Single-RNA-level analysis of full-length HIV-1 RNAs reveals functional redundancy of m6As

doi:10.21203/rs.3.rs-2679540/v1

Download PDF

Article

Single-RNA-level analysis of full-length HIV-1 RNAs reveals functional redundancy of m⁶As

https://doi.org/10.21203/rs.3.rs-2679540/v1

This work is licensed under a CC BY 4.0 License

Journal Publication

published 11 Apr, 2024

Read the published version in Nature Microbiology →

Version 1

posted

You are reading this latest preprint version

HIV-1 exploits every aspect of RNA, a versatile macromolecule that undergoes various post-transcriptional modifications, to maximize its replication. Although the importance of chemical modifications on RNA has been recognized, their evolutionary benefits and precise roles in HIV-1 replication remain unclear. Most studies have provided only population-averaged values of modifications for fragmented RNAs at low resolution and have relied on indirect analyses of the phenotypic effects of perturbing host effectors, neglecting site-specificity and intra-RNA heterogeneity^1–9. Here, we developed a new RNA-library-preparation method for full-length direct RNA sequencing and analyzed HIV-1-specific modifications at the single-RNA level. Our analysis revealed that the HIV-1 modification landscape is unexpectedly simple, showing three predominant N6-methyladenosine (m⁶A) modifications near the 3' end. More densely installed in viral mRNAs than in genomic RNAs, these m⁶As play a crucial role in maintaining normal levels of RNA splicing and translation. We also discovered that HIV-1 generates diverse RNA subspecies with distinct ensembles of the m⁶As and that these m⁶As regulate splicing independently of each other. Our single-RNA-level study demonstrates that HIV-1 tolerates functionally redundant m⁶As to provide stability and resilience to viral replication while minimizing the risk of unpredictable mutagenesis – a novel RNA-level strategy similar to bet-hedging in evolutionary biology.

Biological sciences/Microbiology/Virology/Retrovirus

Biological sciences/Biological techniques/Epigenetics analysis/Methylation analysis

Biological sciences/Biotechnology/Sequencing/RNA sequencing

RNAs are highly structured macromolecules with various post-transcriptional modifications, including 3’ polyadenylation, splicing, and chemical modifications. Since the late 1950s, more than 300 types of chemical modifications (epitranscriptomes) have been identified^{10, 11}, adding another layer of complexity to RNA biology. These modifications control a wide range of cellular and viral processes and are associated with more than 100 human diseases^11–13. Studying these modifications, however, has been slow and laborious due to technical limitations inherent in the sequencing of “true” (native) RNAs¹.

HIV-1 has a significantly higher number of chemical modifications on its RNAs than typical cellular transcripts^{6, 14}. However, the evolutionary benefits and HIV-1-specific roles of these modifications in viral replication and various RNA functions remain unclear and sometimes even controversial, showing both pro- and anti-viral effects depending on the virus type, replication stage or tested cell type^{3, 9, 15–18}. Most RNA modification studies to date have relied on indirect analyses of the phenotypic effects of perturbing host effectors (known as writers, erasers, and readers)^1–9, neglecting the potential site-specific and context-dependent roles of chemical modifications^{12, 19–24}. Although studies using short-read sequencing have mapped several common modifications onto the HIV-1 genome, including m⁶A, 5-methylcytosine, 2’-O-methylation and N⁴-acetylcytidine, they have only provided low-resolution and population-average values of modifications of a given type for fragmented RNAs^1–9. The site-specific roles of individual modifications and their ensembles on the same RNA strand remain largely unknown.

Nanopore direct RNA sequencing (DRS) is a powerful tool that can analyze individual strands of native RNAs as they continuously pass through nanopores^25–27. This unique technology allows for a simultaneous evaluation of key features of RNAs at the single-molecule level, including RNA sequences, chemical modifications, splicing isoforms, 3’ polyadenylation, and absolute quantitation and profiling of a heterogeneous pool of RNA transcripts^27–30. DRS is also free from the experimental biases associated with current short-read sequencing methods^{2, 31}. However, DRS faces challenges when analyzing long RNA molecules and RNAs that cause motor enzyme stalls^{25, 32–36} and when analyzing chemical modifications at the single-RNA level. Here, we present several technical innovations, including full-length DRS and read-level binary classification methods, that maximize the potential of the DRS technology for the study of HIV-1 RNA biology. We demonstrate dominant and site-specific m⁶A modifications at three distinct locations on the RNA genome and characterize their functional significance in regulating HIV-1 replication at the individual RNA level.

Multiplex reverse-transcription improves the DRS of full-length HIV-1 RNA.

DRS of long and complex RNA molecules, such as premature transcripts (13–18 Kb), cellular Xist and mitochondrial mRNAs, plant and virus RNAs, has been challenging^{25, 32–36}. In our study, initial conventional DRS procedures failed to generate more than 13 reads (0.01% recovery) of full-length HIV-1 sequences in 8 out of 9 runs (Fig. 1a,b and Extended Data Fig. 1). Given that HIV-1 RNAs are 2–30 times more modified than typical cellular mRNAs^{4, 14} and have complex secondary and tertiary structures^{37, 38} – features known to stall reverse-transcription^{31, 39, 40} (RT, an optional step of DRS known to improve the read throughput^{26, 36}) – we reasoned that inefficient linearization of HIV-1 RNA by RT might be the cause of the failure. To alleviate this, we established a multiplex RT using oligonucleotide primers specific to different parts of the HIV-1 genome to improve full-length DRS (Fig. 1a,b and Extended Data Fig. 1). Under optimized conditions using 111 different primers, we generated a total of 810, 1,797, and 2,655 reads of full-length unspliced (US; ~ 9 Kb), partially spliced (PS; ~ 4 Kb) and completely spliced (CS; ~ 2 Kb) intracellular HIV-1 RNAs, respectively, as well as a total of 3,985 reads of full-length virion RNAs (Supplementary Table 1). The improved DRS with multiplex RT enabled a comprehensive analysis of individual reads of HIV-1 RNAs in virions and in virus-producing cells.

Unexpected simplicity in the modification landscape points to the site-specificity of m⁶As.

