Cellular immune responses persist and humoral responses ... - Nature

© 2000 Nature America Inc. • http://medicine.nature.com

ARTICLES

Cellular immune responses persist and humoral responses decrease two decades after recovery from a single-source outbreak of hepatitis C AKINOBU TAKAKI1, MANFRED WIESE2, GEERT MAERTENS3, ERIK DEPLA3, ULRIKE SEIFERT1, ANKE LIEBETRAU2, JEFFERY L. MILLER4, MICHAEL P. MANNS1 & BARBARA REHERMANN1,5


1

Department of Gastroenterology and Hepatology, Medizinische Hochschule Hannover, Carl-Neuberg-Str. 1, 30625 Hannover, Germany 2 II. Klinik für Innere Medizin, Städtisches Klinikum St. Georg, Delitzscher Str. 141, 04129 Leipzig, Germany 3 Hepatitits Program, Innogenetics, Industriepark Zwijnaarde 7, Box 4, 9052 Gent, Belgium 4 Laboratory of Chemical Biology and 5Liver Diseases Section, NIDDK, National Institutes of Health, 10 Center Drive, Rm 9B16, Bethesda, Maryland 20892, USA A.T. present address: The First Department of Internal Medicine, Okayama University Medical School, 2-5-1 Shikata-cho, Okayama, 700-8558 Japan Correspondence should be addressed to B.R.; email: [email protected]

As acute hepatitis C virus (HCV) infection is clinically inapparent in most cases, the immunologic correlates of recovery are not well defined. The cellular immune response is thought to contribute to the elimination of HCV-infected cells1 and a strong HCV-specific T-helper-cell (Th) response is associated with recovery from acute hepatitis C (ref. 2). However, diagnosis of resolved hepatitis C is based at present on the detection of HCV-specific antibodies and the absence of detectable HCV RNA, and detailed comparison of the humoral and cellular immune response has been hampered by the fact that patient cohorts as well as HCV strains are usually heterogeneous and that clinical data from acute-phase and long-term follow-up after infection generally are not available. We studied a cohort of women accidentally exposed to the same HCV strain of known sequence3 and found that circulating HCV-specific antibodies were undetectable in many patients 18–20 years after recovery, whereas HCV-specific helper and cytotoxic T-cell responses with an interferon (IFN)-γ-producing (Tc1) phenotype persisted. The data indicate these HCV-specific CD4+ and CD8+ T cells are biomarkers for a prior HCV exposure and recovery. Because of undetectable antibodies against HCV, the incidence of self-limited HCV infections and recovery may be underestimated in the general population. We analyzed the HCV-specific humoral immune responses of 43 recovered and 34 chronically infected patients 18–20 years after documented exposure to an accidentally HCV (genotype 1b)-contaminated human Rh (Rhesus) immunoglobulin (IgG) (refs. 3,4). In contrast to chronically infected patients, a high percentage of recovered patients (18 of 43; 42%) tested negative for HCV antibodies by a commercial third-generation enzyme immunoassay that contains HCV structural and nonstructural proteins and is routinely used for the diagnosis of present or past HCV infection. Similarly, chronically infected, but not recovered patients had antibodies against HCV envelope (E) 1 and E2 proteins (P < 0.05) (Fig. 1a). To map and compare the epitope pattern of the humoral immune response in both patient groups, we tested partially overlapping peptides, 20–40 amino acid long, from the immunodominant re578

