Obligatory intracellular pathogens, such as viruses, are known for their ability to manipulate host cells both at the molecular and cellular levels. Studies have shown that the host cellular transcriptome is regulated by HIV-1 infection, as part of the cells’ response to pathogen insult [60, 61]. HIV-1 infection results in differential regulation of specific cellular genes in target cells such as CD4+ T cells and monocytes/macrophages suggesting that virus infection alters host cellular proteins either to evade the immune system and/or for optimal viral replication. Considering the prominent role played by miRNA in the regulation of gene expression, we hypothesized that analysis of miRNA in samples from HIV-1 infected individuals may provide information about the molecular network and correlate with the disease status in the host. The novel features of our present study are the following: (i) We have utilized samples from HIV-1 infected hosts characterized for viral load (HVL and LVL), together with appropriate age matched controls; (ii) The samples were derived from a large number of HIV-1 infected subjects; (iii) ours is one of the first in vivo studies to compare miRNA and mRNA simultaneously in PBMCs isolated from a cohort of HIV-1 infected and uninfected subjects. Our results show that HIV-1 regulates the cellular transcriptome significantly through miRNAs, resulting in an inverse correlation of miRNA-mRNA expression. Furthermore, we have also established that there is an interrelationship between multiple miRNA and mRNA interactions and vice versa within the infected individuals.
The majority of miRNAs are transcribed from noncoding DNA, however a number of miRNAs are also known to be expressed along with mRNA suggesting a specific role for mRNA regulation [62–64]. Our comparative analysis of miRNA and mRNA indicate that HIV-1 infection alters both miRNA and mRNA expression profiles in HIV-1 infected subjects compared to uninfected controls. However, the dysregulation of both the RNA species is less prominent in LVL subjects compared to the HVL subjects indicating that viral load has a significant effect. Differentially expressed miRNA between groups indicates both overlapping and group specific miRNA expression. For instance, HIV-1 infection independent of viral load altered expression of only 4 miRNAs, whereas 191 miRNAs are significantly regulated in the high viral load group compared to uninfected controls. Previous studies using PBMCs or CD4+ T cells from infected viremic, elite controllers and multiply exposed uninfected (MEU) subjects indicate that HIV-1 infection and/or virus exposure resulted in signature miRNA expression profile further supports the role of HIV-1 induced miRNA regulation [41, 46, 48].
Although a number of commonly dysregulated miRNAs were noted in these studies, there is also a greater degree of variation between these studies including the current study [46, 48]. The differences observed between these studies could be due to a number of reasons including array platforms, the patient population, sample size, software used and type of analyses, cut-off set by each group and selection criteria of study subjects. Our study has a large sample size (n = 30 or more) and all of our LVL donors were receiving Highly Active Antiretroviral Therapy (HAART) and had no or undetectable viral load for more than 5 years prior to sample collection. Another possible explanation is the cell types used in these studies (total PBMC vs. purified CD4+ T cells), as these are known to exhibit distinct cell specific miRNA profiles [65, 66]. It is also important to point out that our HVL group has very low CD4+ T cell count (<200 cells), suggesting that the differentially regulated miRNAs could be the result of non CD4+ T cell types in response to HIV-1 infection.
Results from our analyses identified several miRNA that are uniquely regulated by HIV-1; however the functional consequences of many of these miRNAs are not fully understood. Further analyses on cell specific miRNAs expressed during virus replication suggest a role for miR-132, -125b, -155, -150, -223 -382 and Let-7 in CD4+ T cells [67–69] and miR-198, -28-3P, 125b, -198, -223 and -382 in monocytes/macrophages [40, 70, 71], respectively. In our analyses using PBMCs, we noted differential regulation of miR-132, Let-7, -28-3P, -125b, -223, -382, but no difference in others. Additionally, miRNAs (miR-214, -223, 132, -128, -150, -32, 24, -93, -198, -296, 351 and -17-92 cluster) are known to target viral RNA and possess antiviral activity and innate immune functions [72, 73]. Among these miRNAs, miR-214 and miR-223 are significantly down regulated in HVL subjects, whereas miR-32, miR-198, miR-128, miR-150 and miR-15 did not show any change. MiR-146a has been linked to innate immunity and immune activation [74, 75] and miR-155 is involved in T cell proliferation and Treg regulation [76, 77]. These miRNAs are significantly upregulated in HVL subjects compared to seronegative and LVL subjects, suggesting that inflammatory factors are highly upregulated in subjects with HVL supporting hyper activation of immune cells during AIDS. Our studies have shown that the LVL group is similar to the uninfected controls at the mRNA level as evident in few gene interactions in the STRING pathway analysis of the mRNA array data similar to previous studies . On the other hand, several host cellular pathways and networks including cytokine and chemokine interactions, signaling pathways, IFN induced genes, metallothionein genes and pro-inflammatory cytokines were dysregulated in high viral load in comparison to uninfected controls and low viral load samples. Although we have highlighted immune cell related functions in this analyses further analyses indicate that additional pathways such as apoptosis and cell death are that are distinctly regulated in HVL subjects compared to LVL and control further supporting the CD4 loss observed in this group.