To identify modification sites on a whole-genome scale, we used a two-step signal-refinement process that involved the use of in vitro transcribed (IVT) HIV-1 RNAs as a non-modified RNA control (see Supplementary Note 1 for details). With our optimized conditions, the Tombo analysis⁴¹ generated highly reproducible per-read modification signals (p values) in our repeated experiments (Extended Data Fig. 2). The results from Tombo and other tested software tools, including Eligos2⁴², Nanocompore⁴³, and xPore⁴⁴, consistently revealed a small number of prominent modification signals on the 3’ side of the HIV-1 genome (Fig. 1c and Extended Data Fig. 4).

To identify the most likely modification sites, we compared the modification signals generated by three different tools – Tombo, Eligos2, and Nanocompore^41–43, each analyzing different aspects of DRS signals, such as ionic current levels, base-calling error rates, and dwell time⁴⁵. We selected the top 149, 167, and 156 modification signals, respectively, from these tools and cross-compared them (Supplementary Note 1.6). Despite the high reproducibility of all three tools (Extended Data Fig. 4), we identified only seven common peaks, reflecting the variable detection efficiencies when using different DRS signal features⁴⁵. Notably, among the seven peaks, five were located at or adjacent to the known m⁶A motifs (DRACH: D = A/G/U, R = A/G, H = A/C/U). One DRACH peak at position A17 was excluded because we found the signals near the end of the reads to be inherently unstable (see Supplementary Note 1.2.3). The remaining four DRACH sites (A8079, A8110, A8975, and A8989) consistently exhibited strong modification signals across all our tests (Supplementary Note 1) and were highly conserved among HIV-1 subtype B (Fig. 1e). These sites also coincided with the major m⁶A peaks in previous short-read sequencing studies^{6, 9}. Considering 242 DRACH sites present in the HIV-1 genome, the predominant modification signals in these few DRACH sites suggest their strong site-specificity in terms of m⁶A functions or its installation.

Moreover, the 25 common modification peaks detected by both Tombo and Eligos2 were also significantly enriched near the DRACH sites (Fig. 1e) and in the m⁶A-reader binding sites (Extended Data Fig. 5a)^{6, 9}. In contrast, we did not find any notable associations between these common peaks and other modifications, such as the 2’-O-methylation, 5-methylcytosine, or N⁴-acetylcytidine sites^{2, 4, 5, 7, 8}(Extended Data Fig. 5a). Given these modifications are several-fold more frequent or at least as common as m⁶As on the HIV-1 genome^{6, 14}, the results suggest they may be less site-specific than m⁶As. The signals from non-site-specific modifications are likely diluted in this population-based analysis, making them insignificant or undetectable.

DRS-detected dominant m⁶As are confirmed at the single-nucleotide resolution.

To evaluate the existence of the four most probable m⁶As, we first analyzed HIV-1 virion RNA after in vitro treatment with an m⁶A eraser, ALKBH5⁴⁶. All of the four m⁶A sites showed varying levels of signal reduction in the Tombo, Eligos2, Nanom6A⁴⁷ and dwell time⁴⁸ analyses (Fig. 2a and Extended Fig. 5b). Next, we introduced a point mutation to each of the four most probable m⁶A sites (A8079G, A8110G, A8975C, and A8989T) of an HIV-1 provirus plasmid (pNL4-3). All mutants showed a complete absence of modification signals when compared to IVT RNAs with identical mutations (Fig. 2b). Lastly, we confirmed the two prospective m⁶As at positions A8975 and A8989 by oligonucleotide liquid chromatography coupled to tandem mass spectrometry (oligonucleotide LC–MS/MS) (Fig. 2c)⁴⁹. As the significant modification signals on A8079 site were consistent in all our tests, we selected A8079, A8975, and A8989 to investigate their site-specific roles. m⁶A at A8110, however, remains unclear.

Knocking out all three dominant m⁶As, but not individual m⁶A, affects HIV-1 fitness.

The functions of m⁶A modifications are believed to be determined by host effectors that catalyze, recognize and remove such modifications (known as writers, readers and erasers, respectively)¹². Mounting evidence also suggest that these modifications have site-specific and context-dependent roles, controlling local RNA-protein interactions by modulating the RNA structures where the interactions occur^{12, 19–24}. m⁶As play important roles in regulating various aspects of RNA biology, including RNA structure, splicing, translation, metabolism, and translocation within cells¹² and promote HIV-1 replication in general^{3, 6, 8, 9, 17}. However, our current understanding is primarily based on indirect analyses of the phenotypic effects of perturbing m⁶A writers, readers, or erasers in host cells, which overlook the potential site-specific roles of the modifications. Some findings remain controversial, showing inconsistent results depending on the replication stages, cell types, and assays used in the studies^{3, 9, 15–18}.

To directly analyze the functions of m⁶As on HIV-1 RNA, we generated m⁶A-knockout viruses using site-directed mutagenesis and evaluated key steps of viral replication in WT host cells (Fig. 3a). Although the m⁶As were effectively removed (Fig. 2b), none of the single mutations resulted in significant reductions in any of the tested replication steps, including total HIV-1 RNA production, US RNA production, viral protein expression (Gag, Vif and envelope gp41), virion production (extracellular p24), and infection of reporter cells (Fig. 3 and Extended Data Fig. 6). However, the triple mutation of all three m⁶A sites significantly reduced US RNA levels (Fig. 3d). It is known that the loss or reduction of US RNA results in a drastic reduction in viral fitness⁵⁰ because US RNAs are essential for producing the structural proteins (Gag/Gag-Pol) and genomic RNA. As expected, all subsequent steps, including p24 Gag production, virion release, and viral infectivity, were also significantly reduced (Fig. 3b,e,f).

The triple m⁶A mutation induces over-splicing of HIV-1 RNA, lowering US RNA levels.

Given the critical importance of RNA splicing in HIV-1 replication, particularly in controlling US RNA levels⁵⁰, we investigated the roles of the three m⁶As in HIV-1 alternative splicing (Fig. 4 and Extended Data Fig. 7). HIV-1 produces over 50 different forms of spliced RNA, an extraordinarily high level of alternative splicing^{17, 51}. While DRS can effectively disentangle complex RNA isoforms^{28, 32, 36, 52}, analysis of HIV-1 RNAs has remained impractical due to poor full-length sequencing. Here, using the new multiplex RT method, we were able to reproducibly generate full-length reads of approximately 2 Kb of CS, 4 Kb of PS, and 9 Kb of US RNA, with recovery rates of 54.1%, 31.5% and 34.9%, respectively, (Fig. 4a,b). We successfully assigned 94.8% of these full-length reads to 186 exon combinations, including 53 major isoforms⁵³, without any notable ambiguity (Fig. 4a). The read counts were generally consistent with the densitometric quantification of PCR amplicons of the CS and PS isoforms (Fig. 4c and Supplementary Table 2).

