Human liver transplantation as a model to study hepatitis C virus pathogenesis


  • Michael G. Hughes Jr.,

    1. Department of Surgery, Medical University of South Carolina, Charleston, SC
    Search for more papers by this author
  • Hugo R. Rosen

    Corresponding author
    1. Division of Gastroenterology and Hepatology, Department of Medicine, University of Colorado Health Sciences Center, Denver, CO
    2. Division of Liver Transplantation, Department of Medicine, University of Colorado Health Sciences Center, Denver, CO
    3. National Jewish Hospital, Denver, CO
    4. Denver VA, Denver, CO
    • Division of Gastroenterology and Hepatology, B-158, University of Colorado Health Sciences Center, Academic Office Building 1, Room 7614, 12631 East 17th Avenue, PO Box 6511, Aurora, CO 80045
    Search for more papers by this author
    • Telephone: 303-724-1858; FAX: 303-724-1891


Hepatitis C is a leading etiology of liver cancer and a leading reason for liver transplantation. Although new therapies have improved the rates of sustained response, a large proportion of patients (∼50%) fail to respond to antiviral treatment, thus remaining at risk for disease progression. Although chimpanzees have been used to study hepatitis C virus biology and treatments, their cost is quite high, and their use is strictly regulated; indeed, the National Institutes of Health no longer supports the breeding of chimpanzees for study. The development of hepatitis C virus therapies has been hindered by the relative paucity of small animal models for studying hepatitis C virus pathogenesis. This review presents the strengths of human liver transplantation and highlights the advances derived from this model, including insights into viral kinetics and quasispecies, viral receptor binding and entry, and innate and adaptive immunity. Moreover, consideration is given to current and emerging antiviral therapeutic approaches based on translational research results. Liver Transpl 15:1395–1411, 2009. © 2009 AASLD.

As described in many excellent articles in Liver Transplantation,1–3 liver disease related to hepatitis C virus (HCV) is the single leading indication for liver transplantation throughout the world, and its significance as a clinical problem cannot be overstated. In this review, we highlight what has been learned in the past decade about liver transplantation as a model for studying HCV pathogenesis, including important insights into the roles of viral kinetics and quasispecies, hepatitis C receptor binding and viral entry, and innate and adaptive immunity and how these insights might be applied to novel preventative and therapeutic approaches.


One factor limiting the development of HCV therapies is the paucity of animal models for HCV infection that simulates human infection. Although chimpanzees have been used to study HCV biology and treatments, their cost is quite high, and their use is strictly regulated; indeed, the National Institutes of Health no longer supports the breeding of chimpanzees for study.4 The severe combined immunodeficient (SCID)/albumin–urokinase plasminogen activator mouse has emerged as the current gold standard of small animal models of HCV infection. After transplantation with human hepatocytes in the first few weeks of life, the subacute liver failure induced by the transgene leads to a strong proliferative stimulus for hepatocytes.5 The native murine hepatocytes are inhibited from responding, and this leaves the human hepatocytes (protected from xenograft rejection by the SCID status) to proliferate and achieve up to 90% repopulation of the liver.6 The mice can then be infected with HCV of defined origin (eg, H77 or JFH-1) or from clinical serum samples, maintaining high-level infection titers for many months.7, 8 Improvements in methods of hepatocyte generation, adjuvant immune interventions, and improved breeding strategies have markedly reduced earlier limitations on the number of mice that can be produced, rendering larger studies more practical than studies with other in vivo models of HCV infection, but there are only a few laboratories in the world that can generate and maintain these mice. Moreover, the SCID status of these mice precludes immunological analyses9 unless cells are added back (eg, adoptive immunotherapy).


BSA, bovine serum albumin; CTL, cytotoxic T lymphocyte; E1, first envelope protein; E2, second envelope protein; ELISPOT, enzyme-linked immunosorbent spot; HCV, hepatitis C virus; HDL, high-density lipoprotein; HLA, human leukocyte antigen; HVR1, highly variable region 1; I-R, ischemia-reperfusion; IFN, interferon; KIR, killer cell immunoglobulin-like receptor; LCL, lymphoid cell line; LDL, low-density lipoprotein; LDLr, low-density lipoprotein receptor; MAAD, maximal amino acid diversity; MFI, mean fluorescence intensity; NK, natural killer; NT, natural T; PBS, phosphate-buffered saline; PCR, polymerase chain reaction; PD-1, programmed death 1; RT-PCR, reverse-transcriptase polymerase chain reaction; SCID, severe combined immunodeficient; SR-BI, scavenger receptor type B-class I; STAT-C, specifically targeted antiviral therapy for patients with chronic hepatitis C virus.


The human liver transplantation model provides a unique opportunity and research framework for examining HCV pathogenesis for a number of reasons (Table 1).10

Table 1. Advantages and Disadvantages of the Human Liver Transplant Model
  • The explant liver contains a high number of hepatitis C virus–specific T cells.

  • The posttransplant natural history is accelerated.

  • The time of infection is known, and this allows viral kinetics and host immune studies.

  • Sequential serum specimens and biopsy specimens are typically available.

  • Immunosuppression

  • Human leukocyte antigen mismatch (and its effect on antigen recognition)

  • Extrapolation to acute infection is confounded by the fact that the recipient is not naïve to hepatitis C virus infection.

Liver Explants Are Enriched with HCV-Specific T Lymphocytes

HCV infections that follow a chronic course are usually marked by low frequencies of antigen-specific T cells targeting few epitopes.11 Most studies of the intrahepatic compartment to date in humans and chimps have relied on nonspecific expansion to yield a sufficient number of cells for analysis. Because the whole organ is removed at the time of liver transplantation, it is possible to characterize intrahepatic cells directly ex vivo without in vitro expansion. As shown in Fig. 1, the liver is enriched for HCV-specific cytotoxic CD8+ T cells [cytotoxic T lymphocytes (CTLs)]. Intrahepatic lymphocytes typically demonstrate distinct phenotypic profiles associated with exhaustion, including up-regulation of programmed death 1 (PD-1)12, 13 and down-regulation of CD127.14 Understanding the molecules associated with T cell exhaustion within the hepatic compartment provides insights and a rationale for novel therapeutic targets. For example, blockade of the PD-1/programmed death ligand pathway restores the functional competence of HCV-specific CTLs. A number of ongoing studies are targeting this pathway either by blocking interactions between the receptor and its ligand(s) or by down-regulating PD-1.

