|Home | About | Journals | Submit | Contact Us | Français|
Integrated HIV-1 genomes are found within actively transcribed host genes in latently infected CD4+ T cells. Readthrough transcription of the host gene might therefore suppress HIV-1 gene expression and promote the latent infection that allows viral persistence in patients on therapy. To address the effect of host gene readthrough, we used homologous recombination to insert HIV-1 genomes in either orientation into an identical position within an intron of an active host gene (HPRT). Constructs were engineered to permit or block readthrough transcription of HPRT. Readthrough transcription inhibited HIV-1 gene expression for convergently orientated provirus but enhanced HIV-1 gene expression when HIV-1 was in the same orientation as the host gene. Orientation had a >10 fold effect on HIV-1 gene expression. Due to the nature of HIV-1 integration sites in vivo, this orientation-dependent regulation can influence the vast majority of infected cells and adds another complexity to the maintenance of latency.
Highly Active Anti-Retroviral Therapy (HAART) can reduce HIV-1 viremia to below the clinical limit of detection in many infected individuals and dramatically reverse their progression to AIDS (Perelson et al., 1997; Gulick et al., 1997; Hammer et al., 1997). However, HIV-1 persists in resting memory CD4+ T cells in the form of a stably integrated, transcriptionally silent provirus (Chun et al., 1995; Chun et al., 1997). Latently infected cells are rare, with a frequency of 1/106 resting CD4+ T cells (Chun et al., 1997; Finzi et al., 1997). These cells do not appear to produce any viral proteins and are thus unaffected by the antiretroviral drugs or the host immune system (Hermankova et al., 2003; Lassen et al., 2004b). Nevertheless, upon cellular activation, replication-competent viruses can be quickly released and rekindle the infection. Moreover, HIV-1 latency inadvertently exploits the long-lived nature of the resting memory CD4+ T cells. The decay rate of the latently infected resting CD4+ T cells is extremely slow (Finzi et al., 1999; Siliciano et al., 2003). Replication-competent viruses can be recovered from this latent reservoir even in patients whose viral loads have been undetectable for seven years (Siliciano et al., 2003). As such, this small pool of latently infected cells present in each infected individual serves as a lifelong reservoir for the virus. This latent reservoir is considered the major barrier to viral eradication under current regimens.
Understanding the molecular mechanisms by which HIV-1 latency is established and maintained is essential for developing strategies to “purge” the latent HIV-1 reservoir. Although resting CD4+ T cells with integrated HIV-1 DNA can be detected in vivo (Chun et al., 1995; Chun et al., 1997), the direct infection of resting cells does not generally proceed to integration (Stevenson M et al., 1990; Zack et al., 1990; Zhou et al., 2005). Rather, the phenotype of cells harboring latent HIV-1 suggests that they arise from infected CD4+ T lymphoblasts that have reverted to a resting memory state (Pierson et al., 2000; Chun et al., 1997). Thus, the establishment of latency appears to be the accidental consequence of HIV-1 tropism for activated CD4+ T cells. When the activated CD4+ T cells revert back to a resting state as memory cells, they undergo a profound change in state, and many proposed mechanisms of HIV-1 latency reflect aspects of the intracellular microenvironment that become suboptimal for HIV-1 gene expression in resting CD4+ T cells (Lassen et al., 2004a). Mechanisms to explain HIV-1 latency include: (1) proviral integration into sites that are or that become repressive for transcription (Jordan et al., 2001; Winslow et al., 1993), (2) the absence, in the nucleus of resting CD4+ T cells, of crucial host transcription activators for HIV-1 expression (Bohnlein et al., 1988; Duh et al., 1989; Ganesh et al., 2003; Nabel and Baltimore, 1987; Tong-Starksen et al., 1987), (3) presence of cellular transcriptional repressors (Jiang et al., 2007; Tyagi M, 2007; Williams SA et al., 2006; Coull et al., 2000; He and Margolis, 2002), (4) histone modifications that mediate repression of integrated HIV-1 gene expression (Williams SA et al., 2006; du Chene I et al., 2007; Marban et al., 2007), (5) premature termination of HIV-1 transcription due to the absence of viral protein Tat and Tat-associated host factors (Adams et al., 1994; Herrmann and Rice, 1995; Kao et al., 1987), (6) failure to export singly spliced and unspliced HIV-1 RNAs in the absence of the HIV-1 Rev protein (Malim et al., 1989), (7) nuclear retention of multiply spliced HIV-1 RNA in resting CD4+ T cells (Lassen et al., 2006), and (8) inhibitory cellular microRNAs expressed in resting CD4+ T cells (Huang et al., 2007). The net effect of multiple mechanisms is the profound but reversible silencing of HIV-1 gene expression in resting CD4+ T cells. Interestingly, although HIV-1 latency is not the sole result of any single mechanism, removing any one of the multiple restrictions on HIV-1 gene expression can lead to virus production in experimental settings, probably because of a strong positive feedback loop involving HIV-1 Tat.
