Both anti-inflammatory and antiviral properties of novel drug candidate ABX464 are mediated by modulation of RNA splicing

ABX464 is a first-in-class, clinical-stage, small molecule for oral administration that has shown strong anti-inflammatory effects in the DSS-model for inflammatory bowel disease (IBD) and also prevents replication of the HIV virus. ABX464 which binds to cap binding complex (CBC) has demonstrated safety and efficacy in a phase 2a proof-of-concept clinical trial in patients with Ulcerative colitis. Previously, with limited technologies, it was not possible to quantify the effect of ABX464 on viral and cellular RNA biogenesis. Here, using RNA CaptureSeq and deep sequencing, we report that ABX464 enhances the splicing of HIV RNA in infected PBMCs from six healthy individuals and also the expression and splicing of a single long noncoding RNA to generate the anti-inflammatory miR-124 both ex vivo and in HIV patients. While ABX464 has no effect on pre-mRNA splicing of cellular genes, depletion of CBC complex by RNAi leads to accumulation of intron retention transcripts. These results imply that ABX464 did not inhibit the function of CBC in splicing but rather strengthens it under pathological condition like inflammation and HIV infection. The specific dual ability of ABX464 to generate both anti-inflammatory miR-124 and spliced viral RNA may have applicability for the treatment of both inflammatory diseases and HIV infection.

ABX464 is a novel drug candidate for treating patients infected with human immunodeficiency virus (HIV) and patients with ulcerative colitis (ABIVAX, data in file). Despite the successful control of viremia, many HIV-infected individuals treated with ART exhibit residual inflammation associated with non-AIDS-related morbidity and mortality. Several reports have shown that measures of inflammation and immune activation are the best independent predictors of disease progression in HIV-infected individuals. Thus, the anti-inflammatory activity of ABX464 is potentially relevant for the intended use in treating HIV patients, in whom the inflammation around viral reservoirs was shown to substantially contribute to adverse cardiovascular and tumorigenic effects despite long-term ART-treatment. In addition, ABX464 protects mice from the lethal effects of DSS (Dextran Sulphate Sodium), which is a key animal model for inflammatory bowel disease 1 . Patients with UC may benefit from ABX464 which has demonstrated safety in phase 2 clinical trial (ABIVAX, data in file) and has a mode of action different from classical medications including corticosteroids, immunomodulators and biologic treatments.
ABX464 is a small molecule that binds to the cap binding complex (CBC) 2 , a complex at the 5′-end of the pre-mRNA transcript that promotes the initial interaction with transcription and processing machinery [3][4][5] . The CBC recruits several factors to m7G-modified transcripts to mediate processing events and is required for efficient cellular and viral pre-mRNA splicing 3 . The interaction of CBC with the U1 snRNP at the 5′ splice site of the first intron in the transcript 4,6 and direct interaction of CBC with proteins in U4/U5/U6 particles enhances the formation of spliced mRNAs 5,7 . Although CBC is not essential for viability in either yeast or humans 8,9 , its deletion results in a reduction in the recruitment of several splicing factors to the nascent transcript, resulting in inhibition of cotranscriptional spliceosome assembly 5 . The CBC complex has also been shown to affect microRNA biogenesis [10][11][12] . miRNAs are transcribed by RNA pol II as primary (pri)-miRNAs, which carry the m7G cap 13 . During nuclear and cytoplasmic processing events, the pri-miRNA loses the m7G cap, and the mature, 21-23-nucleotide-long miRNA is incorporated into RISC (RNA-induced silencing complex) to guide RNA silencing 12 . Since a large fraction of miRNA genes are located in introns 14,15 , the CBC complex may be involved in the interplay between the processing of intronic pre-miRNAs and pre-mRNAs 16,17 .
ABX464 inhibits viral replication by affecting the biogenesis of viral RNA 2 but its effect on cellular and viral RNA biogenesis has not been analyzed in detail. ABX464 will only act on viral replication once proviral DNA was integrated to cellular DNA. This is important as the viral genome, once integrated in infected cells, requires both activation and inhibition of precursor mRNA splicing 18,19 . Successful infection and production of new infectious HIV particles requires the balanced expression of seven viral proteins (Rev, Tat, Nef, Vif, Vpr, Vpu and Env) that are produced by splicing of the HIV-1 primary 9 kilobases (kb) transcript; among these, the Tat and Rev factors are essential for viral gene expression at the transcriptional and posttranscriptional levels in infected cells 18,19 . The HIV-1 primary transcript serves not only as genomic RNA for progeny virus but also as the mRNA that encodes the viral Gag and Gag-Pol proteins 18,19 . While most cellular unspliced RNAs are retained in the nucleus, where they are degraded 20 , nuclear export of the unspliced viral RNAs is facilitated by the Rev protein through binding to the Rev responsive element (RRE) [21][22][23][24] and interaction with CRM1-dependent export machinery 23,25 . Therefore, inefficient alternative splicing is required to maintain a balance between HIV gene expression and viral production 18,26 . This balance is thought to be mediated by the HIV long terminal repeat (LTR) and the presence of suboptimal viral 5′ and 3′ splice sites (5′ and 3′ ss), which are positively regulated by regulatory sequences and their recognition by cognate trans-acting cellular factors [26][27][28][29][30][31][32] . By binding the CBC complex, ABX464 has been shown to interfere with Rev-mediated export of unspliced RNA 2 . However, the underlying mechanisms behind modulation of viral and cellular splicing and/or miRNA biogenesis by the binding of ABX464 to CBC are presently unknown.