Figure 1.

The liver is enriched for antigen-specific CTLs; 10.9% of CTLs stain for the HCV NS3-1073 epitope. (A) Flow cytometry dot plot gated on CD3+CD8+ T cells showing HCV-specific A2-1073 CTLs (R4) enriched within the hepatic compartment. The expression of PD-1 on antigen-specific (71.74%) cells with respect to bulk CTLs (6.02%) is shown in the histogram. CTLs shown in R5 may include cells reactive against other regions of HCV. (B) Total HCV-specific PD-1 expressed as the percentage of pentamer+ CD8+ T cells (MFI). Chronic HCV infection is associated with a greater percentage of PD-1+ HCV-specific CTLs in the periphery (n = 17 patients, 45 pentamer responses) and in the liver (n = 9 patients, 29 pentamers). The data are shown for 7 patients (21 pentamers) with resolved infection. The intensity of PD-1 staining is also higher in the cells of chronically infected patients, with the most concentrated PD-1 expression being shown in the liver.14Abbreviations: CTL, cytotoxic T lymphocyte; HCV, hepatitis C virus; MFI, mean fluorescence intensity; PD-1, programmed death 1; PERCP, peridinin chlorophyll protein.

Accelerated Natural History Allows the Definition of Distinct Disease Outcomes Within a Relatively Short Period of Time

In the immunocompetent setting, chronic HCV infection is a very slowly progressive disease, and this makes a prospective evaluation of its natural history very problematic.15 Consequently, long periods of time are required to document any clear-cut evidence of progressive liver injury. In contrast, the proportion of HCV-positive liver transplant recipients who develop cirrhosis at 5 years ranges from 21% to 35%.16–18 Accordingly, the median and mean rates of fibrosis development (which are nonlinear) are significantly higher than those observed pre-transplantation (P < 0.0001).19 As a result, annual protocol liver biopsies are recommended in HCV-positive liver allograft recipients.20 The telescoped natural history of HCV has allowed the identification of specific factors associated with disease progression; donor age21, 22 and the use of T cell depletion for the treatment of rejection (eg, OKT3)23, 24 have consistently been associated with more rapid disease progression. Interestingly, prospective analyses of nontransplant patients have shown that age at index biopsy (rather than duration of infection) is an important predictor of progression, suggesting that HCV becomes more fibrogenic with advancing age.25 Conversely, as shown in the immunoglobulin anti-D cohort, young age appears to be protective.26 The importance of the T cell response in mediating HCV clearance has been underscored in a number of studies14, 15, 27, 28 (also discussed later in this article).

Prospective Tracking of Patients: Serum, Peripheral Blood Mononuclear Cells, and Liver Biopsy

One of the main limitations of studies involving patients with HCV is the cross-sectional or retrospective nature of most analyses. An inherent advantage of the transplant model is that the time of infection of the allograft is known and patients are prospectively followed as part of their clinical care; this allows the collection of blood and serial analyses of liver tissue. Because acute infection of the allograft invariably occurs, there is an opportunity to study the innate and adaptive immune responses triggered in the early stages of infection. Moreover, studying the role of HCV quasispecies in disease progression usually requires some perturbation in the immune response either by the immune system in the acute setting or under treatment-induced pressure.14 Thus, although the immunocompetent setting is associated with stable viral replication that does not vary to a significant degree over months to years, liver transplantation is characterized by a marked increase (∼16- to 20-fold)29 in circulating viral titers, allowing both viral kinetics and viral sequence evolution analyses.

The major disadvantages of the liver transplantation model are related to the large number of variables [eg, ischemic injury, immunosuppression, degree of human leukocyte antigen (HLA) mismatch, and donor genetic factors], which preclude simple interpretation. Moreover, acute infection of the allograft occurs in a patient who is not immunologically naïve to HCV, and therefore extrapolation to the acute infection setting is not perfect. Nonetheless, as outlined next, the liver transplantation model has provided us with important insights into the roles of viral kinetics and quasispecies, hepatitis C receptor binding and viral entry, and innate and adaptive immunity.


Viral kinetics refers to the serial quantification of the amount of virus in the serum of infected patients. This has been used for multiple purposes, such as the measurement of disease progression and response to antiviral therapy and the prognosis following liver transplantation. The serum viral load at the time of measurement likely reflects a complex interaction between viral production by infected cells and clearance by the host immune system. After liver transplantation, the relative contribution of each of these factors likely differs, depending on the point in time at which the patient's serum is sampled. Although often used to evaluate the degree of liver infection, measurements of the amount of virus in the serum may not be as relevant as measurements of the amount of virus in the liver. Intrahepatic and membrane-associated virus, rather than freely circulating virus, likely causes liver injury; therefore, the liver viral load may better reflect the magnitude of infection than the serum viral load. As patients differ in rates of viral clearance, it seems reasonable to suppose that liver-derived and serum-derived viral loads may differ, although this is not clearly defined. Whereas Terrault et al.30 found that serum and liver viral loads differed widely (the ratio of the liver viral load to the serum viral load ranged from 17 to 286), Sreekumar et al.31 demonstrated that serum and liver viral loads correlated well (r = 0.77–0.93, P < 0.01), though intrahepatic levels were always higher (79-fold on average). Both groups compared serum and liver viral loads after liver transplantation using the same technique, and these reported differences may have resulted from the narrow dynamic range of detection for their assays. These early generation branched DNA assays discriminated a 3-log range of concentrations, in contrast to current polymerase chain reaction (PCR) techniques, which routinely can quantify virus over a 6-log viral concentration.32 How well serum and liver amounts correlate with current PCR methodology normalized to known standards in immunocompromised transplant patients remains unknown.