Because access of the transcriptional machinery to the integrated provirus is a critical prerequisite for HIV-1 gene expression, one of the most widely discussed hypotheses concerning HIV-1 latency is that the non-productive nature of infection in resting CD4+ T cells reflects proviral integration into chromosomal sites that are, or that become, repressive for transcription (Jordan et al., 2001; Jordan et al., 2003). The model (J-Lat) is based on results of elegant studies in cell lines selected for a latent phenotype. In contrast, when cell lines are infected with HIV-1 in vitro without selection for a latent phenotype, integration sites are generally found within actively transcribed host genes (Mitchell et al., 2004; Schroder et al., 2002). Further characterization of the HIV-1 integration sites in the inducible J-Lat cells revealed that the majority of integration sites located in genes, while alphoid repeats in the centromere, gene deserts and highly expressed genes were favored compared to the constitutively expressed sites (Lewinski et al., 2005). Importantly, in vivo studies of the bulk of HIV-1 integration sites in resting CD4+ T cells from patients on suppressive HAART revealed that most latent viral genomes resided within the introns of active host genes, although technical limitations have restrained the separation of replication competent proviruses from the much larger proportion of defective ones (Han et al., 2004). Notably, a new study of the integration sites of the expression-competent proviruses from a primary cell derived in vitro latency model has shown the preferentially integration into gene regions at a similar frequency as found in the in vivo study (unpublished data). Therefore, latency is not simply due to the inaccessibility of the integrated proviruses to the transcriptional machinery.
A direct consequence of the nature of HIV-1 integration sites in vivo is that HIV-1 gene expression may be decreased by transcriptional interference (TI). TI is a direct cis effect of one transcriptional process on a second transcriptional process (Adhya and Gottesman, 1982; Callen et al., 2004; Eszterhas et al., 2002; Greger et al., 1998; Mazo et al., 2007; Shearwin et al., 2005; Petruk et al., 2006). In general, transcription from an upstream promoter suppresses gene expression from a downstream promoter. TI is observed in special situations in which transcription from the upstream promoter is not terminated before reaching the downstream promoter. In the case of integrated HIV-1 genomes, polymerase complexes initiating at the host gene promoters will be continuously running through the HIV-1 genome. This has two interesting consequences. First, the entire HIV-1 genome becomes incorporated into an intron of the primary transcript of the host gene from which it is presumably spliced out and degraded. This prediction was confirmed experimentally (Han et al., 2004). Second, the read-through transcription may alter the level of HIV-1 transcription initiated from the viral promoter and contribute to latency. In this study, we attempted to verify this hypothesis and quantitatively measure the influence of host gene readthrough transcription on expression of viral genes from the integrated HIV-1 provirus.
Our previous analysis of the sites of integration of HIV-1 proviruses in resting CD4+ T cells from patients on HAART showed that HIV-1 DNA is usually integrated into introns of active cellular genes (Han et al., 2004). Therefore, HIV-1 is essentially a “genome within a gene”, and the regulation of HIV-1 gene expression must be understood in this context. Based on studies in other systems, it is reasonable to assume that readthrough transcription of the host gene will impact HIV-1 gene expression, but this issue has not been directly studied. Effects of readthrough transcription have been analyzed for endogenous nested genes, integrated transgenes, and non-coding RNAs (Jaworski et al., 2007; Strathdee, 2006; Martianov et al., 2007). Proudfoot and colleagues have elegantly demonstrated that the binding of critical transcription factors to HIV-1 LTR is suppressed by transcriptional interference from a closely adjacent tandem HIV-1 promoter (Greger et al., 1998). However, since HIV-1 integration sites map throughout the length of the host genes and not necessarily adjacent to a cellular promoter (Han et al., 2004), and since the average size of human genes is ~27 kb and some genes can be megabases in length (2001), it is important to analyze the impact of readthrough transcription in a natural host gene setting. Therefore, we established a model to measure and compare HIV-1 expression levels under conditions where readthrough transcription could be turned on or off (Fig 1) in order to evaluate the hypothesis that readthrough transcription of the host gene affects HIV-1 gene expression. Given the extraordinary difficulty of carrying out mechanistic studies on the rare cells that constitute the latent reservoir in vivo, we set up an in vitro model using the HCT116 cell line. We inserted the HIV-1 genome into a precisely determined intronic position about 20kb downstream from the host gene promoter. We chose the Hypoxanthine-guanine phosphoribosyltransferase (HPRT) gene, an actively transcribed housekeeping gene located on the X chromosome. In male cells (such as HCT116), this gene is present in a single copy per cell. Karyotyping confirmed that the HCT116 cells had a single X chromosome (not shown). To generate matched sets of cells in which readthrough transcription of the HPRT gene either occurs [Readthrough (+)] or does not occur [Readthrough (−)], we included a triple repeat of a strong poly adenylation signal sequence and a spacer sequence to stop host gene transcription upstream of the HIV-1 genome. This triple poly-adenylation sequence was excised using Cre recombinase to generate Readthrough (+) cells. With the exception of the poly-adenylation sequence and an adjacent β-geo selection cassette (Lobe et al., 1999), the Readthrough (+) and Readthrough (−) cells are identical. Because no obvious bias in terms of the orientation of HIV-1 integration relative to the host genes was observed in vivo (Han et al., 2004), two pairs of cell lines were generated in which HIV-1 was inserted in either forward or reverse orientation with respect to HPRT (Fig 1).