In this study, we elucidate the mechanism of action of ABX464 as small molecule to treat patients with HIV infection and IBD. Using RNA capture and deep sequencing we demonstrate that both effects are mediated through the same mechanism i.e. enhanced pre-mRNA splicing. ABX464 not only enhances pre-mRNA splicing of HIV viral RNA to block HIV replication but also triggers the splicing of a long non-coding RNA which houses one of the loci for the anti-inflammatory miR-124 and thereby increases the expression of the anti-inflammatory microRNA, miR-124. Our findings may open up new treatment options for patients with HIV and inflammatory diseases.

ABX464 generates spliced HIV RNA variants.
To profile viral transcriptional events modulated by ABX464 and thereby assess the full depth of the HIV transcriptome, we employed a recently described targeted RNA capture and sequencing strategy (RNA CaptureSeq). This strategy involves the construction of tiling arrays across the HIV genome, against which cDNAs are hybridized, eluted and sequenced. RNA CaptureSeq is similar to previous in-solution capture methods 33 and exome sequencing approaches 34 , but when combined with deep-sequencing technology, provides saturating coverage and permits the robust assembly of rare and unannotated HIV transcripts 35 .
PBMCs from 6 HIV-negative donors were infected with the YU-2 strain, followed by treatment with ABX464. Our protocol resulted in mild infection that did not lead to cell death at 8 days postinfection (dpi), and viral replication was inhibited by more than 70% with 5 µM of ABX464 (Fig. 1B). After cDNA capture from infected cells that were untreated or treated with ABX464, libraries were prepared and sequenced using an Illumina sequencing facility. To highlight potential new splicing events induced by ABX464, we used a custom bioinformatics pipeline with the main step involving the assembly of putative transcripts from targeted RNA sequencing reads to construct contigs (Illumina paired-end 2 × 75 bp) (Fig. S1A,B). The putative transcripts (contigs) were then mapped to the HIV genome. The HIV profiling datasets used in our analysis yielded high sequence coverage (2 to 30 million reads for the 9 kb genome), allowing deep quantification of HIV splice variants. We were able to reliably quantify each HIV alternative splicing over a wide range of transcript abundances and show that while ABX464 treatment resulted in enhanced splicing, ABX464 did not favor generation of any splice variant over the others (Fig. 1C, Table S1).
To test whether enhanced splicing by ABX464 generates new viral RNA variants, the splicing events were filtered and analyzed. After mapping the contigs to the genome of the YU-2 strain, the latter was used as an anchor for additional clustering and assembly of new HIV transcripts. The partial gene structures generated by the spliced alignment were merged whenever they shared consecutive splicing sites spanning an intronic viral sequence. The superstructures so formed correspond to all possible HIV gene structures for which each complete exon is supported by at least one of the alignments (Supplementary Materials, Materials and Methods). At the HIV transcript level, this led to the formation of superassemblies, where each superassembly is a merge of initial contigs matching a predicted superstructure. Notably, the linking of contigs to the HIV-1 genomic sequence and the requirement that all splicing sites in merged contigs coincide, precluded the formation of spurious superassemblies. Figure 1C and Table S1 present the distribution of the number of contigs that were merged to form one superassembly for each donor. While, in all samples treated by ABX464, most of the assembled contigs (90%) corresponded to spliced RNAs (Fig. 1C, Table S2 and Fig. S2) or small RNAs containing gag-pol sequences that can be released from introns, in untreated samples, spliced RNAs represented a minor fraction (less than 24%), and the majority of contigs corresponded to full-length unspliced viral RNA (more than 74%) (Fig. 1C, Table S2 and Fig. S2). This result demonstrated that ABX464 preferentially generated spliced HIV RNA variants in infected PBMCs, which would compromise the subsequent synthesis of full-length HIV-1 pre-mRNA and assembly of infectious particles, thereby leading to inhibition of viral replication. Interestingly, one of the spliced variants generated by ABX464 treatment (Fig. 1A,D) was present in cells infected by both the YU-2 and Ada-M strains, and no polymorphism in its sequence was detected in the HIV-1 B and C subtypes that were strongly inhibited by ABX464. Furthermore, this new splice variant was detected using long read sequencing, which points to the additional accuracy that this method brings in delineating complex and rare spliced isoforms and estimating their relative abundances 36 . This new RNA variant can generate immunogenic peptides or being toxic to cells harboring proviral DNA. This is consistent with the diminished viral loads and reduced viral DNA levels (ABIVAX, data in file) observed in HIV patients treated with ABX464 37 . Thus, ABX464 not only enhanced the splicing of HIV RNA but also generated a new HIV splice variant. ABX464 does not affect cellular splicing. To ensure that ABX464 acted specifically on HIV splicing and did not significantly