During the liver transplant operation and early post-transplant, there are likely important differences in serum and liver viral loads. Garcia-Retortillo et al.33 showed that the allograft acts as an absorption column for the virus. The serum viral load rapidly decreases with reperfusion (Fig. 2) of the allograft, presumably as the liver removes the virus from the circulation and the intrahepatic viral amount increases (the intrahepatic viral load was not determined in this study). The rate of viral removal or efficiency of absorption during reperfusion was variable between patients, and this suggests that some livers are more capable of binding the virus. Those livers with significant ischemia-reperfusion (I-R) injury appeared less capable of associating with the virus, as the viral half-life was prolonged in comparison with clearance for livers without I-R injury. This may be due to cell surface receptor injury and impaired hepatocyte interaction with HCV. Although I-R injury may diminish viral uptake by the liver, its overall effect may be to hasten recurrence due to other mechanisms (eg, impaired regenerative capacity of the liver). This is suggested by data demonstrating that increasing warm ischemia time contributes to increased rates of viral recurrence34 and that preservation injury has a more pronounced impact on outcomes for HCV-infected patients than for HCV-uninfected patients.35, 36 By demonstrating that early after transplant allografts differ in the number of infected cells (immunohistochemical staining for the viral antigen), prior studies37, 38 support the notion that allografts differ in their ability to bind the virus. Characterization of the factors contributing to the rate of viral clearance by the liver might allow pretransplant identification of allografts more susceptible to infection and prevent the use of organs likely to have a higher affinity for HCV.

Figure 2.

Schematic of perioperative viral kinetics. The serum viral load remains fairly stable during hepatectomy as the liver continues to release virus into the circulation. During the anhepatic phase, the viral load begins to decrease as the virus-producing liver has been removed. The rate of viral decrease during this stage is related to blood loss and dilution from resuscitation. With reperfusion of the allograft, the virus is rapidly removed from the circulation by the previously uninfected allograft. It appears that allografts are variable in their ability to bind and remove the virus as rates of viral clearance differ markedly. The viral load reaches a nadir (likely due to saturation of cell surface receptors for the virus) and then starts to increase with established infection and viral replication. Adapted from Hepatology.33

Viral-specific factors also contribute to the variability in viral absorption by the liver. It has been shown that only a portion of the virus in a recipient's bloodstream infects the liver, and the fraction of the virus capable of infection differs between patients.39, 40 A larger infectious fraction might result in more rapid viral clearance as a greater proportion of the viral inoculum is bound and removed from the serum by the liver but this remains unknown. Furthermore, a greater inoculum size should predict worse outcomes. With primary HCV infection, inoculum size predicts the likelihood of seroconversion41, 42 and severity of subsequent infection43 in vivo and the amount of virus internalized within cells in vitro.44 By analogy, the level of viremia at the time of allograft reperfusion may predict worse outcomes following liver transplantation. This remains unclear as early studies measuring the pretransplant viral load45, 46 quantified the viral amount prior to initiation of the transplant operation, not immediately prior to reperfusion. Therefore, these results are likely confounded by blood loss and ongoing resuscitation during this intervening period. No recent studies are available to clarify this potential discrepancy. Furthermore, secondary sites of viral infection (eg, lymph nodes47) may also contribute to variability in the amount of virus available to infect the liver and in the net rate of viral clearance. Both the inoculum size and the proportion that is infectious should contribute to how much virus infects the liver, and this may more accurately predict outcomes than currently measured predictors.

The serum viral load reaches a nadir 8 to 24 hours after reperfusion, and this likely represents saturation of cell surface receptors for HCV in the allograft.33 The subsequent increase in the viral load should represent established infection and production of new virus by the infected allograft. During the first 7 days following transplantation, the viral kinetic appear highly variable between individuals and may be related to attenuated immunological responses of the recipients. In this respect, Garcia-Retortillo et al.33 observed that 5 of 6 patients with a second-phase decline in the HCV viral level did not receive corticosteroids, whereas only 1 of 13 patients who received corticosteroids as part of their immunosuppressive regimen showed a second-phase decline in the viral level. Fukumoto et al.48 similarly demonstrated that for patients with steroid induction, the serum viral load increases early (typically by postoperative day 2). These observations are consistent with studies demonstrating that viremia increases with corticosteroid administration for both immunocompetent patients with chronic hepatitis49–51 and immunosuppressed liver transplant patients treated for rejection.31, 52 This may represent a steroid-specific effect rather than diminished immunocompetence, as cyclosporine A administration in chronically HCV infected patients does not increase viremia53 (although there is controversy as to whether cyclosporine has antiviral action). It is likely that the steroid effect impairs immunological clearance rather than stimulating viral production by the allograft, as steroid administration does not appear to alter rates of viral replication in livers after transplantation.54

Once infection of the allograft is established, the serum viral level may more accurately reflect the intrahepatic amount of the virus, and this may explain how the early serum viral load can predict subsequent outcomes. In the first month post-transplant, the serum viral load increases to as much as 20 times pretransplant levels,45 possibly because the allograft is better able to support viral replication than the now removed cirrhotic liver.55 How high the viral load rebounds early (<30 days) can predict the severity of disease recurrence56; however, data are mixed regarding the predictive power of the serum viral load beyond the first month of transplant.31, 48, 57–59 Therefore, the severity of the initial infection likely plays a significant role in subsequent outcomes, and later viral load patterns become more complex to interpret when multiple factors (ie, immunosuppression, robustness of the host immune system, viral fitness for propagation, and allograft support of replication) likely determine viral replication.

What Is the Differential Contribution of the Allograft in Supporting Viral Replication?

Certain allografts may serve as more efficient hosts for viral replication than others. Negro et al.54 demonstrated that rates of viral replication in allografts [as determined by antigenomic, strand-specific reverse-transcriptase polymerase chain reaction (RT-PCR)] appear to differ between patients and seem unrelated to immunosuppression. This study was limited by its semiquantitative nature and lack of consistent, defined protocol biopsy time points. Aside from this study, rates of viral replication in the liver have not been well studied after transplantation, likely because of the challenges of antigenomic or negative strand–specific RT-PCR. The ability of the positive strand to self-prime the negative-strand replicative intermediate has necessitated the use of less sensitive thermostable reverse transcriptases and therefore sensitive but poorly quantitative Southern blotting to visualize the product.39 With the development of strand-specific real-time RT-PCR allowing for more reliable and discriminatory quantification of replication,60, 61 future studies evaluating differences in viral replication between allografts is warranted.