We used homologous recombination to introduce the HIV-1 constructs into the third intron of the HPRT gene (Figure 2A). Two large matching constructs (~23kb) were generated, each consisting of six parts: two flanking HPRT homology arms, an internally deleted HIV-1 genome (in either orientation), a selection cassette consisting of a β-galactosidase-neomycin (β-geo) fusion gene, an SV40 triple poly-adenylation signal, and spacer DNA. Notably, because recent work indicates that RNA polymerase II can continue past the polyadenylation site (West et al., 2004; Teixeira et al., 2004) and may move beyond 1 kb before falling off the DNA (Proudfoot et al., 2002; Dye and Proudfoot, 2001), a 1.5 kb segment of spacer DNA without homology to human sequences was introduced following the stop sequence to allow elongating polymerase to disengage before reaching the HIV-1 genome. RT-PCR analysis showed the nascent transcripts of the 3’ region of the spacer were significantly decreased in the readthrough (−) clone, confirming the effectiveness of the stop signal (Supplementary Fig 2). The β-geo positive selection marker and the triple poly-adenylation signal were flanked by LoxP sites, which can be recognized by CRE recombinase and thus removed (Lobe et al., 1999). The internally deleted HIV-1 genome was derived from HIV-EGFP-HSAΔE vector (Reiser et al., 2000) which has two intact LTRs, the critical accessory genes like tat and rev, and GFP and HSA coding sequences in the env and nef genes, respectively (Fig. 2B). After transfection of these constructs, surviving HCT116 clones were selected with G418 (400 ug/ml) and 6-thioguanine (6-TG, 10ug/ml), for the presence of β-geo and the disruption of HPRT, respectively. The frequency of correct homologous recombination events was extremely low (~1 per 2–6 million cells). For clones with correctly inserted HIV-1 genomes, an aliquot of cells was transfected with Cre recombinase to excise the poly-adenylation signal and restore HPRT readthrough. These clones could be selected in HAT medium which selects for HPRT expression, indicating HIV-1 sequences do not efficiently truncate HPRT sequences. The end result was two sets of clonal cell lines in which read-through transcription of the host gene either proceeds through or stops before reaching the HIV-1 genome, which was present in either same or the reverse (convergent) orientation with respect to HPRT.
To verify that the cell clones obtained using the above selection protocols contained correctly inserted HIV-1 genomes, we used genomic DNA PCR (Fig 2A, C). PCR amplification using a 5’ primer in HPRT but outside the homology arm and a 3’primer in ;-β-geo gave a band in Readthrough (−) cell lines but not the Readthrough (+) cells from which the β-geo and poly-adenylation signal were excised by Cre recombinase (Fig. 2A, C). Using another 3’primer in the HIV-1 genome, PCR products of the correct size were obtained only for Readthrough (+) cells. Direct sequencing of joints provided further confirmation that the constructs were correct. We also carried out confirmatory Southern blot analysis. As shown in Supplementary Fig. 3, restriction enzymes were chosen so as to give a unique and identifiable pattern of fragments for each type of insert. All four of the constructs gave the expected pattern of bands. In addition, we carried out X-gal staining to verify the removal of β-geo by Cre/LoxP recombination in both Readthrough (+) cell lines (Fig 2D). Most importantly, we analyzed HPRT transcription to determine whether the stop cassette caused premature termination of upstream transcription (Figure 2E). Realtime RT-PCR from purified mRNAs was used to measure the HPRT mRNA levels in each clone. The realtime RT-PCR primers were placed in the two HPRT exons (exons 3 and 4) flanking the intron containing the HIV-1 genome. In the Readthrough (−) clones containing the triple poly-adenylation signal, the HPRT transcription was not detected using the exon 3-exon 4 primer pair. In contrast, Cre-mediated excision of the stop signal in the Readthrough (+) clones restored HPRT transcription through this region to levels seen in wild type cells (Fig 2E). It was thus possible to evaluate the effects of readthrough transcription on HIV-1 gene expression.
Using this system, we first analyzed the HIV-1 gene expression in cell clones in which HIV-1 lies in the reverse transcriptional orientation (convergent orientation) with respect to the host gene. We isolated poly-adenylated mRNAs from each clone to measure the steady-state levels of mature HIV-1 transcripts, thus avoiding interference from HIV-1 sequences present in introns spliced from the host gene. Realtime RT-PCR amplification of a segment of the HIV-1 gag gene was carried out to quantify the amount of HIV-1 transcripts. 5’-RACE using the same 3’-primer confirmed the transcriptional start site at the HIV-1 LTR (Supplementary Fig. 4). By comparing the HIV-1 transcription levels in the Readthrough (+) and Readthrough (−) clones, we discovered that HIV-1 transcription was decreased by approximately four fold when HIV-1 and the host gene were transcriptionally convergent (Fig. 3a). The reduction was statistically significant (P=0.03). In addition, the GFP expression levels analyzed by flow cytometry showed a similar pattern of reduction by the host transcriptional readthrough (Fig. 3b). Notably, the FACS plots suggested that even within clonal cells, the population of cells showed a broad expression profile. However, when the high- or low-expressing cells within the same cell population were sorted and subject to expansion, each population could reproduce the pattern of the entire population from which the cloned cell was sorted, suggesting that each clonal population exhibits significant variation in expression around a mean (Eszterhas et al., 2002).