gions of individual HCV proteins. Among chronically infected patients, the most frequent and strongest antibody reactivities targeted epitopes in the HCV core and nonstructural (NS) 4 proteins (Fig. 1b) as well as the recombinant NS3 protein (not shown). In contrast, these and all other epitopes tested were only recognized by plasma samples from a minority of recovered patients. To determine whether HCV-specific antibodies were lost after clearance of HCV or whether they had never been present, we analyzed cryopreserved sera obtained 10 years after exposure, which were available from 10 of the 43 recovered patients. Using the standard third-generation enzyme immunoassay, all 10-year samples had antibodies against HCV, but the ‘signal-to-cut-off’ ratio (with a cut-off of the mean + 3 standard deviations of ten negative controls) of sera from 9 of 10 patients decreased substantially between 10 and 18 years after exposure, resulting in a signal below the cut-off in 5 of 10 cases 18 years after infection (Fig. 1c). The data support the idea that the HCV-specific antibody response gradually decreased and eventually disappeared after recovery from HCV infection. Although it is possible that these patients are still able to mount an amnestic response upon re-exposure to or re-infection by HCV, the findings indicate that hepatitis C can be resolved without a ‘scar’ in the humoral immune response; that is, without any humoral evidence of prior infection. This observation is comparable to the loss of antibodies against early Epstein-Barr virus antigens after recovery from infection with Epstein-Barr virus5. Despite the reduction in HCV-specific humoral immune responses, HCV-specific CD4+ and CD8+ cellular immune responses were still detectable in the blood of most recovered patients (Figs. 2 and 3). To analyze the CD4+ T-cell response, we stimulated peripheral blood mononuclear cells (PBMCs) from 29 recovered and 20 chronically infected patients with a panel of six recombinant HCV proteins. Using an experimental cut-off derived from the analysis of 15 uninfected controls, a proliferative response with a stimulation index equal to or greater than 3 was considered positive. Based on these criteria, PBMCs from 23 of the 29 recovered patients (79%) NATURE MEDICINE • VOLUME 6 • NUMBER 5 • MAY 2000


ARTICLES

a

b

c


Fig. 1 Humoral immune response. a, Antibody responses against recombinant E1 and E2 proteins. [S/C] (vertical axis), ‘signal-to-cut-off’ ratio. b, Antibody responses against immunodominant core and NS4 peptides. Horizontal axis, peptides identified by their position in the HCV polyprotein. c, Anti-HCV responses to HCV core–NS3 fusion protein and nonstructural proteins. In boxes, patient identification numbers. [S/C] (vertical axis), ‘signal-to-cut-off’ ratio.

recognized at least one HCV protein, compared with PBMCs from only 5 of the 20 chronically infected patients (25%) (P < 0.001; Fig. 2a), whereas the response against tetanus toxoid as a recall antigen was similar in both groups (Fig. 2b). In total, 54 of 174 assays against individual HCV proteins (31%) were above the cut-off in the recovered patients, and the stimulation index was often high. In contrast, only 5 of 120 individual assays (4%) from patients with chronic hepatitis C produced positive results (P < 0.001), and the stimulation indices were usually low (Fig. 2a). The two patients with the greatest decrease (R1 and R8) and

a

b

NATURE MEDICINE • VOLUME 6 • NUMBER 45 • MAY 2000

three (R6, R9, R20) of four (R6, R9, R20, R24) patients with complete loss of antibodies against HCV (Fig. 1c) who were tested for cell-mediated immune responses showed vigorous HCV-specific cellular responses (Fig. 2a). These results demonstrate that loss of antibodies against HCV was not accompanied by loss of cellular immune responses. As HCV-specific T cells have been detected in the blood of HCV-antibody-negative, HCV-RNA-negative blood donors (ref. 6) and in HCV-seronegative, healthy persons frequently exposed to HCV (refs. 7 and 8), our data indicate that these individuals may have recovered from a clinically inapparent HCV infection. Based on their HLA-A2 haplotype, we selected 18 patients for analysis of their CD8+ T-cell response, and used ten HLA-A2-restricted peptides with a sequence identical to that of the predominant source virus (HCV isolate AD78, AJ 32996, genotype 1b) to determine the frequency of CD8+ T cells by direct ex vivo cytokine enzyme-linked immunospot (ELISPOT) analysis. (Fig. 3a). Ten of eleven recovered patients (91%) mounted recall interferon (IFN)-γ responses of more than 10 specific spots per 3 × 105 PBMCs to at least one HCV peptide (Fig. 3a). In total, 42 of 106 IFN-γ ELISPOT assays (40%) were positive in the recovered patients, compared with 0 of 40 in the chronically infected patients (P < 0.001). Despite this readily detectable, antigen-specific recall response, however, the precursor frequency of HCV-peptide-specific T cells in the blood was low (maximum of 0.016% of all PBMCs). In contrast to IFN-γ, interleukin (IL)-5 was produced only in response to the control antigen tetanus toxoid, but not in response to any HCV peptides, indicating an HCV-specific Tc1 recall response (data not shown). We next investigated whether HCV-specific T cells could be