Regarding total HIV-1 RNA production, we observed no significant differences between the WT HIV-1 and triple-mutants (Fig. 4d-i). As expected from the molecular biology tests described above, the fraction of US RNA was significantly lower in the triple mutants than in the WT (Fig. 4d-ii). Since cells rarely tolerate unspliced or incompletely spliced transcripts, HIV-1 must heavily suppress its RNA splicing in order to maintain sufficient levels of US RNA⁵⁰. However, triple mutants showed a significantly increased usage of D1 donor (Fig. 4d-iii, which occurs in all spliced RNA), A7 acceptor (Fig. 4d-iv, which occurs for all CS), and all other donors and acceptors (Fig. 4d-v). Consequently, the “over-splicing” by the triple mutants significantly reduced US RNAs while relatively increasing the CS portion (Fig. 4d-vi). Single-mutant viruses also showed an increase in CS RNA but maintained a considerably higher level of US RNA than did the triple mutants (Fig. 4d-vi).

Triple mutations reduced HIV-1 protein translation without affecting poly(A) tail lengths.

In addition to its role in RNA splicing, m⁶A has been associated with RNA translation, metabolism, and 3’ polyadenylation^{3, 54}. Both viral Vif and envelope proteins are mainly translated from PS RNA. Although the PS RNA levels (Fig. 4d-vi) and mRNAs for Vif and envelope (Fig. 4e) were maintained at relatively similar levels among cells producing any mutants, intracellular Vif and envelope gp41 proteins were significantly reduced by triple mutations, but not by any of the single mutations (see Fig. 3c), indicating that inefficient translation of Vif and envelope mRNAs by triple mutants compared to those by single mutants or WT. The length of 3’ poly(A) tails is known to have significant implications for RNA translation and metabolism³⁰. Our analysis confirmed that there were no notable differences in poly(A) tail lengths between the WT and mutant RNA isoforms in this regard (Fig. 4f and Extended Data Fig. 8). These results support the regulatory roles of these m⁶As in viral RNA translation.

Read-level binary classification identifies RNA subspecies with distinct m⁶As.

To investigate the functions of the three m⁶As at the single-RNA-molecule level, it is crucial to determine the presence of m⁶As accurately and without bias for each read and at different sites. We have developed new read-level binary classification methods that are specific to each of the three m⁶As (m⁶Arp models; see Supplementary Note 2 for details). These methods are based on the consistent read-level p-values surrounding these sites (Fig. 5a-ii and Extended Data Fig. 9e). The generation of per-read p-values is highly reproducible under our optimized conditions (r² > 0.999 with > 25K IVT reads), and the p-values remained consistent in repeated experiments (Extended Data Fig. 2c,d).

Our pre-trained m⁶Arp models showed an area under the Receiver Operating Characteristics curve (AUROC) ranging from 0.95 to 0.97, with false positive rates of 8.40–10.80% and false negative rates of 8.90–12.40% for the three m⁶As (Supplementary Table 3). Our models outperformed Nanom6A⁴⁷ and m6Anet⁵⁵, which are k-mer-based methods optimized for whole transcriptome analysis (Fig. 5a-iii). Moreover, the performance of our models, which evaluate m⁶A presence one read at a time, is unaffected by the number of reads or the data composition of test samples (i.e., data sparsity and imbalance problems)^{44, 45, 55, 56}. These features of our models enabled us to accurately determine RNA subspecies with distinct ensembles of the three m⁶As (subspecies A-H) (Fig. 5a-iv), and to compare these RNA subspecies in various settings.

The three m⁶As are more densely present on HIV-1 mRNAs than on genomic RNA.

Our models also demonstrated a strong linearity of quantification (r² > 0.9982) (Extended Data Fig. 9g). Consistent with a recent report demonstrating reduced m⁶A levels in the genomic RNAs¹⁵, we found the stoichiometry of the three m⁶As were significantly higher on translating mRNAs (CS and PS RNA) than that on genomic (virion) RNA (Fig. 5b). The estimates from other tools, including Nanom6A, m6Anet and Tombo, were consistent with our findings (Extended Data Fig. 9h). These m⁶As were most frequently detected on CS, showing average 88.5% (±1.8 SD), 78.7% (±3.6 SD) and 47.2% (±3.6 SD) of m⁶A modifications at A8079, A8975, and A8989, respectively.

Interestingly, read-level analyses of RNA subspecies revealed that virtually all CS and PS reads had at least one of these m⁶As (subspecies A-G, collectively accounting for 97.5% and 96.3% of all CS and PS reads, respectively), while the fraction dropped to 82.1% in virion RNA (Fig. 5c). Moreover, RNA subspecies with multiple m⁶As (subspecies A-D) accounted for a predominant portion in the CS and PS RNAs (80.7% and 76.4%, respectively), whereas the portion was substantially lower in the virion RNAs (47.2%). US RNAs (consisting of both translating mRNA and genomic RNA types^{57, 58}; see Fig. 4b) showed a mixed character of mRNAs and virion RNAs as expected. The group H (lacking the three m⁶As) was mostly not spliced and highly enriched in genomic RNAs. These results, therefore, further support the important roles of these m⁶As in splicing and translation. Having a relatively lower number of m⁶As on the genomic RNA may be favored during the virion packaging and during viral RT where m⁶As are reported to be inhibitory^{9, 15}.

Given the genetic and structural differences between HIV-1 mRNAs and genomic RNAs^{57, 58}, the differential levels of the three m⁶As between these two types of RNAs may be controlled co-transcriptionally^{59, 60}. A recent report also suggests that m⁶As on the genomic RNA are selectively demethylated by a Gag-FTO complex prior to virion packaging¹⁵. While the exact modes of action remaining unclear, the fine-tuning of these m⁶As by HIV-1 emphasizes the critical importance of these m⁶As in viral replication.

