HIV-1 transcription start sites usage and its impact on unspliced RNA functions in people living with HIV

ABSTRACT

HIV-1 unspliced RNA serves two distinct functions during viral replication: it is packaged into particles as the viral genome, and it is translated to generate Gag/Gag-Pol polyproteins required for virus assembly. Recent studies have demonstrated that in cultured cells, HIV-1 uses multiple transcription start sites to generate several unspliced RNA species, including two major transcripts with three and one 5′ guanosine, referred to as 3G and 1G RNA, respectively. Although nearly identical, 1G RNA is selected over 3G RNA to be packaged as the virion genome, indicating that these RNA species are functionally distinct. Currently, our understanding of HIV-1 transcription start site usage and the functions of RNA species is based on studies using cultured cells. Here, we examined samples from people living with HIV to investigate HIV-1 transcription start site usage and its impact on RNA function. Using paired samples collected from the same participants on the same date, we examined the HIV-1 unspliced RNA species in infected cells (PBMC) and in viruses (plasma). Our findings demonstrate that in people living with HIV, the virus uses multiple transcription start sites to generate several unspliced transcripts, including 3G and 1G RNA. Furthermore, we observed an enrichment of 1G RNA in the paired plasma samples, indicating a preferential packaging of 1G RNA in vivo. Our study illustrates the complex regulation of HIV-1 unspliced RNA in people living with HIV and provides valuable insights into how HIV-1 unspliced RNAs serve their functions in vivo.

IMPORTANCE

HIV-1 virions must contain unspliced RNA and its translation products to maintain infectivity. How HIV-1 unspliced RNA fulfills these two essential and yet distinct roles in viral replication has been a long-standing question in the field. In this report, we demonstrate that in people living with HIV, the virus uses multiple transcription start sites to generate several unspliced RNA species that are 99% identical in sequence but differ functionally. One of the RNA species, 1G RNA, is selected over other HIV-1 unspliced RNAs to be packaged into viral particles. These findings are consistent with previous cell-culture-based observations and provide insights into how HIV-1 regulates its unspliced RNA function in people living with HIV.

INTRODUCTION

HIV-1 is a pandemic agent that causes AIDS. As a member of the Retroviridae, HIV-1 carries RNA genome in the virions and converts the RNA genome into viral DNA upon infecting a new cell. HIV-1 integrates its DNA into host chromosomes to form a provirus and relies on host machinery to express the proviral genome (1). Host RNA polymerase II (Pol II) transcribes the provirus to generate viral RNA, which is modified by the host machinery including adding a modified guanosine cap and polyadenylation. HIV-1 transcripts undergo complex alternative splicing to generate multiple types of spliced RNAs to express several viral proteins, while some transcripts remain unspliced (2–4). Unspliced HIV-1 RNA (referred to as HIV-1 RNA hereafter) serves two distinct functions: it is packaged into virions as the viral genome, and it is translated to generate Gag/Gag-Pol polyproteins.

Recent studies demonstrate that Pol II uses neighboring sequences to initiate transcription, generating several HIV-1 RNA species that are nearly identical but varied by a few nucleotides (nt) at the very 5′ end of the transcripts (5–7). There are three consecutive guanosines at the junction of U3 and R, and Pol II can initiate transcription at all three guanosines, resulting in RNA containing three (3G), two (2G), or one (1G) guanosine at the 5′ end. In cells infected with a lab-adapted molecular clone NL4-3, 3G RNA is the most abundant HIV-1 RNA; however, in virions produced by these cells, 1G RNA is the predominant HIV-1 RNA species (6). Thus, 1G RNA is selected over 3G RNA to be packaged into viral particles, and these two nearly identical RNAs differ functionally in the cell culture system (5–12). Furthermore, HIV-1 mutants with altered transcription initiation patterns that mostly express 3G RNA or 1G RNA exhibit replication fitness defects compared to the wild-type NL4-3 virus (8). Thus, maintaining the heterogeneous usage of transcription start sites is important to HIV-1 replication fitness in cell culture systems (8).

Our current understanding of HIV-1 transcription start site usage and its impact on RNA functions is mainly based on experiments performed in cell culture systems or probing RNA structures using in vitro transcribed RNAs (5–11, 13, 14). However, the environment for the virus in people living with HIV is far more complex than cell culture systems. Infected T cells may be at different stages of differentiation and located in various body compartments. Additionally, host defense systems, including immune response and innate immunity, can exert pressures on HIV-1. It is not known whether the virus uses multiple transcription start sites or packages a specific RNA species in people living with HIV. Here, we sought to bridge this knowledge gap and address these questions. For this purpose, we analyzed peripheral blood mononuclear cells (PBMC) and plasma samples collected from the same study participants and compared HIV-1 transcription start sites in these paired samples. Our results demonstrated that in people living with HIV, the virus uses multiple transcription start sites and preferentially packages 1G RNA.