It has been shown that Huh 7.5 cells, which have a single point mutation in the double-stranded RNA sensor retinoic acid–inducible gene I, have impairment of innate antiviral defenses within the hepatocyte and are more permissive to viral replication, yielding viral titers that are approximately 50-fold higher and more efficient spread of the infection.62 The possibility that differential interferon (IFN) signaling, including the phosphorylation and nuclear translocation of IFN regulatory factor 3, may play an important role early after transplantation is an important hypothesis that has yet to be tested. It was first suggested 10 years ago63, 64 that cytokine gene polymorphisms within the allograft might mediate the inflammatory and antiviral microenvironment early post-transplantation, but whether there is coupling between innate immunity within the hepatocyte (eg, IFN production) and subsequent expansion of innate and adaptive lymphocytes remains unexplored (discussed later). If associations between genetic markers and HCV-related allograft injury are confirmed, these results would have important implications for the identification of optimal donors (particularly in the living donor setting, in which there would be time to screen for these markers) and for the identification of high-risk patients earlier, which would allow the initiation of antiviral therapy either preemptively or with the first signs of histological recurrence.


Because of the genetic variability of the virus, the viral envelope proteins that mediate attachment to liver cells mutate rapidly and frequently, and this results in multiple quasispecies within an individual.65 The second viral envelope protein (E2) is thought to be the primary interface with host cells.66, 67 Within this region of the viral envelope is a highly variable region [highly variable region 1 (HVR1)] consisting of 27 amino acids68–71 that is thought to derive its variability from a combination of immunological pressure71 and the high error rate of the RNA-dependent RNA polymerase that the virus uses for replication. These different amino acid sequences within each individual each represent a quasispecies variant. The role of quasispecies evolution with established infection is controversial, and most studies have focused on quasispecies diversity as a surrogate for immunological pressure (ie, more pressure equals more diversity, and less pressure equals less diversity). Although Sullivan et al.72 found that higher levels of diversity correlated with less severe recurrence (presumably because the immune system mounted a stronger humoral response to the virus), Pessoa et al.73 showed that immunosuppressed transplant patients had greater quasispecies diversity than immunocompetent nontransplant patients. In fact, the quasispecies nature of established infection with all RNA viruses has been called into question.74

Regardless of the state of flux for viral mutation and immunological pressure, certain HVR1 protein sequences appear to be selected out by the liver as it is infected with the virus. A complex population of quasispecies generated in the liver and secondary sites of infection such as lymph nodes47 is available to bind and infect the liver. Hughes et al.39, 40 demonstrated that only a portion of this infectious inoculum present in patient serum prior to reperfusion of allografts went on to infect the liver and that this quasispecies selection began immediately upon reperfusion. There was a sudden shift in the relative frequency of certain quasispecies variants over others prior to any detectable viral replication in the liver. The authors obtained serum samples prior to reperfusion of liver allografts and liver biopsies after reperfusion (but prior to completion of the operation) and 5 to 10 days after transplantation. By sequencing the PCR product, they then determined which quasispecies from the infecting serum were also present in liver samples.

As demonstrated for 1 representative patient (Fig. 3), the HVR1 quasispecies that were recovered from the preperfusion serum were quite diverse, and not all were infectious. Of the 14 variants identified, only 3 were recovered from the postperfusion liver, and only 1 of the variants selected out by the liver went on to infect the allograft at week 1. Though usually considered an immunological decoy, HVR1 has been shown to have properties that could allow it to interface with the host cells. Despite its high rate of nucleotide variability, the chemical-physical properties and conformation of this exposed region on the viral envelope are highly conserved, with basic residues consistent with a role in protein receptor and glycosaminoglycan binding.75, 76 Selection for HVR1 variants has also been observed in vitro77–79 and in chimpanzees,80 and antibodies directed against HVR1 have prevented infection.67, 81–84 Therefore, HVR1 must play some role in viral attachment to cells.

Figure 3.

Quasispecies selective nature of allograft infection. As demonstrated in a single, representative patient, only a portion of HVR1 quasispecies binds and subsequently infects the liver. New quasispecies are rapidly generated with allograft infection and intrahepatic viral replication. The first sequence represents the consensus sequence. Each letter of the sequence represents the amino acid at the given position. Hyphens represent conserved amino acids. The number next to each sequence represents the number of times that sequence was identified in the given sample. Abbreviation: HVR1, Hypervariable Region 1. Reprinted with permission from American Journal of Transplantation.39 Copyright 2005, John Wiley & Sons, Inc.

The results of Hughes et al.'s study39 may partially explain the variable rates of viral clearance by allografts during reperfusion reported by Garcia-Retortillo et al.33 In addition to the patient from Fig. 3, 6 other patients demonstrated variable degrees of quasispecies selectivity, with different infectious fraction sizes (number of variants in the reperfusion liver/number of variants in the pretransplant serum) between patients. It seems likely that a patient with a large infectious fraction would have a faster rate of viral clearance during reperfusion and a greater magnitude of infection than one with a smaller infectious fraction. It is also possible that persistence of a predominant variant from the pretransplant serum to the postperfusion liver would result in higher rates of clearance and a greater magnitude of infection. This may explain the results reported by Gretch et al.,85 who found that persistence of a predominant serum variant from the pretransplant serum to the posttransplant serum was associated with recurrent HCV disease, whereas failure of predominant variants to persist post-transplant was associated with no early recurrence. For the predominant variant to persist and lead to recurrent disease, it may represent the most infectious variant leading to a greater infectious fraction and magnitude of infection, although the relationship between quasispecies selection and viral kinetics has not been assessed and therefore that conclusion cannot be made. When evaluated overall, the postperfusion liver selected out a population of variants more closely related to one another than those in the infecting serum (Fig. 4). This selection for a closely related population of quasispecies (out of a diverse population of potentially infectious quasispecies) by the liver suggests that there is an as yet unidentified common property to infectious HVR1 sequences that could be targeted to decrease the magnitude of the initial allograft infection.

Figure 6.

