|Home | About | Journals | Submit | Contact Us | Français|
The aim of this study was to investigate the quasispecies heterogeneity of hepatitis C virus (HCV) in the plasma, cryoprecipitate, and peripheral lymphocytes of chronically infected HCV patients with mixed cryoglobulinemia (MC). We studied 360 clones from 10 HCV-positive patients with MC and 8 age-, gender- and HCV genotype-matched subjects with chronic HCV infection but without MC. A partial nucleotide sequence encompassing the E1/E2 region, including hypervariable region 1 (HVR1), was amplified and cloned from plasma, cryoprecipitates, and peripheral blood mononuclear cells (PBMC), and the genetic diversity and complexity and synonymous and nonsynonymous substitution rates were determined. Heterogeneous selection pressure at codon sites was evaluated. Compartmentalization was estimated by phylogenetic and phenetic (Mantel's test) approaches. The patients with MC had 3.3 times lower nonsynonymous substitution rates (1.7 versus 5.7 substitutions/100 sites). Among the subjects with HCV genotype 1, the MC patients had significantly less complexity than the controls, whereas the diversity and complexity were similar in the genotype 2 patients and controls. Site-specific selection analysis confirmed the low frequency of MC patients showing positive selection. There was a significant correlation between positive selection and the infecting HCV genotype. The quasispecies were less heterogeneous in PBMC than in plasma. Significant compartmentalization of HCV quasispecies was observed in the PBMC of four of nine subjects (three with MC) and seven of nine cryoprecipitates. In one subject with MC, we detected a 5-amino-acid insertion at codons 385 to 389 of HVR1. Our results suggest reduced quasispecies heterogeneity in MC patients that is related to a low selection pressure which is probably due to an impaired immune response, the HCV genotype, and/or the duration of the infection. The frequent HCV quasispecies compartmentalization in patients' PBMC suggests a possible pathogenetic significance.
Hepatitis C virus (HCV) is thought to be the causative agent of mixed cryoglobulinemia (MC) (2, 20), a systemic vasculitis caused by cold-precipitable serum proteins and clinically characterized by a classical triad of symptoms (purpura, asthenia, and arthralgia) (33), which has been suggested to be a low-level malignant B-cell lymphoproliferative disorder (35). MC patients have an increased risk of developing non-Hodgkin lymphoma (NHL) (36, 51), and it has been suggested that HCV infection itself may be associated with an increased risk of NHL (19, 48, 63). The trigger mechanism of MC and MC-associated lymphomas may be the HCV antigen-driven proliferation of particular lymphocyte clones (36), but the more recent identification of HCV sequences in lymphoid cells (particularly B cells) has also suggested the possibility of direct infection of the cells involved in the lymphoproliferative process leading to MC and possibly NHL (21, 29, 30, 60, 62).
Like other RNA viruses, HCV is characterized by a high degree of genetic heterogeneity (40, 41). In particular, a domain of 27 amino acids in the N terminus of the E2 gene (hypervariable region 1 [HVR1]) is the most heterogeneous region of the entire HCV genome. As in the case of the other RNA viruses fulfilling the predictions of Eigen's theory on the evolution of prebiotic RNA elements (13, 24), the word quasispecies has been adopted to describe its high propensity for variation.
It has been shown that HVR1 is involved in binding the putative cell receptor of HCV and is one of the main targets of anti-HCV neutralizing antibodies (47, 61). Mutations in this region are therefore potentially important for viral persistence, since they affect cell tropism and viral escape from immune defenses (31).
Studies of quasispecies and the natural history of hepatitis C suggest a correlation between the degree of viral diversity and the clinical course of the disease (16, 31). Nevertheless, little information is available concerning role of HCV quasispecies in the pathogenesis of HCV-associated extrahepatic conditions. Gerotto et al. (23) have recently reported the presence of a single amino acid insertion in the HVR1 of some type II cryoglobulinemic patients, thus raising questions about the existence of specific cryoglobulinemia-associated mutations in HCV genome.
The aim of the present study was to investigate the genetic heterogeneity of HCV in the plasma, cryoprecipitate, and peripheral lymphocytes of patients with chronic HCV infection, with or without MC.