Redundant roles of each of the three m⁶As in regulating RNA splicing.

We analyzed splicing patterns of these RNA subspecies to evaluate the roles of each of these three m⁶As and their ensembles. Consistent with splicing patterns on a total population scale, all WT subspecies showed substantially lower donor (e.g. D1, D4, A5) and acceptor (e.g. A7) usages than the triple mutants (Fig. 6a), pointing to the suppressive roles of these m⁶As. Among the subspecies A-G, however, there were only moderate differences in splicing patterns and donor or acceptor usages. Having at least one of the three m⁶As, regardless of the position or the number of m⁶As installed, was sufficient for these RNAs to control splicing events and produce all major splicing isoforms (Fig. 6a).

Having multiple of these m⁶As on individual RNAs minimizes the risk of mutational loss.

Given the functional redundancy of these m⁶As, we then asked why HIV-1 maintains excessive m⁶As on its RNAs. First, having multiple m⁶As may be necessary to sustain normal levels of viral replication. In such a scenario, our single mutations would show either a selection for RNAs with multiple m⁶As or an increase in m⁶A stoichiometry at intact DRACH sites to compensate for the loss of a dominant m⁶A. However, we found indistinguishable or only moderate differences in m⁶A stoichiometry (Fig. 6b,c) and m⁶A landscape (Extended Data Fig. 10a) between the single mutants and the WT HIV-1. Like WT RNAs, all RNA subspecies of single mutants, regardless of m⁶A status, showed no significant bias toward the utilization of certain splicing donors or acceptors (Extended Data Fig. 10b-d).

Given that there is no clear evidence of adaptive selection or changes in m⁶A functions, a more favorable scenario may be that HIV-1 spreads the risk of losing m⁶A functions⁶¹. HIV-1 may tolerate these multiple redundant m⁶As to minimize the risk of unpredictable random mutagenesis (10^− 5 to 10^− 3 mutations/bp/cycle for HIV-1⁶²) that may knockout a major m⁶As, analogous to bet-hedging in evolutionary biology reducing risks against unpredictably fluctuating environments⁶¹. Interestingly, all the single mutants maintained at least one of the three m⁶As in most of their CS (91.5–95.8%) and PS (89.1%-95.0%) RNAs, levels comparable to the WT HIV-1 (Fig. 6d). Despite causing substantial loss of m⁶A at the population level, single mutations had only a marginal effect on overall viral fitness. The loss of all three, however, eroded the potential of RNA communities to sustain their control over splicing and translation, and adversely affected the various stages of the HIV-1 life cycle (Fig. 3).

Perspective

In this study, we made significant strides in understanding the HIV-1 epitranscriptome through technological innovations enabling a full-length and individual-RNA-level analysis of long and complex RNAs. Our analysis revealed that HIV-1 maintains multiple functionally redundant m⁶As at a few positions near the 3’ end of its RNAs to control splicing and translation. Given the high mutation rates of HIV-1⁶², this strategy may be necessary to minimize the risk of losing the crucial m⁶As during viral replication. Moreover, we discovered that the three m⁶As are significantly more densely installed in translating mRNAs than in genomic RNAs. HIV-1 may need to fine-tune m⁶A levels on genomic RNAs to reduce the potential inhibitory effects of the m⁶As during virion packaging and reverse transcription^{9, 15} while maintaining a certain level of m⁶As to evade innate immune responses in the new host^{63, 64}.

We show here that a full-length, single-molecule-level analysis using DRS can provide new opportunities to untangle the complexity of the RNA biology. With the full-length DRS data, we demonstrated a systemic and comprehensive analysis of key features of complex HIV-1 RNA isotypes, including epitranscriptomic modifications, alternative splicing and poly(A) tails from the same molecules, otherwise impossible with conventional assays. Our data reveal remarkable epitranscriptomic variations among HIV-1 RNAs, which adds another layer of complexity to HIV-1’s already diverse genetics and structural characteristics^{17, 37, 50, 51, 65}. Our discoveries, in turn, generate new questions about the functions and roles of HIV-1’s versatile RNA molecules. The new methods and analytical standards presented herein, can serve as a useful reference for future investigations of RNAs of interest and various RNA viruses in the ever-expanding RNA virosphere⁶⁶.

Extraction of HIV-1 virion RNA from pNL4-3 transfected 293T cells:

293T cells were transfected with the HIV-1 proviral DNA construct pNL4-3 by polyethylenimine (PEI) as described⁶⁷. The cell culture medium was exchanged with fresh medium at 6 h post-transfection and the supernatant was harvested at 72 h for virion RNA extraction and p24 particle release measurement by the HIV-1 p24 ELISA kit (Abcam). Total RNA from the cells was extracted by TRI® reagent(Sigma-Aldrich) following the manufacturer’s instructions. Total viral particles were concentrated and centrifuged for 1 h 40 min at 28,000 g at 4°C with a 10% sucrose gradient. After discarding the supernatant, the pellet was resolved using 160 µL 1X HBSS, followed by DNase treatment for 30 min at 37°C. The sample was incubated in 1 mL TRIzol for 5 min, followed by the addition of 200 µL chloroform and shaking for 30 seconds. The tube was then centrifuged for 10 min at 12,000 g at 4°C. The clear upper aqueous layer, which contains RNA, was transferred to a new 1.5 mL tube, and 0.7 mL of isopropanol was added. After 10 min incubation at room temperature, the tube was centrifuged for 10 min at 12,000 g at 4°C. The supernatant was discarded, and the then pellets were resuspended in 1 mL of 70% ethanol, followed by another centrifugation for 5 min at 7,500 g. The RNA pellet was air-dried in the hood and then resuspended in 30 µL of DEPC-treated water.

Generation of in vitro transcribed (IVT) RNA from pNL4-3 plasmid:

The pNL4-3 plasmid (100 ng) was used for in vitro transcription template generation by PCR (Platinum™ PCR Supermix High Fidelity, Invitrogen) with each pNL4-3-specific PCR primer set followed by agarose gel purification and additional PCR purification. In vitro transcription (RiboMax Large Scale RNA production systems, Promega) and RNA purification (RNAClean XP beads, Beckman Coulter) were followed. The PCR primer sequences are listed in Supplementary Notes 1 and 2.

