Reader Comments

Post a new comment on this article

Four observations and a suggestion

Posted by kwitwer1 on 05 Oct 2011 at 16:07 GMT

The authors are commended for publishing this interesting article, addressing important questions about the nature and consequences of the PBMC innate immune response to HIV infection--in particular, changes in miRNA and mRNA levels. The authors correctly note that their study is not directly comparable to published in vivo studies, the results of which appear to differ substantially. As constructive criticism, I would offer several additional observations. First, there is tremendous variation across samples in the raw data for four endogenous small RNA normalization controls (Table 1, mistakenly identified in the text as "two miRNA and two small RNA controls"): from 50-fold for U6 to 150-fold for RNU48. This variation is not associated with infection, and although control RNAs may vary somewhat in complex biological samples, cultured cells are unlikely to display such large fluctuations. Nor does Table 1 present the entire range of fluctuation in the study: only five of 16 donors are shown, and the authors state that samples from two donors were discarded based upon even more divergent values. Second, in the abstract and elsewhere it is written in error that 21 miRNAs were significantly (p<0.05) differentially regulated. Only 16 miRNAs with p<0.05 are listed in Table 3 (along with five at p=0.05 or greater). Furthermore, it is not clear that the reported p values were corrected for multiple comparisons; one presumes that this standard correction was made, but, if not, it is likely that few or none of the reported differences would remain significant following correction (whether by Bonferroni or less conservative methods). Third, it is difficult to interpret the qRT-PCR validation of nine miRNAs with significant differential expression and five with no significant changes according to array analysis. The error bars (Figure 3) are not explained, and no p values are provided. Hence, it is unclear if any of the displayed changes is significant. Although Figure 3 does not identify the two classes, cross-checking with Table 3 suggests that many of the control miRNAs were upregulated as much as or more than the putatively significantly upregulated miRNAs. Furthermore, of the nine miRNAs with array p values reportedly less than 0.05, the actual p values for three were 0.05 or greater. Only six significant miRNAs were assayed, then, and with unclear outcome. Of these six, it would appear that one miRNA was either confused with its hairpin partner in the validation assay or mislabeled: miR-33b* is listed in Table 2, while miR-33b is shown in Figure 3. Fourth, Gupta, et al. write that a study of brain miRNA during HIV infection and/or HIV dementia by Tatro, et al. (PLoS One, 2010) found that "of the 19 miRNAs that were dysregulated in HIV-1 positive brain tissue compared to uninfected brain tissue, six were also dysregulated in HIV-1 infected PBMCs compared to uninfected cells," but Tatro et al. report dysregulation of 97 miRNAs, not 19, in HIV infection and/or HIV dementia and do not compare brain tissue with PBMC. None of the miRNAs reported by Gupta, et al. is found among the 97 dysregulated brain miRNAs of Tatro, et al. It is thus unclear why Gupta, et al. present Tatro's work as "in line with" their findings. The authors are encouraged to address these observations to enhance the community's understanding of their study. The striking variation of endogenous controls, particularly, suggests that exploring a more global normalization strategy may be desirable. Perhaps the authors could test these and/or deposit their array results in a public database such as GEO to facilitate alternative analyses by others. Until then, one might interpret with caution the extensive and perhaps somewhat conjectural (albeit fascinating) pathway analyses in this report.

No competing interests declared.