Fig. 2 CD4+ T-cell response. a, HCV-specific CD4+ proliferative T-cell response of 29 patients with resolved hepatitis (R1–R29) and 20 patients with chronic hepatitis (C1–C20). A stimulation index greater than 3 (hatched bars) is considered positive. *, For patient R25, the stimulation index against protein 1 (HCV core) is 44. Horizontal axes; HCV proteins 1–6: 1, HVC core; 2, NS3; 3, NS3–helicase; 4, NS4; 5, NS5A; 6, NS5B. In boxes, patient identification numbers b, Frequency of CD4+ proliferative T-cell responses to HCV proteins 1–6 (horizontal axes: 1, HCV core; 2, NS3; 3, NS3–helicase; 4, NS4; 5, NS5A; 6, NS5B) or tetanus toxoid (TT). 579


ARTICLES cally infected patients mounted CTL responses against peptides that were not tested and/or restricted by HLA alleles other than HLA-A2, three HLA-B7- or B35-restricted epitopes presented by autologous phytohemagglutinin-stimulated T-cell blasts were not recognized either (data not shown). To determine whether the CTL response detected long after clinical recovery was mediated by recently activated (DR+) cells, which may indicate frequent stimulation by replicating virus, we stimulated separate cultures of CD8+DR–, CD8+DR+ or total CD8+ T cells with peptide-pulsed, adherent antigen-presenting cells (Fig. 3c). We used the most frequently recognized peptide 5 to expand antigen-specific cells in each subfraction. After 3 weeks in vitro culture, the DR– T-cell population showed a high cytotoxicity similar to that of unseparated CD8+ T cells in all four patients analyzed, whereas the CTL reactivity of the DR+ population was either absent (patient R12) or weak (patient R2) (Fig. 3c). What is the immunological mechanism that maintains the cellular immune response? Whereas, CD4+ T-helper cells may become activated by residual viral antigens trapped and presented by follicular dendritic cells in the regional lymph nodes, the induction of cytotoxic T cells generally requires endogenous antigen synthesis in virus-infected cells. A vigorous, polyclonal and multi-specific CTL reTotal DR– DR+ Total DR– Total DR– Total DR– DR+ sponse was also maintained in the blood of patients who recovered from hepatitis B (ref. 9). These HBVFig. 3 CD8+ T-cell response. a, Direct ex vivo analysis (IFN-γ ELISPOT) of peripheral specific CTLs expressed activation markers indicablood T-cell responses of recovered patients (R1–R12) and chronically infected pative of recent stimulation by trace amounts HBV, tients (C1–C6). Data represent the number of antigen-specific IFN-γ spots (spots in the presence of antigen—spots in the absence of antigen). IL-5 spots were only seen detectable in PBMCs (ref. 9). In contrast, expansion of HCV-specific CTLs from DR– rather than DR+ prein response to tetanus toxoid, not in response to any HCV peptides (not shown). cursors indicates long-term memory after a remote , not tested. b, HCV-specific cytolytic activity of in vitro stimulated T-cell cultures of recovered patients (R1–R12) and chronically infected patients (C1–C6). The infection in the distant past. Small resting memory CTL response index represents the sum of the specific