Since HIV-1 transcription can be activated by TNF-α through NFκB signaling (Swingler et al., 1994), we tested whether TNF-α could reverse this transcriptional interference. After three hours of TNF-α treatment, the nuclear NFκB levels were significantly increased in treated cells (Supplementary Fig. 5), and HIV-1 gene expression was increased in both Readthrough (+) and Readthrough (−) clones (compare Fig 3a and 3b). However, a 3 fold inhibitory effect of transcriptional interference was still observed in the Readthrough (+) clones (Fig. 3c). In both Readthrough (+) and Readthrough (−) clones, levels of HIV-1 gene expression were higher than in untreated cells (Fig. 3a), indicating that TNF-α did have the expected upregulatory effect on HIV-1 transcription. However, TNF-α stimulation did not completely abrogate the effect of transcriptional interference. In addition, since the viral protein Tat is a crucial transcriptional activator, we analyzed whether addition of supplementary Tat by transient transfection could reverse the observed transcriptional interference. As was the case with TNF-α, supplementary Tat alone or in combination with TNF-α stimulated HIV-1 transcription in both Readthrough (+) and Readthrough (−) clones but did not completely abrogate the effects of transcriptional interference (Fig. 3c).
Our analysis of HIV-1 integration in patients did not reveal any bias with regard to orientation (Han et al., 2004), and we expected a similar degree of transcriptional interference for HIV-1 genomes in the same orientation as the host gene. However, instead of transcriptional interference, we found that HIV-1 transcription was enhanced by ~4 fold when HIV-1 and the host gene were in the same orientation (Fig. 4a). The enhancement was statistically significant (P=0.03). Flow cytometric analysis confirmed at the protein level the increase in GFP expression in Readthrough (+) clones (Fig. 4b). Again, we activated each cell clone with TNF-α, supplementary Tat, or both. In all conditions, the levels of gene HIV-1 expression were increased in each clone relative to levels observed in the absence of stimulation. However, the enhancing cis effect of readthrough transcription persisted even when the levels of HIV-1 gene expression were increased by TNF-α, Tat or both (Fig. 4c).
Current models for transcriptional interference (Shearwin et al., 2005) relevant to our system include: (1) “sitting duck” interference, (2) collision, and (3) promoter occlusion. The sitting duck mechanism proposed by Callen et al. suggests that convergently transcribing elongation complexes read through the target promoter and render it slow to transit from the open complex to the elongation complex (i.e. slow to “fire”) (Callen et al., 2004). The model was defined in studies of prokaryotic promoters, and the relevance of such a mechanism in eukaryotes is not yet clear. Head-on collision, occurring only for converging elongation complexes, can lead to the premature termination of the transcriptional progress of one or both complexes (Shearwin et al., 2005). Promoter occlusion, originally defined by Adhya and Gottesman in 1982 (Adhya and Gottesman, 1982) in λ phage, is the process through which one elongating complex blocks the assembly of another initiation complex. The interfering and the target promoters can be either convergent or in same orientation (Shearwin et al., 2005). Because this mechanism is a general one, we set out to analyze the role of promoter occlusion in the orientation-dependent effect of readthrough transcription on HIV-1 gene expression.
Using chromatin immunoprecipitation (ChIP) followed by realtime PCR amplification of the HIV-1 promoter region (−116 to +4), we compared the binding of RNA polymerase II, general transcriptional factors, and the other transcription factors crucial for HIV-1 transcription to the HIV-1 promoter in Readthrough (+) and Readthrough (−) clones. The ChIP primers used capture both the 5’ and 3’ LTRs and thus measure the net effect of readthrough on LTR occupancy. Fig. 5a shows the results of ChIP in the set of cell clones in which HIV-1 and the host gene are in the convergent orientation. Total RNA polymerase II (pol II) can be readily detected at the HIV-1 promoter in the Readthrough (−) cells using the N20 antibody, which recognizes an epitope outside the carboxyl-terminal domain (CTD). Host gene readthrough reduced the occupancy of total pol II by 35%, consistent with the observed decrease in HIV-1 gene expression (Fig. 3). As a control in this and other ChIP experiments involving pol II and general transcription factors, we examined occupancy at the GAPDH promoter and found no differences between Readthrough (+) and Readthrough (−) clones. Phosphorylation of the CTD, particularly at serine 5 within the heptad repeat, enhances early transcriptional elongation (Ho and Shuman, 1999). Using another pol II antibody that specifically recognizes phosphorylated serine 5 in the heptapeptide repeat sequence of the CTD, a 45% decrease in the binding of the active pol II was observed in the Readthrough (+) clones (Fig. 5a). TATA-Binding Protein (TBP) and TFIIH are two important general transcription factors. Using antibodies specific to TBP and p62 (a subunit of TFIIH), we demonstrated a similar reduction (32% for TBP and 42% for p62) in their occupancies on the HIV-1 promoter. ChIP of three other host proteins involved in HIV-1 transcription, SP1, NFκB p65 and CDK9, demonstrated reductions by 42%, 40% and 36%, respectively, in Readthrough (+) clones. For these factors, controls were done by analyzing the occupancy of each factor at a control host promoter known to bind the relevant factor. In each case, occupancy was similar in Readthrough (+) and Readthrough (−) clones. Thus the decrease in occupancy at the HIV-1 promoter in Readthrough (+) clones reflects a specific effect of readthrough transcription on the HIV-1 promoter. Such a consistent decrease in occupancy of multiple factors involved in HIV-1 transcription suggests that promoter occlusion plays an important role in the transcriptional interference caused by readthrough in the convergent orientation. The impaired accessibility of individual factors to the HIV-1 promoter can collectively inhibit the assembly of the initiation complex necessary for transcription.
