Detection of active hepatitis C virus and hepatitis G virus/GB virus C replication in bone marrow in human subjects

Hepatitis C virus (HCV) and the recently described hepatitis G virus (HGV), known also as hepatitis GB virus C, are similarly organized positive-strand enveloped viruses. As the nomenclature of the latter agent has not been decided yet, for the purpose of this article it will be referred to as “HGV.” Genomes of both viruses contain 5′ and 3′ untranslated regions, flanking a single open reading frame encoding structural and nonstructural proteins at its 5′ and 3′ ends, respectively.1 2 The translated single long polyprotein is subsequently cleaved by viral and host enzymes into a number of structural and nonstructural proteins. It is assumed that both HCV and HGV replicate through the RNA negative strand, the presence of which could be regarded as a direct evidence of viral replication.

Whereas there is little doubt that HCV replicates primarily in the liver, the presence of extrahepatic replication sites remains controversial. In support of such a possibility come studies reporting the relatively common detection of HCV RNA negative strand in peripheral blood mononuclear cells (PBMCs).3,4 This evidence, however, has been questioned, as commonly used techniques are limited in their ability to discriminate between positive and negative RNA strands.5 Importantly, in several studies that used assays carefully optimized for strand specificity, HCV RNA negative strand was not detected in PBMCs from infected patients.6-8Similarly, although the presence of active replication in bone marrow (BM) was suggested by in situ detection of viral RNA and viral antigens,9 it was not confirmed by an investigation using strand-specific assay.10 However, the latter study was relatively small, as it included only 6 patients.

HGV does not seem to be a primary hepatotropic virus. Although originally associated with hepatitis, it is currently clear that, in the absence of concomitant infection with another hepatotropic virus, there is usually no liver injury and viral replicative forms were not detected in the liver.8,11 The major replication sites of HGV are not well defined; however, we have recently found viral replicative forms in BM, spleen, and lymph nodes.12,13 In support of the likely role of hematopoietic cells in supporting HGV replication comes a recent report on successful infection of human PBMCs in vitro.14 

Nevertheless, strand-specific assays were not applied previously to the study of HCV and HGV replication in BM in a larger group of subjects, and only those found positive for markers of infection in serum were included in past studies. Here we report on the results of our study analyzing the presence of HGV and HCV viral replicative intermediaries in BM from 48 patients of a hematologic outpatient clinic. The viral negative strand RNAs were detected using highly strand-specific Tth-based reverse transcriptase-polymerase chain reaction (RT-PCR).

The subjects were selected from consecutive patients seen between May and July 1998 by the same consulting hematologist (E.K.) at an outpatient hematology clinic in a large industrial city in central Poland. The inclusion criteria were as follows: availability of BM sample (BM aspiration biopsy was performed as part of clinical investigation), age older than or equal to 18 years, no serologic evidence of HIV-1 infection, and no antiviral therapy at the time of the study. Forty-eight patients (28 men, 20 women, mean age 56.8 ± 11.7 years) satisfied the above criteria and were included in the study. Although many subjects have had chronic liver disease (Table 1), they were not preselected with regard to markers of hepatitis. The high representation of the latter patients among studied patients was due to the fact the hematology clinic was serving a large liver disease clinic. The study protocol adhered to local Institutional Review Board (IRB) requirements.

BM samples were washed once in phosphate-buffered saline (PBS) to lower contamination by serum and stored at −80°C until analysis. RNA was extracted from serum and BM samples by means of a modified guanidinium thiocyanate-phenol/chloroform technique using a commercially available kit (RNAzol, Gibco/BRL, Rockville, MD). One microgram of total RNA, (as determined by spectrophotometry) was routinely used for RT-PCR. In the case of serum, the amount of extracted RNA loaded into the reaction corresponded to 20 μL. However, all initially negative serum samples were retested using RNA extracted from 100 μL of serum.