RESULTS

Determine the amounts of HIV-1 RNA required to study the 5′ end of HIV-1 RNA

We have previously established an accurate and reliable next-generation sequencing (NGS)-based 5′ rapid amplification of cDNA ends (RACE) to study the 5′ end of HIV-1 RNA (8). In this method, a primer that annealed to the gag gene was used to prime cDNA synthesis, thereby ensuring only unspliced HIV-1 RNA species are studied. We have used the NGS-based 5′RACE to study HIV-1 transcription initiation in cell culture systems successfully (8) and have identified multiple HIV-1 species initiated near the U3-R junction (Fig. 1A). In our cell culture studies, we generally use more than 1 × 10⁵ copies of HIV-1 RNA in each reaction to determine the transcription start site usage by NGS-based 5′RACE. However, clinical samples are limited in the amount of HIV-1 RNA compared to cell culture systems. Furthermore, HIV-1-infected cells are minor portions of PBMC, and infected cells may express different levels of HIV-1 RNA. To better assess the sensitivity of our assay, we determined the amount of HIV-1 RNA required to reliably detect various HIV-1 RNA species. For this purpose, we used RNA isolated from T cells infected with NL4-3 and quantified the amounts of unspliced HIV-1 RNA by quantitative real-time PCR using primers located in the gag gene. We used different amounts of HIV-1 RNA and supplemented with RNA isolated from uninfected cells to maintain the total amount of RNA and performed NGS-based 5′RACE. To determine the reproducibility of the assay, we performed three independent 5′RACE experiments (marked as 1, 2, and 3 in Fig. 1B) using the same set of RNA samples. As shown in Fig. 1B, our assay reliably determined HIV-1 transcription start site usage when there were 500 copies of unspliced HIV-1 RNA in the reaction. We have also performed the same dilution experiment without adding the uninfected cellular RNA to mimic conditions in plasma samples and obtained similar results (Fig. 1C). As all biochemical assays have minor experimental variation, we used 1,000 copies of HIV-1 unspliced RNA as a cutoff value and only examined samples containing HIV-1 RNA exceeding this amount.

Fig 1.

Determine the range in which transcription start site usage can be reliably detected in the NGS-based 5′RACE assay. (A)The general structure of a partial HIV-1 provirus and HIV-1 RNA species initiated near the U3-R junction. The DNA sequence of R is shown in blue. For clarity, the first of the three consecutive guanosines is designated as position 1, and position numbers are shown on top of the DNA sequence. (B) HIV-1 RNA species were detected in a dilution experiment in which total RNA is maintained by adding RNA isolated from uninfected cells. (C) HIV-1 RNA species were detected in a dilution experiment in which samples were diluted with water. Proportion of HIV-1 RNA species are indicated in Y axis, and HIV-1 RNA copy numbers are indicated in X-axis; three independent 5′RACE reactions (denoted as 1, 2, and 3) were performed using the same RNA sample.

Analyses of PBMC and plasma samples from people living with HIV

To determine HIV-1 transcription start site usage in vivo, we examined the 5′ ends of HIV-1 RNA in PBMC samples collected from people living with HIV. To determine whether there are preferences of specific HIV-1 RNA species packaged into particles, we compared the 5′ end of the HIV-1 RNA from PBMC and plasma samples from the same donors. All samples were collected when donors were not on anti-HIV-1 treatments; information on study participants is listed in Table 1. Using established procedures, HIV-1 RNA from virions can be isolated from plasma samples (15, 16). We have identified PBMC and plasma samples from three people living with HIV, with each sample containing more than 1,000 copies of HIV-1 RNA, and the paired samples were collected on the same date. Subtype B HIV-1 was identified in all three individuals. In all three PBMC samples, we observed HIV-1 RNA initiated from several transcription start sites, with 3G RNA being the most abundant transcript, followed by 1G RNA (Fig. 2, marked as PBMC). In all three paired plasma samples (Fig. 2, marked as plasma), 1G RNA is the predominant transcript. When compared with the PBMC samples, 1G RNA is enriched in plasma samples, whereas 3G RNA is reduced (Fig. 2). These results are similar to those observed in the cultured cells (6, 8). We have identified a second plasma sample from donor AVBI02-18 that was collected 3 months prior to the paired PBMC and plasma samples; results from this sample (Fig. 2; marked as plasma*) are similar to those of the other plasma samples. In addition to paired PBMC and plasma samples, we have also examined seven plasma samples without paired PBMC samples. In all seven plasma samples, 1G RNA is the predominant RNA species (Fig. 3), providing additional support that 1G RNA is the packaging substrate for HIV-1 in vivo.