or globally affect the splicing events of human genes, we used a high-throughput RNAseq approach. Many genome-wide expression studies of HIV infection are based on analyses of total peripheral blood mononuclear cells (PBMCs) 38,39 , which consist of over a dozen cell subsets, including T cells, B cells, NK cells and monocytes. To avoid the dilution of the specific gene expression signals of particular cell subsets by those from the other cells and thus a reduction of the specificity of this approach, we used purified CD4+ T cells from the PBMCs of 4 donors. The CD4+ T cells were uninfected or infected with the YU-2 strain and were untreated or treated for 6 days with ABX464, followed by high-throughput RNAseq. Each raw dataset of the samples contained between 44 and 105 million single-end reads (50 bp), with an average of approximately 60 million raw reads per sample ( Fig. 2A). More than 97% of the bases had a quality score of ≥Q20. Approximately 98% of the total raw reads were mapped to the human genome sequence (GRCh38), giving an average of 60 million human reads per sample for further analyses. The reads that were correctly mapped (approximately 98% of total input reads) to the gene and transcript locations (GTF annotation file) ( Fig. 2A) were subsequently analyzed using in-house package suites for transcript abundance normalization and evaluation. Multidimensional scaling analysis (MDS) is an unsupervised global analysis approach and is useful for reducing the dimensionality of gene expression data (Fig. 2B). MDS minimizes dimensions, preserving the distances among data points, and allows for projecting multidimensional gene data onto just two or three new dimensions that explain most of its variance. As a result, it is possible to visually interpret major trends in the data, for example, in terms of the similarities among various  data points. The MDS of our gene expression data showed, without any outliers, that the different donors separated well and distributed into the DMSO (untreated) and ABX464 treatments that were infected or uninfected. The displayed variance was donor-dependent (clustered by donor) but treatment-independent (no data structure related to the different treatments), which suggests that the ABX464 molecule did not induce a major difference in CD4+ T cell gene expression (Fig. 2B). For a gross estimation of the alternate splicing events modulated by ABX464 in infected and uninfected CD4+ T cells, we compared junction read counts that originated from exon-exon boundaries across different samples (Fig. 2C). While the alignment of reads to exon junctions was purely coincidental, we did not observe gross differences in the total number of reads corresponding to exon-exon boundaries between treated and untreated CD4+ T cells, whether infected or uninfected, across any of the samples (Fig. 2C). For a more statistically qualified assessment of alternate splicing in these CD4+ T cells, we followed a method for differential splicing analysis across multiple conditions named SUPPA (super-fast pipeline for alternative splicing analysis) 40 . The alternate splicing events were classified into five major groups: Alternative 5′ splice site (A5SS), Alternative 3′ splice site (A3SS), Alternative first and last exons (Alt exons) and Retained exon (RI) (Fig. 2C). A "percent splicing index (psi)" score (ψ-score) was calculated for each of the transcript variants for each sample 41 . Using a stringent cutoff, the difference of ψ-score from untreated sample was fixed at 0.4 and p-value was fixed at 0.05 for differential splicing events induced upon treatment with ABX464 to be considered significant. The number of significant events for each of the 5 possible AS events is presented in Fig. 2C and Table 1. No switch-like events were detected where the difference in psi-score was exactly 1 or −1. We next calculated the ψ-scores of transcripts in uninfected versus infected CD4+ T cells (Fig. 2C). No switch-like events were detected in infected or uninfected T CD4+ samples when compared to ABX464-treated samples (Fig. 2C). The exact numbers of common and differential splicing events between infected and uninfected CD4+ T cells treated with ABX464 were very low (fewer than 10 events, Table 1), while depletion of NCBP1 (Nuclear Cap Binding protein subunit 1), a component of the CBC, in stem cells by only 50% gave rise to large variations, with 81 Alt exon, 12 A 5′ SS, 10 A 3′ SS and 206 IR events (Fig. 2C, Data File S1). The high level of IR indicated that CBC complex is a major component preventing accumulation of unspliced RNA as most of unspliced transcripts will be degraded by nonsense mediated decay (NMD). Comparison of exon coverage reads of a common highly expressed gene, B2M, between the ABX464 and DMSO conditions in the 4 donors revealed that ABX464 did not increase splicing events in B2M (Fig. 2D). In summary, ABX464 treatment did not induce alternate splicing of transcripts, and therefore, ABX464 had no potential to dramatically alter gene expression in activated CD4+ T cells. Consistent with this result, FACS analysis of purified activated CD4+ T cells or PBMCs from 7 donors after 6 days of ABX464 treatment did not reveal any changes in the CCR6/CXCR3 or CD45/CCR7 (Th17/Th1 and effector memory cells, respectively) subpopulation (Fig. S3).

