Search tips
Search criteria 


Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
Mol Cell. Author manuscript; available in PMC 2011 March 31.
Published in final edited form as:
PMCID: PMC3068534

HIV-1 mRNA 3’ end processing is distinctively regulated by eIF3f, CDK11, and splice factor 9G8


A genetic screen previously identified the N-terminal 91 amino acids of the eukaryotic initiation factor 3 subunit f (N91-eIF3f) as a potent inhibitor of HIV-1 replication. Overexpression of N91-eIF3f or full-length eIF3f reduced the level of HIV-1 mRNAs in the infected cell. Here we show that N91-eIF3f and eIF3f act by specifically blocking the 3’ end processing of the HIV-1 pre-mRNA both in vivo and in vitro. Furthermore, the results suggest that eIF3f mediates this restriction of HIV-1 expression through the previously unsuspected involvement of a set of factors that includes eIF3f, the SR-protein 9G8 and the cyclin dependent kinase 11 (CDK11). We propose that eIF3 affects HIV-1 3’ end processing by modulating the sequence-specific recognition of the HIV-1 pre-mRNA by 9G8.

Keywords: HIV-1 restriction, RNA 3’ end cleavage, 9G8 binding sites


The generation of mature mRNAs in eukaryotes requires multiple processing steps, including capping, splicing, and polyadenylation, that are coupled to ensure proper processing (Maniatis and Reed, 2002). The mature 3’ ends of most eukaryotic mRNAs are generated by endonucleolytic cleavage of the primary transcript followed by addition of the poly(A) tail (Colgan and Manley, 1997; Wahle and Ruegsegger, 1999). In mammals, these reactions are catalyzed by a large multicomponent complex that is assembled on specific sequences in the pre-mRNA. The cleavage and polyadenylation factor (CPSF) recognizes the highly conserved hexanucleotide AAUAAA, whereas the cleavage stimulation factor (CstF) binds a more degenerate GU- or U-rich element downstream of the poly(A) site. In addition, the cleavage reaction requires mammalian cleavage factor I (CFIm), mammalian cleavage factor II (CFIIm), and poly(A) polymerase (PAP). Sequences outside the core poly(A) site may contribute to the efficiency of mRNA 3’ end processing. Upstream stimulatory elements (USE) and USE-dependent 3’ end processing have been defined for a number of genes that are involved in important physiological processes (Brackenridge and Proudfoot, 2000; Danckwardt et al., 2004; Hall-Pogar et al., 2005; Moreira et al., 1998).

HIV-1 replication requires the 3’ end processing of multiple alternatively spliced transcripts, as well as unspliced mRNA. The use of a single poly(A) site for both spliced and unspliced mRNAs might involve a unique manipulation of the host 3’ end processing machinery. The HIV-1 poly(A) site resides within the repeat region (R) that is present at both the 5’ and 3’ ends of the viral pre-mRNA. These R regions, which are part of a highly conserved hairpin structure, contain several controlling functional motifs, (Berkhout et al., 1995). The entire 3’ LTR of HIV-1 contains many regulatory sequences, namely the putative AAUAAA poly(A) signal, a G+U-rich region, and the USE element, all required for efficient cleavage and polyadenylation of the viral transcripts.

Among the mRNA 3’ end processing factors that interact with the splicing machinery is CFIm (Awasthi and Alwine, 2003; Millivoi et al, 2006). CFIm is an essential complex composed of a small subunit of 25 kDa and one of the large subunits of 59, 68, or 72 kDa (Ruegsegger et al., 1996). The preincubation of an RNA substrate with CFIm reduces the lag phase of the cleavage reaction performed in vitro (Ruegsegger et al., 1996). The CFIm 59- and 68-kDa proteins have a domain organization that is reminiscent of spliceosomal serine/arginine rich (SR) domains (Graveley, 2000). Members of the SR family of splicing factors contain one or more N-terminal RNA recognition motifs (RRMs) that function in sequence-specific RNA binding and a C-terminal domain rich in alternating arginine and serine residues, referred to as RS domain, which is required for protein-protein interactions with other RS domains (Smith and Valcarcel, 2000). SR proteins comprise a family of essential splicing factors that are often interchangeable. They bind multiple sites, including exonic enhancers, in pre-mRNA, and some remain bound to spliced mRNA. CFIm has been shown to interact directly with U2AF65 (Millivoi et al, 2006), an SR-family protein that participates in 3’ splice site recognition, and the SR proteins 9G8, SRp20, and hTra-2 (Dettwiler et al., 2004).

The interaction of CFIm with U1 snRNP (Awasthi and Alwine, 2003), U2AF65 (Millivoi et al, 2006) and the SR proteins Srp20, hTra-2b and 9G8 (Dettwiler et al., 2004) suggests a role for CFIm in the coordination of 3’ end processing and pre-mRNA splicing. Such a role is supported by the identification of CFIm as a component of purified spliceosomes (Rappsilber et al., 2002; Zhou et al., 2002; Chen et al., 2007). Polyadenylation is functionally linked to and enhances the splicing of RNA (Colgan and Manley, 1997; Scott and Imperiale, 1996). This observation suggests that the binding of CFIm to the pre-mRNA may be an early step in the assembly of the 3’ end processing complex, analogous to the role of SR proteins in spliceosome assembly. The observation that CFIm associates with the transcription elongation complex throughout the transcription unit is consistent with such a role (Venkataraman et al 2005). Processing factors involved in pre-mRNA splicing and 3’ end formation can influence each other positively (Gunderson et al., 1994; Kyburz et al., 2006; Wassarman and Steitz, 1993) and were found to be necessary for efficient 3’ end processing (Danckwardt et al., 2007).

