RNA sequencing of CD4 T-cells reveals the relationships between lncRNA-mRNA co-expression in elite controller vs. HIV-positive infected patients

Introduction

Human immunodeficiency virus (HIV) has been studied for more than 30 years, and research has found that HIV invades host CD4+ cells, integrates its own DNA into the host genome, and establishes a reservoir in the early stage, followed by massive replication, which destroys the normal immune system function of the host, and triggers various concurrent symptoms (Sabin & Lundgren, 2013; Eisele & Siliciano, 2012). Ultimately, HIV infection can cause the host to die. Although highly active antiretroviral theraphy (HAART) can almost completely reconstitute the immune function of the host, it cannot eradicate the latent HIV pool in the host, which always maintains a low level of virus replication (Murray et al., 2016). However, among HIV-infected individuals, a small number of patients are found to be inherently resistant to viral replication, and spontaneously induce suppression of the latent pools without HAART, resulting in an undetectable viral load in plasma for a long period (Migueles & Connors, 2010; Gonzalo-Gil, Ikediobi & Sutton, 2017). These patients, termed elite controllers (ECs), have attracted significant research interest. Studying the characteristics of the autoimmune factors of ECs is anticipated to identify important factors that control virus replication. Such valuable information could lead to novel methods to treat and alleviate HIV infection.

Long-noncoding RNAs (lncRNAs) are transcribed RNA molecules greater than 200 nt in length that regulate gene expression by diverse, but as yet not completely understood, mechanisms (Wilusz, Sunwoo & Spector, 2009). Although the function of most lncRNAs is unknown, several have been shown to regulate gene expression at multiple levels, from DNA to phenotype (Wapinski & Chang, 2011). Through a variety of means, including cis (near the site of lncRNA production) or trans (co-expressed with their target gene) mechanisms, lncRNAs play a vital role in many biological processes (Wang & Chang, 2011). Studies on lncRNAs have become a hotspot in current non-coding RNA research.

Recent studies focused on the role of lncRNAs in HIV pathogenesis, especially the relationship between lncRNA regulation of gene expression and viral infection, replication, and latency (Trypsteen et al., 2016; Lazar, Morris & Saayman, 2015; Saayman et al., 2014; Nair, Sagar & Pilakka-Kanthikeel, 2016). A few lncRNAs have been characterized and proven to be closely associated with HIV, for example, the knockdown of host lncRNA NEAT1 enhanced virus production by increasing nucleus-to-cytoplasm export of Rev-dependent instability element (INS)-containing HIV-1 mRNAs (Zhang et al., 2013). The knockdown of lncRNA NRON enhanced HIV-1 replication through increased activity of nuclear factor of activated T-cells (NFAT) and the viral long terminal repeat (LTR) (Imam et al., 2015). Similarly, lncRNA MALAT1 releases epigenetic silencing of HIV-1 replication by displacing the polycomb repressive complex 2 from binding to the LTR promoter (Qu et al., 2019). Furthermore, an HIV-encoded antisense lncRNA, ASP-L, was proven to promote latency HIV (Saayman et al., 2014; Kobayashi-Ishihara et al., 2012). In addition, HEAL is a broad enhancer of multiple HIV-1 strains because depletion of HEAL inhibited X4, R5, and dual-tropic HIV replication, which was rescued by HEAL overexpression (Ti-Chun et al., 2019). Although the mechanism of lncRNA function is sometimes elusory and unpredictable, many current studies, which are limited to the screening and functional prediction stage based on chip or sequencing results, also provide us with valuable information and from the basis for mechanistic research.

In the present study, we conducted a transcriptomics investigation for HIV ECs, to identify differences in the transcriptional expression profiles between ECs and HIV-positive infected patients as compared with healthy controls. The obtained lncRNA-mRNA co-expression network revealed the possible role of the functional relationships between lncRNAs and mRNAs in the ability of ECs to inhibit viral replication.

Methods

Results

Subjects

A total of 196 individuals infected with HIV were enrolled in our cohort, including patients under treatment (181, 92.4%), HIV-positive infected patients (13,6.6%) and elite controllers (2, 1.0%). In addition, we recruited two healthy individuals from the out-patient department of the hospital. Most patients were infected with the virus through heterosexual contact, followed by intravenous infection, and homosexual transmission. Based on the basic principle of intra-group identity, we determined three men and three women who were assigned to each of the three study groups as study subjects. Their ages ranged from 39 to 54 and were free of other diseases, such as tuberculosis, diabetes, or hepatitis. Specific details of the clinical characteristics of the participants are shown in Table S1. Descriptive statistics are reported as counts (percentage) for dichotomous and categorical variables, and the median (the 25th percentile and 75th percentile) for continuous variables. The experimental design and analysis process of this research are shown in a flowchart (Fig. 1).