ABX464 does not alter cellular gene expression.
To determine whether ABX464 induced a quantitative change in the transcriptome of infected cells, we measured transcript levels (see Fig. S1 for pipeline) from the RNAseq data. More than 60 000 different transcripts were detected among all samples. After counts per million normalization (CPM), transcripts of each sample were filtered out with a coverage cutoff value of CPM >5; that is, a gene was considered to be expressed in a sample if it had at least five counts for each million mapped reads in that sample. Since this experiment used four donors, four replicates were available for each condition (infected/ uninfected cells with/without ABX464 treatment), and a gene was considered to be expressed if it was expressed to at least five counts for each million mapped reads in all replicates. Following these selection criteria, we ended up with a list of transcripts common to all samples, including 11 700 different transcripts with an average of around 3 000 raw counts per transcript. Using the CPM values of the qualified genes, a global expression plot was produced for each donor. Under the conditions of mild infection used in our study, we did not detect strong changes in gene expression, with only 15 genes being downregulated in response to infection in untreated samples ( Fig. 2E top panel, Table 2). ABX464 treatment resulted in the upregulation of 9 genes in the infected samples, and 6 downregulated and 7 upregulated genes in the uninfected samples ( Fig. 2E middle and lower panels, respectively; Table 2), demonstrating a mild effect of ABX464 treatment on gene expression (Fig. 2E). The six genes that were upregulated following ABX464 treatment were shared between infected and uninfected samples, suggesting that this upregulation was mediated by ABX464 treatment and was independent of infection. Three of the six genes, TOR1AIP2, SCML1 and PPP1R2, had fold changes above 3 with p-values of 10 −5 . Interestingly, TOR1AIP2, a protein thought to regulate protein folding as well as intracellular trafficking, was recently shown to constrain the late steps of HIV replication 42 and may, therefore, be an effector of ABX464-induced changes.
ABX464 upregulates miR-124. Since the CBC complex, which is targeted by ABX464, is involved in the biogenesis of small noncoding RNAs, and our global analysis thus far did not include these, we decided to evaluate if miRNAs or small nucleolar RNAs (snoRNAs) were differentially regulated by ABX464. We performed a microarray analysis for these RNAs from the PBMCs of 6 donors. Cells that were infected with the YU-2 strain, followed by treatment with ABX464 were compared with uninfected and untreated controls. A total of 104 human miRNAs and 40 snoRNAs were significantly differentially expressed in infected PBMCs, when compared to uninfected PBMCs (data file S2), with a false discovery rate lower than 0.05 and fold change higher than 1.5. and DMSO conditions in the 4 donors, we confirmed that ABX464 did not increase splicing events in B2M. (E) Volcano plot of DMSO_I vs DMSO_NI (upper panel), DMSO_NI vs 464_NI (middle panel), DMSO_I vs 464_I (lower panel). The gene expression variation generated by ABX464 treatment was very low in infected (6 downregulated genes) and uninfected (6 downregulated genes) samples. The cluster analyses revealed complete separation of the infected and uninfected samples based on the expression profiles of the differentially expressed miRNAs (Fig. 3A). While infection led to this substantial variation in the expression of small noncoding RNAs (Fig. 3A, left panel), ABX464 treatment induced reproducible upregulation of a single microRNA, miR-124, in infected and uninfected cells (Fig. 3A, middle and right panels respectively, data files S3 and S4). To confirm this result, we used another method of profiling microRNA, the TaqMan Low Density Array (TLDA), which is based on reverse transcription real-time PCR, with a screening capacity for 760 miRNAs (Fig. S4A). Again, only miR-124 was observed to be upregulated by ABX464. Furthermore, quantitative PCR of total RNA isolated from infected and uninfected PBMCs of 5 donors demonstrated that infection led to a slight reduction in miR-124 expression (Fig. 3B, also visible in the total analysis in Fig. 3A), but treatment with ABX464 resulted in significant upregulation of miR-124 in both infected and uninfected cells (Fig. 3B). Next, we determined whether the observed miR-124 response to ABX464 treatment in PBMCs could be attributed to specific cell types. Using purified CD4+, CD8 and macrophages cells, we found that ABX464 treatment resulted in upregulation of miR-124 in lymphoid cells (Fig. 3C) but not in monocyte-derived macrophages, where the expression of miR-124 was undetectable (Figs 3C and S4B). Finally, we proved that this upregulation of miR-124 is specific to the ABX class of small molecules, since other antiretrovirals such as maraviroc, efavirenz, darunavir and AZT did not upregulate the expression of miR-124 in PBMCs, whereas ABX530, a molecule that has the same properties as ABX464, induced upregulation of miR-124 to a similar extent as ABX464. (Fig. 3D).