The SR regions are the target of extensive phosphorylation (Graveley, 2000). A wealth of evidence established the crucial role of phosphorylation and dephosphorylation events in the control of constitutive and alternative splicing (Shin and Manley, 2004). Phosphorylation of SR proteins directly affects their subcellular localization (Misteli et al., 1998), protein-protein interactions (Xiao and Manley, 1997), and splicing activity (Graveley, 2000). Many protein kinases have been identified that phosphorylate general transcription factors or splicing factors in vivo and in vitro, notably including cyclin-dependent kinases (CDKs). These kinases represent a family of protein serine/threonine kinases that become active upon binding to a regulatory partner, usually a cyclin. CDKs were first identified as regulators of cell cycle (Sherr and Roberts, 2004), but recently have been implicated in transcription and RNA processing.

The SR protein 9G8 is a general splicing factor that shuttles between the nucleus and the cytoplasm and can promote nucleocytoplasmic export of mRNA (Huang and Steitz, 2001). The phosphorylation status of 9G8 is crucial for its splicing and export functions (Swartz et al., 2007). Recently it was shown that 9G8 co-immunoprecipitates with CDK11p110 and is a CDK11 substrate (Hu et al., 2003), indicating that CDK11 is likely involved in the regulation of pre-mRNA splicing. CDK11 is a ubiquitously expressed member of the CDK family. The human CDK11 is encoded by two distinct but highly similar genes– CDC2L1 and CDC2L2 (Cell division control 2 Like 1 and 2). Differential splicing of transcripts from these genes generates more than 20 CDK11 mRNAs that encode at least two related proteins, CDK11p110 and CDK11p58. During apoptosis, a third CDK11p46 isoform is generated by caspase cleavage of the p110 and p58 proteins (Shi et al., 1994).

We have previously described the establishment of a genetic screen to identify cellular proteins possessing anti-HIV activity (Valente and Goff, 2006; Valente et al., 2009). The rationale for these experiments was that exogenous overexpression of a particular host gene (or portions of it) might result in the inhibition of viral infection or expression of a reporter retrovirus. In such a screen we identified the N-terminal 91 amino acids of eIF3f (N91-eIF3f) as an inhibitor of virus gene expression (Valente et al., 2009). eIF3f was first identified as a component of the multi-subunit eIF3 translation factor which promotes binding of eukaryotic initiation factor 2 (eIF2), GTP and Met-tRNAi with the 40S ribosome to form a 43S pre-initiation complex (Hinnebusch, 2006). We showed that overexpression of N91-eIF3f or full-length eIF3f drastically restricted HIV-1 activity by reducing nuclear and cytoplasmic viral mRNA levels, targeting the 3’ long terminal repeat (3’ LTR) in the viral mRNA. Additionally, we demonstrated that the 3’ end processing of HIV-1 mRNA precursors was specifically reduced in N91-eIF3f expressing cells. Our results suggested a previously unsuspected role of eIF3f in HIV-1 mRNA 3‘ end processing. Here we further define the mechanism by which eIF3f specifically affects HIV-1 mRNA 3’ end formation. We show that eIF3f specifically interacts with 9G8 and CDK11 and together they alter the 3’ end processing of the HIV-1 pre-mRNA in a sequence-specific manner.


N91-eIF3f blocks wt HIV-1 replication

We previously have previously shown that a particular cDNA, dubbed H2, encoding the N-terminal 91 residues of the eukaryotic initiation factor 3 subunit f (N91-eIF3f) could potently block HIV-1 mRNA synthesis (Valente and Goff, 2009). To test its effect on wild-type HIV replication, we introduced the H2 cDNA into HeLa cells expressing both CD4 and CXCR4, the receptors used for HIV entry, and established HeLa CD4-CXCR4 clones expressing H2. The cells were then exposed to VSV-HIV-Puro reporter viruses. Clones expressing H2 cDNA consistently restricted infection, whereas control cells expressing the empty vector (pBabe-HAZ, (Gao et al., 2002)) were highly susceptible (Figure 1A). These clones were next challenged with the wild-type HIV strain pNL4-3 at three different concentrations. Aliquots of the viral culture supernatants were harvested for 10 days post infection and the reverse transcriptase (RT) activity of the virus in the culture medium was determined. We observed a drastic reduction of RT activity in the medium collected from cells carrying H2 cDNA, even at the highest concentration of virus (Figure 1B). Upon infection, empty vector control cells produced virus, formed syncytia, and detached from the tissue culture plate, resulting in a drastic drop of RT activity. The formation of syncytia in cells expressing H2 was rarely seen at any time (data not shown).

Figure 1
(A) One HeLa-CD4 cell clone expressing empty vector control pBabe-HAZ and three independent H2 cDNA expressing clones were tested for susceptibility to infection with VSV-HIV-Puro. (B) Replication of wt HIV-1 in resistant cell lines. HeLa-CD4-HAZ and ...

We further assessed the restriction of wild-type virus replication with a more sensitive assay (Figure 1C). We used HeLa CD4-CXCR4 cells stably expressing a luciferase gene driven by the 5’LTR promoter of HIV-1 as a reporter cell line responsive to Tat protein expressed by an incoming virus. The culture supernatants of the infected HeLa lines on day three post-infection (from Figure 1B) were used to infect this reporter cell line, and luciferase activity was determined 48 hours post infection. An almost 10-fold reduction of luciferase activity was observed when the supernatants originated from cells carrying H2 cDNA. Together these results clearly document the potent ability of H2 cDNA to block wild-type HIV replication, even when the amounts of virus are sufficiently high to kill empty vector control cells.

To test whether N91-eIF3f affected cell division or viability, we measured the distribution of cells through the cell cycle (Figure 1D) and their metabolic activity (Figure 1E). Overexpression of N91-eIF3f did not significantly compromise cell growth or metabolism.