This study included 10 HCV-positive patients (eight females and two males; median age, 67.5 years [range, 51 to 78]) with symptomatic MC, and 8 age- and gender-matched patients with chronic (CH) type C hepatitis (six females and two males; median age, 65 years [range, 52 to 72]) without any signs or symptoms of MC. MC was diagnosed on the basis of the reported manifestations of Meltzer and Franklin's triad (purpura, asthenia, and arthralgia) (33), and the repeated demonstration of a cryocrit level of ≥2% (including the determination made immediately before the study). The other criteria for inclusion in the cryoglobulinemic group were the presence of HCV-RNA in plasma, a first diagnosis of HCV infection more than 1 year before study entry, a liver biopsy no more than 2 years before the start of the study, and no history of interferon therapy. The exclusion criterion was steroid therapy in the 3 months preceding the study.
The criteria for inclusion in the control group were repeated nondetectability of serum cryoglobulin by the standard method (see below), no history of purpuric manifestations or serious arthralgia, no history of interferon therapy, a liver biopsy no more than 1 year before study entry, and an HCV geno/subtype similar to that of the MC patients.
Total RNA was extracted from 200 μl of plasma/concentrated cryoglobulins or 3 × 106 peripheral blood mononuclear cells (PBMC) by using a commercially available kit based on affinity column purification (High-Pure Viral RNA Kit and High-Pure RNA isolation kit; Roche Diagnostics, Mannheim, Germany).
The plasma of the cryoglobulinemic subjects was incubated at 37°C for 30 min. before RNA extraction. Concentrated cryoglobulins were prepared by incubating cryoglobulinemic plasma for up to 3 days at 4°C, followed by centrifugation at 7,000 × g for 20 min at 4°C. The precipitates were then washed three times with 1 ml of phosphate-buffered saline at 4°C, resuspended in 1 ml of phosphate-buffered saline for 30 min at 37°C, and then immediately subjected to RNA extraction.
HCV-RNA was detected by means of reverse transcription (RT) nested-PCR for the amplification of 5′ noncoding region (NCR) sequences, and its quantity in plasma and cells was estimated by limiting dilution PCR as described elsewhere (60). The HCV-RNA titers were expressed as the highest dilution giving a positive result in 1 ml of plasma or 105 cells (i.e., PCR units [PU]).
The HCV quasispecies were molecularly characterized by amplifying a 351-nucleotide sequence encompassing the E1/E2 region (including HVR1). RT nested-PCR was performed by using two pairs of primers recognizing the main HCV genotypes circulating in Italy: outer primers 5′-GGDCAYCGMATGGCNTGGGA-3′ (positions 1284 to 1303) and 5′-GGNGSRTARTGCCAGCARTANGG-3′ (positions 1813 to 1791) and inner primers 5′-GCTTGGGATATGATGATGAACTGGTC-3′ (positions 1296 to 1321) and 5′GGTGTGGAGGGAGTCATTGCAGTT3′ (positions 1646 to 1623).
The sequences encompassing the E1/E2 obtained by using RT nested-PCR were inserted in the plasmid vector pGEM (pGEM-T Easy Vector System II; Promega Corp., Madison, WI) and transfected into Escherichia coli JM109 competent cells. After overnight incubation at 37°C in agar medium, insertion was checked by PCR with HCV envelope inner primers on white colonies. A mean of 10 HCV-positive clones for each sample were then bidirectionally analyzed by means of automated sequencing (ABI Prism 3100 Genetic Analyzer; Applied Biosystems Division, Foster City, CA) in the presence of the specific inner primers described above, and the ABI Prism BigDye terminator cycle sequencing ready reaction kit (Applied Biosystems).
Each sample was phylogenetically analyzed by using a 228-nucleotide sequence (positions 1359 to 1586) (10) encompassing the E1/E2 regions including HVR1 (nucleotides 1491 to 1571).
A total of 360 DNA sequences from the cryoglobulinemic patients and controls were aligned with nine HCV sequences retrieved from the GenBank database and representing the HCV subtypes commonly observed in Italy. The reference sequences (with accession numbers) were HCV1a (M62321), HCV1b (D10074), HCV2a (D00944), HCV2b (D10988), HCV2c (AF142392 and D31972), HCV2e/f (D49757), HCV3a (D14311), and HCV4a (Y11604) (9, 11, 42-44, 54, 56).