Despite the consistent occupancy changes for multiple factors observed when the HIV-1 genome was in the convergent orientation, a more complex pattern was observed in the cells with HIV-1 in the same orientation as HPRT. As shown in Fig. 5b, the amount of total RNA pol II on the HIV-1 promoter was increased by approximately 2 fold in Readthrough (+) clones. This finding is consistent with the observed increases in HIV-1 gene expression. Some of the pol II captured in this assay may be part of elongation complexes transcribing the HPRT gene, although the occupancy of pol II at the spacer region upstream of the HIV-1 LTR was very low as revealed by ChIP experiments with primers in that region (data not shown), suggesting the elongating pol II transcribing the host gene does not represent a major proportion of the amount of pol II observed around the HIV-1 promoter. The binding of CTD-phosphorylated active pol II was likewise increased by 2.1-fold. The occupancies of TBP and TFIIH showed increases of 1.6-fold and 1.3-fold, respectively. However, the occupancies of other host proteins involved in HIV-1 transcription showed a broader profile of changes, with a 15% decrease in SP1 occupancy, a 25% reduction in CDK9 occupancy, and relatively little change for NFκB. Thus promoter occlusion contributes to a decrease in HIV-1 gene expression when HIV-1 is present in the opposite orientation to the host gene but is not observed when HIV-1 is in the same transcriptional orientation.
In an attempt to find another mechanism to explain the impact of orientation, we used ChIP to analyze the histone modifications in the vicinity of the integrated provirus. Several recent studies have identified particular histone modifications as landmarks for the transcriptional status of genes. For example, H3K9,14Ac was found enriched around the promoter of active genes (Guenther MG, 2007), H3K36me3 was associated with elongation (Guenther MG, 2007), and H3K27me3 was linked to silencing (Schuettengruber et al., 2007). The nucleosome positions around the HIV-1 transcriptional start site have been well characterized by Verdin and colleagues (E Verdin, 1993). Since the nucleosome located at the HIV-1 promoter-enhancer region could be disrupted during transcriptional activation (E Verdin, 1993), we examined the region (+437 to + 543) downstream of the transcriptional start site of HIV-1.
In the convergent orientation, Readthrough (+) cells exhibit higher levels of histones bearing the H3K36me3 modification. Histones bearing the H3K9,14Ac and H3K27me3 modifications were increased as well. When HIV-1 was inserted in the same transcriptional orientation as the host gene, there were similar increases for the H3K9,14Ac and H3K27me3 modifications, whereas the amount of H3K36me3 was little affected by the readthrough. These results suggest that changes in histone modification may not be a major contributor to the orientation-dependent regulation we observed.
Our system allows quantitation of the effect of host readthrough transcription on the integrated HIV-1 genome in each orientation with or without readthrough. Because HIV-1 is generally integrated within actively transcribed host genes, transcriptional readthrough is the normal state. For HIV-1 genomes located at an identical site within an actively transcribing host gene, orientation relative to the host gene has a >10 fold effect and is statistically significant (P=0.02) on the steady-state level of HIV-1 transcription (Fig. 7a). LTR-driven GFP expression demonstrated similar magnitude of difference (Fig. 7b). Therefore, host gene transcriptional readthrough regulates the expression of integrated HIV-1 proviruses in an orientation-dependent manner (Fig. 7c). When HIV-1 and the host gene are in the same orientation, host readthrough enhanced the integrated HIV-1 expression, while in the convergent orientation, such an effect in cis is inhibitory.
Using a system in which HIV-1 proviruses are inserted in precisely the same position within an active host gene in either orientation with and without readthrough transcription, we demonstrate for the first time that there is orientation-dependent cis regulation of transcription of integrated HIV-1 by the readthrough transcription of the host gene. Transcriptional interference is observed when HIV-1 is inserted in the opposite orientation as the host gene, while enhancement of viral gene expression occurs when HIV-1 is in the same orientation as the host gene. Because of the nature of HIV-1 integration sites in vivo (Han et al., 2004; Liu et al., 2006), HIV-1 gene expression in the vast majority of infected cells is likely to be influenced by this interaction between the host gene and the provirus at the transcriptional level.
Transcriptional interference between convergent promoters has uniformly been found to be inhibitory (Callen et al., 2004; Elledge and Davis, 1989; Eszterhas et al., 2002; Hongay et al., 2006; Imamura et al., 2004; Prescott and Proudfoot, 2002; Ward and Murray, 1979). Similarly, we found a 4-fold reduction in the HIV-1 gene transcription by the host gene readthrough when HIV-1 was in the opposite orientation as the host gene. In addition, ChIP analysis revealed a consistent 30~40% reduction in the occupancy of RNA pol II and other relevant transcription factors at the HIV-1 LTR in this situation. Although the recruitment of various proteins of the pre-initiation complex still occurred, it was significantly down-regulated by readthrough transcription. In addition to a reduced occupancy of critical transcription factors at the HIV-1 promoter, other mechanisms might also be involved. For example, collisions between converging elongation complexes have been demonstrated to lead to the premature termination of the transcriptional progress of one or both complexes, with the distance between the two convergent promoters affecting the strength of the interference (Callen et al., 2004). In certain experimental systems, the ability of an elongating polymerase to read through a DNA-bound protein roadblock is enhanced by increasing the number of elongation complexes (Epshtein V, 2003). Since we did not detect an equal reduction in the steady state level of HPRT transcripts (Fig 2E) when HIV-1 was in the opposite orientation to HPRT, collision does not appear to be a major contributor to the observed interference. However, it is possible that such a mechanism may not be reciprocal (Callen et al., 2004). In principle, due to the convergent direction of the two promoters, the generation of both sense and anti-sense RNAs may induce the degradation of HIV-1 sequences, resulting in a decrease in the steady-state level of HIV-1 transcripts. To test this possibility, we transiently transfected the cell lines with an siRNA targeting Dicer1, a critical component of the RNA interference machinery. No obvious change in the steady-state HIV-1 transcripts level was observed by transiently knocking down Dicer1 (Supplementary Fig 6).