Strand specificity of our RT-PCR for the detection of HCV and HGV negative RNA strands was ascertained by conducting complementary DNA (cDNA) synthesis at a high temperature using the thermostable enzyme, Tth. The sensitivity and strand specificity of these reactions was established using synthetic RNA as templates. A detailed description of both strand-specific assays and sequence of used primers was published previously.7 11 In brief, the cDNA was generated in 20 μL of reaction mixture containing 50 pmol/L of sense primer, 1 × RT buffer (Perkin Elmer), 1 mmol/L MnCl₂, 200 μmol/L (each) dNTP, and 5 units Tth (Perkin Elmer). After 20 minutes at 65°C, Mn²⁺ were chelated with 8 μL of 10 × EGTA chelating buffer (Perkin Elmer), 50 pmol/L of antisense primer was added and the volume was adjusted to 100 μL, and the MgCl₂ concentration was adjusted to 2.2 mmol/L. The amplification was performed in Perkin Elmer GenAmp PCR System 9600 thermocycler as follows: initial denaturing for 1 minute at 94°C, 50 cycles of 94°C for 15 seconds, 58°C for 30 seconds and 72°C for 30 seconds, followed by a final extension at 72°C for 7 minutes. Twenty microliters of the final product were analyzed by agarose gel electrophoresis and Southern hybridization with a³²P-labeled internal oligoprobe.

The strand-specific assays were capable of detecting approximately 100 genomic equivalent (eq) molecules of the correct strand, while unspecifically detecting more than or equal to 10⁷ to 10⁸ genomic eq of the incorrect strand. The addition of 1 μg of total cellular RNA extracted from human tissues would lower the the sensitivity of the reaction by no more than 1 log, whereas the specificity of the assays was not affected.7 11 Thus, both strand-specific assays were capable of detecting approximately 10³ viral genomic eq in 1 μg of RNA. In case of serum, the approximate detection limit was 10³ eq per 1 mL.

MMLV RT-based detection of HCV and HGV has been described in detail elsewhere.7 11 These assays were capable of detecting approximately 10 genomic eq of the correct synthetic template but were not strand specific. Similary to Tth-based assay, the addition of cellular RNA would slightly lower the sensitivity by up to 1 log. The established detection limit was approximately 100 genomic eq per micrograms of total RNA or 1 mL of serum.

When deemed necessary, titers were determined by analyzing 10-fold serial dilutions of the RNA template. Appropriate measures, described elsewhere,11 were used to prevent and detect contamination. All RT-PCR runs included positive controls consisting of end-point dilutions of respective RNA strands and negative controls included normal BM, PBMCs, and normal sera. In addition, to account for RNA extraction efficiency, synthetic template was routinely “spiked” into the negative control samples. The presence of anti-HCV and E2 antibody to the HGV envelope protein was determined by commercially available tests (UBI HCV EIA, United Biochemical and Quantinine Human GBV-C EIA, R&D Systems, Minneapolis, MN). The infecting HCV genotypes were determined as described previously.15 

The distribution of viral infection markers in different groups of patients is shown in Table 1. When serum samples were analyzed, 18 (38%) patients were found to be HCV RNA positive and 6 (13%) were found to be HGV RNA positive. Three patients were positive for both viral sequences. Anti-HCV was demonstrated in 24 (50%) patients and anti-HGV was found in 19 (40%) patients. However, although all patients who were HCV RNA positive were also anti-HCV positive, only 1 patient who was HGV RNA positive had respective antibodies detectable in serum. Thus, the overall prevalence of infection markers in serum was 50% for both viruses.

When BM samples were analyzed, the presence of HCV RNA was detected in 15 (31%) patients, and the presence of HGV RNA was detected in 9 (19%) patients. The concordance between the presence of viral sequences in serum and BM was not absolute. In 3 patients, HGV RNA was detectable in the BM but not in the serum; 2 of these patients were also anti-HGV negative and the presence of viral sequence in BM was the only evidence of infection. In 3 patients, HCV sequences could be demonstrated in serum but not in BM, and in 1 patient, HCV RNA was detectable in BM but not in serum. However, the latter patient, in whom the condition was diagnosed as chronic hepatitis C, was anti-HCV positive.

Altogether, the presence of HGV RNA and HCV RNA, either in serum or BM, was demonstrated in 9 (19%) and 19 (40%) patients, respectively, and at least 1 marker of infection was present in 26 (48%) and 24 (50%) patients, respectively.