In recent years, miR-124 has emerged as a critical modulator of immunity and inflammation. miR-124 is known both as a key regulator of microglia quiescence in the central nervous system and as a modulator of monocyte and macrophage activation 43,44 . miR-124 also plays a critical role in both innate and adaptive immune response [45][46][47][48] . In particular, miR-124 was shown to be a critical mediator of cholinergic anti-inflammatory action by reducing IL-6, TNF-α and MCP-1 production 49 . Interestingly, IL-6, TNF-α and MCP-1 were upregulated in the commonly used DSS-induced experimental mouse model of colitis, whereas ABX464 reduced the expression of these proinflammatory cytokines 1 . Persistent immune activation and systemic inflammation also play central roles in the pathogenesis of HIV disease 50,51 . Many HIV-infected individuals treated with antiretroviral therapy (ART) exhibit residual inflammation associated with non-AIDS-related morbidity and mortality 52 the observed downregulation of miR-124 expression upon HIV infection of PBMCs or CD4+ T cells, and subsequent upregulation with ABX464 treatment, we decided to examine if similar modulation of miR-124 expression was seen in HIV patients undergoing antiretroviral therapy (ART). Since the chronic inflammatory state in patients receiving combination ART is primarily related to the extent of damaged gut-associated lymphoid tissue [54][55][56] , miR-124 expression was monitored in rectal biopsies from HIV-infected patients undergoing ART and treated with ABX464 (N = 9). miR-124 expression was downregulated in HIV patients treated with ART, in comparison to its expression in colon biopsies from healthy donors (N = 10). Treatment with ABX464 for 28 days restored the expression of miR-124 to the level of healthy donors (Fig. 3E). When ABX464 treatment was stopped for 28 days, the expression of miR-124 decreased to reach the level seen before treatment (Fig. S4C,D), indicating that the changes observed in expression were specifically due to ABX464. Thus, ABX464 modulates the expression of miR-124 both in vitro and in HIV patients.
ABX464-induced splicing of a long noncoding RNA at the miR-124-1 locus results in upregulation of miR-124. miR-124 is encoded from three independent genes, miR-124-1, miR-124-2, and miR-124-3, located on human chromosomes 8 and 20 (Fig. 4A). To determine which of these genes is induced by ABX464, we employed targeted RNA CaptureSeq to assess the full depth of the transcriptome. The increased sequencing depth of RNA CaptureSeq together with ab initio transcript assembly was employed to determine which locus is affected by ABX464 treatment (Fig. 4A,B). We reconstructed all transcripts assembled within the precapture RNA-Seq data with a similar uniformity of transcript coverage (100% of transcript chains reconstructed; Fig. 4A,B). The total number of reads for untreated as well as ABX464 treated samples varied between 2 and 30 million in infected and uninfected PBMCs (Table S3). The number of reads from the three loci encoding miR-124 in the treated samples were 4-to 12-fold higher than in the untreated samples (Fig. 4B, Table S3), confirming that ABX464 led to a large increase in miR-124. In contrast, ABX464 had no effect on the expression of miR-429 (Fig. 4B), located outside the miR-124 region, again showing the specificity of ABX464 in targeting miR-124. Most of the aligned reads from the 10-kb regions of each locus originated from miR-124-1 and miR-124-3, whereas the number of aligned reads from miR-124-2 was quite low even after ABX464 treatment (Fig. 4B). The effect of ABX464 on the expression of miR-124 was most pronounced from the miR-124-1 locus (Fig. 4B). All mapped reads from the miR-124-1 locus in the treated samples aligned to miR-124 and a 2 kb region surrounding it. Close inspection of this region indicated that the miR-124 sequence is embedded in a long noncoding RNA (lncRNA 0599-205) at the miR-124-1 locus (Fig. 4A,C).
The sequencing depth of RNA CaptureSeq permitted us to assemble ab initio transcripts exhibiting a complex array of splicing patterns. This approach revealed that splicing of lncRNA 0599-205 is activated by ABX464, and this splicing was not present in untreated samples (Fig. 4C,D). Most of the contigs in the untreated samples aligned with unspliced lncRNA 0599-205. Mapping the reads at the splice junctions (J1, J2, J3 and J4) and exon-exon junctions (J5 and J6) of this lncRNA allowed quantification of spliced and unspliced RNA in the treated samples (Fig. 4D). After 8 days of treatment with ABX464, which we assumed would be enough time for the transcripts to reach steady-state levels, the level of unspliced RNA was nine times higher than that of spliced RNA. Since spliced RNA is more stable than unspliced RNA, we must consider that spliced lncRNA 0599-205 is a source of miR-124 upregulation and therefore becomes less stable. Consistent with this prediction, we found that miR-124 production (reads that mapped to only the 85 bp miR-124 region) compensated for the lack of spliced lncRNA 0599-205 (Fig. 4D).

Splicing of lncRNA 0599-205 is required for the production of miR-124.
