structural proteins and the precore, polymerase, and X nonstructural viral proteins (reviewed in reference 131). One of the major HBV transcripts is a 3.5-kb greater-than-genome-length RNA that is translated to produce the viral core and polymerase proteins and also serves as a pregenomic RNA that is encapsidated with the polymerase by the core protein in the cytoplasm of the hepatocyte (131). Viral replication occurs entirely within these capsids by reverse transcription of the pregenomic RNA to produce a single-strand DNA copy, which serves as the template for second-strand DNA synthesis, producing a circular double-stranded DNA (131). Viral capsids containing double-stranded DNA traffic either to the nucleus, where they amplify the viral cccDNA genome, or to the endoplasmic reticulum, where they engage the viral envelope proteins, bud into the lumen, and exit the cell as virions that can infect other cells (131).
In contrast, the HCV life cycle is entirely cytoplasmic. The HCV genome is a 9.6-kb, uncapped, linear, single-stranded RNA (ssRNA) molecule with positive polarity that serves as template for both translation and replication. Translation of the plus-strand RNA initiates at an internal ribosomal entry site, resulting in production of a single polyprotein precursor that is processed into structural (C, E1, E2, p7) and nonstructural (NS2, NS3, NS4A, NS4B, NS5A, and NS5B) protein subunits by host and viral proteases (4, 37, 50, 51, 62, 85, 87, 98, 117, 123, 134, 146, 147, 154, 166). HCV replication occurs via a minus-strand intermediate within a membranous compartment in the cytoplasm of the cell (99), yielding double-stranded RNA (dsRNA) intermediates as essential components of its life cycle. Thus, while HBV RNA, RNA/DNA hybrids, and DNA replicative intermediates are sequestered entirely within capsid particles (131), the HCV ssRNA and dsRNA replicative intermediates are likely exposed to the dsRNA sensing machinery of the cell (99). It would not be surprising, therefore, if these two viruses induce very different innate cellular responses following infection. We will discuss the potential significance of this observation when we discuss the cellular genomic responses to HBV and HCV infection later in this review.
As mentioned above, another interesting difference between HBV and HCV is the rate at which they expand in the liver after inoculation, and the peak viral titers that they ultimately produce. For example, experimentally infected chimpanzees inoculated with 108 genome equivalents (GE) of HBV display a prolonged lag phase before viral DNA displays logarithmic expansion in the liver or serum, and the peak titers are routinely in the 109 GE per ml range (58, 152). In contrast, when the same animals are subsequently inoculated with HCV, the viral RNA expands logarithmically within the first 2 weeks and reaches peak titers that rarely exceed 106 GE per ml (Fig. (Fig.1)1) (149). Since the onset of the cellular immune response to both of these viruses is detectable 3 to 4 weeks after inoculation, i.e., before the logarithmic spread of HBV but after the log phase of HCV (references 149 and 152 and see below), the two viruses could represent very different antigenic challenges to the immune response, and this could have an important impact on the outcome of the infection. Once infection is established, differences in HBV and HCV antigenic burden could also have an impact on viral clearance. In particular, it has been shown by the use of highly sensitive and specific immunohistochemical reagents that HBV can infect up to 100% of the hepatocytes (58, 152). Importantly, high-level HBV antigen expression (especially HBcAg) is easily detectable within infected cells (58, 152), which could contribute to their high visibility to the adaptive immune system and, therefore, the usual outcome of viral clearance in immunologically competent adults. In contrast, there is considerable uncertainty about the number of hepatocytes that are infected by HCV. Indeed, estimates based on HCV RNA quantification in chimp liver biopsies during HCV infection suggest that either very few hepatocytes replicate HCV at high levels or that low-level HCV replication occurs throughout the liver (12). Both situations would be expected to result in inefficient antigen recognition by the adaptive immune system, which could contribute to the usual outcome of viral persistence in HCV infection. Finally, because of the lack of proofreading activity in the viral RNA-dependent RNA polymerase (NS5B), the mutation rate of HCV is very high (10−3 per nucleotide per year) (106). This results in the rapid evolution of a viral quasispecies in each infected subject, presumably due to immune selection pressure. Since hepadnaviruses replicate by reverse transcription of an RNA pregenome, they are thought to have a high mutation rate (10−5 per nucleotide per generation) (48, 115), albeit one that is 100 times lower than that for HCV, which may account for the fact that the emergence of viral variants during HBV infection is much less frequent than for HCV infection (24, 121).