Nanopore direct RNA sequencing (DRS) of full-length HIV-1 RNA by multiplex RT:

For HIV-1 RNA DRS, 1 ug of virion RNA or 10 ug of total cellular RNA in 9 µL was used for DRS library preparation following the manufacturer’s protocol (the Oxford Nanopore direct RNA sequencing, SQK-RNA002) with a modification of the RT step. Mixtures of 111 HIV-1 sequence-specific DNA oligomers at a copy number ratio of 1:30 (HIV-1 RNA : each oligomer) were annealed to the RNA at 65°C for 5 min and proceeded to the RTA ligation step. The DNA oligomers used are listed in Supplementary Table 5. The library was run on FLO-MIN106D flow cell for 48 h on a MinION device (Oxford Nanopore Technologies). In vitro transcribed RNA fragments were sequenced the same way except 4 µg RNA was used for the input.

In vitro demethylation of m⁶A on HIV-1 RNA:

Recombinant ALKBH5 (Active Motif) was used for in vitro treatment of HIV-1 virion RNA as described⁶⁸. The reaction mixture contained KCl (100 mM), MgCl₂ (2 mM), RnaseOUT (Invitrogen), L-ascorbic acid (2 mM), α-ketoglutarate (300 µM), (NH₄)₂Fe(SO₄)₂·6H₂O (150 µM), and 50 mM of HEPES buffer (pH 6.5). The mixture was incubated for 1.5 h at room temperature and then stopped by the addition of 5mM EDTA.

m⁶A dot immunoblotting:

The extracted HIV-1 virion RNA was directly used for dot-blot assays, as previously described^{63, 67}. Briefly, 50 ng of virion RNA, diluted in 1mM EDTA (total 100uL), were mixed with 60 µL of 2X SSC buffer (Invitrogen) and 40 µL of 37% formaldehyde (Invitrogen). The mixture was incubated at 65°C for 30 min. The nitrocellulose membrane (Bio-Rad) and nylon membrane (Roche) were both soaked with 10X SSC for 5 min before loading the RNA samples. Samples were loaded equally on nitrocellulose and nylon membrane followed by washing with 10X SSC buffer. The nylon membrane was washed with 1X TBST (25 mM Tris, 0.15 M NaCl, 0.05% Tween 20) and stained with methylene blue for shaking 2–5 sec and washed with ddH2O. The nitrocellulose membrane was UV cross-linked and then blocked with 5% milk in 1X TBST 1 h. m⁶A levels were detected by using an m⁶A-specific antibody (Abcam). Images were analyzed by ImageJ software, and the relative RNA m⁶A levels were normalized to methylene blue staining.

LC-MS/MS sample preparation:

The oligo mixture, including a biotin-labeled target-specific DNA oligomer (1:100) and oligos covering other sites(1:30), was annealed to HIV-1 virion RNA at 65°C for 5 min and then cooled down to room temperature. Samples were digested by nuclease S1 (Invitrogen) for 2 h at room temperature followed by phenol:chloroform purification as previously described⁶⁹. The biotin-labeled target DNA:RNA duplex was recovered by using Dynabeads™ MyOne™ Streptavidin C1(Thermofisher Scientific) following the manufacturer’s instructions. For RNase T1 digestion, about 200 ng of denatured (95^oC for 2 min and snap cooling at 4^oC) HIV-1 RNA (obtained by modified RNase protection assay) was digested with 50 units of RNase T1 (Worthington) at 37^oC for 2 h and dried in speedvac system (Thermofisher Scientific).

LC-MS/MS:

The LC-MS/MS analysis was performed using a BEH C18 column (1.7 µm, 0.3x150 mm, Waters) with Ultimate 3000 UHPLC (Thermo Scientific, Ultra High-Performance Liquid Chromatography) coupled to the Synapt G2-S (Waters) mass spectrometer. The gradient chromatography was performed at 5 µl min^− 1 flow rate using Mobile phase A (8 mM TEA and 200 mM HFIP, pH = 7.8 in water) and mobile phase B (8 mM TEA and 200 mM HFIP in 50% methanol) at 60°C. The gradient consists of an initial hold at 3% B for sample loading, followed by ramping to 55% B in 70 min, 99% in 2 min with 5 min hold before re-equilibration (30 min) at 3% B for initial conditions. The resolved digestion products in the chromatographic eluent were detected in negative ion mode through electrospray ionization (ESI) on a Synapt G2-S (Quadrupole Time-of-Flight, Q-TOF) mass spectrometer operating in sensitivity mode (V-mode). The ESI conditions included 2.2 kV at capillary, 30 V at sample cone while maintaining source and desolvation temperatures at 120 ^oC and 400 ^oC, and gas flow rates at 3 and 600 L h^− 1, respectively. A scan range of 545–2000 m/z (0.5 sec) and 250–2000 m/z (1 sec) was used for first (MS) and second (MS/MS) stage data acquisition. The top three most abundant ions in the first stage were selected for fragmentation for MS/MS using m/z dependent collision energy profile (20–23 V at m/z 545; 51–57 V at m/z 2000) before excluding them for 60 sec using the dynamic exclusion feature.

LC-MS/MS data processing:

The m/z values of the RNase T1 digestion products (and their fragment ions) of a 40-base long HIV-1 RNA sequence were predicted using Mongo Oligo mass calculator. Manual identification and assignment of m⁶A modification was made by scoring for ~ 14 Da mass shift of the theoretically expected oligonucleotides (following cleavage at the 3’end of guanosine) in the modified RNA compared to the unmodified version. A set of controls was used to assign the m⁶A modification at positions 8975 and 8989, respectively.

Site-directed mutagenesis:

gBlocks (Integrated DNA Technologies) with single or combination mutations of m⁶A sites were introduced to the HIV-1 vector pNL4-3 for the mutant plasmids. For each mutant plasmid, 500 ng pNL4-3 and 100 ng gBlocks were digested with NcoI-HF and BamHI-HF and ligated for 30 minutes at room temperature using T4 quick ligation (NEB). The sequences of the mutant plasmids were confirmed by Sanger sequencing.