In spite of the universal inhibitory effect between two active convergent promoters, results for two promoters transcribing in the same direction vary in different systems. Inhibition of the transcription from downstream promoters has been widely observed in models involving closely adjacent tandem promoters or in tandemly associated transgenes (Yahata et al., 2007; Eszterhas et al., 2002; Greger et al., 1998; Corbin and Maniatis, 1989; J Eggermont and N J Proudfoot, 1993; Yahata et al., 2007). Proudfoot et al. demonstrated that in cell lines with two tandemly integrated HIV-1 promoters, transcription from the upstream LTR had a negative impact on the downstream one (Greger et al., 1998). However, an unexpected and novel finding of our system is that upstream transcription could indeed enhance HIV-1 gene expression for HIV-1 proviruses that are in the same orientation as the host gene. Interestingly, consistent with our results for HIV-1 inserted in the same orientation as the host gene, several studies have shown that upstream transcription can increase the transcription from the downstream promoter. For instance, introducing an active promoter upstream to the silent human endogenous retrovirus (HERV)-K18 promoter activates its transcription in cis (Leupin O et al., 2005). In addition, noncoding transcription from the upstream mouse T cell receptor-α locus can activate Jα promoters located several kilobases downstream, and the blockage of the upstream elongation abolished the downstream transcription (Abarrategui I, 2007). Furthermore, it has been reported that the same transgene inserted at the same position can be expressed in one orientation but silenced in the other orientation in insect, mouse, and cultured human cell lines, although the underlying mechanisms were not clearly known (Alami et al., 2000; Francastel et al., 1999; Sabl and Henikoff, 1996)
Several potential mechanisms may explain how readthrough transcription increases expression of HIV-1 present in the same orientation as the host gene. One potential mechanism for the enhancing phenotype is a change in histone modification of chromatin (Williams SA et al., 2006; du Chene I et al., 2007; Marban C, 2007). However, when we analyzed three histone markers (H3K9,14Ac, H3K36me3, H3K27me3) that are characteristic of promoter activation, active elongation and silencing (Guenther MG, 2007), (Schuettengruber et al., 2007), we did not observe major differences in post-translational modification of histones whether HIV-1 was in the same or opposite orientation as the host gene. Other potential explanations can also be considered. The integration site of HIV-1 in our system was in the third intron of HPRT, about 300 bp downstream from the end of the third exon of HPRT. Although HIV-1 is present at the identical site within the host gene in both sets of cells, the relative distances between the HIV-1 promoter and the adjacent HPRT exon are different due to the length of the HIV-1 construct itself. As mRNA processing reactions, such as splicing, occur cotranscriptionally, the splicing complex might have an effect on an adjacent promoter. In addition, removal of repressors upstream of the HIV-1 transcription start site might occur differentially depending on orientation. Finally, alterations in DNA topology upon readthrough elongation might positively effect transcription from other promoters in the same orientation.
Since HIV-1 randomly inserts into different positions within active host genes (Han et al., 2004), it is possible that the level of the orientation-dependent regulation will be variable and may depend on the relative rate of HIV-1 promoter clearance and the rate of the host gene elongation (Callen et al., 2004; Mazo et al., 2007). Nevertheless, our results reveal another layer of complexity of the HIV-host gene interaction that is likely to affect the vast majority of integrated proviruses and that may play a role in HIV-1 latency and in the regulation of HIV-1 gene expression in productively infected cells. Orientational preferences were not apparent in the bulk of HIV-1 integration sites in patient resting CD4+ T cells (Han et al., 2004) and inducible J-Lat cells (Lewinski et al., 2005). However, since integration in a convergent orientation inhibits HIV-1 expression, it will be interesting to determine whether there is a bias towards such an orientation in the subset of infected resting CD4+ T cells that harbor replication-competent HIV-1, thus contributing to HIV-1 latency in vivo. Interestingly, in the human genome, the activities of human endogenous retroviruses (HERV) are very low (Smit, 1999). For those that reside within genes, HERVs in the convergent orientation with respect to the host gene are over-represented by 5 fold. The orientation-dependent positive and negative effects on HIV-1 gene expression were not abrogated by stimulation with TNF-α or Tat. Thus in activated cells, the enhancing effect may lead to greater HIV-1 gene expression in activated T cells that have a provirus integrated in the same orientation as the host gene. At the population level, this may compensate for transcriptional interference observed in cells with proviruses integrated in the convergent orientation.