The 3’ end of HIV-1 mRNA is processed inefficiently by nuclear extracts of N91-eIF3f expressing cells

Earlier work showed that the 3’ LTR was a critical target of eIF3f repression of HIV-1 mRNA formation, suggesting that precursor RNA 3’ end processing might be affected (Valente and Goff, 2009). We therefore assessed the specific activity of nuclear extracts from TE671 or HeLa cells expressing N91-eIF3f, or an empty vector control, for in vitro cleavage of an HIV-1 RNA at the site of polyadenylation (Figure 2A). A long (338 nt) 32P-labeled uncleaved precursor RNA was incubated with extracts in vitro, and cleavage was assessed by gel electrophoresis and autoradiography (Figure 2B). When proper cleavage occurred, a 237 nt 5’ cleaved product was detected. Extracts from TE671 or HeLa cells expressing N91-eIF3f consistently produced significantly less cleavage product as compared to extracts of control cells expressing empty vector. To confirm the specificity of the cleavage, extracts of HeLa cells expressing HAZ were incubated with the wild-type HIV RNA precursor (HIVwt) or a mutant with a deletion in the core poly(A) hexamer (HIVΔHex). The cleavage was dependent on the presence of the poly(A) hexamers as expected for the authentic reaction.

Figure 2
Processing efficiency of HIV-1 poly(A) site in vitro

We further tested the effect of adding recombinant eIF3f or N91-eIF3f to the cleavage reactions (Figure 2C-left). As before, extracts from HeLa-N91-eIF3f produced less cleavage product as compared to extracts of empty vector control cells (Figure 2C, lanes 1 and 2). The addition of recombinant eIF3f-GST or N91-eIF3f-GST protein, but not GST alone, to extracts of empty vector control cells (compare lanes 1, 3, 5 and 7) significantly reduced cleavage efficiency. The presence of N91-eIF3f or eIF3f reduced cleavage activity by 4 fold, as judged by densiometry (Figure 2C-right). Addition to HeLa-N91-eIF3f extracts also reduced the already low cleavage activity.

To test whether the reduced cleavage activity of HeLa H2 extracts was specific to RNAs encoded by the HIV-1 3’ LTR, we tested their ability to process 3’ end sequences of other RNAs. We tested RNAs from Drosophila melanogaster Notch 1, Adenovirus L3, and SV40 early region, containing canonical poly(A) signals, as well as RNAs from the human VTI1B (vesicle transport through interaction with t-SNAREs homolog 1B), Nab1 (NGF1-A binding protein1) and GluL (glutamine synthetase) genes, containing non-canonical 3’ poly(A) signals (Figure 2D-left). Precursor RNAs containing these different sequences for cleavage and polyadenylation were prepared and cleavage reactions with HeLa HAZ or HeLa H2 nuclear extracts were performed. As before, H2 nuclear extracts were consistently poor in processing the HIV 3’ LTR (compare lanes 1 and 2). Reduced cleavage was also observed for the adenovirus L3, Nab1 and GluL poly(A) sites (compare lanes 5 and 6, 11 and 12, 13 and 14). In contrast, equivalent cleavage activity between the two nuclear extracts was observed for the 3’ end processing of Notch1 and SV40 poly(A)s (compare 3 and 4, 7 and 8). Notably, increased cleavage efficiency was seen for the VTI1B poly(A) site. Percursor RNAs with or without the core poly(A) hexamer were included as controls. We conclude that N91-eIf3f was able to interfere with or enhance the efficiency of cleavage of the various 3’end regions in a complex sequence-specific manner.

Heterologous poly(A) signals downstream of HIV 3’ LTR relieve restriction by N91-eIF3f

To confirm that a 3’ end cleavage deficiency was the functional basis of N91-eIF3f mediated HIV restriction, we tested whether a second heterologous poly(A) signal placed downstream from the 3’ LTR in tandem could provide a functional site and so rescue the expression of the reporter in N91-eIF3f expressing cells. The BGH and SV40 poly(A) signals were cloned downstream from the 3’ LTR of the HIV-Puro vector (Figure 3A), and TE or H2 cells were transfected with these DNAs or control expression constructs. The number of puromycin resistant colonies was scored after growth in media containing puromycin (Figure 3B-left). Indeed, heterologous poly(A) sites downstream from the 3’ LTR were able to rescue expression of the constructs in H2 cells (compare construct 1 with 2 and 3). The same experiment was performed with cells expressing full-length eIF3f or empty vector control (Figure 3B-right). The same pattern of expression was found in eIF3f as in N91-eIF3f expressing cells: as before, supplementing the 3’ LTR with the heterologous poly(A) sites rescued expression. The rescued expression was similar to the total substitution of the 3’ LTR by heterologous poly(A) site (constructs 4 and 5). These results support the previous data that cleavage at the HIV-1 3’ LTR poly(A) site does not occur efficiently in the presence of N91-eIF3f.

Figure 3
Processing efficiency of HIV-1 poly(A) site in vivo

We next characterized the viral messages produced in the nucleus of N91-eIF3f expressing cells when a heterologous poly(A) site was placed downstream from the 3’ LTR. For this study we introduced constructs 2 and 3 (from Figure 3B) into cells expressing full-length eIF3f, or the empty vector control, selecting for transformants with low concentrations of puromycin to prevent selection of cells with lower eIF3f expression (Valente et al., 2009). Total RNA was extracted from these cells and cDNA was prepared using oligo dT as primer for cDNA synthesis. The ratio of 3’ LTR poly(A) site usage versus BGH poly(A) usage (Figure 2C-left) or SV40 poly(A) usage (Figure 3C-right) was determined by quantitative real time PCR (qRT-PCR) using specific primers to distinguish the two different messages. In cells overexpressing eIF3f there was an increased usage of the downstream poly(A) site as compared to the HIV-1 poly(A) site, which agrees with the previous results of inefficient cleavage at the 3’ LTR poly(A) site. However, this increase in heterologous poly(A) usage was not very substantial, with approximately 2 fold higher usage of downstream poly(A) site in eIF3f cells as compared to empty vector control. We propose that the very close proximity of a downstream functional poly(A) site promotes the usage of the upstream HIV-1 poly(A) site, and that it could be sufficient to recruit the cleavage machinery to the 3’ LTR, increasing the cleavage at both poly(A) addition sites.