cytotoxicities of eight replicate cells with long intermitotic periods can persist in cultures per peptide; hatched bars, CTL response indexes more than 3 s.d. above the the absence of any specific or cross-reactive antigen mean of healthy, uninfected controls. Sera of all recovered patients and PBMCs of and are capable of rapid effector functions11–13. most recovered patients (except for patients R2 and R11) are negative for HCV RNA Consistent with those reports11–13, HCV RNA was not by nested RT–PCR. c, HCV-specific cytotoxic activity is mediated by cells expanded in detectable in the PBMC compartment of three of vitro from DR– T cells and is substantially higher than in cultures expanded from DR+ T four patients with a mainly DR– phenotype of CTLs cells. DR-depleted cell fractions contain less than 3.2% DR+ cells, DR-enriched cells + (Fig. 3c). In fact, HCV RNA was undetectable by contain 80.7% (patient R2) and 99% (patient R12) DR cells. Sera of all patients and PBMCs of all patients, except for patient R2, lacked HCV RNA by nested RT–PCR. nested PCR in the PBMCs of 10 of 12 recovered patients (82%) with a detectable CD8+ T-cell response (Fig. 3b). Although it is possible that low level viral expanded in vitro and whether they exerted peptide-specific cy- replication still occurs in the liver or other tissues of the recovtotoxic activity. The sum of the percent specific cytotoxicity of ered patients, these virological and immunological data indieight replicate microwell cultures stimulated with the same pep- cate that this may occur at a much lower level than after tide was designated the cytotoxic T-lymphocyte (CTL) response recovery from HBV. Furthermore, recovery from hepatitis B index for the peptide, as described6,9,10. CTL response index val- and C seems to be associated with differential regulation of not ues were considered substantial at more than 3 standard devia- only the cellular but also the humoral immune response: virustions above the mean CTL response index values in the specific antibodies persist for decades, maybe for life in pauninfected controls (Fig. 3b, hatched bars). Using these criteria, tients who recover from hepatitis B, but disappear gradually 11 of 12 recovered patients (92%) had a CTL response to at least after recovery from hepatitis C. Although the factors and mechanisms that induce, mediate one HCV epitope, and six patients had multi-specific CTL responses against two to four epitopes from different HCV pro- and regulate the immune response in the earliest phase after inteins (Fig. 3b). In total, 21 of 120 CTL assays (17.5%) were fection need to be studied in detail by prospective analyses of positive. In contrast, only one of six chronically infected pa- acutely infected patients, our data indicate that distinct cellular tients showed a substantial CTL response against one of these immune response patterns may contribute to a more precise dis10 peptides, corresponding to one of 60 (1.5%) positive CTL as- tinction between naive subjects and those who have recovered says. Although the possibility cannot be excluded that chroni- from HCV.