As will be described later in this review, viral clearance and disease pathogenesis with both HBV and HCV infection are mediated largely by the immune response. For these noncytopathic viruses to persist, they must either not induce a response or they must overwhelm, evade, or counteract it. All of these scenarios have been shown to be operative in chronic HBV and HCV infection. Interestingly, as we will see later, it appears that HBV “evades” the innate response by simply not inducing it to act as a stealth virus in this regard (171). On the other hand, HBV appears to employ active evasion strategies that target the adaptive immune response (see below), which is well known to be activated by this virus (8, 10, 28, 29, 91, 97, 118-120, 150, 152). This suggests that the adaptive response plays an important role in the control of HBV infection while, interestingly, the innate response does not. Importantly, as will be discussed later in this review, both of these ideas are supported by genomic analysis of the liver in acutely HBV-infected chimpanzees (171). In contrast to HBV, HCV displays several cunning evasion strategies that target the innate and the adaptive immune responses (see below), suggesting that it activates and must, therefore, evade both during infection. Importantly, this concept is also supported by the gene expression profiles of the acutely HCV-infected liver (see below and references 12 and 142). The known HBV and HCV evasion strategies will now be briefly discussed.
Mutational inactivation of B-cell and T-cell epitopes occurs in chronic HBV infection (9, 21, 121, 165), but it is much more common in HCV (24, 36, 169; reviewed in reference 137). Indeed, mutations that abrogate recognition by antibody, CD4 T cells, and CD8 T cells in chronically HBV- and HCV-infected humans (7, 9, 20, 24, 36, 65, 82, 92, 121, 132, 153, 157) and chimpanzees (36, 169) have been described. The T-cell epitope mutations often occur in epitope residues that bind to major histocompatibility complex (MHC) molecules, thereby precluding antigen presentation (24, 36). Less frequently, inactivating mutations in residues that flank T-cell epitopes and impair the ability of those epitopes to be processed by the proteasome (132) or transported by the TAP protein into the endoplasmic reticulum (133), also precluding antigen presentation, have been described. More often, mutations occur in epitope residues that are engaged by the T-cell receptor (TCR). Inactivating mutations of this sort preclude antigen recognition (9, 36, 49, 68, 92), which makes the cells that are infected by the mutant virus invisible to the T cells containing the corresponding TCR. In addition, epitope mutations in which the variant residue remains visible to the TCR but antagonizes it have been described (7, 9, 24, 72, 157); with these mutations, the TCR no longer recognizes its cognate wild-type epitope, making cells infected by the wild-type virus and the mutant virus invisible to that TCR-bearing population of cells. All of these mechanisms have been described and probably contribute to viral persistence in both HBV and HCV infection, although they are much more common in HCV infection, presumably because of the higher mutation rate for HCV.