Identification of DE lncRNAs between ECs and HPs

Among all transcripts, 3,543 official gene symbols that matched in the NONCODE database had corresponding lncRNA transcripts. Three main categories of lncRNAs accounted for the majority, including antisense lncRNAs (2537/14853, 17.1%), intronic lncRNAs (8152/14853, 54.9%), and long intergenic noncoding RNAs (lincRNAs; 3344/14853, 22.5%). There were 151 upregulated lncRNAs and 160 downregulated lncRNAs in ECs vs. HPs according to the threshold of fold change >2 and p < 0.05. In both the mRNA and lcnRNA clusters (Fig. 2), the samples showed marked intra-group correlations and inter-group differences, which further confirmed the specificity of the samples. In terms of the overall profiles of mRNAs encoding proteins with specific biological functions, the HC and EC groups showed closer clustering and were relatively distant from the HP group. The lists of FPKM values and regulatory situations of all DE mRNAs and lncRNAs in ECs vs. HPs are shown in Table S5.

LncRNA-mRNA co-expression network analysis

Almost 0.9 million co-expression relationships were discovered, and the exact form was expressed as a linear correlation whereby one lncRNA targets one mRNA.Based on previous DE analysis, 1109 DE lncRNAs targeting DE mRNAs were identified In all correlations (Table S6). Figure 3 shows the network constructed using these relationships, in which 142 DE lncRNAs were co-expressed with 127 DE mRNAs. To investigate the importance ranking of these DE mRNAs, and find the genes at the center of the functions of gene expression, we construted a PPI network. Figure 4 shows the top 50 hub genes that playing an important role in interaction of 127 DE mRNAs, in which many genes functioned centrally and could be frequently regulated by other transcripts. Considering the differences in the outputs of different algorithms, we identified those mRNAs that were commonly detected by all 12 algorithms and focused on those mRNAs that were regulated by lncRNAs and their functional enrichment analysis. Table S7 shows the list of 24 intersecting hub mRNAs among the top 50 calculated using the 12 algorithms. In Fig. 3, we identified 16 of 24 intersecting hub mRNAs in the diagram, which helped us to quickly locate the target protein coding genes and easily access the enriched functions of these genes. Next, we selected the 16 hub mRNAs and their associated lncRNAs from Fig. 3 and constructed Fig. 5. We then identified the top 10 GO terms and pathways of all DE mRNAs most closely related to infection and immunity in Fig. 3. Many significant functions are co-enriched, such as immune response-regulating signaling pathway (GO:0002764) and cellular response to interleukin-1 (GO:0071347) (Fig. 6). It should also be pointed out that when there were various many-to-many regulatory relationships, the most statistically significant regulatory relationship was selected as the main study object.

Dicussion

In the present study, we systematically analyzed the expression profiles of lncRNAs and mRNAs from CD4+ T cells from ECs, HPs, and HCs, and constructed a co-expression network based on the relationships among DE transcripts and database annotations. The functions of most lncRNAs have not been well characterized. Therefore, the gene functions in this research mainly referred to those of mRNAs and their expression products. The potential functions of candidate lncRNAs and the lncRNAs-associated biological processes in ECs were predicted using the lncRNA-mRNA network and functional enrichment analysis. We identified certain genes that play a significant role in the ability of ECs to inhibit the replication of latent virus. Most of these genes were downregulated in the HCs and ECs and upregulated in the HPs or vice versa, which was consistent with their clinical characteristics, in that ECs maintained a similar gene expression pattern to the HCs, which distinguished them from the HPs.