b

c

580

CTL response index

% specific cytotoxicity


Specific spots / 3 × 105 PBMCs

a

NATURE MEDICINE • VOLUME 6 • NUMBER 5 • MAY 2000


ARTICLES Methods


Patient population. Thirty-four chronically infected patients who had HCV RNA in their serum on repeated tests and 43 patients who seemed to have recovered, on the basis of lacking HCV RNA in at least three tests during the last 4 years of follow-up and having normal alanine aminotransferase levels, were chosen for this study based on their willingness to participate. Patient demographics and clinical spectrum of the acute initial course in 1978–1979 were representative of the larger cohort of 266 patients followed in Leipzig, Germany 14. None of the chronically infected patients had clinical evidence of cirrhosis at the time of this study. Liver biopsies done in 20 patients histologically documented the absence of cirrhosis. Serum alanine aminotransferase levels were slightly increased in chronically infected patients (that is, within three times the upper limits of normal) and serum HCV RNA levels ranged from 0.002 × 106 to 1.947 × 106 copies/ml. None of the patients had received antiviral therapy for hepatitis C. Fifteen healthy individuals without any history of HCV infection served as controls. All subjects gave informed consent for this study. HLA types were determined by complement dependent microcytotoxicity (One Lambda, Los Angeles, California) and molecular HLA-A2 subtyping (Dynal, Lake Success, New York). Antibody analysis. A commercially available third-generation enzyme immunoassay (AxSYM HCV Version 3; Abbott) was used to detect antibodies against an HCV core–NS3 fusion protein (amino acids 1–150 and 1,192–1,457) and nonstructural proteins (c200, amino acids 1,192–1,931; c100-3, amino acids 1,569-1,931; NS5, amino acids 2,054–2,995). Enzyme immunoassays with recombinant E1, E2 (INNOTEST HCV E1Ab and E2Ab prototype; Innogenetics, Gent, Belgium) and NS3 proteins (amino acids 1,188–1,465) were done as described15. A ‘signal-to-cut-off’ ratio greater than 1 (that is, greater than the mean + 3 standard deviations of ten negative controls) was considered positive. HCV core, E1, E2, NS4 and NS5 proteins were analyzed for antibody reactivity using a series of peptides 20–40 amino acids in length (genotype 1b). Peptides were sprayed on a membrane as separate lines, incubated overnight at room temperature with a dilution of 1:100 of the serum or plasma, followed by alkaline-phosphatase-conjugated antibody against human IgG. Reactivity was detected using NBT/BCIP (nitro blue tetrazolium/5-bromo-4-chloro-3-indolyl phosphate) as substrate16. Synthetic peptides and HCV antigens for T-cell analysis. Ten HLA-A2restricted peptides at the HCV amino-acid position of previously identified epitopes6,17, but with the predominant HCV sequence in the infectious source (AJ 32996, HCV isolate AD78) (genotype 1b), were used: 1, HCV core35, YLLPRRGPRL; 2, HCV core131, ADLMGYIPLV; 3, HCV core178, LLALLSCLTI; 4, HCV-E1-257, AIRRHVDLLV; 5, HCV-NS3-1073, CVNGVCWTV; 6, HCV-NS3-1169, LLCPSGHVV; 7, HCV-NS4B-1789, SLMAFTASV; 8, HCV-NS4B-1807, LLFNILGGWV; 9, HCV-NS4B-1851, ILAGYGAGV; 10, HCV-NS5B-2727, KLQDCTMFV. Recombinant HCV proteins were purchased from Mikrogen (Munich, Germany): 1, HCV core (amino acids 1–115); 2, NS3 (amino acids 1,007–1,534); 3, NS3–helicase (amino acids 1,207–1,488); 4, NS4 (amino acids 1,616–1,863); 5, NS5A (amino acids 2003–2,267); 6, NS5B (amino acids 2,600–2,868). Proliferation assay. Without prior in vitro expansion, replicates of 2 × 105 PBMCs per well of a 96-well flat-bottom plate (Costar, Cambridge, Massachusetts) were directly subjected to a 5-day proliferation assay in the presence of either 1 µg/ml HCV protein, 50 µg/ml tetanus toxoid (Behring, Marburg, Germany), phytohemagglutinin, medium alone or control buffer18. The stimulation index was calculated as the ratio of the average number of counts per minute of five replicate cultures in the presence of antigen as compared to control buffer or medium. A stimulation index greater than 3, which was higher than the average stimulation index + 3 s.d. in 15 healthy, uninfected controls, was considered positive. IFN-γ and IL-5 elispot assays. Ninety-six-well plates (Millititer; Millipore/ELISPOT, Bedford, Massachusetts) were coated overnight at 4 °C with the primary antibody against human IFN-γ or IL-5 (Endogen, Woburn, Massachusetts), washed four times with sterile phosphatebuffered saline and blocked for 1 h at 25 °C with Roswell Park Memorial Institute (RPMI) medium and 1% bovine serum albumin (Sigma). NATURE MEDICINE • VOLUME 6 • NUMBER 45 • MAY 2000