Phylogeny was reconstructed by using PHYLIP version 3.5 (18) and Molecular Evolutionary Genetics Analysis (MEGA) version 2.1 software (27). The distance matrices were calculated by using Kimura's two-parameter model (25) and an empirical transition/transversion ratio of 2. Distances are expressed in terms of substitutions per site (or 100 sites in the text). The unrooted phylogenetic trees were constructed by means of the neighbor-joining algorithm. The obtained topology was statistically evaluated by means of the bootstrap method (17) with 1,000 resampling replications. The internal branches with bootstrap values of ≥65% were considered significant. The trees were displayed by using the Treeview program (45).
Synonymous (dS) and nonsynonymous (dN) distances were estimated for the HVR1 sequences alone by using the Nei and Gojobori method (39) with the Jukas-Cantor correction.
Selection usually affects only a few amino acids, and so the mean dN/dS ratio (ω) over an entire coding sequence can only rarely be significantly >1. For this reason, we used three models allowing for heterogeneous dN/dS ratios among sites implemented in the CODEML program included in the PAML package (58): M0 (one ratio), which assumes one dN/dS ratio for all sites; M1 (neutral), which assumes proportions of conserved and neutral sites (ω = 0 or ω = 1); and M2 (selection), which adds a further class of sites under positive selection (ω > 1). The performances of the models were compared by using a likelihood-ratio test that allows comparison of the likelihood of nested models using a chi-squared distribution with the degree of freedom equal to the difference in the number of parameters between the two models. A F3x4 codon frequency model was assumed, and the transition/transversion rate (k) was estimated by the program. An empirical Bayesian approach made it possible to identify specific sites under positive selection with a posterior probability of >90% (significant) or >95% (highly significant).
The compartmentalization of the HCV quasispecies was evaluated by using phylogenetic and phenetic approaches.
The phylogenetic analysis concentrated on the E1/E2 region was aimed at evaluating the evolutionary relationship between HCV quasispecies on the basis of the compartmental origin of the clones. The significance of the obtained topology was evaluated by means of bootstrap (see above) and maximum-likelihood analyses.
Likelihood mapping is a quartet puzzling method based on the maximum-likelihood approach designed to investigate a priori the phylogenetic information contained in a sequence alignment (without computing an overall tree). The likelihood mapping program is implemented in TREE-PUZZLE (53). The method is based on calculating the likelihood of all of the possible fully resolved tree topologies for each analyzed quartet (groups of four randomly chosen sequences). Each quartet has three possible topologies: the likelihood of each is estimated by using the maximum-likelihood approach, and the likelihood values are represented as a dot inside an equilateral triangle. Each corner of the triangle represents one of the three fully resolved trees; the center of the triangle being a completely unresolved tree topology (star-like evolution) and the sides being two equally possible topologies. The percentage of dots in each area is a measure of the probability of the topologies: dots falling in a corner support the corresponding topology, whereas a high percentage of dots in the center indicate an unresolved topology (all of the topologies are equally possible). When analyzing more than four sequences, it is possible to group them in four different subsets (a, b, c, and d), and the method indicates the most likely topology among them.
The likelihood mapping analysis was made by grouping the sequences on the basis of their compartment of origin (a = plasma, b = PBMC/cryoglobulins). When suggested by the results of the distance phylogenetic analysis, we divided the cryoprecipitated clones into two further subsets on the basis of the distance between them and plasma clones (b = more closely related, c = more distantly related).
The mathematical models and parameters for the maximum-likelihood analysis were estimated by using Modeltest 3 (version 3.06) (49) and PAUP* 4.0 (D. L. Swofford, Sinauer Associates, Inc., Sunderland, MA).
Molecular phylogenesis reconstructs ancestral relationships among sequences, whereas phenetic analysis statistically determines their degree of genetic similarity. We used Mantel's test (a permutation analysis for the comparison of two distance matrices) to determine on a robust statistical basis whether the sequences detected in one body compartment are more similar to each other than to the sequences in other compartments by comparing the intra- and intercompartment Kimura distances between E1/E2 sequences (1, 50).