Digital Quantitative PCR for total HIV-1 RNA:

Viral RNA production was measured by RT-PCR and DRS analysis. An equal amount of total cellular RNA for each sample was used for reverse transcription and cDNA generation. The QuantStudio™ 3D Digital PCR System (Applied Biosystems™) was used with appropriate consumables provided by the manufacturer, including the QuantStudio™ 3D Digital PCR Master Mix v2 and TaqMan™ 5’-6 FAM probe (Applied Biosystems™). The primers used are listed in Supplement Table 5.

Western blot analysis:

Harvested cells were lysed in RIPA 1X buffer (Abcam) with a protease inhibitor cocktail (Sigma-Aldrich) and incubated for 30 mins on ice. The cell lysates were centrifuged at 12,000 xg for 15 mins at 4°C and the supernatant was transferred to a fresh tube and mixed with an equal volume of Laemmli 2x buffer (Bio-rad). The mixture was then incubated at 95°C for 10 mins. The proteins were separated on SDS polyacrylamide gels and then transferred to a nitrocellulose membrane (Bio-rad). The membranes were washed in 1X PBST(10 mM sodium phosphate, 0.15 M NaCl, 0.05% Tween 20) and blocked in 5% milk in 1X PBST for 1 h. Primary and secondary antibodies were diluted at 1:1000 and 1:5000, respectively, each in 5% milk in 1X PBST. The signals were visualized by chemiluminescence. The primary antibodies used were HIV-1 p24 (AIDS Reagent Program), gp41 (AIDS Reagent Program), vif (AIDS Reagent Program), GAPDH (Abcam), and anti-mouse (Promega) for the secondary antibody.

HIV-1 infectivity measurement:

An equal amount of virus stock (pg) was used to infect GHOST cells in 6-well plates⁷⁰. After 48 h post-infection, the cells were washed with PBS and fixed. The GFP expressions for all samples were acquired by the Attune™ NxT flow cytometer (Thermo Fisher Scientific) and analyzed by the FlowJo software (BD Biosciences).

Data preprocessing and quantification of HIV-1 viral RNA and cellular RNA:

Multi-fast5 reads were base-called by guppy (version 3.2.8 or higher) using the fast base calling option. The base-called multi-fast5 reads were then converted to single-read fast5s using the Oxford Nanopore Technologies API, ont_fast5 (v3.3.0). Fastqs were aligned to the HIV-1 genome reference sequence AF324493.2 from the NCBI or the human reference sequence (Human genome assembly GRCh38.p13) with the options “-ax map-ont” using minimap2 (v2.24). Unmapped reads were discarded using SAMtools (v.1.6). For in-depth analysis of the 3’-end region, short reads (read length < 2,000) were filtered out using NanoFilt (v.2.7.1).The sequence read length was extracted by aligning sequences against the HIV-1 transcriptome reference using minimap2 (v.2.24), retaining multiple secondary alignments (parameters -p 0 -N 10) and counting the number of unique read IDs among mapped alignments. Reads were required to be over 8,000 nucleotides to be classified as full-length HIV-1 gRNA. Sequencing read coverage depth was calculated using bedtools genomcov of version 2.25.0, and visualized for 1-nt binning size in the plots.

RNA modification site calling:

Various tools were used to detect native modifications in viral RNA datasets (see Supplemental Note 1). Default options were used for all software used in this study, except where noted. Dwell time was extracted for per-read and per-position levels using the Tombo Python API.

Nucleotide sequence conservation near the major m⁶A sites:

The HIV-1 B subtype sequences corresponding to A8079, A8110, A8975 and A8989 of the NL4-3 strain (RNA) were extracted from the HIV sequence database (https://www.hiv.lanl.gov/). The sequence logo plots were generated using the ggseqlogo R package.

HIV-1 RNA splicing analysis:

Reads were aligned against the HIV genome AF324493.2 from the NCBI in spliced mapping mode using a kmer size of 14. Unmapped reads, alignments with a quality lower than 30, and alignments were discarded using SAMtools (v.1.6). Reads were screened for potential splicing donor (SD) and splicing acceptor (SA) sites by identifying exon start/end positions from BED files. SD and SA sites considered in this study are presented in Fig. 4a. Reads presenting the same combination of splice junction and exotic sequences were grouped together and counted. If the positions of potential SD/SA sites were not in Fig. 4a, the reads were annotated as unclassified. Annotation was performed according to the convention established in⁵³ or, for potential spliced isoforms, to the open reading frame encountered in the read. The relative levels of spliced viral isoforms were quantified using absolute counts compared to wild type. Splicing donor and splicing acceptor usage were calculated using absolute counts after screening of known SD/SA sites from BED files. The reads were classified as spliced when their splice junction include either D1 or D1c splicing donor site. If they have an A7 splicing acceptor site, they were further classified into completely spliced (CS). Then, the rest were classified into partially spliced (PS). The poly(A) tail length of each classified group was estimated using the Nanopolish (v.0.14.0) polya module with default parameters.

IVT read depth analysis:

Six sets of IVT data were generated by randomly choosing 0.5K, 1K, 5K, 10K, 20K, 30K, 40K reads from a pool of a total 50k half-length IVT reads. Each of these IVT sets was tested as a canonical control for Tombo analysis (Fishers 0). 384 reads of HIV-1 virion RNA were used as a sample dataset. The averaged p-values for each position were tested with nonparametric pairwise comparison using the ggstatsplot R package. A p value < 0.05 was considered significant.

Statistical analysis

All statistical details are listed in the figure legends. All averaged data include error bars that denote standard deviation, with single data points shown. A p-value < 0.05 was considered significant.

