Transitory Viremic Surges in a Human Immunodeficiency Virus-Positive Elite Controller Can Shift the Cellular Transcriptome Profile: a Word of Caution for Microarray Studies

Abstract

Human immunodeficiency virus (HIV) is capable of infecting diverse blood leukocytes, through which interaction between HIV and cellular transcriptome can be unveiled. Recently, the transcriptional profiles of primary CD4+ and CD8+ T cells from HIV-positive individuals have been shown to distinguish progression from nonprogression (3). This study, together with our study on cell surface antigens using a protein microarray (8), has demonstrated that the CD8+ T-cell transcriptome and its surface are more informative in differentiating HIV disease groups than the CD4+ T cells. This is consistent with our recent findings, which showed that the whole-genome transcriptome in primary CD8+ T cells was able to segregate HIV-positive viremic (VIR) patients on highly active antiretroviral therapy (HAART) from therapy-naive long-term nonprogressors (LTNPs) (7). Here, we wish to highlight how a subtle surge in plasma viremia in HIV-positive LTNPs can affect the clustering pattern of the cellular transcriptome, which should be an imperative consideration for genome-wide microarray studies on primary cells from HIV-positive patients. We used the Illumina Sentrix Human-6 Expression BeadChip to generate genome-wide expression profiles of CD8+ T cells at two time points over 12 months from four LTNPs along with the VIR group and negative controls (Table ​(Table1).1). Cluster analysis at time point 1 revealed a distinct cluster of LTNPs away from the VIR group (Fig. ​(Fig.1A).1A). Strikingly, 1 year later, the cluster analysis at time point 2 segregated all LTNPs but one (L4), which coincidently grouped with the VIR group (Fig. ​(Fig.1B).1B). Retrospective tracking of L4 plasma viremia revealed 1,000 copies/ml. Interestingly, a similar exception was also noted in Hyrcza's study (3), where one nonprogressor patient (L4) with 269 copies/ml formed a distinct cluster with a progressor (C4). These unexpected findings from two independent studies clearly provide several lessons which may prove to be pertinent to all microarray studies. FIG. 1. Clustering analysis of global gene expression profiles of CD8+ T cells from LTNPs at two time points over 12 months, along with those from the viremic patients and negative controls. Panels A and B show time points 1 and 2, respectively. Groups ... TABLE 1. Clinical details of study patients First, plasma viremia is a key determinant of the clustering pattern between the VIR and LTNP groups, as evident from our study that the low viremic surge could shift the whole transcriptome profile despite the therapy effects. This shift could represent the effects of viral replication resulting in the change of certain CD8+ T-cell subsets (1), which was reflected at the transcriptome level. This shift also highlighted the greater effects of viral replication on the cellular transcriptome as opposed to other factors, including HAART (2, 5). Second, the most current viral loads (VLs) should be determined prior to microarray studies, as intermittent viremia surges do occur in HIV-positive patients responding well to HAART (4) and therapy-naive LTNPs, and they should be seriously considered. Third, stringent clinical criteria should be adopted for selecting and/or defining HIV-positive LTNPs for microarray studies, as suggested recently (6). Finally, a VL threshold may exist, above which the transcriptome profile can shift. This postulation came from our and Hyrcza's data that the transcriptome profile from LTNPs with VLs of <50 copies/ml clustered together, whereas the transcriptome profile in nonprogressors with VLs of 269 to 1,000 copies/ml shifted (3). Should this threshold be determined, it may serve as a reference guide for future microarray studies.