Cryopreserved PBMCs (3 × 105) were thawed and cultured in RPMI 1640, 5% AB serum and 2 mM L-glutamine with individual HCV peptides at a concentration of 10 µg/ml. After 30 h for IFN-γ elispot assays and 48 h for IL-5 elispot assays, the plates were washed seven times and incubated overnight with 100 µl biotin-conjugated secondary antibody against IFN-γ (Endogen, Woburn, Massachusetts) or IL-5 (PharMingen, San Diego, California). After being washed four times and incubated 2 h with streptavidin–alkaline phosphatase (1:2,000 dilution; Dako, Carpinteria, California), the plates were washed again four times with phosphate-buffered saline and developed with freshly prepared NBT/BCIP solution (nitro blue tetrazolium/5-bromo4-chloro-3-indolyl phosphate; BioRad, Richmond, California). The reaction was stopped by rinsing with distilled water and the spots were counted manually. Cytotoxicity analysis of in vitro expanded HCV peptide-specific T cells. PBMCs were separated on Ficoll–Histopaque gradients, and eight replicate cultures of 4 × 105 cells per well were stimulated in a 96-well plate with 10 µg/ml individual HCV peptides in RPMI, 10% AB serum, 2 mM L-glutamine, 100 U/ml penicillin and 100 µg /ml streptomycin. Recombinant human IL2 (20 U/ml; Chiron, Emeryville, California) was added every 3 d, and cultures were re-stimulated with 1 × 105 irradiated autologous PBMCs and 10 µg/ml peptide every 7 d and were tested for cytolytic activity on days 20–24 (refs. 10,17). HCV-specific cytotoxicity was determined by a standard 4hour split-well 51Cr-release assay against 3 × 103 Epstein-Barr-virus-transformed B cells (JY-EBV, HLA-A2.1, B7, Cw7) pulsed overnight with 10 µg/ml synthetic peptide10,17. The sum of the percent cytotoxicity of eight replicate cultures stimulated with the same peptide was designated the CTL response index for that peptide, as described9,10,17. CTL response index values more than 3 s.d. above the mean in HLA-A2+ controls were considered positive. Cytotoxicity analysis of HCV-peptide-specific T cells expanded from DR+ and DR– T-cell subpopulations. To assess the HCV-specific CTL activity of activated (DR+) and non-activated (DR–) T-cell subsets, CD8+ T cells were purified from PBMCs with anti-CD8 magnetic beads (Dynal, Lake Success, New York), stained with phycoerythrin-labeled antibody against DR (Becton Dickinson, San Jose, California) and sorted into DR+ and DR– subpopulations on a Coulter flow cytometer. Purity was confirmed by analysis after sorting. DR-enriched, DR-depleted or unfractionated CD8+ T cells (5 × 104) were then stimulated with 1 × 105 irradiated (3,000 rad) autologous macrophages/monocytes that had been isolated from PBMCs by plastic adherence, 10 µg/ml peptide 5 and 10 ng/ml IL-7 (PeproTech, Rocky Hill, New York). Cells were re-stimulated twice at 7day intervals with irradiated adherent cells, 10 ng/ml IL-7 and 10 µg/ml peptide 5 and assayed for cytotoxic activity after 20–24 d of culture1,17. Detection and quantification of HCV RNA. Total RNA was extracted from 100 µl sera using Trizol LS Reagent (Life Technologies) and also from 2.5 × 106 PBMCs of patients R1–R12 to correlate the presence or absence of HCV RNA with the activation state of CTLs (Fig. 3b). RT–PCR was done as described19. PCR products were separated by 1.5% agarose gel electrophoresis and detected by staining with ethidium bromide. Serum HCV RNA was quantified with the Amplicor HCV Monitor Test (Hoffmann LaRoche, Grenzach-Wyhlen, Germany). Statistical analysis. A standard chi-square test was used to assess differences in the frequency of humoral or cellular immune responses in both patient groups (Figs. 1a and b, 2a and 3a and b). Differences with a P value of