Alfonzo, J.D. et al. A call for direct sequencing of full-length RNAs to identify all modifications. Nat Genet 53, 1113-1116 (2021).
Zhang, Y., Lu, L. & Li, X. Detection technologies for RNA modifications. Exp Mol Med 54, 1601-1616 (2022).
Baquero-Perez, B., Geers, D. & Diez, J. From A to m(6)A: The Emerging Viral Epitranscriptome. Viruses 13 (2021).
Courtney, D.G. et al. Epitranscriptomic Addition of m(5)C to HIV-1 Transcripts Regulates Viral Gene Expression. Cell Host Microbe 26, 217-227 e216 (2019).
Tsai, K. et al. Acetylation of Cytidine Residues Boosts HIV-1 Gene Expression by Increasing Viral RNA Stability. Cell Host Microbe 28, 306-312 e306 (2020).
Kennedy, E.M. et al. Posttranscriptional m(6)A Editing of HIV-1 mRNAs Enhances Viral Gene Expression. Cell Host Microbe 19, 675-685 (2016).
Ringeard, M., Marchand, V., Decroly, E., Motorin, Y. & Bennasser, Y. FTSJ3 is an RNA 2'-O-methyltransferase recruited by HIV to avoid innate immune sensing. Nature 565, 500-504 (2019).
Lichinchi, G. et al. Dynamics of the human and viral m(6)A RNA methylomes during HIV-1 infection of T cells. Nat Microbiol 1, 16011 (2016).
Tirumuru, N. et al. N(6)-methyladenosine of HIV-1 RNA regulates viral infection and HIV-1 Gag protein expression. Elife 5 (2016).
Davis, F.F. & Allen, F.W. Ribonucleic acids from yeast which contain a fifth nucleotide. J Biol Chem 227, 907-915 (1957).
Boccaletto, P. et al. MODOMICS: a database of RNA modification pathways. 2021 update. Nucleic Acids Res 50, D231-D235 (2022).
Jiang, X. et al. The role of m6A modification in the biological functions and diseases. Signal Transduct Target Ther 6, 74 (2021).
Abebe, J.S., Verstraten, R. & Depledge, D.P. Nanopore-Based Detection of Viral RNA Modifications. mBio 13, e0370221 (2022).
McIntyre, W. et al. Positive-sense RNA viruses reveal the complexity and dynamics of the cellular and viral epitranscriptomes during infection. Nucleic Acids Res 46, 5776-5791 (2018).
Pereira-Montecinos, C. et al. Epitranscriptomic regulation of HIV-1 full-length RNA packaging. Nucleic Acids Res 50, 2302-2318 (2022).
N'Da Konan, S. et al. YTHDC1 regulates distinct post-integration steps of HIV-1 replication and is important for viral infectivity. Retrovirology 19, 4 (2022).
Tsai, K. et al. Epitranscriptomic addition of m(6)A regulates HIV-1 RNA stability and alternative splicing. Genes Dev 35, 992-1004 (2021).
Lu, W. et al. N(6)-Methyladenosine-binding proteins suppress HIV-1 infectivity and viral production. J Biol Chem 293, 12992-13005 (2018).
Shi, H., Wei, J. & He, C. Where, When, and How: Context-Dependent Functions of RNA Methylation Writers, Readers, and Erasers. Mol Cell 74, 640-650 (2019).
Liu, N. et al. N(6)-methyladenosine-dependent RNA structural switches regulate RNA-protein interactions. Nature 518, 560-564 (2015).
Konig, J. et al. iCLIP reveals the function of hnRNP particles in splicing at individual nucleotide resolution. Nat Struct Mol Biol 17, 909-915 (2010).
Llibre, J.M. et al. Genetic barrier to resistance for dolutegravir. AIDS Rev 17, 56-64 (2015).
Sun, L. et al. RNA structure maps across mammalian cellular compartments. Nat Struct Mol Biol 26, 322-330 (2019).
Wu, B. et al. Molecular basis for the specific and multivariant recognitions of RNA substrates by human hnRNP A2/B1. Nat Commun 9, 420 (2018).
Jain, M., Abu-Shumays, R., Olsen, H.E. & Akeson, M. Advances in nanopore direct RNA sequencing. Nat Methods 19, 1160-1164 (2022).
Garalde, D.R. et al. Highly parallel direct RNA sequencing on an array of nanopores. Nat Methods 15, 201-206 (2018).
Foord, C. et al. The variables on RNA molecules: concert or cacophony? Answers in long-read sequencing. Nat Methods 20, 20-24 (2023).
Gleeson, J. et al. Accurate expression quantification from nanopore direct RNA sequencing with NanoCount. Nucleic Acids Res 50, e19 (2022).
Subramanian, A. et al. Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc Natl Acad Sci U S A 102, 15545-15550 (2005).
Passmore, L.A. & Coller, J. Roles of mRNA poly(A) tails in regulation of eukaryotic gene expression. Nat Rev Mol Cell Biol 23, 93-106 (2022).
Helm, M. & Motorin, Y. Detecting RNA modifications in the epitranscriptome: predict and validate. Nat Rev Genet 18, 275-291 (2017).
Li, R. et al. Direct full-length RNA sequencing reveals unexpected transcriptome complexity during Caenorhabditis elegans development. Genome Res 30, 287-298 (2020).
Soneson, C. et al. A comprehensive examination of Nanopore native RNA sequencing for characterization of complex transcriptomes. Nat Commun 10, 3359 (2019).
Kim, D. et al. The Architecture of SARS-CoV-2 Transcriptome. Cell 181, 914-921 e910 (2020).
Zhang, S. et al. New insights into Arabidopsis transcriptome complexity revealed by direct sequencing of native RNAs. Nucleic Acids Res 48, 7700-7711 (2020).
Workman, R.E. et al. Nanopore native RNA sequencing of a human poly(A) transcriptome. Nat Methods 16, 1297-1305 (2019).
Tomezsko, P.J. et al. Determination of RNA structural diversity and its role in HIV-1 RNA splicing. Nature 582, 438-442 (2020).
Watts, J.M. et al. Architecture and secondary structure of an entire HIV-1 RNA genome. Nature 460, 711-716 (2009).
Guo, L.T., Olson, S., Patel, S., Graveley, B.R. & Pyle, A.M. Direct tracking of reverse-transcriptase speed and template sensitivity: implications for sequencing and analysis of long RNA molecules. Nucleic Acids Res (2022).
Kwok, C.K. & Balasubramanian, S. Targeted Detection of G-Quadruplexes in Cellular RNAs. Angew Chem Int Ed Engl 54, 6751-6754 (2015).
Stoiber, M. De novo Identification of DNA Modifications Enabled by Genome-Guided Nanopore Signal Processing. . bioRxiv (2017).