Using the Tth-based strand-specific assay, the presence of negative strand HCV RNA and HGV RNA was detected in BM from 4 and 5 patients, respectively, whereas all serum samples were negative (Table 1). These results were confirmed in 2 independent experiments using 2 separate extraction procedures. To exclude the possibility that these results represent false-positives because of the presence of high titer positive strand RNA, the viral titer was estimated by serial dilutions and was found to range from 10³ to 10⁵ genomic eq per microgram of total RNA, thus remaining well within the specificity limits of our strand specific assay. The patients with evidence of HCV and HGV replication in BM were not significantly different from the rest of the infected patients with respect to age, gender, or history of blood transfusion; their infecting HCV genotypes were either 1b or 1a, which was also not different from the rest of the patients. Some clinical and virologic data on these patients are presented in Table 2.

In this study, the presence of HCV RNA negative strand in BM was demonstrated in 4 of 24 (17%) patients who were HCV positive and the titer of positive strand was 1 to 2 logs higher than the titer of the negative strands, which is a proportion similar to that expected for a flavivirus at its replication site. However, the actual prevalence of active infection could be higher, as the replication in other subjects could have been below the sensitivity limit of our strand-specific assay. Extrahepatic HCV replication is more likely to be detected in the presence of immunosuppression: The presence of HCV RNA negative strand was found in lymph nodes and BM from HIV-coinfected subjects,15 as well as in hematopoietic cells derived from patients who were HCV positive and transplanted into the severe immunodeficiency mouse.16 However, even in these studies, the presence of viral negative strand was by no means universal and the titers were very low. This contrasts with the reported common detection of HCV RNA and viral antigens in BM and PBMCs by in situ techniques.9,17,18 Because the strand-specific assays are relatively sensitive, our RT-PCR was capable of detecting approximately 10³ viral genomic eq in 1 μg of total RNA, this suggests a very low rate of replication or rapid degradation of negative strand RNA. HCV invasion of hematopoietic cells may not be benign; HCV infection was associated with such lymphoproliferative disorders as non-Hodgkin's lymphoma and cryoglobulinemia.19 Although the effect of HGV on BM is unclear, one study reported a possible link between HGV infection and low-grade non-Hodgkin's lymphoma.20 

Interestingly, in our previous study, which used the same Tth-based assay, we did not detect HCV replication in PBMCs from HIV-negative subjects.7 Similarly, other studies6,8 failed to demonstrate the presence of HCV RNA negative strands in PBMCs when using highly strand-specific RT-PCR assays. One possible explanation for this discrepancy is that HCV may be particularly apt at infecting CD34⁺ hematopoietic progenitor cells.18 It is also possible that HCV replication becomes more efficient in proliferating cells. In support of this concept, our recent observations are that PBMCs from subjects with chronic hepatitis C occasionally become positive for the presence of HCV RNA negative strand after stimulation with phytohemagglutin (PHA) mitogen.21 

HCV circulates as a number of closely related but not identical genomes, referred to as quasi-species, and the presence of this dynamic mutant reservoir could facilitate viral adaptation to replication in various secondary cells. HGV does not seem to be as variable,13 and BM cells may represent the primary, not the secondary, site of replication. BM replication may not be that unusual for flaviviridae, for example, the hog cholera virus was found in megakaryocytes,22 whereas dengue virus was shown to replicate efficiently in BM progenitors and hematopoietic cell lines.23 Interestingly, we identified 3 patients in whom HGV RNA was detectable in the BM but not in the serum; 2 of them were anti-HGV negative in serum. The latter 2 patients were probably in the early phase of the infection when antibodies have not yet developed.24 Although it supports the notion of HGV replication in BM, it also demonstrates that studying HGV RNA and specific antibodies in serum may underestimate the true prevalence of the infection. However, this may also be occasionally true for HCV infection as one of the patients studied was HCV RNA positive in BM but not in serum. This could probably be explained by the fact that viral levels in serum can fluctuate during chronic infection, becoming occasionally undetectable by RT-PCR.25 Whether this was the case is unclear, as follow-up serum samples were unavailable for analysis.

In summary, we have described the presence of active HCV and HGV replication in BM from human subjects. For HGV, the presence of viral RNA in BM was occasionally the only evidence of infection.