Several mechanisms are probably responsible for viral persistence during HBV and HCV infection (reviewed in references 28, 64, 116, and 137). As will be discussed later in this review, chronic HBV and HCV infection are characterized by absent, weak, or narrowly focused CD4 and CD8 T-cell responses to the corresponding viral antigens (78, 97, 104, 112, 120, 145, 168). The responsible mechanisms are not entirely clear, but T-cell deletion, anergy, exhaustion, and ignorance have all been reported to occur in HBV- and HCV-infected humans and chimpanzees (reviewed in references 28, 29, and 137). Interestingly, at least two virus-specific tolerogenic mechanisms that can be ascribed to the HBV precore and surface proteins appear to be operative in chronic HBV infection The HBV precore protein is not required for HBV replication or infection (23, 25, 155, 156), but it is processed and secreted as HBeAg, a 17-kDa protein that is small enough to cross the placenta and induce neonatal tolerance, at least in HBV transgenic mice (96). In addition, HBeAg has been shown to suppress the antibody and T-cell response to HBcAg in adult T-cell receptor transgenic mice (26), functioning either to delete or anergize HBcAg/HBeAg cross-reactive T cells, depending on their functional avidity for the tolerogenic epitope. Thus, HBeAg may suppress immune elimination of infected cells by HBcAg-specific T cells and, thereby, contribute to viral persistence in chronically infected adults. Clinical evidence supports this notion, since viral mutations that preclude the production of HBeAg are often associated with exacerbations of liver disease and, sometimes, even with viral clearance in chronically infected patients (17, 47). Thus, although the precore protein has no known role in the viral life cycle, it may function as an HBV-specific immunosuppressive factor that protects the virus against immune attack. The hepatitis B surface antigen (HBsAg) might also suppress immune elimination of infected cells by functioning as a high-dose tolerogen, since extremely high serum HBsAg titers, in the mg per ml range, are often seen with chronically infected patients (122, 167). In keeping with the high antigen load, chronically infected patients display absent or subnormal levels of HBsAg-specific CD8+ T cells; however, these CD8+ T-cell populations display abnormal HLA/peptide tetramer binding properties in contrast to the few HBcAg-positive T cells that are detectable in these patients, which are functionally normal in this regard (122). In addition, it has been reported that the HBV X protein, a transcriptional transactivator that is required for initiation of infection (177, 180), can inhibit cellular proteasome activity when it is overexpressed (66). Thus, the HBx protein has the potential to inhibit antigen processing and presentation if it is overexpressed to a comparable degree in infected cells. In all instances, therefore, HBV attempts to make itself invisible to the various effector limbs of the immune response, functioning as a stealth virus in this regard.
While HBV evasion strategies are focused primarily on the adaptive immune response, HCV has developed mechanisms that appear designed to enable it to escape both the innate and the adaptive response. As shown in Table Table1,1, mutational escape from the adaptive response due to its high mutation rate is common in HCV-infected patients and chimpanzees (35, 36, 52, 136, 153, 162, 163, 169, 170). In addition, however, because it replicates via a dsRNA intermediate, HCV activates protein kinase R (113), interferon regulatory factor 1 (IRF-1) (42, 113), and IRF-3 (42, 143), and downstream antiviral genes that were originally shown to be activated by these factors in different viral infections (reviewed in references 14 and 141). In keeping with this notion, as we will describe later in this review, HCV induces a large number of interferon (IFN)-inducible genes in the liver during acute and chronic infection (12, 13, 142). Nonetheless, HCV appears to be resistant to these antiviral pathways (128, 141). Indeed, several HCV structural and nonstructural proteins (E2 and NS3 and NS5, respectively) have been shown to inhibit nonoverlapping functions of the innate immune response. E2 and NS5A have been shown to bind to the kinase domain of PKR and inhibit IRF-1 activation (46, 113, 148). The NS3/4A protease has been shown to cleave the toll-like receptor 3 adaptor protein (TRIF) (84) and to disrupt RIG-I signaling (39), thus blocking the phosphorylation and effector action of IRF-3 (40). In addition, some investigators have demonstrated that overexpression of the HCV core protein can inhibit Fas-mediated apoptosis (93, 144), although this is controversial since it has also been shown to stimulate apoptosis by others (178, 179), and overexpression of the HCV nonstructural protein E2 inhibits natural killer cell function by binding to CD81 (32, 158). It is important to point out, however, that all of these effects have been demonstrated in transfection and replicon systems where the HCV proteins are overexpressed. Therefore, these functions should be considered speculative until they can be tested with a cell culture model of natural HCV infection.