TABLE 1.

Characteristics of study participants

Participant identifier	Sample	Date	Age in years	Sex at birth	HIV risk	Prior ART if anya
FMGA-013	PBMC and Plasma	8/6/2001	44	Male	MSM	None
FMGA-023	PBMC and Plasma	1/20/2004	21	Male	MSM	None
AVBIO2 −18	PBMC and Plasma	6/2/2014	56	Male	MSM	Yes
AVBIO2 −18	Plasma	3/7/2014	55
IDFU-166	Plasma	8/18/2003	38	Female	Unknown	None
FMGA-006	Plasma	10/10/2000	40	Male	MSM	None
FMGA-016	Plasma	9/4/2001	31	Male	Bisexual	None
FMGA-018	Plasma	9/5/2002	43	Male	MSM	None
SHRTST-06	Plasma	5/28/2003	62	Male	MSM	Yes
FMGA-022	Plasma	2/27/2004	30	Male	MSM	None
AVBIO-136	Plasma	5/23/2005	37	Male	MSM	Yes

^a Whether participant had received antiretroviral therapy prior to sample collection.

Fig 2.

HIV-1 RNA species detected in PMBC and plasma samples from people living with HIV. Participant identifiers are listed on the top; each set of paired PBMC and plasma samples were collected on the same date. The sample marked as plasma* was collected 3 months prior to the paired PBMC and plasma samples from the same donor (AVBIO2-18). Proportions of HIV-1 RNA species are shown in Y-axis, and the types of samples are labeled in X-axis.

Fig 3.

Distribution of HIV-1 RNA species in plasma samples from people living with HIV. Proportions of HIV-1 RNA species are shown in Y-axis, and the patient identifiers are shown in X-axis.

Analyses of HIV-1 sequences known to regulate transcription start site usage

We have previously shown that HIV-1 transcription start site usage is regulated by the nucleotide identities of the start sites and the distance between three consecutive guanosines and a 6-nt sequence CATATA with overlapping CATA and TATA boxes, referred to as CATA/TATA box (8). The usage of transcription start sites of the three PBMC samples is similar to that of NL4-3 (Fig. 1 and 2), with 1G and 3G RNA being the two major HIV-1 RNAs. We detected 3G, 2G, and 1G RNA in all three PBMC samples, indicating these samples harbor HIV-1 with three consecutive guanosines at the junction of U3 and R. To better understand how HIV-1 regulates transcription start site usage in vivo, we determined the distance between the CATA/TATA box and the three guanosines in HIV-1 from the three PBMC samples. For this purpose, we sequenced a portion of U3-R using RNA isolated from PBMC samples and the consensus sequences obtained are listed in Fig. 4. The U3 region is duplicated by intramolecular plus-strand DNA transfer during reverse transcription (17); thus, in most proviruses, the 5′ and 3′ U3 have the same sequence (18). In our experiments, the 3′ U3-R sequences were analyzed and used to infer the 5′ U3-R sequence. In all three samples, the CATA/TATA box is preserved, and the distance between CATA/TATA box and the three guanosines is the same as NL4-3; furthermore, the sequences between CATA/TATA box and three guanosines are mostly the same. We then analyzed 2,064 subtype B HIV-1 sequences from the Los Alamos database and found that 90% of the sequences have the intact CATA/TATA box, 98% of the sequences have the three consecutive guanosines, and 93% of the sequences have the same distance as NL4-3 between these two elements (Fig. 4). These findings are consistent with the observation that transcription start site usage is important for the replication fitness of HIV-1.

Fig 4.

Sequence logo showing the conservation of sequences near the U3-R junction of subtype B HIV-1. Black underline, the overlapping CATA/TATA box; red underline, three consecutive guanosines at the start of the R region.

DISCUSSIONS

HIV-1 RNA serves as both the virion genome and a translation template. How these two distinct functions are regulated has been a long-standing question in the field. In this report, we demonstrated that in people living with HIV, the virus uses multiple transcription start sites to generate several HIV-1 RNA and preferentially packages 1G RNA similar to that observed in cell culture system. These findings extend our current understanding of HIV-1 replication mechanism from cell culture systems to people living with HIV.