Briefly, the Kimura two-parameter distance matrix was compared to the compartment distribution matrix (Mc), which had the same dimensions as the former and was obtained by replacing Mc(i, j) with 0 (if sequence i is from the same compartment as sequence j) or 1. The correlation coefficient was computed, including all pairs except the diagonal of both matrices, and up to 100,000 permutations of rows and columns were made. The proportion of times the original correlation coefficient was exceeded by the permutated values was the exact significance (P) of the correlation observed. Mantel's test was performed by using “zt” software for the simple Mantel's test (7), a freely available program (http://www.psb.rug.ac.be/~erbon/mantel/) that facilitates the analysis of matrix population models and the simulation of stochastic processes.
The distribution of the quantitative variables (such as complexity or distances) was described by use of median values and ranges. The comparisons of patients and controls, subjects infected by HCV genotype 1 or 2, and subjects with a high or low viral loads were made by using analysis of variance (ANOVA) or the Mann-Whitney nonparametric test for independent samples; the comparisons of different body compartments were made by using Wilcoxon's nonparametric test for paired samples. P values of <0.05 were considered significant.
All of the analyses were performed by using the SPSS package (SPSS v. 11; SPSS, Inc., Chicago, IL).
The sequences were submitted to GenBank under accession numbers DQ067633 to DQ067825.
The clinical, laboratory and virological characteristics of the patients and controls are shown in Table Table1.1. Seven MC patients had type II cryoglobulins, and three had type III.
All of the patients had chronic type C hepatitis with different degrees of liver involvement; median serum alanine aminotransferase (ALT) levels were not significantly different in the two groups (MC patients, 46.5 IU/liter [range, 31 to 81]; controls, 53 IU/liter [range, 37 to 65]; P = 0.8 [Mann-Whitney test]).
All of the MC patients were HCV 5′-NCR PCR positive in all of the analyzed compartments (plasma, PBMC, and cryoprecipitate), whereas one of the controls did not have any detectable sequences in his PBMC (subject 11).
Plasma HCV-RNA levels (measured by means of limiting dilution PCR of the 5′-untranslated region) were similar in the patients and controls, with median log10 titers of 3.5 (range, 2 to 6) versus 4 (range, 3 to 6) (P = 0.6 [Mann-Whitney test]) (Table (Table1).1). There was no significant difference in HCV-RNA loads between patient plasma and cryoprecipitate samples (median log10 cryoprecipitate titer, 3.0; range, 3 to 4). The PBMC of the MC patients and controls showed small amounts of HCV-RNA (median titer, 7.3 PU/105 cells; range, 0.3 to 150).
E1/E2 sequences were detected in the plasma samples of all of the subjects, the PBMC of five MC patients (patients 2, 4, 5, 6, and 8) and four controls (subjects 13, 15, 16, and 17), and the cryoprecipitates of nine MC patients (all except patient 4) (Table (Table1).1). A total of 360 clones were analyzed: 180 from plasma samples, 90 from PBMC, and 90 from patient cryoprecipitates.
In order to evaluate the accuracy of the quasispecies measure based on the analysis of 10 clones per sample, the investigation was extended to 20 clones in plasma of four randomly selected cases (patients 1, 4, 5, and 6). No significant differences in terms of complexity, master sequence, genetic diversity, or synonymous or nonsynonymous substitution rates were observed when the results were compared to those of the 10-clone analysis (data not shown).
Patient 6 had a 15-nucleotide insertion at codon 385 that corresponded to the second residue of the HVR1 (Fig. (Fig.1).1). A BLAST search of the primary sequence databases did not retrieve any homologous insertions in the HCV sequences recorded in GenBank (as of February 2005). The insertion was present with only minimal changes in the clones obtained from all three body compartments (plasma, cryoprecipitate, and PBMC). Alignment of the predicted amino acid structure showed that the insertion mainly consisted of polar amino acids (S, S/A, H/R, V, and Q/H), thus increasing the hydrophilicity of the N terminus of the HVR1.
Figure Figure22 shows the unrooted phylogenetic neighbor-joining tree of the E1/E2 region obtained by including the clones and the prototype E1/E2 sequences of different HCV genotypes. The clones obtained from 10 subjects (seven patients and three controls) clustered with reference sequences of type 2c HCV, whereas the clones of control 13 were clearly separate from those of all of the other subjects, forming a highly significant cluster (bootstrap value, 100%) with an HCV-2e/f isolate of Jakarta. The remaining subjects clustered with prototype 1b. The clustering of the clones obtained from each individual was significant (between-sample bootstrap values were always ≥75).