Thus, because it is so visible to the innate immune system, HCV appears to employ several evasion strategies that contribute to its high propensity to persist. There is a problem with this hypothesis, however, at least in terms of evasion of the interferon response. Since activated IRF-1 and IRF-3 are both known to induce beta interferon (IFN-β) gene expression (45, 164), one would not expect interferon-regulated genes to be strongly induced in the liver during HCV infection if E2, NS3, and NS5A inhibit their activation. Nonetheless, genomic analysis of the livers of several HCV-infected chimpanzees (see below) indicates that a strong correlation exists between the level of HCV viremia and the intrahepatic expression of a large number of interferon-regulated genes (12, 13, 142), which shouldn't occur if HCV proteins inhibit interferon induction. Clearly, more studies are needed to understand the basis for these apparently contradictory observations.
Until then, we favor the notion that HCV circumvents the interferon signaling cascade not by blocking its induction but by inhibiting the antiviral effector functions of interferon-induced target genes. Alternatively, if IRF-1 and IRF-3 activation are partially blunted by HCV, instead of being totally blocked, the interferon induction might be at a sufficient level to keep viral titers relatively low without completely preventing HCV infection. Thus, despite the interesting and provocative reports about HCV-induced IRF-1 and IRF-3 blockade, the potential impact of these processes on HCV infection will not be testable until the development of cell culture and small animal models of HCV infection.
The fact that most adult-onset HBV infections are self-limited while nearly all neonatal HBV infections persist serves as strong evidence that host factors play a critical role in the outcome of HBV infection (28, 29). This relationship is reinforced by many studies showing a close association between the vigor, diversity, and effector functions of the cellular immune response to HBV and HCV and the outcomes of these infections, not only in naturally infected humans (81, 104, 112, 151) but also in transgenic mice (56, 175) and experimentally infected chimpanzees (6, 31, 149, 152). These observations will now be considered, using the host pathways that contribute to these differential outcomes as the starting point for discussion.
Neonatal tolerance to HBV is probably responsible for viral persistence following mother-infant transmission (29). On the other hand, the basis for the inadequate immune response that is characteristic of adult-onset chronic HBV and HCV infections is not well understood and may, in fact, be multifactorial in origin. For example, while viral clearance during self-limited HBV and HCV infection is characterized by a vigorous, polyclonal CD4 and CD8 T-cell response to these viruses (31, 91, 120, 138, 149, 151, 152), primary failures to establish a CD4 and CD8 T-cell response in a patient who was studied prospectively after accidental needle stick exposure to HCV (151) and in HCV-inoculated chimpanzees (149) have been clearly described, suggesting that primary immunological nonresponsiveness to HCV probably led to persistent infection in those cases. Whether, as has been shown with other viral systems, the failure to produce a vigorous T-cell response to HCV was due to the possible negative impact of antigen overload during immunological priming (reviewed in references 34 and 130), to virus-induced defects in antigen presentation (reviewed in reference 176), to the hyperinduction of regulatory T cells (reviewed in reference 127), to a genetically determined restriction of the virus-specific T-cell repertoire (reviewed in reference 105), or occurs for other reasons remains to be determined. It is also thought that the diminished T-cell responses in chronically infected patients could be due to the induction of anergy and/or the exhaustion of an initially vigorous T-cell response by high viral load, presumably reflecting excessive antigen stimulation of virus-specific T cells. If this occurs, it would contribute to the maintenance of persistent infection, although it would obviously be a secondary event. There is a precedent for this notion, since Lechner et al. (80) and Ulsenheimer et al. (160) have shown that CD8+ and CD4+ T-lymphocyte responses are induced during acute HCV infection but are not sustained during progression to chronicity. There is also evidence that the ability of HCV-specific CD8+ T cells to produce gamma interferon following in vitro exposure to antigen is compromised (“stunned”) as a consequence of antigen overstimulation in vivo during persistent infection (53, 81, 151, 168). Furthermore, Boni et al. (16) have shown that antiviral treatment can overcome CD8+ T-cell hyporesponsiveness in subjects with chronic HBV infection, suggesting that the T cells are present in these subjects but are suppressed. Thus, primary and secondary immunological hyporesponsiveness to both viruses can occur in persistent HBV and HCV infection.