To directly assess the contribution of lncRNA 0599-205 to the production of miR-124, we cloned the genomic sequence of chromosome 8 from 9 903 167 through 9 904 210 of lncRNA 0599-205 into a plasmid vector (Fig. 5A). Both the spliced and unspliced transcripts of lncRNA 0599-205 could be readily detected in the transfected HeLa cells (Fig. 5B). However, we failed to detect an upregulation of miR-124 in transfected HeLa cells in response to ABX464 treatment since splicing is maximized in HeLa cells and the CBC complex would have an effect at the site of transcription (Fig. 5B). However, transient transfection of the plasmid into HeLa cells resulted in a dramatic (1 250-fold) increase in miR-124 levels (Fig. 5C, left panel). In contrast, when the splice sites of the lncRNA 0599-205 are mutated only traces of miR-124 are detected (Fig. 5C, left panel). Consistent with the fact that unpliced lncRNA 0599-205 are less stable, the amount of mutated lncRNA 0599-205 is less than wild type (Fig. 5C, right panel). All together, the results demonstrate that splicing of the lncRNA 0599-205 is a prerequisite for the production of mir-124.

Discussion
Our investigation provides new insights into the mechanism of action of ABX464 in stopping HIV infection and upregulating the anti-inflammatory miR-124 in infected patients undergoing ART treatment. ABX464 binds CBC (ABIVAX, data in file), a complex that stimulates processing reactions of capped RNAs, including their splicing, 3′-end formation, degradation, and transport. Given that the CBC is believed to bind all classes of m7G-capped RNAs, including precursors and mature forms of mRNAs, stable long noncoding RNAs (lncRNAs), nonadenylated histone RNAs, and precursors of spliceosomal small nuclear RNAs (snRNAs), it is important to understand how the binding of ABX464 to CBC complex achieves its specificity. Both the variation of splicing of HIV RNAs and lncRNA 0599-205 are robustly induced by ABX464. We don't know the reason for that. It could be that weak splice sites that characterizes both transcripts, is at the origin of the induction of these splicing events by ABX464-CBC complex. Before to be exported to cytoplasm most of human protein coding genes are spliced, if not they are degraded in the nucleus, and in the cytoplasm a pioneer round of translation is important for mRNA quality control. The lncRNA 0599-205 like any RNA in the cell will be degraded or stay at its site of transcription which is the case of many long coding RNA. The miR-124 sequence is at the 3 rd exon of lncRNA 0599-205, this  is also a peculiar situation because most of microRNA are generally classified as "intergenic" or "intronic" based upon their genomic location. Intergenic miRNAs are known to be transcribed as independent transcription units, while intronic miRNAs are believed to be processed from the introns of their hosting transcription units and hence share common regulatory mechanisms and expression patterns with their host genes. Since ABX464 treatment did not induce any variation of the microRNA other than miR-124 and very little variation of expression of genes that might host microRNA, we believe that ABX464-CBC interaction induces a specific effect of viral RNA and miR-124 biogenesis. This latter point is also demonstrated by the fact that miR-124 is transcribed by 3 loci but only the miR-124.1 locus is affected by ABX464 treatment and also the splicing of lncRNA 0599-205 is required for the production of miR-124 (Fig. 5). In keeping with this, the mouse miR-124.1 locus also harbor a lncRNA (Rncr3) which serves as pri-miR124 to generate miR-124 and this expression is modulated at post-transcriptional level by the splcing factor PTB1 which inhibits the production of miR-124 by binding the Rnc3 RNA 57 . Nuclear cap functions are mediated by the CBP80 and CBP20 proteins, which associate cotranscriptionally with the nascent RNA 8,58,59 . CBP20 interacts directly with the m7G cap through its classical RNA recognition motif (RRM), while CBP80 ensures high-affinity binding of the full CBC and provides a platform for  an enhancer of viral but not cellular splicing. Binding of ABX464 may change the conformation of the CBC to allow efficient interaction of CBP80 with a splicing factor and thereby render recognition of the viral splicing sites more efficient. It is known that HIV RNA has suboptimal splicing sites to allow inhibition of splicing of the 9 kb primary transcript at later stages of infection 18,62 . In contrast, most cellular genes require splicing for expression, and unspliced intron-containing transcripts are retained in the nucleus where they are degraded 20 . Depletion of NCBP1 by 50% leads to upregulation of unspliced IR transcripts confirming the implication of CBC in NMD. ABX464 is making sure that viral HIV RNA is spliced (since it is transcribed from integrated pro viral DNA resembling any gene in the cell), exported and unspliced viral RNA is degraded (presumably by NMD). ABX464 is not changing any of CBC functions therefore it is not expected that ABX464 by binding the CBC complex change the general RNA biogenesis in the cell. However, very little is known about CBC complex under HIV infection or immune cells activation. The finding that ABX464 binds to CBC and induce specific changes in immune cells; miR-124 overexpression by modulating the splicing 0599-205 long coding RNA and enhancement of splicing of HIV RNA, will be of great help to understand the involvement of CBC under HIV infection and inflammation. A prediction of splice site strength using maximum entropy (MaxEnt) revealed that the 5′ splice site between exon 2 and intron 2 of lncRNA 0599-205 is very weak compared to the other splice site (Table S4). Since unspliced lncRNA 0599-205 cannot exit the nucleus due to quality control machinery and CBC-mediated control, spliced lncRNA 0599-205 constitutes a miRNA storage form, possibly in addition to other functional properties of the intact spliced transcript. This storage may be maintained through low transcriptional and degradative activity of unspliced lncRNA 0599-205 and producing only low levels of mature miR-124 release under normal conditions. Indeed, when lncRNA 0599-205 is unspliced, its fate is similar to that of other short-lived transcripts 63 , susceptible to a CBC complex called CBC-NEXT 64 , which promotes RNA degradation via the nuclear RNA exosome 65 . Consistently, the splicing of lncRNA 0599-205 is a prerequisite for the production of miR-124. The splicing of lncRNA 0599-205 is important to stabilize the mature transcript for recognition by the microRNA processing machinery and thereby produce miR-124. ABX464 treatment could, thus, enable the rapid release of a large amount of miR-124 through lncRNA 0599-205 splicing without requiring transcriptional activation of the lncRNA 0599-205 locus. In HIV patients this could be the mechanism by which miR-124 is up-regulated following treatment with ABX464 (Fig. 3E).
