Binding of U1 snRNA to 5'ss D4 within a subgenomic env mRNA has turned out to substantially increase the env mRNA steady-state level. Therefore, the presence of D4 not only has been a prerequisite for splicing but also for the nuclear export of unspliced RNA through the action of Rev. To analyze whether 3'ss A7 also contributes to the steady-state level of the glycoprotein mRNA we inactivated 3'ss A7 and the two upstream minor 3'ss, A7a and A7b 234, by silent point mutations (A7 ^-, Fig. 2A) enabling analysis of these mutations in the glycoprotein-mediated syncytia assay.

To verify that the introduced mutations did not lead to activation of a cryptic 3'ss we additionally compared the subgenomic HIV-1 transcripts by Northern blot analysis of the respective poly(A)⁺ RNA fractions following transient transfection of HeLa-T4⁺ cells with either the subgenomic env expression vector SV E/X tat^- rev^- or SV E/X tat^- rev^- A7^- (Fig. 2B). Due to mutations of the tat and rev ATG translational initiation codons, these vectors express neither Tat nor Rev. Thus, in the absence of Rev transfection with SV E/X tat^- rev^- led almost exclusively to detection of spliced mRNA (Fig. 2B, lane 1). In contrast, after cotransfection with a Rev-expressing plasmid the majority of the detected mRNA was the unspliced poly(A)⁺ glycoprotein mRNA (Fig. 2B, lane 1'). As expected, mutations of all three 3'ss, A7, A7a and A7b, led to complete loss of any detectable spliced transcript indicating that no cryptic splice acceptor was significantly activated (lane 3). However, in the presence of Rev the amount of unspliced poly (A)⁺ env mRNA was unaffected by the presence or absence of a functional splice acceptor (cf. lane 1' with 3') demonstrating that the 3'ss mutations did not decrease the pool of unspliced poly (A)⁺ transcripts. This contrasted the previously shown 5'ss dependency of spliced and unspliced transcripts 28 (cf. lane 1 with 4 and 1' with 4'), i.e. the lack of U1 snRNA-binding to the 5'ss leads to env RNA degradation (see hGH detectability in lanes 2, 4, 2', and 4'). The results of the Northern analysis were confirmed by glycoprotein expression analyzed by Western blot and syncytium formation (data not shown). Together these results demonstrate that a 3'ss is dispensable for Rev-mediated env expression. Moreover, these results show that the protective function of U1 snRNA binding is independent of the recognition of the 3'ss during progression of spliceosome formation.

To exclude the possibility that the requirement for U1 snRNA complementarity for protection of the transcript was caused by an RNA surveillance mechanism detecting a functional 3'ss in the absence of a 5'ss we mutated both the 5'ss (D4^-, Fig. 2A) and 3'ss (A7^-) and analyzed the steady-state levels of total poly(A)⁺ RNA. In the absence of both the 5' and 3'ss RNA could still not be detected irrespective of the presence or absence of Rev (Fig. 2B, lanes 2 and 2'). This indicates that the U1 snRNA dependency for the expression of this subgenomic env mRNA was not due to an unpaired/cryptic splice site but was intrinsic to the transcript sequence.

Increasing the complementarity between the 5'ss D4 and U1 snRNA did not lead to a decrease in env expression, indicating that even in the presence of a strong 5'ss Rev-regulated env mRNA transport was not impaired 2329. To specifically investigate the influence of the strength of the 3'ss on Rev-mediated glycoprotein expression we improved the strength of 3'ss A7 in its context of a subgenomic glycoprotein expression vector. To achieve this, the suboptimal BPS was attenuated and a new BPS with higher complementarity to U2 snRNA was created further downstream. Additionally, the canonical AG dinucleotides of the cryptic sites A7a and A7b were mutated to prevent an interference with potentially binding splicing factors and the pyrimidine content of the PPT (in the region between the new BPS and the intron/exon border) was increased from 48 % to 77 %. All these nucleotide changes were introduced as silent mutations except for one (Val to Ala at position 703), which was not expected to influence the fusogenic activity of the glycoprotein (Fig. 2A, A7⁺).