It is widely believed that the cytotoxic T-lymphocyte (CTL) response clears viral infections by killing infected cells. CTL killing is an inefficient process, however, requiring direct physical contact between the CTLs and the infected cells. Thus, it may not be possible for CTLs to kill all infected cells if the CTLs are greatly outnumbered, as occurs during these infections, in which as many as 1011 hepatocytes can be infected. Thus, although the liver disease in these infections is clearly due to the cytopathic activity of the CTL response, viral clearance may require more efficient CTL functions than killing. Important insights into the pathogenetic and noncytopathic antiviral functions of the CTL response have come from studies in HBV transgenic mice that develop an acute necroinflammatory liver disease after adoptive transfer of HBsAg-specific CTL clones (1, 56, 103). In that model, the CTLs rapidly enter the liver and recognize viral antigen, triggering two events, (i) apoptosis of the hepatocytes that are physically engaged by the CTLs and (ii) secretion of IFN-γ, which noncytopathically inhibits HBV gene expression and replication in the rest of the hepatocytes (54, 56) by preventing the assembly of the RNA-containing capsids in the cytoplasm (172) in a proteasome- (126) and kinase-dependent (125) process. During this remarkable process, the viral nucleocapsids disappear from the cytoplasm of the hepatocytes (56, 172) and the viral RNAs are destabilized by a SSB/La-dependent mechanism in the nucleus (59-61, 159), yet the hepatocytes remain perfectly healthy (54, 159). As a result, all of the viral gene products and virions decrease in the liver and the serum (56), inhibiting further viral spread. This antiviral process is completely blocked by the administration of antibodies to IFN-γ before the CTLs are injected, and it is not induced by HBsAg-specific CTLs derived from IFN-γ knockout mice whose cytopathic functions are perfectly normal (56), indicating that IFN-γ production by the CTLs is responsible for the noncytopathic antiviral effect. Furthermore, HBsAg-specific, Fas ligand-deficient, and perforin-deficient CTL clones that do not cause hepatitis in these animals do inhibit viral replication (56), proving genetically that the cytopathic and antiviral functions of CTLs are completely independent of each other. These results suggest that a strong intrahepatic CTL response to HBV can suppress viral gene expression and replication noncytopathically. In addition, it may even “cure” infected hepatocytes of the virus, provided that the HBV transcriptional template (cccDNA) is also eliminated from infected cells. This could not be tested with HBV transgenic mice, because they do not produce cccDNA (57). However, as will be discussed below, the kinetics of cccDNA elimination during experimental HBV infection in chimpanzees suggests that cccDNA could, at least partially, be eliminated from hepatocytes by a noncytolytic mechanism (58, 173).
Interestingly, it has been shown that HBV replication is also suppressed by the antiviral effects of alpha/beta interferon (55, 95, 172). Indeed, in the transgenic mouse model, at least, HBV replication is inhibited by any stimulus that induces IFN-γ or IFN-α/β in the liver, including CD4+ T cells (41), NK and NKT cells (70), and other hepatotropic viral (22) and parasitic infections (94, 108). This raises the possibility that HBV infection can be controlled by many arms of the immune response and perhaps explains why HBV infection is almost always self-limited in immunologically normal adults.