We and others have shown that the 5′ UTRs of 3G and 1G RNA fold into distinct ensembles of conformations and the RNA structures regulate the RNA functions (6, 7). The 1G RNA tends to fold into structures with exposed dimerization signal and Gag binding sites, facilitating its selection as viral genomes (6). Our finding that 1G RNA is selected over 3G RNA as the virion genome in people living with HIV suggests that these RNA structural differences also exist in vivo to allow such selection. We have also observed that based on its 5′ UTR structures, 3G RNA is translated more efficiently in vitro and in cultured cells using a polysome profiling assay (19). Because of assay sensitivity, we could not perform the same assay to examine the translation efficiency of HIV-1 RNA using samples from people living with HIV. Given the similarity between the results from cell culture and in vivo studies described above, it is likely that 3G RNA is preferentially translated in people living with HIV. Current experimental evidence illustrates that 1G and 3G RNA differ functionally, and 5′ UTR structures play an important role in RNA functions. However, the detailed mechanisms of how 1G RNA is selected over 3G RNA to be packaged as virion genomes, or how 3G RNA is translated more efficiently, are not fully understood. Further studies are needed to dissect these mechanisms; such studies will provide insights into the regulation of genome packaging and gene expression, as well as the intricate regulation of HIV-1 through the utilization of heterogenous transcription start sites. We have previously shown that HIV-1 transcription start site usage is regulated by a short region of HIV-1 LTR, from CATATA box in the U3 to the three consecutive guanosines at the beginning of R (8). Our analyses of HIV-1 harbored by three sets of clinical samples and the subtype B sequences in the Los Alamos database showed that three features of this region are highly conserved, the CATA/TATA box, the three consecutive guanosines, and the distance between these two elements. Interestingly, the fifth nt of the CATATA sequence exhibit low levels of variation (10%; Fig. 4, line marked as Database); this mutation destroys the TATA box but leaves the CATA box intact. It has been shown that HIV-1 mainly used the CATA box instead of the TATA box as the core promoter sequence where pretranscription complex assembles (20), providing an explanation for the tolerance of abolishing the TATA box. Our analyses showed that the three consecutive guanosines feature is highly conserved among subtype B viruses, which is consistent with a report indicating this feature is shared among different subtypes of HIV-1 (14). There are also low levels of variations in the distance between CATATA box and three consecutive guanosines, which is known to alter transcription start site usage (8). We have previously identified two founder viruses, CH058 and CH106 (21), with a 1-nt deletion between the CATATA box and the three guanosines that altered transcription start site usage (9). These viruses do not express 3G RNA; two of their major transcripts are the 1G RNA and C5 RNA, which is initiated from position 5 of the R region. We hypothesize that in these viruses, the function of the 3G RNA is replaced by C5 RNA.

Taken together, we have shown that multiple transcription start site usage and preferential packaging of 1G RNA exist in people living with HIV. These two features also exist in many strains of simian immunodeficiency virus (SIV), including those that are closely related to the progenitor viruses transmitted into human population that became HIV-1 Group M, N, and O viruses (9). Therefore, using transcription start sites to regulate RNA functions is a common mechanism for HIV-1 and related viruses. However, not all retroviruses generate multiple unspliced transcripts. For example, an SIV that infects red-capped mangabey (SIV_rcmGAB1) only generates one major unspliced transcript, which is packaged (9). It is worth noting that all retroviral unspliced RNAs need to serve as both the viral genome and as a translation template. Little is known about how retroviruses other than HIV-1 and related viruses regulate unspliced RNA functions. For example, we do not know whether it is common for other retroviruses to use heterogeneous transcription start sites. It has been illustrated that the murine leukemia virus has two pools of unspliced RNA, one for translation and one for packaging (22). However, the mechanism by which these two pools of unspliced RNAs are maintained is not known. Further studies are needed to reveal the mechanism(s) retroviruses use to achieve the intricate balance of having unspliced RNAs serve two distinct functions.