As expected, analysis of HeLa-T4⁺ cells transfected with this vector revealed that the introduced mutations improved the efficiency of A7 as evident by a dramatic increase in the amount of spliced transcript (Fig. 3A and 3B, cf. lane 1 with 2). This was also confirmed by in vitro splicing experiments with the respective splicing constructs (data not shown). In the presence of Rev however, almost no unspliced poly(A)⁺ message was observed (Fig. 3A, B, cf. lane 1' with 2'), suggesting that splicing, enhanced by the strength of the 3'ss, competes with Rev activity.

To address the question of whether a suboptimal 5'ss could compensate for an efficient 3'ss in Rev function we combined a 5'ss of intermediate complementarity to U1 snRNA (-1G3U, Fig. 2A) 28 with the efficient 3'ss A7⁺. In agreement with our previous results, in the presence of A7 this intermediately strong 5'ss led to a 2–3 fold decrease in the amount of RNA (Fig. 3A, cf. lanes 1 and 3, 1' and 3'). However, while the ratio of spliced to unspliced transcripts (Fig. 3A, s/us) was altered only 3-fold, in the presence of A7⁺ this ratio increased up to 25-fold irrespective of the strength of the 5'ss (Fig. 3). This finding demonstrates that Rev activity is specifically and inversely dependent on the efficiency of the 3'ss A7.

To determine the sequence requirements of a 3'ss compatible with Rev function in more detail, we constructed a single-intron splice reporter based on a truncated HIV-1 tat/rev intron harbouring the RRE (Fig. 4A) and analyzed 3'ss A5 because of its complexity. A5 exhibits a discontinuous pyrimidine stretch and overlaps with the competing alternative 3'ss 4c, 4a and 4b. Moreover, ten BPSs have been experimentally mapped in this region, five of which are associated with splicing at 3'ss A5 1430 (see Fig. 5, constructs A4cab and A5). Since the AG-dinucleotides and BPSs can compete for binding of splicing factors we mutated them consecutively (Fig. 4A): First the AG dinucleotides of 3'ss A4c, a and b were changed to CG (AG^-) to exclude splicing at these positions. Next, the complementarity between the 5' BPS (named BPS1 in Fig. 4A) and U2 snRNA was reduced while the complementarity of the 3' BPS (BPS2) was enhanced (b1^- b2⁺). Thirdly, the pyrimidine content was increased from 52% in the wild type 3'ss A5 to 60% (Py⁺) and 72 % (Py⁺⁺), respectively.

Following transient transfection of HeLa-T4⁺ cells with these constructs the poly(A)⁺ RNA was analyzed by Northern blot. Neither the mutations of the upstream AGs (Fig. 4B, lane 2) nor of the branch sites (lane 3) led to splicing at the 3'ss A5 but efficiently allowed Rev-dependent detectability of the unspliced transcript (lanes 2' and 3').

Spliced RNA was not detected until the pyrimidine content was further increased (lane 4 and 5). Remarkably, a pyrimidine content of 60% (Py⁺) was still compatible with a low-level of Rev function (lane 4') but in contrast, a highly efficient 3'ss due to a further increase in the pyrimidine content of only 12% (Py⁺⁺) was not (lane 5').

Removing the improvement of BPS2 (SA5 b1^- AG^- Py⁺⁺) reduced splicing efficiency 3-fold (cf. lane 5 with 6) and concomitantly restored Rev-compatibility (cf. lanes 5' and 6') in spite of the high pyrimidine content. This suggests a comprehensive effect of overall 3'ss strength on Rev activity.

Interestingly, we found no indication that the suboptimal BPS 1 and BPS 2 were competing with each other. Splicing was less efficient than in the construct with an optimal branch site (cf. lane 7 with 5), but enhanced compared to the construct with only one predicted suboptimal branch site (cf. lane 7 with 6). Reconstruction of the AGs of 3'ss A4c, a, b further decreased the level of spliced transcripts (cf. lane 7 with 8) but increased the level of the Rev-dependent unspliced RNA (cf. lane 7' with 8'). In general, the amount of unspliced transcript in the presence of Rev (Fig. 4B and 4C, lanes 1'–8') was inversely proportional to that of the spliced transcript. This confirms our findings shown in Fig. 3, that splicing efficiency driven by the 3'ss competes with Rev function.