Levels of CD56+ lymphocytes in chronic HCV patients and HCV-negative liver disease patients prior to liver transplantation as well as normal healthy controls. (A) Multiparameter flow cytometry analysis was used to estimate the levels of CD56 lymphocytes, NK (CD56+CD3) and NT (CD56+CD3+) cells, in chronic HCV infection prior to liver transplantation. (B) Total CD56+ lymphocyte levels were significantly decreased in all chronic HCV patients compared to normal uninfected control subjects. This reflects a deficiency in both NK and NT cells. A decrease in total CD56+ cells was also observed for control non-HCV chronic liver disease patients because of a significant reduction in NT cells but not NK cells. Chronic HCV patients were stratified into 2 groups depending on the severity of disease recurrence post–liver transplantation. Of interest, all CD56+ lymphocyte populations were decreased in the patient group with subsequent severe outcome in comparison with those who had mild recurrence of HCV liver disease. In comparison with non-HCV liver disease, this reduction was significant for total CD56+ and NK (but not NT) cells only for the severe group. *P < 0.05. Abbreviations: HCV, hepatitis C virus; NK, natural killer; NT, natural T; PerCP, peridinin chlorophyll protein. Reprinted with permission from Liver Transplantation.132 Copyright 2008, American Association for the Study of Liver Diseases.

Figure 4.

Hypervariable region 1 quasispecies selected out by the liver are more closely related than the pool of quasispecies in the pretransplant serum inoculum. With the generation of new quasispecies, diversity increases. Gray lines represent individual patients. The black line represents mean values. Abbreviation: MAAD, maximal amino acid diversity. Reprinted with permission from American Journal of Transplantation.39 Copyright 2005, John Wiley & Sons, Inc.

After infection is established in the allograft and viral replication and mutation proceed, quasispecies continue to evolve and likely do so independently of the role that HVR1 has in cell attachment. A variant that represents a small proportion of the virus infecting the liver may have a survival advantage over other variants. This variant may go on to represent a majority of quasispecies within an individual as infection progresses, as small fitness differences may contribute to more efficient replication and propogation.86 The rapid increase noted in HCV viral loads after transplantation proves the high capacity of HCV to adapt to a new environment. In particular, viral escape from a dominant immune response early after liver transplantation could play a central role in viral persistence by enhancing viral survival when it is most susceptible to immune selection (ie, during massive infection of the allograft).87 Transplantation of an HCV-infected liver into an HCV-positive recipient represents a model of superinfection,88 and Vargas and colleagues89 demonstrated that superinfection of the liver recipient by the donor's strain was associated with significantly milder disease than the recipient strain becoming (or remaining) dominant. In addition, genotype 1 or 1b consistently predominated over non-1 or non-1b genotypes in recipients of infected grafts, and this suggested replicative differences among viral strains.

Despite the coexistence of virus-specific immune responses, HCV is able to persist for a virtually indefinite period of time in a tug of war with the host as a complex of heterogeneous and dynamic genomes. Most HCV quasispecies analyses in liver transplant patients have focused on HVR1 located at the N-terminus of the E2/NS1 region, and the results have been conflicting. As mentioned previously, Gretch et al.85 showed that successful propagation of pretransplant major quasispecies was associated with a more severe form of HCV disease recurrence, and this finding was subsequently confirmed by Doughty and colleagues.90 In contrast, Pessoa et al.73 found that, in a subset of patients with fibrosing cholestatic hepatitis, divergence of quasispecies was enhanced, and this resulted in the emergence of many new variants. However, differences in quasispecies are not in themselves definitive evidence for the existence of immune selection. The assumption that RNA viruses are in mutation-selection equilibrium has recently been called into question; that is, the state of flux in mutants previously ascribed to immune pressure may depend more on the relative fitness of viral subpopulations.74 In this model, lower viral loads in patients with epitopic sequence variation may simply reflect compromised replicative activity of the variant. These considerations are particularly relevant in the orthotopic liver transplantation setting, in which HCV may have a direct viral cytopathic effect and in which no protective role for virus-specific antibody responses has ever been established. Moreover, studies of anti-HCV immunoglobulins prepared from a pool of highly immune human plasma have failed to prevent graft infection in HCV-infected liver transplant recipients.91


CD81, expressed on the hepatocyte cell surface, likely contributes to HCV infection of allografts after transplantation. This widely expressed 25-kD protein in the tetraspanin superfamily is involved in cell adhesion and signal transduction.92 The protein spans the cell membrane 4 times and forms 2 extracellular loops that are exposed on the cell surface. Although the intracellular and transmembrane segments of the protein are highly conserved across multiple species, the extracellular loops are diverse and conserved only for humans and chimpanzees (species permissive for HCV infection).93, 94 The major extracellular loop has been shown to bind E2 of HCV, and antibodies that prevent infection in vivo (chimpanzees) neutralize binding of HCV to CD81.93

E2 can bind CD81 independently of any other cell surface receptors, and several areas (including HVR1) within E2 have variable impacts on binding affinity.94 The role of HVR1 within E2 is controversial. Conflicting studies have demonstrated that changes to HVR1 do not appear to alter the conformation of E2 or the ability of E2 to bind CD8195, 96 and that HVR1 is a key determinant of E2 binding of CD81.97

The amount of CD81 on the cell surface determines the amount of E2 that can bind the cell,94 and cells may need a certain threshold level of the receptor for infection.98 By analogy, allografts that remove virus from the circulation during reperfusion at a faster rate than other allografts33 may have a greater density of CD81. In the only study evaluating CD81 expression in liver transplants, the authors demonstrated that allografts differ in CD81 density at the time of transplantation.39 Figure 5 demonstrates anti-CD81 staining in an allograft from 1 representative patient at the time of transplant and up to 4 months following transplant. Although allografts appeared to express high levels of CD81 by 1 month, they differed in the amount at the time of reperfusion. Whether or not these differences resulted in variable magnitudes of infection remains unknown, but it is reasonable to hypothesize that some allografts are more susceptible to infection than others on the basis of the amount of CD81 expressed on the cell surface.

Figure 5.

CD81 expression (brown staining) after liver transplantation: (A) the explant liver, (B) the allograft 2 hours after reperfusion, (C) the allograft 3 days post-transplant, (D) the allograft 1 week post-transplant, and (E) the allograft 2 months post-transplant. Reprinted with permission from American Journal of Transplantation.39 Copyright 2005, John Wiley & Sons, Inc.