The degree of heterogeneity of the HCV quasispecies was measured in terms of genetic complexity (mutant frequency) and diversity (genetic distances) and compared to the patient characteristics and organic compartment.
The HCV quasispecies in the plasma samples had a median complexity of 0.65 (range, 0.3 to 1.0) and a diversity of 2.2 (range, 0.4 to 13.4) substitutions/100 sites in the whole E1/E2 region (Table (Table2).2). Most (57%) of the complexity was due to HVR1 variability (median complexity, 0.4 [range, 0.2 to 0.9]), and the divergence was 1.6 times higher than in the whole E1/E2 region (median diversity, 3.6 [range, 0.3 to 32.0] substitutions/100 sites). The MC patients showed a median genetic divergence between plasma HVR1 sequences of 3.3 substitutions/100 sites (range, 0.3 to 16.8) and a mutant frequency of 0.35 (range, 0.2 to 0.9), whereas the controls showed a median diversity that was 1.5 times as much (4.8 substitutions/100 sites [range, 1.0 to 32.0]) and a median mutant frequency of 0.45 (range, 0.3 to 0.6). These differences were not statistically significant. In terms of synonymous and nonsynonymous substitution rates, the MC patients had a 3.3-times lower median dN than the controls (1.7 versus 5.7 substitutions/100 sites), and the dS was similar in the two groups (6.3 versus 5.8 substitutions/100 sites [Table [Table22]).
Table Table33 shows the median values (diversity, complexity, and dN and dS) of patients and controls after stratification by the infecting genotype. The MC patients with type 1 HCV showed significantly less quasispecies complexity in both the E1/E2 and the HVR1 regions than the controls with the same genotype (P = 0.041 and P = 0.004 [ANOVA]). There were no significant differences in quasispecies diversity and complexity between the MC patients and controls with HCV genotype 2.
Interestingly, we observed an 11.4-fold-greater increase in dN in the controls with genotype 2 than in those with genotype 1 (12.5 versus 1.1 [Table [Table3]),3]), whereas the MC patients with genotype 2 showed a smaller increase in dN than those with genotype 1 (4.7 versus 1.1) and a greater increase in dS (11.5 versus 1). More generally, the median distance of HVR1 was significantly less in all of the subjects (MC plus CH) infected by HCV genotype 1 than in those infected by genotype 2 (1.4 versus 7.7; P = 0.037 [Mann-Whitney test]) due to a threefold reduction in dS and an eightfold reduction in dN (Table (Table33).
No significant differences in the median complexity and diversity of the HVR1 or E1/E2 sequences were found after the study population was stratified on the basis of HCV-1 RNA levels (data not shown).
The analysis confirmed a low level of selection both in patients and controls. The median ω was 0.24 (range, 0.0001 to 11.42) among the MC patients and 0.59 (range, 0.064 to 216.7) in the controls: only two of ten patients (20%) and four of eight controls (50%) had an ω of ≥1.
The selection model (M2) was significantly favored over M1 or M0 in three of the eight controls (37.5%) and two of the ten MC patients (20%). In particular, among the subjects infected with genotype 1 HCV, one of four controls (25%) and none of the MC patients showed signs of positive selection; the corresponding figures among the type 2-infected subjects were two of four (50%) and two of seven (28.6%).
The proportion of sites under selection varied from 8 to 24%, and the number of sites with a >90% posterior probability varied from 2 to 13, without any significant differences between the two patient groups. Figure Figure33 shows the selected sites that were identified. The vast majority (20 of 22 [91%]) were included in the HVR1. The greatest pressure was found in the central part of HVR1 (amino acids 392 to 405) with two peaks at 394 and 401. Patient 6 also had sites under selection in the insertion (amino acids 386S, 387H, and 389Q). Finally, only two selected codons with a 90% posterior probability (both in HVR1) were identified in type 1 strains, whereas 20 sites under selection were found in type 2 strains (Fig. (Fig.33).
The HCV quasispecies in the PBMC of both the MC and the CH subjects were less heterogeneous than those found in their plasma samples. In particular, the E1/E2 complexity and HVR1 diversity were significantly less (P = 0.015 and P = 0.038 [Wilcoxon nonparametric test for paired samples; Table Table4]).4]). The proportion of synonymous substitutions was also significantly less in PBMC than in plasma (P = 0.011), and the difference in dN was of borderline significance (P = 0.06 [Table [Table44]).