HeLa cells, which contain a large excess of splicing factors, efficiently produce HIV viral RNA splice products 66 and a large amount of miR-124 from transfected HIV and lncRNA 0599-205 constructs, respectively. Thus, ABX464 does not impede the normal function of CBC in splicing but renders it more efficient. As viral but not cellular genes require the unspliced state for viral replication, and splicing of lncRNA 0599-205 can affect miR-124 expression, ABX464 inhibits viral replication and induces anti-inflammatory miR-124 expression.
Large numbers of studies in different cells and systems have found that miR-124 is a critical modulator of inflammatory and immunological responses. The finding that ABX464 upregulates miR-124 expression in infected patients could be a major advantage in treating residual inflammation associated with non-AIDS-related morbidity and mortality in HIV patients [52][53][54] . In many cases, miR-124 functions as a negative regulator for inflammatory signals, providing negative feedback to help maintain homeostasis. Furthermore, Ulcerative colitis patients treated with ABX464 also demonstrate elevation of miR-124 and show evidence of a robust and consistent efficacy signal of ABX464 (Unpublished results). Thus, ABX464 is a small molecule with therapeutic potential not only against HIV, but also against inflammatory diseases.

Materials and Methods
Human samples. All  Cell culture, infection and transfection. Buffy coats from HIV-negative individuals were obtained from the local blood donation center in Centre de transfusion sanguine Montpellier. Human peripheral blood mononuclear cells (PBMCs) were isolated by Ficoll (Histopaque, Sigma) gradient centrifugation. CD4+ and CD8+ T cells were purified from PBMCs after Ficoll gradient centrifugation by Human CD4+ beads positive selection and CD8+ beads positive selection, respectively (Miltenyi). PBMCs, CD4+ and CD8+ T cells were grown at 37 °C and 5% CO 2 to a density of 1.5 × 10 6 cells/ml in RPMI GlutaMAX medium (Life Technologies) supplemented with 10% fetal calf serum (FCS) (Thermo Fisher), 40 U/ml of IL2 (PeproTech) and 5 µg/mL of PHA (Roche) for 2 days. Three days later, cells were resuspended at 1.5 × 10 6 cells/ml in RPMI supplemented with 10% FCS and 40 U/ml IL2 and separated into two parts (infected and uninfected cells). Cells were infected with 80 ng of p24/10 6 cells of the YU-2 strain for 4 to 6 hours and then rinsed with PBS before medium renewal. Both infected and uninfected cells were then centrifuged and resuspended to a density of 1.5 × 10 6 cells/ml in medium supplemented with diluted drug solubilized in DMSO (Sigma) to a final 0.05% DMSO concentration or antiretroviral compounds according to the manufacturer's instructions. Cells were treated for 6 days with a medium renewal at day 3. HIV p24 titration was performed by ELISA from cell culture supernatants with Innotest kit (Ingen) according to the manufacturer's instructions.
To generate monocyte-derived macrophages, monocytes were isolated using CD14+ positive microbeads (Miltenyi) and cultured in X-VIVO 10 medium (Lonza) supplemented with 10 mM HEPES, 1 mM sodium pyruvate, 1% non-essential amino acids, 10% FBS, 10 ng/ml GM-CSF and 100 ng/ml M-CSF for 7 days with medium renewal at day 3. After 7 days of differentiation, the cells were washed, plated in 6-well plates and incubated for 24 hours in medium without cytokine addition. Cells were treated for 6 days with a partial medium change at day 3.