To the best of our knowledge, the present study was the first to predict lncRNA-mRNA interactions for HIV ECs, in which many genes had not been previously reported to function in HIV-associated biological processes. The MAPK (mitogen activated protein kinase) family was found to be associated with the reactivation and replication of the virus in many studies (Wang et al., 2017; Prasad & Tyagi, 2015) , and the HIV-1 virus inhibition activity of MAPK-NF-κB/AP-1 pathway inhibitors was the result of the negative regulation of HIV-1 LTR promoter activity (Gong et al., 2011). In a recent study, HDAC6 (histone deacetylase 6) overexpression significantly enhanced the expression of pro-inflammatory cytokines, such as TNF-α (tumor necrosis factor alpha), interleukin (IL-1)β, and 289 IL-6, with a concomitant reduction in acetylated α-tubulin. HIV-1 trans-activator of transcription (Tat) can upregulate the expression of various chemokine genes including IL-1, IL-6, CCL2 (C-C motif chemokine ligand 2), CXCL8 (C-X-C motif chemokine ligand 8), and CXCL10 (C-X-C motif chemokine ligand 10). In addition, HDAC6 overexpression increased the activation of MAPK factors including extracellular signal-regulated kinase (ERK), Jun terminal kinase (JNK), and mitogen-activated protein kinase 14 (p38). Similarly, HDAC6 overexpression resulted in activation of the nuclear factor kappa B (NF-κB) and activator protein 1 (AP-1) signaling pathways (Youn et al., 2016; Nookala & Kumar, 2014). In our results, low expression levels of HDAC6 and MAPK8 and downregulation of IL-1 and IL-6 associated receptors in ECs indicated suppressed HADC6- MAPK-NF-κB/AP-1 signaling and fewer immune response disorders. Inhibition of histone deacetylase contributes to reverse the latency of HIV. Ataxia Telangiectasia Mutated (ATM) is also involved in the NF-kappa B signaling pathway (hsa04064) by activating p38MAPK and p53 sequentially to induce a pro-apoptotic pathway. Ariumi & Trono (2006) indicated that ATM could enhance HIV-1 replication by stimulating the action of the Rev viral posttranscriptional regulator, and Lau et al. (2005) demonstrated that a novel and specific small molecule inhibitor of ATM kinase activity, KU-55933, is capable of suppressing the replication of both wild-type and drug-resistant HIV-1. Interestingly, some studies considered that HIV-encoded protein Vpu contributes to the attenuation of the anti-viral response by partial inactivation of interferon regulatory factor 3 (IRF3) while host cells undergo apoptosis (Park et al., 2014). By contrast, another study reported that Vpu affects interferon expression by inhibiting NF-κB activity without affecting IRF3 levels or activity (Manganaro et al., 2015). Thus, the role of IRF3 in HIV infection and replication remains controversial. The above hub mRNAs are related to annotated lncRNAs including C3orf35, TMEM254-AS1, ARRDC3-AS1, LINC00202-1, TRAF3IP2-AS1 and to unannotated lncRNAs (Fig. 5), which might control HIV replication by regulating the expression of these genes. To date, no study has been published showing that these lncRNAs function in the pathways identified in the present study.

Alternatively, some hub genes (Fig. 4) were not regulated by lncRNAs but interacted with each other and participated in biological processes such as HIV life circle, inflammation and immune activation and cytokines. Lysine demethylase 6A (KDM6A; also called UTX-1) enzymatic activity is required for the viral protein Tat to remove a repressive mark H3K27me3 in the HIV-1 long terminal repeat (LTR) and promoted HIV-1 gene expression by enhancing both NF-κB p65′s nuclear translocation and its binding to the HIV-1 LTR. H3K27 demethylase activity is required for increased HIV-1 transactivation induced by KDM6A (Zhang et al., 2016). In our study, the low expression level of KDM6A in the ECs might be a possible causes of their low viral loads. TRAF3IP2 encodes TRAF3 interacting protein 2, wich is involved in regulating responses to cytokines by members of the Rel/NF-kappaB transcription factor family. TRAF3IP2 interacts with TRAF proteins (tumor necrosis factor receptor-associated factors) and either I-kappaB kinase or MAP kinase to activate either NF-kappaB or Jun kinase. We found that lncRNA TRAF3IP2-AS1 was an anti-sense non-coding RNA of TRAF3IP2 that was upregulated in the ECs, and was co-expressed with downregulated MAPK8. This suggested there may be a regulatory relationship among TRAF3IP2-AS1-TRAF3IP2-MAPK8 that plays an important role in the suppression of viral replication.

HAART cannot eradicate the latent virus in the host; therefore, the genes identified in the present study might play a non-negligible regulatory role in the relationship between the latent virus and the host. Latent virus is the biggest obstacle to the elimination of the disease. Thus, understanding the processes that contribute to its persistence, such as inflammation and immune activation, are crucial for the remission and cure of HIV (Dahabieh, Battivelli & Verdin, 2015; Massanella, Fromentin & Chomont, 2015) .

There are also some limitations to this study. First, because of the limited amount of RNA in the samples, the sample size of this study was relatively small. Although we selected eight and six mRNAs for verification using qRT-PCR for 16 genes and 40 samples three times, respectively and found that the expression change trends of 13 of them were consistent with the RNA sequencing results, these results lack of sufficient statistical significance because of the small sample size. Part of the verification analysis is shown in Fig. S2. Second, we only assessed gene expression of CD4+ T cells, which, although they are the main targets of HIV and are vitally important, they are only one of the many cell types that are infected by HIV. Further research should be performed to validate the function and mechanism of the genes in a larger sample and in more cell types.

Conclusions

This study identified several important differentially expressed genes associated with the elite controller phenomenon, using RNA-sequencing and bioinformatics analysis to explore how ECs spontaneously control virus replication. Some differentially expressed genes were identified and enriched for meaningful GO terms and KEGG pathways related with viral infection and immune responses. After filtering lncRNAs and co-expressed mRNAs using the Pearson correlation test, the functions of the lncRNAs were predicted using an lncRNA-mRNA network comprising the functional annotations of mRNAs. This study forms the basis for subsequent cellular and molecular studies, and provides new targets for gene-targeted therapy in the future.

Supplemental Information