The forward and reverse orientation constructs used for homologous recombination, p203MJHIV-F and p203MJHIV-R, respectively, were assembled in multiple steps by cloning HIV-1 sequences (digested by PsiI from the pHIV-EGFP-HSAΔE-YH vector) and a piece of spacer DNA into the AscI site of pWA203 using linker-mediated ligations. The murine spacer DNA was amplified by PCR from the genomic DNA of the BA/F3 cell line using primers: MJ-1F: GCGATCGCGGGTTGCCATAAGTGAAC and MJ-1R: GCGATCGCTGTGTGGGTGTATTGGTG. The PCR conditions were as follows: denaturation at 94°C for 3 minutes, followed by 30 cycles of 94°C for 30s, 60°C for 30s, and 68°C for 1.5min. pHIV-EGFP-HSAΔE-YH was derived from pHIV-EGFP-HSAΔE (Reiser et al., 2000) with truncations between two NdeI sites. pWA203 (pBSHPRT-bgeo) contains the following sequence elements in a pBluescriptSK(−)(Stratagene) backbone: a 5' homologous arm of 5.3 kb from the human HPRT gene (Accession No. M26434, nts 10425–15724), a floxed selection cassette (including a CAG promoter, β-geo coding sequence and three tandem copies of SV40 poly-adenylationdenylation signal), and a 3' homologous arm of 1.6 kb from the human HPRT gene (nts 15725–17283). Both HPRT homologous arms were derived from pE3deltaNeo (a gift from Dr. Raymond Monnat), and the floxed selection cassette was derived from pQX107 (a gift from Dr. Jeremy Nathans). Constructs involved in each step were amplified in MAX Efficiency Stbl2 cells incubated at 30°C. Colony PCRs amplifying the junctions were done to screen for the correct recombinants, and the amplicons were confirmed by sequencing.
Wild type HCT116 cells (ATCC) were transfected with either p203MJHIV-R or p203MJHIV-F by Lipofectamine™ 2000 (Invitrogen). Readthrough (−) clones, with HIV-1 in the convergent (p203MJHIV-R) or same (p203MJHIV-F) orientation, were generated by homologous recombination. To select for the clones with the correct homologous recombination events, transfected cells were subjected to selection with G418 (400 ug/ml) beginning the day after transfection, and additional selection with 6-thioguanine (6-TG, 10ug/ml) was initiated 3 days after the transfection. McCoy5A media (Invitrogen) with 10% FBS was used as the base media. Transfections were done with 12 × 106 wild type HCT116 cells, the cells were plated in twelve 10-mm culture dishes (Falcon) after transfection so that the cell colonies surviving selection were well separated from each other. After ~20 days of culture, individual colonies were picked under microscope and then continuously expanded in the selective media. An aliquot of the cells were subjected to X-gal staining (Gene Therapy Systems) to verify the positive insertion.
To isolate Readthrough (+) cells, a vector encoding Cre (a gift from Dr. Randall Reed) was transfected into each Readthrough (−) clone by nucleofection (Program #D32, VCA-1003, AMAXA). Transfected cells were subjected to selection for restored HPRT gene expression in the base media plus 1×HAT (Hypoxanthine Aminopterin Thymidine) supplement (GIBCO). Individual colonies were picked and further expanded as above, and underwent X-gal staining (Gene Therapy Systems) to confirm excision of the B-gal cassette by Cre recombinase.
RT PCR for measuring the effectiveness of the disengagement of upstream elongating polymerase was carried out using purified total RNAs from Same Readthrough (+) and Same Readthrough (−) clones. Following PCR primers were used: before-HIV-F: GCTAACCAAAATCATCCCAAA, and before-HIV-R: GAAAATAATTCAGAGGAATCACAGG.
Genomic DNA from each cell clone (WT; Readthrough (−) convergent; Readthrough (−) same; Readthrough (+) convergent; Readthrough (+) same) were extracted using the Gentra Puregene kit.PCR was performed with genomic DNA to verify the correct recombination events using the following primers: Primer A (JB8533): GTGACACACAAATGTCCCATTTTCA; Primer B (CAG1): CTATGAACTAATGACCCCG; Primer C: AGCTTGCTACAAGGGACTTTCC; Primer C’: TGGTACTAGCTTGAAGCACCATCCA. PCR conditions were: denaturation at 94°C for 3 minutes, followed by 30 cycles of 94°C for 30s, 60°C for 30s, and 68°C for 8min. HIFI platinum Taq DNA polymerase (Invitrogen) was used for the amplification. The resulting PCR products were further sequenced to verify the correct junctions.
Genomic DNA was isolated from 3 × 10 6 cells from the four cell lines in addition to the wild-type HCT 116 parental cell line (PureGene). DNA was quantitated using a NanoDrop spectrophotometer (Fisher Scientific). A double restriction digest with 25 ug genomic DNA for each of the cell lines was carried out overnight at 37° C with SphI and BlpI. The digests were then electrophoresed on an 0.8% agarose gel overnight at 40 volts. The gel was briefly stained with ethidium bromide to visualize markers. Prior to transfer, DNA was depurinated with 0.5 M HCL for 8 minutes followed by denaturation with 0.5 M NaOH, 1 M NaCl for one hour ( 2 × 30 minute washes). The gel was neutralized with 1 M Tris, 3 M NaCl, pH 7.4(2 × 30 minutes washes). The gel was gently rocked at room temperature for all washes. DNA was then transferred onto Hybond- N+ membrane (Amersham Biosciences) overnight in 10X SSC buffer. After transfer, DNA was crosslinked onto the membrane (Stratalinker, Stratagene). The membrane was then baked for one hour at 80°C in a vacuum oven and then prehybridized in 5X SSC, 5X Denhardt’s Solution, 0.5% SDS, 0.1 mg/ml sheared salmon sperm DNA for four hours at 65°C. The EGFP DNA probe used in the Southern analysis was prepared by carrying out a double restriction digest of pHIV-EGFP- HSAΔE-YH with BmtI and BsrI and then gel purifying the 727 bp fragment. The probe was labeled with [α-32 P]dCTP to a specific activity of 1 × 109 dpm/ug using Ready-To-Go DNA Labelling Beads (-dCTP) (Amersham Biosciences) following the manufacturer’s protocol. The radiolabeled probe was purified using ProbeQuant G-50 Micro Columns (GE Healthcare) following the manufacturer’s protocol. Hybridization with the radiolabeled probe was carried out for 18 hours at 65°C using the same buffer that was used for prehybridization. The membrane was first washed in Buffer I (2X SSC, 0.1% SDS, 4 × 5 minutes washes at room temperature) followed by one wash in Buffer II ( 1X SSC, 0.1% SDS, 1 × 15 minute wash at 65°C). Radiolabeled fragments were detected using the Typhoon 9410 Variable Mode Imager (Amersham Biosciences).
