|Home | About | Journals | Submit | Contact Us | Français|
Recently, Datta et al. proposed that hepatitis B virus (HBV) invariably infects peripheral blood lymphocytes (PBLs) and that the HBV sequences obtained from PBLs differ from those recovered from the serum due to major differences in humoral immune pressure (2). The basis for their conclusion was a high prevalence of a mutation at G145R in the HBV of the PBL compartment, which was not found in sera. We believe that these data result from contamination of the PBL sequences.
Learn et al. proposed guidelines for protecting the integrity of the HIV sequence databases (4). They suggested examining a distance matrix for “high levels of unexpected similarity” that would raise suspicion for contamination.
A BLAST search including sequences from the Datta et al. paper confirmed identity of a number of the PBL sequences but also identified another identical sequence (GenBank accession number DQ084757). Interestingly, that sequence was published in 2006 from the same investigators (1) and was designated X2. The X2 sequence is identical to many of their PBL sequences in the current report, including PB06, PB16, PB24, PB30, PB35, PB38, PB39, PB43, PB45, PB47, and PB54 (Fig. (Fig.1).1). Only mixed bases separate X2 from other PBL sequences, including PB02, PB33, PB41, PB61, PB62, and PB76, indicating that these also most likely contain X2. A single nucleotide change separates X2 from PB17 and PB26, as well as a second cluster involving Y3 and PB07, which, in turn, are only one mixed base different from PB19 and PB32. Furthermore, the G145R mutation, which they attribute to shared immune escape in the PBL compartment, was also found in several sequences from the 2006 paper, including X2 and Y3 (which were, according to Table 1 in the 2006 publication, obtained from siblings). The variability seen in Fig. Fig.11 of that publication also refutes the hypothesis that constraints on HBV evolution at the studied locus explain the homogeneity of their sequences.
Their data seem biologically implausible since they indicate that the HBV in the serum and that in the PBL had different genotypes. For this to occur, one would have to hypothesize that the subjects were infected with two different genotypes but that only one of the genotypes was able to enter PBLs and subsequently disappeared from the circulating quasispecies.
We raise these concerns because we strongly believe that inferences about HBV biology, and the integrity of the HBV database, need to be protected from sequences generated through contamination. Though the authors state that “the possible cross-contamination between samples or reagents was ruled out by including appropriate controls during all of the experimental steps and aptly following the PCR recommendations of Kwok and Higuchi” (3), simply applying these precautions does not always prevent contamination. The HBV community needs to remain vigilant for sequences generated by contamination; this goal would be served by following the principles proposed by Learn et al. (4). Unexpected homogeneity is strongly indicative of tight epidemiological linkage or contamination. In the case of the paper by Datta et al., only the latter seems plausible.
Michalak et al. have already established that hepadnaviruses are strongly lymphotropic and that lifelong occult persistence is inevitable in hepadnaviral infections (19). However, genetic variability data of lymphocyte-associated hepatitis B virus (HBV) was lacking, and this is what we studied.
We strongly disagree with Drs. Thio and Ray that our data originated due to contamination. (i) Initially, we submitted the present paper with direct sequence data. Following reviewers' suggestions, reamplification and multiple clonal analyses from both serum and peripheral blood leukocytes (PBL) were done to exclude the possibility of contamination and to confirm results. (ii) The precautions and controls for maintaining the integrity of HBV sequences were explicitly explained in the paper.
We think that it may not be rational to use the same parameters for validating the HBV sequence data as those specified for HIV in the work of Learn et al. (17) because of the following: (i) HBV evolves at an almost-100-times-lower rate than does HIV (12, 13), and identical sequences among unrelated study subjects are thus common in HBV studies, and (ii) completely identical HBV sequences or single/two-nucleotide differences among unrelated isolates, in the genetic region that we studied and also in other genetic regions (represented in Fig. Fig.1),1), are frequent in HBV literature (1, 2, 4, 10, 15, 18, 21, 22).
Regarding the similarity of two serum isolates (X2 and Y3) from our previous study (7) to that of present PBL isolates, it is important to note what follows. (i) The isolates X2 and Y3 were not siblings and were from different families, a fact which also supports the present data. (ii) It is extremely unlikely that these sequences, isolated 5 years ago, will contaminate solely 24 to 61 current PBL isolates rather than other serum isolates analyzed in the same laboratory during different periods (discussed in the paper). (iii) Possibly the Ae/A2-specific HBV sequences in those serum isolates (X2 and Y3) might have originated from the PBL of the respective subjects; which is in accordance with a report on the woodchuck hepatitis virus (WHV) model showing that lymphatic system-restricted hepadnaviruses may seldom engage hepatocytes (20) and may thus be detectable in the circulation.
The possibility of infection with two different HBV genotypes is not at all biologically implausible, because of the following: (i) the presence of different HBV genotypes without evidence of coinfection and subsequent selection of dominant genotype among mixes (facilitated by host immune status) have been evidenced by recurrence of the original genotype, following reversion to the original immune status (3, 8, 11, 14); (ii) we have also discussed the possibility of circulation of genotype Ae/A2 below the detection limits of our assay; (iii) compartment-specific predominance of different hepatitis C virus (HCV) genotypes in the leukocytes and serum is well documented (9); (iv) recently, Coffin et al. (5, 6) demonstrated differential predominance of HBV variants between liver and peripheral blood mononuclear cells (PBMC), which also supports the compartment-specific evolution and selection of HBV variants under altered immune pressures.
Taken together, our data are in accordance with previous and recent studies, ruling out contamination. Finally, we refer to the work of Korber et al. (16), mentioning that similar sequences may not inevitably indicate experimental error but may also indicate an interesting biological phenomenon. Thus, similar studies of HBV compartmentalization should be encouraged.