RNA CaptureSeq of HIV RNA and cellular microRNAs, sequencing and bioinformatics analysis, are included as Supplementary Materials. miRNA (RT-q)-PCR analysis. miRNAs were extracted from PBMCs, CD4+ or CD8+ T cells using the NucleoSpin miRNeasy kit (Macherey-Nagel) according to the manufacturer's instructions. miRNA reverse transcription (RT) was performed using the miScript II RT kit (Qiagen). The resultant cDNA was used as a template for real-time qPCR performed with the miScript SYBR Green PCR kit (Qiagen), together with the appropriate primers (Qiagen). A specific miR-124 primer (Hs_miR-124a), a control-specific primer (Ce_ miR-39) and two housekeeping gene (miR-26 and miR-191)-specific primers (Hs_miR-26a and Hs_miR-191) were used to perform relative qPCR. The PCR analysis was performed on a LightCycler 480 Instrument II (Roche Molecular Systems, Inc). All reactions were run in triplicate. The relative levels of miR-124 were calculated using the 2 −ΔΔCt method.
Total RNA was extracted from macrophages using the Macherey-Nagel NucleoSpin miRNeasy kit. miRNA reverse transcription (RT) was performed on 2 µl of total RNA according to the Applied Biosystems TaqMan Advanced miRNA assays protocol for TaqMan qPCR. A specific miR-124 primer (Hs_miR-124a), a spike control-specific primer (Ce_miR-39) and housekeeping gene (miR-191)-specific primers (Hs_miR-191) were used to perform relative qPCR. PCR runs were analyzed with Applied Biosystems ViiA 7 software. miR-124 quantification in human biopsies. Quantification of miR-124 was performed on rectal biopsies sample from patients that have been collected in PAXgene Tissue Containers (PreAnalytiX). RNA extractions were performed using the PAXgene Tissue miRNA Kit (PreAnalytiX) following manufacturer's protocol. Briefly, samples were immersed in 250 µl of Buffer TM1 in a 2 ml Safe-Lock microcentrifuge tube containing one 5 mm stainless bead. Samples were disrupted and homogenized using Qiagen TissueLyser apparatus during 2 × 2 minutes at 20 Hz. The rest of the procedure was performed following the manufacturer's protocol. At the end of the procedure, each RNA was eluted into 32 µl of TM4 Buffer. RNA concentration and purity were measured using a NanoDrop ND-1000 spectrophotometer (Thermo Scientific). RNA integrity was assessed using an Agilent 2200 TapeStation with RNA ScreenTape. cDNA templates were prepared using the TaqMan Advanced miRNA cDNA Synthesis Kit (Applied Biosystems) starting from the 10 ng total RNA matrix and following the manufacturer's protocol. qPCR was carried out using two TaqMan Advanced miRNA Assays (Applied Biosystems), one targeting the miR-124-1 miRNA (assay ID: 477879_mir) and the other targeting an endogenous miRNA, miR16 (assay ID: 477860_mir), which was used as a reference for expression data normalization, according to the manufacturer's instructions. QPCR was performed on the LightCycler 480 Instrument II (Roche Molecular Systems, Inc). All assays were labeled using FAM-MGB chemistry.
Scientific RepoRts | (2019) 9:792 | https://doi.org/10.1038/s41598-018-37813-y Statistical data analysis. All statistical analyses and plots were done with R software v3.5. For normalization and processing, we used edgeR v3.22.1, EDASeq v2.14 and DESeq2 v1.20. We decided to use counts per million (CPM) over transcripts per million (TPM) mainly because TPM discards a lot of information about the original count sizes and thus would have produced too much noise for differential testing. The first step was to filter the data. We considered a gene to be expressed at a reasonable level in a sample if it had at least five counts for each million mapped reads in that sample. For marker selection, we used a type 1(*) error α threshold set to 5%, enabling a highly flexible approach to the differential analyses despite the lack of replicates. For the same reason, the log 2 fold change cutoff was set to 1.5 for both positive and negative values, and the cutoff p-value was typically ≤0.05. Those appropriate cutoffs allowed us to to identify differentially expressed genes. The false discovery rate (FDR) values were computed for illustration purposes but were not taken into account in the decision due to the lack of replicates. The first methodology choice was Benjamini and Hochberg. This procedure provides less stringent control of type 1 errors compared to familywise error rate (FWER) procedures (such as the Bonferroni correction). We added an MDS plot for the data with R and edgeR package. The MDS plot produces a plot in which distances between samples correspond to the leading biological coefficient of variation (BCV) between those samples. In this plot, we did not observe segregation between the DMSO-NI, 464-NI, DMSOI and 464-I conditions. We also produced several volcano plots with R and the EDASeq package. These plots were a good representation of the differential expression (DE) testing in the RNA-Seq experiments. These plots showed the log 2 fold-change (FC) and the -log 10 of p-values for all genes investigated in the DE test. The green points highlighted are markers that satisfied our selection criteria (significant in both p-value and log 2 FC). The yellow points highlighted are genes that satisfied the log 2 FC criterion but failed the p-value criterion. We retained these points to investigate further even though they had no decisional value.