eIF3f, CDK11 and 9G8 in vitro association

A potential insight into the mechanism of N91-eIF3f inhibition of HIV mRNA 3’ end processing was provided by the observation that eIF3f interacts directly, via its MPN domain, with the cyclin-dependent kinase CDK11, both in vitro and in vivo (Shi et al., 2003). CDK11, together with cyclin L, interacts directly with both the transcriptionally active hyperphosphorylated form of RNAPII and SR protein splicing factors, and functions in the regulation of alternative splicing (Hu et al., 2003; Loyer et al., 2008; Yang et al., 2004). Both hyperphosphorylated RNAPII (Cramer et al., 2001) and SR proteins (Dettwiler et al., 2004; Maciolek and McNally, 2007) functionally interact with components of the mRNA 3’ end processing machinery. Thus, CDK11 might serve to link eIF3f and mRNA 3’ end processing. Of particular interest in this link is the SR protein 9G8, which interacts directly, both in vivo and in vitro, with CDK11 and cyclin L (Hu et al., 2003; Loyer et al., 2008; Yang et al., 2004). 9G8 also specifically interacts with the mammalian poly(A) site recognition factor CFIm (Dettwiler et al., 2004).

To test whether a previously undescribed interaction between eIF3f and 9G8 exists and to confirm the previously reported interaction between CDK11 and eIF3f in our cell line, TE671, we performed co-immunoprecipitation experiments. TE671 cells were transfected with DNAs to transiently express 9G8-N-FLAG, CDK11p110-C-FLAG and eIF3f-myc, or the empty vector as negative control (Figure 4). The anti-FLAG monoclonal antibody M2 was used to immunoprecipitate the transiently expressed tagged proteins, and the products were analyzed by immunoblotting using either rabbit antibody raised against eIF3f or M2-anti-FLAG antibody. The FLAG-tagged CDK11 protein specifically interacted with endogenous eIF3f as well as overexpressed eIF3f-myc tagged protein (Figure 4, lanes 1 and 4), confirming previous reports (Hu et al., 2003). Furthermore, we observed that FLAG-tagged 9G8 specifically interacts with endogenous eIF3f and myc-tagged eIF3f (Figure 4, lanes 2 and 5). Together these results show that CDK11 and 9G8 both interact with eIF3f.

Figure 4
eIF3f, 9G8 and CDK11

Restriction ability of CDK11 and 9G8

To assess the involvement of CDK11 or 9G8 in the observed restriction of HIV-1, TE and H2 cells were transfected with plasmids expressing CDK11-C-FLAG or 9G8-N-FLAG, and the resulting cellular clones or populations were tested for susceptibility to transduction with VSV-HIV-Puro viruses (Figure 5). Clones expressing CDK11-C-FLAG consistently restricted infection, whereas empty vector control cells were highly susceptible (Figure 5A). Similarly, a population of H2-CDK11-C-FLAG cells and three sub-clones were similarly tested (Figure 5B), and again clones expressing CDK11 showed reduced infectivity as compared with controls. Together these results show that the overexpression of CDK11 has the ability to restrict HIV-VLPs in wild type cells, and to further increase the restriction already observed in H2 cells.

Figure 5
Restriction of HIV-1 mediated by CDK11 and 9G8

Next we assessed the effect of overexpression of 9G8 on HIV-1 infectivity. Populations of TE and H2 cells expressing 9G8-N-FLAG were tested as above for susceptibility to infectivity by VSV-HIV-Puro viruses (Figure 5C). We observed a slight increase in infectivity: approximately 1.6 and 2.5 fold for TE and H2 9G8-N-FLAG populations respectively. These results suggest that increased expression of 9G8 stimulates HIV-1 expression. This effect is not a direct consequence of an increased splicing activity as the reporter genomes used for these experiments are unspliced.

We further overexpressed 9G8-FLAG in HeLa cells, prepared nuclear extracts, and performed both cleavage assays and electrophoretic mobility shift assays (EMSAs) with RNAs spanning the HIV 3’ LTR region. We did not observe any difference in cleavage efficiency or protein-RNA association upon 9G8 overexpression. The increase in infectivity in vivo (Figure 5C) is modest, and may be difficult to see in vitro. The artificial conditions of the lysates are more dilute and so may not respond to changes in concentrations of components.

The 9G8 binding site in the HIV-1 3’ LTR and its function

Inspection of the HIV-1 genome reveals that a sequence related to the consensus 9G8 binding sites, AGACKACGAY and ACGAGAGAY (Cavaloc et al., 1999; Schaal and Maniatis, 1999), is present upstream of the HIV-1 poly(A) site (Figure 6A). In contrast, sequences related to the 9G8 consensus are absent from the Notch and SV40 RNAs poly(A) sites that are unaffected by N91-eIF3f during processing in vitro (Figure 2D). To further study the role of 9G8 and its potential binding site upstream of the HIV-1 poly(A) site, we tested the activity of nuclear extracts of empty vectors and N91-eIF3f expressing cells to bind to this sequence (Figure 6B). As a control two G to C mutations were introduced in the potential upstream 9G8 site (AGACCAATGAC) (Figure 6A). We used nuclear extracts of HeLa-HAZ or HeLa-N91-eIF3f cells and assayed for complex formation by EMSA. We observed the formation of a specific slow migrating larger complex with both nuclear extracts (lanes 1 and 3). We did not observe any significant difference in the binding patterns of HeLa HAZ versus HeLa H2 nuclear extracts. The two point mutations in the potential 9G8 site abolished the ability of the RNA to form the slowly migrating complex (lane 1 versus 2 or lanes 3 versus 4). These results show that a specific protein complex is formed at the potential 9G8 site, and that mutations in this site affect complex formation. We further tested the ability of recombinant 9G8 fused to GST to bind RNA probes containing either the wild-type HIV 3’ LTR or the LTR with the 9G8 site mutation (Figure 6C). A specific RNA-9G8-GST complex formed with the wt probe and was abolished by mutation of the 9G8 site. GST alone did not form of any specific complexes with either probe (data not shown).