To investigate whether these principles apply to the clearance of HBV and HCV infection, we extended these studies to HBV- and HCV-infected chimpanzees (58, 149, 152). In these studies, we showed that the early phase of clearance of HBV was temporally associated with the appearance of CD3, CD8, and IFN-γ mRNA in the liver (152), which reflected the influx of virus-specific CD8+ T cells into the liver (152). But, although HBV replicative intermediates (152, 173) and cccDNA templates (152, 173) decreased as much as 50-fold and 8-fold from peak levels during this time, respectively, there was little or no attendant liver disease, despite the fact that virtually 100% of the hepatocytes were infected (58, 152), suggesting that noncytopathic mechanisms were active during this early phase of viral clearance. Furthermore, we also showed that monoclonal antibody-mediated depletion of CD8+ T cells (but not CD4+ T cells) at the peak of infection delayed the onset of viral clearance and liver disease for several weeks, until the antibody titers waned and virus-specific CD8+ T cells became detectable in the liver (152). Thus, we conclude that the principle of CD8-dependent noncytopathic clearance of HBV, which was discovered by using the HBV transgenic mouse model, is operative in the context of the full-fledged viral infection. Interestingly, several lines of evidence suggest that HCV may also be susceptible to CD8-dependent noncytolytic clearance mechanisms. For example, in a study of the immune response to HCV with five individuals who became infected after accidental needle stick exposure to HCV-positive blood (151), the only subject to clear the virus did so in the context of an IFN-γ-producing HCV-specific CD8+ T-cell response, and clearance occurred without a corresponding surge of liver disease activity (151). Furthermore, viral clearance has been reported to occur in the context of an intrahepatic CD8+ T-cell response in experimentally HCV-infected chimpanzees in the absence of elevated serum alanine aminotransferase activity (a marker of liver cell destruction) (149), and viral clearance has been shown to be delayed by CD8+ T-cell depletion in such animals (138). Lastly, both IFN-γ and IFN-α/β have been shown to inhibit the replication of an HCV replicon without evidence of toxicity in Huh-7 cells in vitro (15, 27, 43, 44), demonstrating that HCV is susceptible to interferon-induced noncytolytic control mechanisms in the replicon system.
The foregoing results suggest that one or more cellular genes that are induced by IFN-γ and/or IFN-α/β are likely to inhibit both HBV and HCV replication. In order to begin to identify these genes, as well as genes that might be transcriptionally regulated by the viruses, global gene expression profiling (Affymetrix gene chip analysis) was performed using liver RNA obtained at multiple time points after both infections (142, 171). As shown in Fig. Fig.2,2, three HCV-infected chimpanzees and three HBV-infected animals were included in these studies (142, 171). As illustrated in Fig. Fig.2,2, three different courses and outcomes of infection were seen with the HCV-infected animals (shown in green). Chimpanzee 96A developed self-limited infection that reached a viral titer of 105 GE per ml and cleared the virus, similar to the outcomes seen with all three of the HBV-infected animals (shown in blue). In contrast, chimp 1581 (which developed the highest HCV titer, 2.5 × 106 GE per ml) initially controlled the infection by 3 to 4 logs before becoming persistently infected, and chimp 1590 (which displayed the lowest viral titer, 104 GE per ml) became persistently infected without a period of initial control. In contrast, all three HBV-infected chimpanzees (1627, 5835, and 1615; shown in blue) developed a self-limited infection, reaching very similar maximal viral titers of 109 GE per ml, and they cleared the virus with remarkably similar kinetics. When the gene expression profiles were established for all of the animals, we searched for virus-induced cellular genes, i.e., genes whose expression levels correlated with the viral titers in all three of the HBV-infected animals or in all three of the HCV-infected animals, and for clearance-related cellular genes, i.e., genes whose expression was correlated with the clearance of HBV in all three of those animals or whose expression correlated with at least a 3 log reduction of viral titer in the corresponding HCV-infected animals.
Initially, we searched for genes whose expression patterns correlated (directly or inversely) with the amount of HCV RNA in the liver over the entire time course profiled, irrespective of the outcome of infection. We reasoned that these genes would be regulated by HCV and that they might reflect activation of the dsRNA sensing machinery of the cell, might be required for HCV to establish and/or maintain itself in the liver, or both. Using Pearson correlation analysis to reduce the chance of misinterpreting random fluctuations in gene expression in individual animals, we searched for transcripts that correlated with the changing HCV RNA titer with a correlation coefficient of at least 0.7 for all three animals (P < 0.05) (shown in red in Fig. Fig.2A).2A). We found 27 unique transcripts that fulfilled these criteria, and we noted that they displayed an average peak fold change (FC) value of 8.8 (142). A complete list of gene names is provided in Table Table2.2. Importantly, consistent with the time course of the IFN-α/β induction in the liver of these chimpanzees as reported by us (149) and in independent studies by Bigger et al. (12, 13), expression of many of these genes is known to be stimulated by IFN-α/β. This suggests that HCV-infected cells probably detect the virus by sensing the presence of dsRNA replication intermediates, thereby inducing the transcription of many IFN-α/β-stimulated genes. Nonetheless, this response apparently fails to eradicate HCV from the infected cells, either for quantitative reasons or because of HCV-induced evasion mechanisms downstream of IFN-α/β-stimulated transcription, as we discussed earlier in this review. These unexpected findings might also suggest that, in contrast to other viruses that are inhibited by these genes, HCV might actually use the product(s) of one or more of these early response genes to facilitate infection. While these questions are not approachable for the chimpanzee, they are readily addressable in vitro with HCV replicon-containing Huh-7 cells, and experiments designed to examine these issues are in progress in many HCV laboratories, including our own, at this time.