Decreasing the ability of CD81 to interact with HCV may alter allograft infection. Ischemia-reperfusion injury (as previously mentioned) may alter allograft affinity for HCV33 and quasispecies selectivity,40 possibly through damage to CD81. Furthermore, antibodies directed against CD81 have been shown to prevent infection in an in vivo SCID mouse/human hepatocyte model.99 These results need to be further validated, as it remains unclear what role CD81 plays in allograft infection.

CD81 likely acts in concert with other cell surface receptors to capture and internalize the virus. In Hughes et al.'s study of quasispecies selection by allografts,39 the authors demonstrated that only a fraction of the virus captured by the allografts went on to infect the allografts. This suggests that multiple receptors bind the virus, but not all receptors internalize the virus. CD81 by itself does not appear to result in infection, as tamarins (species of monkey) that express CD81, which binds E2 with high affinity,100 and mice expressing human CD81101 are both refractory to HCV infection. Despite having a high binding efficiency similar to that of the human immunodeficiency virus gp120-CD4 interaction, only 30% of CD81 internalizes after binding (versus 50%–80% internalization of CD4).102 CD81 may act as an intermediary between attachment and internalization of the virus, as anti-CD81 antibody prevents internalization of pseudoviral particles bound to hepatocytes.103

The low-density lipoprotein receptor (LDLr) may act alone or in concert with CD81 to bind and internalize the virus. HCV has been found to associate with circulating low-density lipoprotein (LDL) in the serum,104–107 bind LDLr on hepatocytes, transfer into cells, and then begin replicating.108–110 Agnello et al.108 demonstrated that the amount of intracellular virus correlated with LDLr density and that anti-LDLr antibodies markedly diminished infection. The fact that anti-LDLr antibodies incompletely prevented infection and that fibroblasts deficient in LDLr were still infected with the virus argues that more than 1 cell surface receptor is involved with infection. It has also been demonstrated that the first envelope protein (E1),111–113 in addition to E2,113 which binds CD81, likely interacts with LDL and leads to internalization of the LDL-HCV complex. Furthermore, CD81 may be a necessary coreceptor for LDLr-mediated internalization of the HCV-LDL complex.113

The recent discovery of other candidate receptors for the virus, human scavenger receptor type B-class I (SR-BI)114 and claudin-1,115 adds to the evidence that HCV infects cells through a complex interplay between other cell surface receptors, viral envelope proteins (including HVR1), and lipoproteins. SR-BI has been found to bind E2 in an HVR1-dependent manner114, 116 requiring coexpression of CD81 for viral infection.116, 117 SR-BI may enhance viral attachment and entry by binding high-density lipoprotein (HDL),116 but it remains unclear how E2/HVR1 interacts with HDL. Evidence suggests that E2 binds SR-BI directly and then is internalized through cholesterol transfer from HDL.118–120 Anti–SR-BI antibodies prevent HCV infection in the presence of HDL,121 but unlike LDL or LDLr, HDL does not appear to associate with HCV and carry HCV into cells.119 Moreover, the recent demonstration that human occludin is an essential HCV cell entry factor whose silencing decreases HCV infectivity opens up a novel target for therapy.122 Future studies assessing the efficacy and safety of antibody cocktails used to block the specific binding and internalization of HCV in the early stages of reperfusion are warranted.


Innate immune lymphocytes are believed to play important roles in the immediate response to viral infections by the production of IFN-γ and/or the recognition of virus-infected cells.123 Natural killer (NK) cells mediate the lysis of virus-infected cells via natural cytotoxicity and antibody-dependent cellular cytotoxicity and are controlled by positive and negative cytolytic signals. Accordingly, injection drug users who remain uninfected with human immunodeficiency virus (IV) 1 despite many years of high-risk exposure demonstrate significantly augmented NK cell lytic activities and cytokine secretion in comparison with HIV-1–infected injection drug users.124

The liver has a uniquely specialized immune system enriched for NK and natural T (NT) cells.123, 125 Studies126, 127 have demonstrated that binding of the HCV envelope protein, HCV-E2, to CD81 directly blocks NK cell functional activation, proliferation, cytokine production, and cytotoxic granule release, instead of giving a costimulatory signal as with T cells. CD81-mediated inhibition of NK cells has been seen for activated and resting NK cells and in NK cells from healthy uninfected donors as well as patients with chronic HCV infection, and this suggests that the inhibitory effect is a general function of NK cell–CD81 ligation and could occur at all stages of infection.123–125 Additional work has shown that NK cells are depleted and phenotypically altered in chronic HCV infection.129–131

In a liver transplantation study composed of 4 patient groups [patients with mild HCV recurrence (n = 9), patients with severe HCV recurrence (n = 10), patients with non–HCV-related liver failure (n = 10), and normal healthy subjects (n = 10)], we found that higher levels of CD56+ lymphocytes are protective; that is, they are associated with milder HCV recurrence (Fig. 6).132 Moreover, HCV is associated with impaired lymphokine-activated killing and natural cytotoxicity and higher expression of the inhibitory receptor NKG2A with respect to HCV-negative controls with liver disease. Taken together, these data showing decreased CD16 expression, decreased circulating frequency of CD8+CD56+ lymphocytes, and increased FasL on CD56+ lymphocytes depict a model of global dysfunction in HCV infection with impaired antibody-dependent cellular cytotoxicity and natural cytotoxicity as well as enhanced apoptosis. A recent prospective study evaluated the population dynamics of innate immune cells for 1 year; there were variations in the frequencies of circulating NK, NT, and γδ T cells occurring within the first week after surgery, but they generally returned to baseline values 1 month later.133 Importantly, HCV RNA levels were statistically associated with the proportion of circulating NK cells. The authors proposed that these observations may be due to a de novo graft infection resulting in an acute inflammatory process and homing of innate and adaptive immune cells to the liver in order to exert their effector function.133

The activity of NK cells is controlled through a complex quantitative and qualitative balance of cell surface activating and inhibitory receptors that react with major histocompatibility complex class I and class I–like molecules.134 The killer cell immunoglobulin-like receptors (KIRs) represent a diverse family of activatory and inhibitory receptors interacting with self–major histocompatibility complex class I ligands. Recently, Espadas de Arias et al.135 explored the role of KIR genotypes and their HLA ligands in HCV disease recurrence and progression in the setting of liver transplantation in a well-defined cohort of 151 donor-recipient pairs. The main findings of this study were as follows: mismatching of KIR–HLA-C ligands between donor-recipient pairs was associated with recurrent hepatitis; the presence of KIR2DL3 in the recipient correlated with fibrosis progression; and in the presence of KIR2DL3, mismatching of KIR–HLA-C ligands favored progression of recurrent hepatitis to fibrosis. Although these results indicating a genetic NK component contributing to allograft injury need to be confirmed independently, they might have important implications in terms of the frequency of protocol biopsies and the threshold for consideration of antiviral therapy. Taken together, the published findings indicate dual roles for NK cells in HCV recurrence: prior to transplant, the frequency of NK cells is associated with protection against severe recurrence, whereas following transplant, mismatching leads to unleashing of NK cells (usually in an inhibited state by default), and this contributes to allograft inflammation and injury.