Figure 6
Processing efficiency of HIV-1 poly(A) site

Next we assessed the ability of HeLa-HAZ nuclear extracts to cleave the HIV-1 poly(A) site in an RNA containing the two point mutations in the 9G8 site (Figure 6D). The mutations in the 9G8 site reduced the cleavage activity by more than 2 fold as judged by measuring the yield of the 5’ cleaved product for each input RNA by densiometry.

Together these results indicate that the binding of SR protein 9G8 within the HIV-1 3’ LTR may contribute to efficient pre-mRNA 3’ end processing. To examine the effects of mutation of the 9G8 site on the expression of the viral genome, we introduced the mutations described above (disrupting the GAC repeats) into the HIV-Puro vector by site-directed mutagenesis. The wt and the mutant RNAs were packaged into VLPs and used to transduce TE cells (Figure 6E). Mutation of the 9G8 site reduced the efficiency of transduction by the virus by about 2 fold, which is close to the fold reduction observed in the cleavage efficiency in vitro (Figure 6D). We observed almost no effect on the efficiency of transduction of H2 cells by HIV-Puro when the 9G8 site was mutated (data not shown). We should note that there may exist some redundancy among the SR family of proteins, and that when 9G8 can no longer bind to the target RNA another SR protein might provide some residual function at another site.


Our results show that expression of a fragment of eukaryotic initiation factor 3 subunit f (eIF3f), or the inappropriate overexpression of the wild-type protein (Valente et al., 2009), can potently block HIV replication (Figure 1), by modulating the host cleavage and polyadenylation machinery to specifically decrease 3’ end processing of HIV-1 mRNA transcripts. The specific cleavage activity of the HIV-1 poly(A) site by nuclear extracts of N91-eIF3f-expressing cells was significantly reduced as compared to empty vector control extracts (Figure 2B). Furthermore, addition of recombinant N91-eIF3f to empty vector control nuclear extracts resulted in a decrease of cleavage efficiency (Figure 2C). The presence of N91-eIF3f, provided in vivo or in vitro, resulted in the reduction of pre-mRNA 3’ end processing. These results strongly argue that eIF3f directly impacts the RNA processing reaction itself, rather than acting indirectly by affecting the translation of other players in the system. It is noteworthy that recently Shi et al. (Shi et al., 2009) reported the purification and characterization of human mRNA 3’ end processing complexes. They identified ~85 proteins, including known and new 3’ end processing factors and over 50 proteins that may mediate crosstalk with other processes. Importantly, among the newly identified factors was eIF3f. These results provide support for the role of eIF3f in RNA processing and highlight the complexity and extent of the molecular architecture of the pre-mRNA 3’ end processing complex.

To functionally assess the inhibition of HIV-1 pre-mRNA 3’ end processing by N91-eIF3f in vivo, we built viral vectors in which the heterologous BGH or SV40 poly(A) site was placed downstream of the HIV-1 poly(A) site (Figure 3B). These viral vectors were expressed in N91-eIF3f-expressing cells at the same levels as in wild-type cells, indicating that the downstream poly(A) site provided the necessary sequences for proper maturation of the viral transcripts and so compensated for the inefficient use of the HIV-1 poly(A) site. These results confirm that the 3’ region of HIV-1 pre-mRNA is the critical target of N91-eIF3f action in vivo. We observed an enhanced, but not exclusive use of the heterologous poly(A) site when it was positioned downstream from the 3’ LTR (Figure 3C). We believe that the close proximity of the two sites is sufficient for the recruitment and efficient stabilization of the cleavage machinery to both poly(A) sites.

We further studied the impact of N91-eIF3f on the in vitro cleavage activity of other poly(A) sites from different genes (Figure 2D). Reduced cleavage activity in the presence of N91-eIF3f was observed for the adenovirus L3 poly(A) site, as well as human non-canonical Nab1 and GluL poly(A) sites. We observed no difference in cleavage activity between nuclear extracts from empty vector and N91-eIF3f expressing cells for mRNA substrates containing the Drosophila melanogaster Notch1 or SV40 early poly(A) sites. The cleavage activity for the VTI1B poly(A) sequence, remarkably, was higher in the presence of N91-eIF3f than in controls. These results suggest that the N91-eIF3f nuclear extract is not simply less active for all 3' end processing. In combination with our in vivo results, these data indicate that N91-eIF3f has a sequence-specific effect on pre-mRNA 3’ end processing.

A mechanism by which N91-eIF3f inhibits HIV mRNA 3’ end processing was suggested by the observation that eIF3f interacts directly with CDK11, both in vitro and in vivo (Shi et al., 2003). CDK11 and eIF3f co-localize within the nucleus, and in vitro, CDK11 specifically phosphorylates eIF3f Ser46 (Shi et al., 2003; Shi et al., 2006), a residue that is included in N91-eIF3f. CDK11 also interacts directly with the SR protein 9G8, both in vivo and in vitro, and 9G8 is an in vitro substrate for CDK11 (Hu et al., 2003; Loyer et al., 2008; Yang et al., 2004). We confirmed that eIF3f interacts with CDK11 and showed that there is a specific interaction between eIF3f and 9G8 in vivo.