The same analysis was performed to identify cellular genes that correlate with HBV DNA titers and might, thus, be induced by HBV infection (171). Accordingly, as we did for the HCV-infected animals, we searched for genes with expression patterns that correlated (directly or inversely) with the amount of HBV DNA in the livers of all three animals over the entire time course profiled, and we restricted our focus to genes whose changing expression levels correlated with the changing viral DNA content in the liver with a correlation coefficient of at least 0.7. As shown in Fig. Fig.2B2B and Table Table2,2, no genes fulfilled these criteria (171). Since virtually 100% of the hepatocytes in all three animals were infected (152), the failure of the virus to induce cellular gene expression as it spread throughout the liver suggests that HBV behaves as a “stealth” virus in that it does not induce an innate response in the cells it infects. This surprising and unprecedented observation may explain why HBV replication is so highly sensitive to the antiviral effects of alpha/beta interferon, since it has evolved without the need to establish any defense mechanisms against this cytokine and, by extension, against any other cytokines that display similar gene expression profiles (e.g., IFN-γ). This is strikingly illustrated by the fact that HBV replication is readily suppressed by antiviral mechanisms induced by toll-like receptor ligands that activate the innate immune response in HBV transgenic mice (69).
To identify genes associated with HBV clearance, we searched for genes whose expression was induced or suppressed only during the phase of viral clearance in the three HBV-infected chimpanzees. To identify similar genes in the HCV-infected animals, we searched for genes whose expression was induced or suppressed during the phase of viral clearance in chimpanzees 96A and 1581 but that were not related to viremia in chimpanzee 1590, since the viral titer never decreased in that animal. We identified 124 genes that were induced during viral clearance in HCV-infected chimpanzees 96A and 1581, but not in 1590, and correlated with HCV-specific T-cell infiltration and IFN-γ expression (149) or IFN-γ-inducible genes, such as RANTES, MIG, and MHC (Fig. (Fig.2C)2C) (142). Similarly, as shown in Fig. Fig.2D,2D, we identified 110 genes that were induced during viral clearance and correlated with the same marker genes in all three HBV animals (152, 171). Table Table33 lists the genes that were induced during clearance of both infections. Not surprisingly, it is heavily weighted towards genes expressed by alpha/beta T cells (e.g., T-cell receptor beta), gamma/delta T cells (e.g., T-cell receptor gamma), CD3, cytolytic effector molecules (granzyme A), and T-cell growth regulatory genes, including the receptor for interleukin 10 (IL-10). In addition, many genes that are known to be induced by IFN-γ are represented on the list, including those involved in antigen presentation (MHC class I and class II, ubiquitin D, and the immunoproteasome subunits LMP2, MECL1, and PA28), chemokines (RANTES, MIG, and MIP-1β) that can recruit antigen-nonspecific inflammatory cells into the liver, the GTPase guanylate binding protein 1 (GBP1) and GBP2, and other IFN-γ-induced genes, e.g., tryptophanyl-tRNA synthetase (TrpRS) (86), solute carrier 7A, and ubiquitin D, whose relevance to the T-cell response and viral clearance are not immediately apparent. Interestingly, some of the proteins in the miscellaneous group are involved in host-pathogen interactions. For example, tyrosine kinase binding protein is involved in NK cell signaling and has been shown to be important in the host defense against murine cytomegalovirus (140). The uncoupling protein 2 limits oxygen radical production and macrophage-mediated immunity (2).