To rule out the possibility that the reduced sequence variability in PBMC may have been due to the small amount of HCV-RNA recovered from this compartment, we analyzed 10 further clones obtained from a single control (subject 15) after a 1:5 dilution of the original PBMC sample. There were no differences in the master sequence, median clone divergence, and mutant frequency between the diluted and undiluted samples.
The diversity and complexity of the HVR1 quasispecies in the plasma and cryoprecipitates of the MC patients positively correlated (the correlation coefficients were as follows: quasispecies divergence, 0.78 [P = 0.014]; complexity, 0.70 [P = 0.035]). They also correlated in terms of the proportion of nonsynonymous substitutions (correlation coefficient of 0.75, P = 0.02).
The compartmentalization of the HCV quasispecies was evaluated in the PBMC of eight subjects (four with MC and four with CH). MC patient 8 had only one mutant largely prevalent in all of the studied compartments (Fig. (Fig.4A),4A), and so he was excluded from further analyses.
Mantel's test results were consistent with significant quasispecies compartmentalization in the PBMC of patients 2 (r = 0.48, P = 10−5), 5 (r = 0.3, P = 0.0002), and 6 (r = 0.25, P = 10−5) but not in the PBMC of patient 4 (r = 0.005, P = 0.5). In contrast, only one of the four CH controls showed significant PBMC compartmentalization (subject 17, r = 0.5, P = 0.0004), whereas subjects 13 (r = 0.045, P = 0.3), 15 (r = 0.0089, P = 0.47), and 16 (r = 0.028, P = 0.52) did not show any quasispecies compartmentalization.
Phylogenetic analysis confirmed PBMC quasispecies compartmentalization in patients 2, 5, and 6 and control 17 (bootstrap values between 95.1 and 66.4 [Fig. 4B and C]) and excluded it in the remaining subjects (Fig. (Fig.4D4D).
The maximum-likelihood method led to identical results. In particular, likelihood mapping of separate groups of plasma and PBMC sequences revealed the significant grouping of PBMC quasispecies in patients 2, 5, and 6 and control 17 (the percentage of quartets supporting this topology varied from 100 to 45.7%).
Analysis of quasispecies distribution in the plasma and cryoprecipitates of nine MC patients revealed three different situations. First, there were patients with identical clones in both compartments (patients 7 and 8, group 1) who had a very homogeneous, substantially identical quasispecies population in all of the analyzed compartments (Fig. (Fig.4A).4A). The Mantel's test results were not significant (P > 0.05) in either of these patients (r = 0.05 and 0.025). Second, there were subjects with clearly different clones in the two compartments (patients 5, 9, and 10, group 2 [Fig. [Fig.4E])4E]) in whom only a minority of the clones derived from plasma clustered with the cryoprecipitate sequences. The highly significant Mantel test results (P < 0.001) in all of these subjects indicated the compartmentalization of the cryoprecipitated and plasma quasispecies (r = 0.68, 1, and 0.66). Third, there were subjects in whom it was possible to recognize two different quasispecies populations in the cryoprecipitate, the first closely related to the plasma quasispecies and the second markedly different (patients 1, 2, 3, and 6, group 3 [Fig. 4B and C]). The Mantel test results showed borderline significance (r = 0.24, 0.17, 0.22, and 0.22 [0.1 < P > 0.03]).
Likelihood mapping confirmed these results, showing no significant compartment-related grouping of quartets in group 1 patients (patient 8 [Fig. [Fig.5]);5]); a high percentage of quartets (>75%) fell at the top of the triangle in group 2 patients (supporting a topology of cryoprecipitated clones that are more closely related to each other than to the plasma clones) (patient 9 [Fig. [Fig.5]).5]). Finally, likelihood mapping performed by separately grouping the cryoprecipitated clones that were more closely (subgroup b) or less closely related to plasma quasispecies (subgroup c) revealed a high percentage (83.3 to 96.1%) of quartets supporting this quasispecies distribution in group 3 patients (patient 2 [Fig. [Fig.55]).
Groups 2 and 3 did not have common plasma and cryoprecipitate master sequences, but patient 2 had a common PBMC and cryoprecipitate master sequence that was absent in plasma.