We propose that 9G8 plays a critical role in promoting cleavage of the 3’ end of HIV-1 RNAs. 9G8 specifically interacts with the mammalian poly(A) site recognition factor CFIm (Dettwiler et al., 2004), and can be UV crosslinked to both nuclear and cytoplasmic polyadenylated RNAs in mammalian cells (Huang and Steitz, 2001). The binding of 9G8 to sequences upstream of the poly(A) site of the avian retrovirus RSV promotes mRNA 3’ end processing in vivo and in vitro (Maciolek and McNally, 2007). A sequence highly similar to previously identified 9G8 binding sites (Cavaloc et al., 1999; Schaal and Maniatis, 1999) is present upstream of the HIV-1 poly(A) site (Figure 6A). We found that mutation of the site abolished specific RNA/protein complex formation, and reduced the cleavage efficiency of the HIV-1 poly(A) site in vitro (Figure 6B and 6C). In contrast, sequences related to the 9G8 consensus are absent from the Notch and SV40 RNAs poly(A) sites that are unaffected by N91-eIF3f during processing in vitro.

Taken together, a network of physical and functional interactions has been established that suggests a mechanism for the specific inhibition of HIV mRNA 3’ end processing by N91-eIF3f. We propose that the endogenous eIF3f and CDK11 regulates 9G8 function and that inappropriately high expression of the eIF3f protein or interaction of N91-eIF3f with the complex alters the normal ability of 9G8 to participate in HIV-1 pre-mRNA 3’ end processing (Figure 7). The observation that N91-eIF3f requires endogenous eIF3f to inhibit mRNA 3’ end processing (Valente et al., 2009) suggests that this inhibition may reflect the disruption of the normal function of eIF3f within the nucleus. In support of this model, we show that overexpression of CDK11 has the same effect on viral restriction as the overexpression of eIF3f or N91-eIF3f. CDK11 induces restriction in TE cells, and further increases restriction of N91-eIF3f expressing cells. Additionally, overexpression of 9G8 has the opposite effect: it increases viral expression in TE cells, and relieves restriction in N91-eIF3f expressing cells. Whether this is an effect of 9G8 directly on cleavage needs to be further studied, as overexpression of 9G8 has been reported to increase export efficiency of viral messages (Jacquenet et al., 2005). These data suggest that the proper interaction of eIF3f, CDK11 and 9G8 is required for efficient HIV-1 pre-mRNA processing and viral gene expression. When this stoichiometry, or perhaps the phosphorylation of 9G8, is disturbed by overexpression of CDK11, eIF3f or N91-eIF3f, the HIV-1 pre-mRNA 3’ end processing is reduced and unprocessed RNAs are degraded in the nucleus.

Figure 7
Proposed model for N91-eIF3f inhibition of HIV-1 mRNA 3’ end cleavage

A SELEX analysis has shown that the CFIm 68/25-kDa heterodimer preferentially binds the sequence UGUAN (N=A>U≥C/G) present upstream of many poly(A) sites and thereby specifically enhances the binding of CPSF to the poly(A) core site (Brown and Gilmartin, 2003; Venkataraman et al., 2005). A distinctive feature of Notch1 and SV40 poly(A) sites is the presence of UGUAA elements, the preferred binding site for CFIm, whereas the HIV and adenovirus L3 poly(A) sites appear to contain fewer or less optimal sequences (Figure 7). This difference in poly(A) site affinity for CFIm may also explain why the 3’ LTR of HIV is more dependent on 9G8 to support the binding of CFIm to the pre-mRNA for a stabilization of the cleavage machinery.

The potential 9G8 binding site appears to be required for efficient poly(A) site processing in vitro (Figure 6). We observed a modest reduction in virus transduction, about 2 fold, when the 9G8 site was mutated (Figure 6E). This inhibition of transduction is not equivalent to the much more dramatic inhibition observed upon N91-eIF3f expression. The modest effect of the binding site mutation on virus infectivity might be explained by the fact that the members of the SR family of proteins may have some degree of redundant functions in vivo. HIV replication is strongly dependent on alternative splicing of the HIV-1 pre-mRNA. It is likely that alternative splice site choices are decided by a combination of splicing factors, including SR proteins, that associate with the pre-mRNA (Graveley, 2000). Therefore, it is possible that other members of the SR family, which includes at least 10 different proteins, may participate in the splicing and polyadenylation of the HIV RNA.

The 9G8 site is located at the 5’ end of the Nef coding region, and it is worth noting here its extreme conservation among different HIV-1 clades (Geyer and Peterlin, 2001) as well as its conservation during sequence variation due to disease progression in two HIV-1 patients studied (Asamitsu et al., 1999). Moreover, this site is present at the 5’ end of the HIV-2 nef gene; a very similar sequence, GAC AGAT GAC is present at the 5’ end of the SIVagm nef gene; and the sequence GAC TCAAAA GAC is present in the U3 region of EIAV (which does not encode the nef gene). This great conservation among different retroviruses may reflect the 9G8 requirement for efficient coordination of splicing-cleavage in the RNA maturation process in the retrovirus life cycle.

In sum, the data provide the first indication that the cleavage at the 3’ end of HIV-1 mRNA is sensitive to specific manipulation by host factors with a large impact on viral production. We also identify sequences in the 3' LTR that can respond dramatically and specifically to particular regulators. Manipulating these factors and their interactions may ultimately offer new means to suppress viral replication.

Experimental Procedures

Production of VSV-HIV reporter viruses

To generate VSV-HIV-Puro virus, 2 μg of the retrovirus vector pSCPW (a generous gift from Dr. Guangxia Gao) was transiently transfected with fugene (Roche) into 293T cells, along with 1 µg of HIV-Gag-pol DNA (p8.91) and 1 µg of pMDG. The culture supernatant was collected at 48 h, 72 h and 96 h, filtered, aliquoted and titrated on TE671 cells.

Titrations using VSV-HIV-1 reporter viruses

105 cells to be tested were seeded in 10-cm plates and the next day cells were infected with serial dilutions of VSV-G-pseudotyped viruses in 8 ml of complete medium. Forty-eight hours post infection, the selective drug was added to the culture media. TE671 cells were cultured with 5 µg/ml Puromycin. Culture media was changed every two days until visible colonies were detected. Colonies were stained with Giemsa and counted.

Real-Time PCR

PCR reactions were as described (Valente and Goff, 2006).