mRNAs from each cell clone were isolated using QIAGEN Oligotex mRNA extraction kit. 5’RACE was performed according to the FirstChoice RLM-RACE protocol (AMBION). HIV-1 specific primers used in the nested PCRs were: First round: 5’ TGAAGGGATGGTTGTAGC 3’; Nested: 5’ AGCTCCCTGCTTGCCCATA 3’.
Reverse transcription reactions from isolated mRNAs were carried out using random hexamers (SuperScript™ III, Invitrogen). Realtime PCR was performed to quantify the steady-state mRNA levels with ABI7000, 7300, or 7900 Real-time PCR Systems (Applied Biosystem). HIV-1 gag primers and probe (Douek et al., 2002) were: GAG-F: 5’GGTGCGAGAGCGTCAGTATTAAG3’ (792nt–814nt), GAG-R: 5’AGCTCCCTGCTTGCCCATA3’ (892nt–910nt), and probe: FAM-AAAATTCGGTTAAGGCCAGGGGGAAAGAA- BHQ (Biosearch Technology). A Taqman expression assay (Hs01003268_g1, Applied Biosystems) was used to quantify the HPRT gene expression. A GAPDH expression assay (Hs00266705_g1, Applied Biosystems) was used to normalize the cell number in each sample. Standards were constructed for absolute quantification of gag copy number. Duplicate reactions were run, and template copies were calculated using the ABI software.
Cells were treated with TNF-α (R&D Systems) at 10ng/ml for three hours and then lysed for mRNA extraction or assayed for nuclear levels of NFκB expression. The HCT116 nuclear fraction was extracted using the nuclear extract kit from Active Motif, and the level of NFκB p65 determined using the TransAM™ NFκB p65 kit (Active Motif). Supplementary Tat was introduced by transfecting pcDNA-TAT-86 (a gift from Dr. Avi Nath) into each cell clones and HIV-1 expression was measured 48 hours later using realtime PCR. Transfection efficiency was measured by SYBR realtime RT PCR (Applied Biosystems), using primers tat86 F: 5’AGTGTTGCTTTCATTGCC3’ and tat86 R: 5’GGTGGGTTGCTTTGATAG3’.
Cells were treated with 1% formaldehyde for 10 min and chromatin was isolated using ChIP-IT™ Express kit (Active Motif). Nuclei were sonicated using 15 bursts of 10 seconds (Output 4, Sonicator 3000, Misonix) to produce DNA fragments ranged from 200–600 bp. Immunoprecipitation of specific proteins and DNA elution were performed as described in the ChIP-IT kit. The antibodies used for immunoprecipitation were as follows: Total RNA polII (N-20), p-CTD polII (3E8), TBP (N-12), TFIIH p62 (H-300), NFκB p65 (C-20), CDK9 (H-169), Sp1 (39058, Active Motif), H3K27me3 (ab6002, abcam), H3K36me3 (ab9050, abcam), H3K9,14Ac (Upstate 06–599). 3E8 was a gift from Dr. Dirk Eick. Antibodies were obtained from Santa Cruz Biotechnology, unless otherwise specified. Realtime PCRs were then carried out using Power SYBR Green PCR Master Mix (Applied Biosystems). Primers used for promoter occupancy were: Pair 1F: 5’AGCTTGCTACAAGGGACTTTCC3’, Pair 1R:5’ACCCAGTACAGGCAAAAAGCAG3’. Primers used to determine histone modifications were: Pair 3F: 5’ATAGTATGGGCAAGCAGG3’, Pair 3R: 5’TGAAGGGATGGTTGTAGC3’ (Kim YK et al., 2006). Occupancy was determined by comparing the Ct value of each sample to that from input DNA by a standard curve with known absolute copy numbers. Samples to which no antibody had been added served as a negative control. Values obtained from these samples were subtracted from each experimental sample to remove the non-specific background signal. Each column represents results from three independent experiments, and error bars, which represent the standard error of the mean, were generated accordingly.
Predesigned siRNA specific to Dicer1 (137012, AMBION) was transfected into cells by Lipofectamine™ RNAiMAX (Invitrogen). Cells were lysed, and mRNAs isolated 48 hrs after the transfection. Knock-down efficiency was verified by the Taqman expression assay (Hs00998582_g1) for Dicer1.
One-tail t-test was performed in Excel to calculate the statistical significance with α=0.05.
We would like to thank Dr. Sarah Wheelan, Dr. Jeffrey Corden, and Dr. Stephen Desiderio for valuable advice. This work was supported by NIH grant 43222 and by the Howard Hughes Medical Institute.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.