A failure to specifically mount an efficient immune response to HCV antigens because of selective defects in the host immune system, because of the effects of immunosuppressive drugs, or because high viral titers affect the normal function of the immune cells might account for why the majority of HCV-infected transplant recipients develop allograft hepatitis. An early study by our laboratory demonstrated that approximately 40% of patients with minimal or self-limited recurrent HCV demonstrated proliferative responses to HCV antigens, whereas none of the patients with severe recurrence did so.136 Subsequent data derived from our laboratory and others137–139 indicate that more sensitive assays [eg, enzyme-linked immunosorbent spot (ELISPOT) and intracellular cytokine staining] do indeed demonstrate viral-specific T cell responses in a higher proportion of patients with end-stage disease and after transplantation than initially appreciated with more conventional assays. CD8+ T cells are the primary effector lymphocytes for the provision of protective immunity against intracellular pathogen infections of parenchymal cells. Moreover, we have provided evidence that T cell responses, particularly HCV-specific CTLs, emerge after liver transplantation, correlating with improved virological response and clinical outcomes (see Table 2).

Table 2. Putative Protective and Adverse Immunological Mechanisms in HCV Recurrence
  1. Abbreviations: HCV, hepatitis C virus; HLA, human leukocyte antigen; KIR, killer cell immunoglobulin-like receptor.

  • HCV-specific CD4+ T cells early post-infection136, 137

  • HCV-specific CD8+ T cells in the setting of antiviral therapy137, 138

  • Pretransplant level of CD56+ lymphocytes132

  • Decreased CD56+ lymphocytes pre-transplant associated with severe recurrence132

  • T cell depletion treatments23, 52

  • Mismatching of KIR–HLA-C ligands between donor-recipient pairs and presence of KIR2DL3 in the recipient135

  • Cytokine gene polymorphisms63, 64

  • Impaired innate interferon signaling within hepatocytes (unknown)

  • Allo-restricted T cell responses (unknown)125, 144

  • Relative allograft expression of CD81 and other HCV receptors with immune properties (unknown)39

As examples, Fig. 7A,B shows 2 HCV-positive recipients who were HLA A2+ recipients of HLA A2+ donor livers, allowing for a detailed comparison of HCV-specific CTL responses with HLA A2–restricted tetramers. Both patients developed severe HCV recurrence. The first patient had high frequencies of multispecific CD4+ T cell responses specific to HCV NS3, NS4, and NS5 antigens on the day of transplantation that decreased in the ensuing months. Levels of tetramer responses to HCV-specific peptides were essentially negative at the first 3 time points but became detectable after initiation of therapy. The CTLs that peripherally reconstituted after transplantation were terminally differentiated (CD28lowCCR7low) memory cells with effector function and clonotypically identical to CTLs present in the explant liver on the day of transplantation. Taken together, these data support a model in which recipient-derived CTLs repopulate the allograft post-transplant, and these efflux into the peripheral blood after the viral load is brought under control.

Figure 7.

Longitudinal analyses of HCV specificity in 2 patients with severe cholestatic HCV who received antiviral therapy. Both patients were HLA A2+ recipients of HLA A2+ donor livers. (A) Reconstitution of HCV-specific cellular immunity in a patient with severe cholestatic HCV recurrence who responded to antiviral therapy (HCV RNA expressed as 106 copies/mL). (Top) IFN-γ enzyme-linked immunosorbent spot responses to HCV recombinant proteins, viral load, and serum bilirubin. (Middle) CD8+ T cell responses to the NS3 1073 tetramer. (Bottom) Amino acid sequence of the NS3 1073–1081 epitope at 4 time points (HCV genotype 1a prototype sequence: CINGVCWTV). (B) HCV-specific immune responses in a patient with severe cholestatic HCV recurrence who failed to respond to antiviral therapy (HCV RNA expressed as 106 copies/mL). (Top) IFN-γ enzyme-linked immunosorbent spot responses to HCV recombinant proteins, viral load, and serum bilirubin. (Middle) CD8+ T cell responses to the NS3 1073 tetramer. (Bottom) Amino acid sequence of the NS3 1073–1081 epitope at 4 time points (HCV genotype 1a prototype sequence: CINGVCWTV) and amino acid substitution (V for I at position 2 was detected but remained stable over time). Abbreviations: E2, second envelope protein; HCV, hepatitis C virus; HLA, human leukocyte antigen; IFN, interferon. Reprinted with permission from Hepatology.137 Copyright 2005, American Association for the Study of Liver Diseases.

In contrast, the second patient failed to respond to antiviral therapy and consequently died 16 months after orthotopic liver transplantation. Recent work in the acute infection (nontransplant) model has shown that priming of CTLs in the absence of vigorous HCV-specific CD4 help is associated with ultimate demise of these cells and lack of viral control.140 By analogy, as shown in Fig. 7B, CD4+ T cell ELISPOT responses were lacking at all but one time; NS3 1073–specific CD8+ T cell frequencies declined progressively, despite the initiation of antiviral therapy, from 0.58% on the day of transplantation to 0.02% by the 10th postoperative month. We excluded the presence of viral escape mutations as a potential cause for the changes in tetramer-specific frequencies by direct sequencing of the epitope coding region at various times, in accordance with prior work showing no association with HCV sequence variation within CTL epitopes and the severity of recurrence.141, 142