MATERIALS AND METHODS

Nucleic acid isolation and RNA quantification

Methods used to isolate RNA from plasma (16) and PBMC (23) have been described previously. Briefly, plasma virion RNA was isolated by first centrifugation of the sample 21,000 × g for 1 h at 10°C; the resulting pellet was lysed with 100 µL of 3M guanidine hydrochloride containing 100 µg of Proteinase K for 1 h at 42°C. Then, 400 µL of 5.7 M guanidine thiocyanate, 50 mM Tris, pH 7.6, 1 mM EDTA, and 175 µg of glycogen (Roche Life Sciences) were added for further disruption and ease of subsequent nucleic acid precipitation. After incubating the lysate at 25°C for 5 min, nucleic acids were precipitated by the addition of 500 µL 100% isopropanol and collected in a microfuge by centrifugation at 21,000 × g for 10 min at room temperature; the pellet was rinsed with 70% (vol/vol) ethanol and dried (16). PBMC RNA was isolated as described in Simonetti et al. (23). Briefly, PBMC were disrupted in 2 mL SPEX SamplePrep tubes with 1.4 mm acid-washed zirconium beads and 1 mL of Trizol at 4°C in a Precellys 24 (Bertin Technologies) tissue homogenizer, homogenizing twice at 6,000 rpm for 30 s, placed on ice for 5 min, and then homogenized twice more at 6,000 rpm for 30 s. RNA was isolated by adding 0.1 mL 1-bromo-3-chloropropane (Molecular Research Center) to the homogenate, followed by homogenization for 15 s, at 5,000 rpm and centrifugation at 14,000 × g for 15 min at 4°C. The aqueous phase, which contained the RNA, was collected and transferred to a new tube containing 240 ng glycogen. The RNA was precipitated with 0.5 mL of 2-propanol, and the pellet was washed with 70% ethanol and dried. RNA samples were precipitated and dissolved in nuclease-free water and then treated with a TURBO DNA-free Kit (Invitrogen). To generate control RNA from HIV-1 infected cells, CEM-SS cells (24) were infected with NL4-3, and viral replication was monitored by p24 ELISA assay (XpressBio). Cellular RNAs were isolated from cells collected 2 days prior to the p24 peak. RNA from cultured cells was isolated using RNeasy Plus Mini Kit (Qiagen). HIV-1 RNA and GAPDH RNA were quantified using quantitative RT-PCR with iTaq SYBR green one-step RT-PCR kit (Bio-Rad). Previously described primers annealed to HIV-1 gag gene (25) and human GAPDH gene (26) were used to quantify HIV-1 RNA and GAPDH mRNA, respectively.

NGS-based 5′RACE, sequences near HIV-1 U3-R junction, and bioinformatic analyses

The NGS-based 5′RACE and bioinformatic pipeline have been described previously (8).

To determine sequences near the U3 and R junction, RNA isolated from PBMC was converted to cDNA using an oligo dT primer and SuperScript III Reverse Transcriptase (ThermoFisher). The cDNA was used to amplify HIV-1 U3-R junction sequences using a pool of primers with overhang Illumina adapter sequences (Table 2). PCR products were purified using magnetic beads (Beckman Coulter), and index primers were added by PCR (Nextera XT Index kit v2 Set B, Illumina) for multiplexing. Libraries were gel-purified using NucleoSpin Gel and PCR Clean-up Mini kit (Macherey-Nagel) and quantified using Qubit dsDNA HS (High Sensitivity) Assay Kit (ThermoFisher), library qualities were checked using TapeStation (Agilent), and multiplexed libraries were run on Illumina MiSeq using a MiSeq Reagent Kit v2 (300 cycles). Following sequence run, reads were processed for quality control, paired reads were assembled, and analysis was performed using a custom Python script. To account for HIV-1 sequence variation in clinical samples, ~500 sequences of the Miseq reads from each sample were randomly selected and aligned to identify major polymorphic sequences. Bioinformatic pipeline was adjusted based on polymorphism identified to analyze reads. The data set was then assembled using Geneious software to generate a sequence logo for the consensus sequence.

TABLE 2.

Primers to amplify cell-associated HIV-1 RNA for U3-R junction sequence^a

Primer name	Primer sequence
HXB9461S-NGS-F1	TCGTCGGCAGCGTCAGATGTGTATAAGAGACAGNNNGCATCCGGAGTACTTCAAGAAC
HXB9461S-NGS-F2	TCGTCGGCAGCGTCAGATGTGTATAAGAGACAGNNNNNNNGCATCCGGAGTACTTCAAGAAC
HXB9461S-NGS-F3	TCGTCGGCAGCGTCAGATGTGTATAAGAGACAGNNNNNNNNNNNNNNGCATCCGGAGTACTTCAAGAAC
NL9584A-NGS-R1	GTCTCGTGGGCTCGGAGATGTGTATAAGAGACAGNNNTTCCCTAGTTAGCCAGAGAG
NL9584A-NGS-R2	GTCTCGTGGGCTCGGAGATGTGTATAAGAGACAGNNNNNNNNNNTTCCCTAGTTAGCCAGAGAG
NL9584A-NGS-R3	GTCTCGTGGGCTCGGAGATGTGTATAAGAGACAGNNNNNNNNNNNNNTTCCCTAGTTAGCCAGAGAG

^a NGS adaptor PCR primers consist of three parts: gene-specific region (underlined), variable linker (N), and overhang adaptor sequence for Illumina sequencing (boldface).