The aims of this study were to test the hypothesis of a role of HCV quasispecies and to estimate the degree of HCV heterogeneity in an extrahepatic systemic HCV-related disease such as MC.
The large number of examined clones showed a clear trend toward less HVR1 heterogeneity in the MC patients, whose median nonsynonymous substitution rate was 3.3-fold lower than that of the controls, whereas the median synonymous substitution rates in both groups were similar. Nevertheless, the tested patients were probably insufficient to reach statistical significance for all of the analyzed parameters. Interestingly, we observed that the subjects infected by HCV genotype 1 (both MC and CH patients) had significantly fewer divergent quasispecies than those infected with genotype 2, mainly due to an eightfold-lower rate of nonsynonymous substitutions.
After the results were stratified on the basis of the infecting genotype, the MC patients, among the subjects infected by genotype 1, were found to have significantly less quasispecies complexity than did the controls. In contrast, both MC and CH patients infected with HCV-2 had similar diversity and complexity values but, although among the controls the higher variability of HCV-2 than HCV-1 was due to a 11 times increase of nonsynonymous substitutions, among MC patients it was largely due to an increase of synonymous mutations, thus suggesting less positive selection in MC subjects than in CH subjects. Since selection affects only a few amino acid sites in a protein, the dN/dS ratios averaged over codon sites is frequently not significantly higher than 1, but new models have been recently developed to allow for heterogeneous dN/dS ratios among sites (58) and can be used to identify the residues of a protein under positive selection. In a recent report, Sheridan et al. (52) showed the importance of this high-resolution analysis when studying the interaction between HCV and the host immune response. Analysis of our data with these models yielded two interesting results. The first was that there was a significant reduction in positive selection in both MC and control subjects infected by genotype 1 compared to those infected with HCV genotype 2. This is the first study that has applied high-resolution analysis to the association between the infecting genotype and positive selective pressure on E1/E2 sites. It has recently been suggested that HCV type 1b is less complex than other HCV genotypes (5, 14), and this may correlate with relatively weak humoral and cell-mediated immune responses against HCV 1b (34, 59). Our findings of the significantly greater diversity of genotype 2 HCV, together with the higher selective pressure at HVR1 level in this genotype, clearly confirm the association between host immune pressure and the HCV genotype.
The second finding was the relatively low level of selection pressure in our study population: only two (20%) of the MC patients (and none of those infected by HCV-1) showed an averaged ω higher than 1 in the E1/E2 region versus four (50%) of the controls (one with HCV-1). The great majority of the sites under positive selection were in HVR1 (both in MC patients and in controls), thus confirming the presence of an immunodominant B-cell epitope in this region (52). A number of studies have suggested that the heterogeneity of HVR1 is related to the host immune response and postulated that viral diversity is driven by immune pressure (15); in particular, it has been observed that immunocompromised hosts (such as agammaglobulinemic or human immunodeficiency virus type 1-infected patients) have a less heterogeneous mutant population (8, 57). Our finding of less quasispecies heterogeneity in MC patients may therefore be due to an impaired immune response in patients with autoimmune disorders, as is also suggested by the low dN values constantly observed among the MC patients.
The results of a phylogenetic study of HCV mutants in donor-recipient pairs led Allain et al. to suggest that the evolutionary rate of the quasispecies is inversely proportional to the duration of the infection (4), whereas other authors (32) have found that E1/E2 genetic drift seems to be largely unrelated to immune pressure, thus suggesting the existence of possible alternative mechanisms. Mixed cryoglobulinemia mainly affects patients aged 40 to 60 years, almost all of whom have no history of acute hepatitis infection. Our case file did not include the period of time in which the subjects were infected. However, except for one CH patient infected by HCV genotype 2e/f, which has been already described in Italy (6), all of our MC patients and controls were infected by HCV genotypes 1b and 2a/c (which have been spreading in Western countries for decades) (46), and we can argue that they have probably been infected for many years. The lower rate of viral replication and evolution in MC subjects with an “old” infection (4) may partially explain the low level of viral diversity and selective pressure observed by us.