Liver transplantation is performed with no regard to specific matching of donor-recipient HLA alleles, and this may be a barrier to the development of protective (ie, antiviral) cell-mediated immunity directed against infected cells within the allograft. Indeed, Belli et al.143 found that full DRB1 donor-recipient mismatch was the only significant variable by multivariate analysis that was associated with progressive fibrosis related to HCV recurrence. These data underscore a protective role for CD4+ T cells and conversely show that impaired antiviral immunity contributes to the occurrence and severity of HCV following transplantation. A study from our group144 has shown the generation of new HCV-specific CD8+ T cells that are restricted by donor HLA alleles yet derived from the recipient's original T cell pool (Fig. 8). For the purposes of this study, we selected HLA A2–negative recipients of HLA A2+ grafts. As shown in Fig. 8, recipient HCV-specific CTL clones cultured with HLA A2+ lymphoid cell lines that had been pulsed with a cognate peptide (NS31406–1415, KLVALGINAV) but not with an irrelevant HCV core peptide elicited a strong immune response by IFN-γ ELISPOT. Moreover, when these recipient CTLs were cultured with recipient-derived (syngeneic) lymphoid cell lines and cognate peptide, there was no appreciable immune response. Thus, these results indicate that these HCV-specific CTLs circulate within the recipient and display functional antiviral activity (ie, they secrete IFN-γ and show cytotoxicity) only when they encounter allograft-derived HLA molecules and viral peptide. These results underscore the plasticity of the T cell receptor and the need to assess both allograft-restricted and recipient-restricted CTLs144 to comprehensively understand the nature of protective immunity to HCV after liver transplantation.

Figure 8.

Recipient-derived T cells that recognize HCV peptides in the context of donor HLA molecules (HLA A2). Taken together, these results suggest that T cells circulate in liver transplant recipients and are not activated until they encounter donor alleles containing HCV peptides. (Top) Enzyme-linked immunosorbent spot assay was performed with 1000 T cells and 20,000 LCLs expressing an A2 allele (top row) or an A3 allele (middle row) or syngeneic (recipient-derived) LCLs (bottom row) cocultured with no peptide, a cognate peptide (NS31406–1415, KLVALGINAV), or an irrelevant HCV core peptide (core35–44, YLLPRRGPRL). Reprinted with permission from Journal of Immunology.144 Copyright 2004, American Association of Immunologists. (Bottom) Demonstration that these HCV-specific clones do not react with allo-HLA alone. LCLs expressing all potential HLA class I alleles from the donor and recipient were used as antigen presenting cells. Briefly, no peptide or cognate peptide (KLVALGINAV) was added to 3 × 106 LCLs, and 1.5 × 106 clones were cocultured for a total of 6 hours (2μM monensin was added after 2 hours). Cells were rinsed twice with PBS and 1% BSA and stained for 30 to 40 minutes with tetramer; cells were rinsed twice, fixed in 200 μL of 4% paraformaldehyde for at least 15 minutes, and then permeabilized with 2 mL of 1× PermWash (Pharmingen) for half an hour at 4°C. Permeabilized cells were stained with monoclonal antibodies to IFN-γ. Stained cells were washed twice in 2 mL of PBS and 1% BSA and fixed in 300 μL of 2% paraformaldehyde. Acquisition was performed within 24 hours of staining; all flow cytometry data were collected with a FACSCalibur flow cytometer (BD Biosciences) and analyzed with CellQuest software (BD Biosciences). (A) No peptide and LCL HLA A2A3B14B44, (B) cognate peptide and LCL HLA A2A3B14B44, (C) cognate peptide and LCL HLA A3A24B62B18, (D) cognate peptide and LCL HLA A30A33B13B14, (E) cognate peptide and LCL HLA A1A3B7B50, and (F) cognate peptide and LCL HLA A3A30B7B13 (recipient-derived). Abbreviations: BSA, bovine serum albumin; HCV, hepatitis C virus; HLA, human leukocyte antigen; IFN, interferon; LCL, lymphoid cell line; PBS, phosphate-buffered saline.


In conclusion, the past 10 years of research have yielded important insights into mechanisms that govern viral replication and mediate HCV-related allograft injury. These results may help identify patients more likely to develop severe HCV recurrence and therefore benefit from current antiviral therapy and provide a rationale for the future use of novel therapeutic approaches. Although a number of promising drugs that specifically inhibit the HCV cell cycle have been withdrawn during early stages of clinical development,145 it is expected that newer agents will become available for the treatment of chronic HCV in the near future. When used as monotherapy, current inhibitors show a low barrier to genetic resistance, a potential problem for antivirals given the high rate and error-prone nature of HCV replication.146 In theory, achieving steady-state levels for peginterferon and ribavirin may reduce the selection and expansion of resistant strains related to specifically targeted antiviral therapy for patients with chronic hepatitis C virus (STAT-C). This challenge is likely greater in liver transplant recipients because of higher viral levels and because a higher proportion of patients require IFN/ribavirin dose reduction (and even cessation). So far, no data are available on the pharmacokinetics and pharmacodynamics of STAT-C in the liver transplant setting.

A number of approaches and their challenges merit further consideration. Although trials with hepatitis C immunoglobulin have been disappointing,91 it is hoped that more comprehensive antibody discovery approaches will provide neutralizing protection in the posttransplant setting as demonstrated in the human liver–chimeric mouse model.99 Antibodies that block molecules involved in viral binding and entry (eg, antibodies directed against CD81)99 will likely need to be of very high affinity (affinity constant ≥ 10−9) to overcome the high affinity binding of E2 to CD81.102 Perfusion of livers with molecules designed to block viral binding ex vivo under cold conditions will likely be a challenge as E1 and E2 binding to cells at 4°C is 7- and 12-fold less than that at 32°C.112 Removal of virus during the anhepatic phase of transplantation, possibly in conjunction with portal or systemic bypass, may provide an innovative approach to reduction in the viral inoculum. This will likely pose a challenge as current technology (eg, double filtration plasmapheresis147–149) does not allow for the high flow rates necessary for the bypass circuit and unfortunately removes critical serum elements such as fibrinogen. An infusion of activated innate lymphocytes (eg, NK cells) derived from the donor might limit HCV replication within the allograft early post-transplant. Adoptive transfer of lymphocytes with high-avidity T cell receptors specifically targeting HCV150 (also M. Nishimura, personal communication, 2009) might represent a future approach that does not elicit alloimmunity.144