Protein expression

Full-length eIF3f or N91-eIF3f DNAs were cloned into pGEX-5X-1 (Amersham), and the encoded proteins were expressed in Escherichia coli as a C-terminal fusion with glutathione S-transferase (GST) and purified from bacterial lysates using Glutathione Sepharose.

Cleavage and Polyadenylation assays

Preparation of RNA extracts, nuclear protein extractions and 3’ end processing reactions were performed as described previously (Gilmartin, 1997; Gilmartin et al., 1995).

Preparation of pSCPW-BGHpoly(A) and pSCPW-SV40poly(A) plasmids

A two step cloning method was performed as described (Valente and Goff, 2006).

Cell Transfection and Immunoprecipitation

Cell transfections were performed using Fugene (Roche) with 15µg of each plasmid per 15cm dish. CDK11-C-FLAG and 9G8-N-FLAG expression vectors were kindly provided by Dr. Jill Lahti. Immunoprecipitations were performed as described (Loyer et al., 2008).


We thank Uriel Hazan, ICGM, Paris and Jill Lahti, St. Jude Children’s Research Hospital, TN for reagents.

STV was supported by Howard Hughes Medical Institute (HHMI), the AmfAR and the Portuguese Ministry of Education. SPG is an HHMI Investigator.


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.


  • Asamitsu K, Morishima T, Tsuchie H, Kurimura T, Okamoto T. Conservation of the central proline-rich (PxxP) motifs of human immunodeficiency virus type 1 Nef protein during the disease progression in two hemophiliac patients. FEBS Lett. 1999;459:399–404. [PubMed]
  • Awasthi S, Alwine JC. Association of polyadenylation cleavage factor I with U1 snRNP. Rna. 2003;9:1400–1409. [PubMed]
  • Berkhout B, Klaver B, Das AT. A conserved hairpin structure predicted for the poly(A) signal of human and simian immunodeficiency viruses. Virology. 1995;207:276–281. [PubMed]
  • Brackenridge S, Proudfoot NJ. Recruitment of a basal polyadenylation factor by the upstream sequence element of the human lamin B2 polyadenylation signal. Molecular and Cellular Biology. 2000;20:2660–2669. [PMC free article] [PubMed]
  • Brown KM, Gilmartin GM. A mechanism for the regulation of pre-mRNA 3' processing by human cleavage factor Im. Mol Cell. 2003;12:1467–1476. [PubMed]
  • Cavaloc Y, Bourgeois CF, Kister L, Stevenin J. The splicing factors 9G8 and SRp20 transactivate splicing through different and specific enhancers. Rna. 1999;5:468–483. [PubMed]
  • Colgan DF, Manley JL. Mechanism and regulation of mRNA polyadenylation. Genes Dev. 1997;11:2755–2766. [PubMed]
  • Cramer P, Srebrow A, Kadener S, Werbajh S, de la Mata M, Melen G, Nogues G, Kornblihtt AR. Coordination between transcription and pre-mRNA processing. FEBS Lett. 2001;498:179–182. [PubMed]
  • Danckwardt S, Gehring NH, Neu-Yilik G, Hundsdoerfer P, Pforsich M, Frede U, Hentze MW, Kulozik AE. The prothrombin 3'end formation signal reveals a unique architecture that is sensitive to thrombophilic gain-of-function mutations. Blood. 2004;104:428–435. [PubMed]
  • Danckwardt S, Kaufmann I, Gentzel M, Foerstner KU, Gantzert AS, Gehring NH, Neu-Yilik G, Bork P, Keller W, Wilm M, et al. Splicing factors stimulate polyadenylation via USEs at non-canonical 3' end formation signals. Embo J. 2007;26:2658–2669. [PubMed]
  • Dettwiler S, Aringhieri C, Cardinale S, Keller W, Barabino SM. Distinct sequence motifs within the 68-kDa subunit of cleavage factor Im mediate RNA binding, protein-protein interactions, and subcellular localization. J Biol Chem. 2004;279:35788–35797. [PubMed]
  • Gao G, Guo X, Goff SP. Inhibition of retroviral RNA production by ZAP, a CCCH-type zinc finger protein. Science. 2002;297:1703–1706. [PubMed]
  • Geyer M, Peterlin BM. Domain assembly, surface accessibility and sequence conservation in full length HIV-1 Nef. FEBS Lett. 2001;496:91–95. [PubMed]
  • Gilmartin GM. In vitro analysis of mammalian cell mRNA 3' processing. In: Richter JD, editor. mRNA formation and function. New York: Academic Press; 1997. pp. 79–98.
  • Gilmartin GM, Fleming ES, Oetjen J, Graveley BR. CPSF recognition of an HIV-1 mRNA 3'-processing enhancer: multiple sequence contacts involved in poly(A) site definition. Genes Dev. 1995;9:72–83. [PubMed]
  • Graveley BR. Sorting out the complexity of SR protein functions. Rna. 2000;6:1197–1211. [PubMed]
  • Gunderson SI, Beyer K, Martin G, Keller W, Boelens WC, Mattaj LW. The human U1A snRNP protein regulates polyadenylation via a direct interaction with poly(A) polymerase. Cell. 1994;76:531–541. [PubMed]
  • Hall-Pogar T, Zhang H, Tian B, Lutz CS. Alternative polyadenylation of cyclooxygenase-2. Nucleic Acids Res. 2005;33:2565–2579. [PMC free article] [PubMed]
  • Hinnebusch AG. eIF3: a versatile scaffold for translation initiation complexes. Trends Biochem Sci. 2006;31:553–562. [PubMed]
  • Hu D, Mayeda A, Trembley JH, Lahti JM, Kidd VJ. CDK11 complexes promote pre-mRNA splicing. J Biol Chem. 2003;278:8623–8629. [PubMed]