Overall, we found only 41 significantly dysregulated mRNAs between uninfected and LVL and all of these mRNAs had a fold change of less than 2. Meanwhile, the comparisons of CT vs. HVL and LVL vs. HVL yielded 309 and 182 significantly regulated probes respectively. Several of these pathways (IFN signaling, innate response, defensins, cytokine/chemokine signaling and T cell signaling) that are dysregulated in the high viral load (compared to low viral load), are consistent with the results of other studies using HIV-1 specific target cell types [4, 42–44, 78]. We also noticed APOBEC3G, a cellular defense factor against retroviruses, to be upregulated in the presence of infection (CT vs. LVL) and high viral load (CT vs. HVL), which parallels Rotger et al. and their finding that the gene is upregulated with increasing viral load . Although the current study utilized total PBMC for miRNA and mRNA profiling, the results from this analysis reflect both the direct and indirect effect of HIV-1 on miRNA and mRNA expression as well as capturing the dysregulated genes from both monocyte and T cell lineages. As it is well established that less than 1% of the cells in total PBMCs are productively infected by the virus, the significant changes observed here may not be result of a direct effect of HIV-1 infection, but may instead be indirect effects. These indirect effects could be due to nonproductively infected cells, exposure of virus particles (both defective and replication competent) to target and non-target cells as well as cellular and viral factors released from the infected cells that could affect bystander cell population. Our results show that it would certainly be worthwhile to perform cell specific analysis on HIV-1 target cells using a focused array to assess the effect that infection on the RNA species within the individual cell types, but this is beyond the scope of this manuscript.
To study the interaction between the miRNA and mRNA, we chose not to simply rely on online interaction databases, instead, we elected to use our array data as a supplement to GroupMiR. Notably, our analysis has predicted pathway-associated targets for many miRNAs, suggesting that miRNAs effectively target host cellular pathways during infection and disease progression. Although several of the predicted miRNA-mRNA pairs showed inverse correlation, we also observed a few positive correlations in mRNA regulation. Specifically, in case of inflammatory factors, IL6, CCL2 and CCL8 are inversely regulated by miR-144, miR-376a, miR-935 and miR-146a, whereas, in our study both expression of miR-512-3p-CCL8 and miR-518d-3p-CCL8 are positively correlated in HVL subjects compared to uninfected controls. Most importantly, our studies have identified several new miRNAs that could potentially regulate RNA transcripts such as AIF1, ALDH1A1, NCR3 and NRGN that warrant further research. Though it is established that the host transcriptome is regulated by HIV-1, it is not clear how transcription is regulated. Based on our results, it is possible to predict that one of the mechanisms that the virus employs is the use of miRNA, as regulation of a single miRNA could modulate multiple mRNAs that belong to specific pathways, which would be beneficial for the virus.
The presence of high viral load has an impact on several mRNA transcripts, however, the effect of other factors such as CD4 count and nadir CD4 has a significant negative correlation for IL8; this is consistent with data in the literature that associates infection with elevated IL8 . This was not surprising because the nadir CD4 counts for some of the HVL subjects were quite low. Furthermore, this paralleled Stone et al.’s findings of elevated IL6 in HIV patients after an immune restoration . We also found significant correlations within LVL for IL6 and NRGN and CD4. Interestingly, we found no significant correlations with the level of viral load, indicating that the presence of virus has an effect on producing differential mRNA expression but the level of the viral load makes no additional difference. Together these observations suggest that HIV-1 infection, irrespective of its severity, has significant impact on host cellular transcripts and gene expression resulting in immune dysregulation. These findings might provide further explanation on why immune function is never restored back to normal in HAART treated HIV-1 subjects with no detectable virus load. Our study of miRNA and mRNA profiles in individuals infected with low and high levels of viral load has significant clinical relevance. The finding of several miRNAs that are specific to a particular viral load could lead to the use of these miRNAs as biomarkers for the determination of prognosis of the disease.