It is important to point out that, because some of these genes are likely to be expressed in the inflammatory cells that infiltrate the liver during viral hepatitis, their role in viral clearance is probably indirect. However, the expression of many genes on this list is not limited to inflammatory cells, and these genes have the potential to exert a more direct effect on these infections by modulating viral replication within the hepatocytes. The unique subunits of the immunoproteasome (LMP2 and LMP7) are in this category, since the proteasome is known to influence various aspects of the life cycle of a number of viruses, including human immunodeficiency virus (129) and HBV (139). The chemokines RANTES and MIG are also in this category, since they are known to be induced in hepatocytes by IFN-γ (71). Furthermore, because of the known ability of IFN-γ to inhibit HBV in vivo (56, 109) and its demonstrated ability to inhibit HCV replication in vitro (27, 44, 56, 76, 109), it is possible that one or more of the genes in this set may contribute to the noncytopathic clearance of these viruses, as with the previously observed antiviral effect on HBV replication in the livers of HBV transgenic mice (54).
Collectively, the results summarized in this review suggest that HBV acts like a stealth virus early in infection, remaining undetected and spreading until the onset of the adaptive immune response several weeks later. We suspect that the relative invisibility of HBV to the innate sensing machinery of the cells reflects its replication strategy, which sequesters the transcriptional template in the nucleus, entails the production of capped and polyadenylated viral mRNAs that resemble the structure of normal cellular transcripts, and the replicating viral genome is sheltered within viral capsid particles in the cytoplasm and, therefore, does not elicit a response. In contrast, HCV activates a strong intracellular antiviral response in the liver, presumably because it replicates via a dsRNA intermediate in the cytoplasm, where it can readily induce the cellular dsRNA sensing apparatus and initiate the signaling cascade. In turn, this results in the induction of many genes, such as the 2′,5′-oligoadenylate synthetase Mx1, that have known antiviral activity (reviewed in references 128 and 141); TRIM14 and 22, members of the same gene family as TRIM5, which has been shown to be part of the host defense against retroviruses (reviewed in reference 11); and ISG15, which has recently been proposed to have a role in innate immunity (124). Impressively, however, HCV cunningly manages to spread through the liver despite induction of these genes, presumably because its E2, NS3, and NS5A proteins can defeat them. On the other hand, both viruses can be controlled when CD8+ T cells enter the liver, recognize antigen, kill whatever infected cells they encounter, and secrete IFN-γ, which triggers a broad-based cascade that amplifies the inflammatory process and has antiviral activity, at least in HBV-transgenic mice and the HBV and HCV cell culture systems. The unexpected finding that HBV does not modulate host cellular gene transcription and apparently does not induce an innate immune response when spreading through the liver raises the possibility that HBV, unlike HCV and other viruses, has evolved to evade innate immunity by not inducing it rather then actively counteracting it; this, in turn, might leave HBV very sensitive to intracellular antiviral mechanisms when they are induced by the adaptive immune system or an unrelated viral infection. Although we suggest that HBV and HCV infections can be inhibited noncytopathically by the cellular genes that are induced during this process, the identities of those genes, the antiviral mechanisms they elicit, and the unique evasion strategies of each virus remain to be determined, marking a new starting point in the quest to understand the cellular and molecular immunobiology of these viral infections.
We thank all of our colleagues who contributed importantly to the work cited in this review, especially Robert Purcell and Jens Bukh (NIAID, National Institutes of Health, Bethesda, MD), Andrew Su and Peter Schultz (Genomics Institute of the Novartis Foundation, La Jolla, CA), and Luca Guidotti (The Scripps Research Institute, La Jolla, CA). We are particularly grateful to B. Rehermann (Liver Diseases Section, NIDDK, National Institutes of Health, Bethesda, MD) and R. Lanford (Department of Virology and Immunology, Southwest Foundation for Biomedical Research, San Antonio, TX) for critical reading of the manuscript.
This work was supported by NIH grants AI20001, CA76403, and CA40489 to F.V.C.
This is manuscript number MEM-17106 from the Scripps Research Institute.