  • Huang Y, Steitz JA. Splicing factors SRp20 and 9G8 promote the nucleocytoplasmic export of mRNA. Mol Cell. 2001;7:899–905. [PubMed]
  • Jacquenet S, Decimo D, Muriaux D, Darlix JL. Dual effect of the SR proteins ASF/SF2, SC35 and 9G8 on HIV-1 RNA splicing and virion production. Retrovirology. 2005;2:33. [PMC free article] [PubMed]
  • Kyburz A, Friedlein A, Langen H, Keller W. Direct interactions between subunits of CPSF and the U2 snRNP contribute to the coupling of pre-mRNA 3' end processing and splicing. Mol Cell. 2006;23:195–205. [PubMed]
  • Loyer P, Trembley JH, Grenet JA, Busson A, Corlu A, Zhao W, Kocak M, Kidd VJ, Lahti JM. Characterization of cyclin L1 and L2 interactions with CDK11 and splicing factors: influence of cyclin L isoforms on splice site selection. J Biol Chem. 2008 in press. [PubMed]
  • Maciolek NL, McNally MT. Serine/arginine-rich proteins contribute to negative regulator of splicing element-stimulated polyadenylation in rous sarcoma virus. J Virol. 2007;81:11208–11217. [PMC free article] [PubMed]
  • Maniatis T, Reed R. An extensive network of coupling among gene expression machines. Nature. 2002;416:499–506. [PubMed]
  • Misteli T, Caceres JF, Clement JQ, Krainer AR, Wilkinson MF, Spector DL. Serine phosphorylation of SR proteins is required for their recruitment to sites of transcription in vivo. J Cell Biol. 1998;143:297–307. [PMC free article] [PubMed]
  • Moreira A, Takagaki Y, Brackenridge S, Wollerton M, Manley JL, Proudfoot NJ. The upstream sequence element of the C2 complement poly(A) signal activates mRNA 3' end formation by two distinct mechanisms. Genes Dev. 1998;12:2522–2534. [PubMed]
  • Ruegsegger U, Beyer K, Keller W. Purification and characterization of human cleavage factor Im involved in the 3' end processing of messenger RNA precursors. J Biol Chem. 1996;271:6107–6113. [PubMed]
  • Schaal TD, Maniatis T. Selection and characterization of pre-mRNA splicing enhancers: identification of novel SR protein-specific enhancer sequences. Molecular and Cellular Biology. 1999;19:1705–1719. [PMC free article] [PubMed]
  • Scott JM, Imperiale MJ. Reciprocal effects of splicing and polyadenylation on human immunodeficiency virus type 1 pre-mRNA processing. Virology. 1996;224:498–509. [PubMed]
  • Sherr CJ, Roberts JM. Living with or without cyclins and cyclin-dependent kinases. Genes Dev. 2004;18:2699–2711. [PubMed]
  • Shi J, Feng Y, Goulet AC, Vaillancourt RR, Sachs NA, Hershey JW, Nelson MA. The p34cdc2-related cyclin-dependent kinase 11 interacts with the p47 subunit of eukaryotic initiation factor 3 during apoptosis. J Biol Chem. 2003;278:5062–5071. [PubMed]
  • Shi J, Kahle A, Hershey JW, Honchak BM, Warneke JA, Leong SP, Nelson MA. Decreased expression of eukaryotic initiation factor 3f deregulates translation and apoptosis in tumor cells. Oncogene. 2006;25:4923–4936. [PubMed]
  • Shi L, Nishioka WK, Th'ng J, Bradbury EM, Litchfield DW, Greenberg AH. Premature p34cdc2 activation required for apoptosis. Science. 1994;263:1143–1145. [PubMed]
  • Shi Y, Di Giammartino DC, Taylor D, Sarkeshik A, Rice WJ, Yates JR, 3rd, Frank J, Manley JL. Molecular architecture of the human pre-mRNA 3' processing complex. Mol Cell. 2009;33:365–376. [PMC free article] [PubMed]
  • Shin C, Manley JL. Cell signalling and the control of pre-mRNA splicing. Nat Rev Mol Cell Biol. 2004;5:727–738. [PubMed]
  • Smith CW, Valcarcel J. Alternative pre-mRNA splicing: the logic of combinatorial control. Trends Biochem Sci. 2000;25:381–388. [PubMed]
  • Swartz JE, Bor YC, Misawa Y, Rekosh D, Hammarskjold ML. The shuttling SR protein 9G8 plays a role in translation of unspliced mRNA containing a constitutive transport element. J Biol Chem. 2007;282:19844–19853. [PubMed]
  • Valente S, Goff SP. Inhibition of HIV-1 Gene Expression by a Fragment of hnRNP U. Mol Cell. 2006;23:597–605. [PubMed]
  • Valente ST, Gilmartin GM, Mott C, Falkard B, Goff SP. Inaugural Article: Inhibition of HIV-1 replication by eIF3f. Proc Natl Acad Sci U S A. 2009 [PubMed]
  • Venkataraman K, Brown KM, Gilmartin GM. Analysis of a noncanonical poly(A) site reveals a tripartite mechanism for vertebrate poly(A) site recognition. Genes Dev. 2005;19:1315–1327. [PubMed]
  • Wahle E, Ruegsegger U. 3'-End processing of pre-mRNA in eukaryotes. FEMS microbiology reviews. 1999;23:277–295. [PubMed]
  • Wassarman KM, Steitz JA. Association with terminal exons in pre-mRNAs: a new role for the U1 snRNP? Genes Dev. 1993;7:647–659. [PubMed]
  • Xiao SH, Manley JL. Phosphorylation of the ASF/SF2 RS domain affects both protein-protein and protein-RNA interactions and is necessary for splicing. Genes Dev. 1997;11:334–344. [PubMed]
  • Yang L, Li N, Wang C, Yu Y, Yuan L, Zhang M, Cao X. Cyclin L2, a novel RNA polymerase II-associated cyclin, is involved in pre-mRNA splicing and induces apoptosis of human hepatocellular carcinoma cells. J Biol Chem. 2004;279:11639–11648. [PubMed]