Finally, although it has been reported that viral heterogeneity directly correlates with serum ALT levels (5), there was no difference in the mean serum ALT levels of our MC patients and controls. One possible confounding factor in the present study is that the group of MC patients included three subjects with cirrhosis, whereas only one of the controls had cirrhosis. However, stratification of the results by the stage of liver disease did not reveal any significantly different quasispecies heterogeneity in the subjects with or without cirrhosis (data not shown).
Taken together, these observations suggest that HCV quasispecies in MC patients are the product of a number of virus-related (genotype, replication activity, and evolutionary rate) and host-related (age, immune response, and period of infection) factors, but longitudinal observations will be required to define the role of each.
The second main objective of the present study was to assess the compartmentalization of HCV in plasma, cryoprecipitate, and PBMC. Although HCV-RNA has been found in the mononuclear blood cells (particularly B lymphocytes) of infected patients (28, 29), including MC patients (60), the ability of the virus to replicate inside these cells is still questioned. We found that the HCV quasispecies in PBMC were significantly less complex and divergent than those in plasma and that both synonymous and nonsynonymous substitution rates were significantly lower in PBMC than in plasma. It can be hypothesized that the reduced complexity and diversity of PBMC quasispecies is due to the slow replication of selected quasispecies “inside” the cells and/or the selective absorption of some variants at the cell surface (1, 12).
The compartmentalization of the HCV quasispecies in PBMC was evaluated by applying phenetic and phylogenetic methods, using Mantel's test, and bootstrap and maximum-likelihood analyses.
Our data confirm the observations of others (1, 12, 38) showing that significant compartmentalization in PBMC is possible, but it was only found in four of nine subjects. Although compartmentalization seems to be a frequent finding in MC, its role in the pathogenesis of the syndrome has not yet been defined.
Analysis of the cryoprecipitates showed that, with the exception of two patients with highly homogeneous quasispecies (a phenomenon previously described by others) (26, 37), all of the other MC patients had at least partially different quasispecies in their plasma and cryoprecipitate samples, and the master sequences were always different. These observations are in agreement with other studies showing the compartmentalization of HCV quasispecies in immunocomplexed and free plasma virus (the last representing the escape mutants) (3, 12).
Aiyama et al. (3) have reported a highly homogeneous quasispecies population complexed with anti-HVR1 antibodies in cryoprecipitates. In our study, the degree of diversity and complexity of the HCV clones in cryoglobulins and plasma significantly correlated. It is probable that antibodies directed against viral antigens other than HVR1 participate in cryoprecipitate formation, thus justifying the HVR1 heterogeneity observed by us and others in cryoglobulin or and immunocomplexes (26).
One of our MC patients had an insertion of five amino acids in HVR1 codons 385 to 389, which was observed with only minimal variations in all of the clones obtained from the patient's plasma, cryoprecipitate, and lymphocyte samples. A number of studies have suggested that HVR1 induces antibodies that can neutralize E1/E2 binding to susceptible cells and, more recently, it has been reported that E1/E2 glycoprotein binds CD81, a tetraspanin largely expressed by human cells (47), and E1/E2 peptide motifs (also included in HVR1) have been identified that can bind the surface of different cell types such as hepatocytes or B cells (22). Sequence variations in this region may affect binding to cell receptors or neutralizing antibodies, as suggested by the observation in our isolate of amino acid sites under positive selection in the insertion. In particular, four of the five newly inserted residues were polar and therefore capable of affecting the protein-protein interactions in the region and its antigenic and binding properties. Gerotto et al. recently found a one-residue insertion/deletion at codons 384 to 385 in about one-third of their MC patients (23), but we did not observe any other case of one-residue insertion or deletion. One reason for the different results of the two studies (which were both performed in Northern Italy) may be the different distribution of HCV genotypes. Further studies are needed to demonstrate the existence of specific mutations characteristic of MC.
In conclusion, the heterogeneity of HCV quasispecies and the positive selective pressure tend to be less in MC patients than in noncryoglobulinemic controls, possibly because of factors related to the impaired immune response, the HCV genotype, and the duration of infection. The mechanisms regulating HCV quasispecies in the PBMC and cryoprecipitates of many MC patients, as well as their pathogenic significance, remain to be defined.
This study was supported by a grant from the Centre for Bio-Molecular Interdisciplinary Studies and Industrial Applications-Centre of Excellence of the Milan University.
We thank Aldo Manzin for helpful advice on phylogenetic analysis.