Jenjaroenpun, P. et al. Decoding the epitranscriptional landscape from native RNA sequences. Nucleic Acids Res 49, e7 (2021).
Leger, A. et al. RNA modifications detection by comparative Nanopore direct RNA sequencing. Nat Commun 12, 7198 (2021).
Pratanwanich, P.N. et al. Identification of differential RNA modifications from nanopore direct RNA sequencing with xPore. Nat Biotechnol 39, 1394-1402 (2021).
Begik, O. et al. Quantitative profiling of pseudouridylation dynamics in native RNAs with nanopore sequencing. Nat Biotechnol 39, 1278-1291 (2021).
Xu, C. et al. Structures of human ALKBH5 demethylase reveal a unique binding mode for specific single-stranded N6-methyladenosine RNA demethylation. J Biol Chem 289, 17299-17311 (2014).
Gao, Y. et al. Quantitative profiling of N(6)-methyladenosine at single-base resolution in stem-differentiating xylem of Populus trichocarpa using Nanopore direct RNA sequencing. Genome Biol 22, 22 (2021).
Stephenson, W. et al. Direct detection of RNA modifications and structure using single-molecule nanopore sequencing. Cell Genom 2 (2022).
Thakur, P., Estevez, M., Lobue, P.A., Limbach, P.A. & Addepalli, B. Improved RNA modification mapping of cellular non-coding RNAs using C- and U-specific RNases. Analyst 145, 816-827 (2020).
Emery, A. & Swanstrom, R. HIV-1: To Splice or Not to Splice, That Is the Question. Viruses 13 (2021).
Stoltzfus, C.M. Chapter 1. Regulation of HIV-1 alternative RNA splicing and its role in virus replication. Adv Virus Res 74, 1-40 (2009).
Depledge, D.P. et al. Direct RNA sequencing on nanopore arrays redefines the transcriptional complexity of a viral pathogen. Nat Commun 10, 754 (2019).
Nguyen Quang, N. et al. Dynamic nanopore long-read sequencing analysis of HIV-1 splicing events during the early steps of infection. Retrovirology 17, 25 (2020).
Slobodin, B. et al. Transcription Dynamics Regulate Poly(A) Tails and Expression of the RNA Degradation Machinery to Balance mRNA Levels. Mol Cell 78, 434-444 e435 (2020).
Hendra, C. et al. Detection of m6A from direct RNA sequencing using a multiple instance learning framework. Nat Methods (2022).
Thabtah, F., Hammoud, S., Kamalov, F. & Gonsalves, A. Data imbalance in classification: Experimental evaluation. Information Sciences 513, 429-441 (2020).
Chen, J. et al. Visualizing the translation and packaging of HIV-1 full-length RNA. Proc Natl Acad Sci U S A 117, 6145-6155 (2020).
Ye, L. et al. Short- and long-range interactions in the HIV-1 5' UTR regulate genome dimerization and packaging. Nat Struct Mol Biol 29, 306-319 (2022).
Huang, H. et al. Histone H3 trimethylation at lysine 36 guides m(6)A RNA modification co-transcriptionally. Nature 567, 414-419 (2019).
Zhou, K.I. et al. Regulation of Co-transcriptional Pre-mRNA Splicing by m(6)A through the Low-Complexity Protein hnRNPG. Mol Cell 76, 70-81.e79 (2019).
Simons, A.M. Modes of response to environmental change and the elusive empirical evidence for bet hedging. Proc Biol Sci 278, 1601-1609 (2011).
Yeo, J.Y., Goh, G.R., Su, C.T. & Gan, S.K. The Determination of HIV-1 RT Mutation Rate, Its Possible Allosteric Effects, and Its Implications on Drug Resistance. Viruses 12 (2020).
Chen, S. et al. N6-methyladenosine modification of HIV-1 RNA suppresses type-I interferon induction in differentiated monocytic cells and primary macrophages. PLoS Pathog 17, e1009421 (2021).
Lu, M. et al. N(6)-methyladenosine modification enables viral RNA to escape recognition by RNA sensor RIG-I. Nat Microbiol 5, 584-598 (2020).
Rambaut, A., Posada, D., Crandall, K.A. & Holmes, E.C. The causes and consequences of HIV evolution. Nat Rev Genet 5, 52-61 (2004).
Dominguez-Huerta, G. et al. The RNA virosphere: How big and diverse is it? Environ Microbiol 25, 209-215 (2023).
Phillips, S., Baek, A., Kim, S., Chen, S. & Wu, L. Protocol for the generation of HIV-1 genomic RNA with altered levels of N (6)-methyladenosine. STAR Protoc 3, 101616 (2022).
Yu, F. et al. Post-translational modification of RNA m6A demethylase ALKBH5 regulates ROS-induced DNA damage response. Nucleic Acids Res 49, 5779-5797 (2021).
Sambrook, J. & Russell, D.W. Purification of nucleic acids by extraction with phenol:chloroform. CSH Protoc 2006 (2006).
Vodros, D. & Fenyo, E.M. Quantitative evaluation of HIV and SIV co-receptor use with GHOST(3) cell assay. Methods Mol Biol 304, 333-342 (2005).
Lorenz, D.A., Sathe, S., Einstein, J.M. & Yeo, G.W. Direct RNA sequencing enables m(6)A detection in endogenous transcript isoforms at base-specific resolution. RNA 26, 19-28 (2020).

There is NO Competing Interest.

Download PDF

Journal Publication

published 11 Apr, 2024

Read the published version in Nature Microbiology →

Version 1

posted

You are reading this latest preprint version

Single-RNA-level analysis of full-length HIV-1 RNAs reveals functional redundancy of m⁶As

Status:

Journal Publication

Version 1

Abstract

Figures

Introduction

Perspective

Methods

Extraction of HIV-1 virion RNA from pNL4-3 transfected 293T cells:

Generation of in vitro transcribed (IVT) RNA from pNL4-3 plasmid:

Nanopore direct RNA sequencing (DRS) of full-length HIV-1 RNA by multiplex RT:

LC-MS/MS sample preparation:

LC-MS/MS:

LC-MS/MS data processing:

Site-directed mutagenesis:

Digital Quantitative PCR for total HIV-1 RNA:

Western blot analysis:

HIV-1 infectivity measurement:

Data preprocessing and quantification of HIV-1 viral RNA and cellular RNA:

RNA modification site calling:

HIV-1 RNA splicing analysis:

IVT read depth analysis:

Statistical analysis

References

Additional Declarations

Supplementary Files

Status:

Journal Publication

Version 1