The observation that the efficiency of the 3'ss competes with Rev function implicates that all HIV-1 3'ss should be inefficient to allow the export of unspliced transcripts necessary for virus replication. Indeed, this has been already reported by O'Reilly and coworkers 27 however, at the time of publication the knowledge of HIV-1 splice site regulation by cis-acting sequences was rather incomplete.

To differentiate between the contribution to the overall splice site strength of the splice site regulating elements in the 3' exonic sequences and the intrinsic strength of the HIV-1 3'ss we used a splice site swapping strategy and analyzed the HIV-1 3'ss with or without their natural downstream exonic sequences (Aex and A, respectively, Fig. 5). Each 3'ss included the experimentally defined or assumed, by complementarity to U2 snRNA, branch point sequence, the polypyrimidine tract and the AG dinucleotide. Because of their functional and spatial overlap the 3'ss A4c, a and b were experimentally considered as an entity. The 3'ss A6, which is located in the tat/rev intron, was not included in this analysis because its activity has been described in isolate HIV HXB2 but not in HIV NL4/3 which was used in this study 231. As reference sequences the non functional (A7^-) and the efficient 3'ss (A7⁺) mutants shown in Fig. 2 and 3 were also included.

Northern blot analyses of these constructs revealed that only 3'ss A2 and A3 led to detection of significant amounts of spliced mRNA in the absence of their natural 3' exonic sequences (Fig. 6A, lanes 2 and 3). Consistent with the results shown in Fig. 3 and 4 these two constructs showed the lowest number of unspliced transcripts in the presence of Rev (Fig. 6A, cf. lanes 2' and 3' with 1', 4'–6'). Thus, these results indicate that 3'ss A2 and A3 are the most efficient core 3'ss, here referred to as the intrinsic efficiency of the 3'ss. For all other 3'ss the intrinsic efficiency was low and significant amounts of unspliced message could be detected in the presence of Rev.

Interestingly, the opposite picture was obtained for the series of constructs where the downstream exonic sequences were included (Fig. 6B and 6C). Compared to their respective intrinsic efficiencies, splicing at A2 and A3 was decreased 3-fold and 1.5-fold in the presence of their downstream exons (Fig. 6C, cf. A2 with A2ex and A3 with A3ex). This is in accordance with the described ESS elements, consisting of three hnRNP A1 binding sites within exon 3 16, and hnRNP H and hnRNP A/B binding sites within exon 4 1832. Therefore, significant amounts of Rev-dependent, unspliced messages are only detectable if the intrinsic strength of these 3'ss is silenced by their downstream exonic sequence (cf. Fig. 6A, lane 2' and 3' with Fig. 6B, lane 2' and 3').

Since the alternative 3'ss A4c, A4a, A4b and A5 are all in close proximity to each other we tested whether these sites are regulated by the same bidirectional enhancer in exon 5 (A4cab5ex), which also leads to efficient splicing of the flanking 3'ss A5 and 5'ss D4 23. Alternatively, additional sequences upstream of this ESE may be sufficient to influence the strength of at least one of the 3'ss A4c, A4a and A4b (A4cabex). The result showed that in the absence of the bidirectional ESE in exon 5 none of these 3'ss could be adequately activated as evident by the absence of any spliced transcript (Fig. 6B, lane 4). Hence, the alternative 3'ss A4c, A4a, A4b and A5 seemed to be moderately activated by the same bidirectional enhancer in exon 5, still allowing Rev-mediated nucleocytoplasmic transport of unspliced transcripts (cf. Fig. 6B, lane 4' with lanes 5' and 6'). Comparison of the amount of spliced transcript from the constructs carrying either the BPS of all 3'ss A4c, A4a, A4b and A5 (Fig. 6C, A4cab5ex) or only the BPS for the 3'ss A4a, A4b and A5 (A5ex) showed a slight increase (30%) in the amount of spliced transcript of the latter. This suggests that competition of the four alternative 3'ss might also contribute to the inefficiency of splicing and that this is also supportive for the Rev-mediated export of the unspliced message.

To date A7 is the only splice site with a known splicing silencer in the intronic region and therefore we cannot distinguish between the impact of the suboptimal PPT and this ISS on the intrinsic inefficiency of this 3'ss (Fig. 6A, lane 6). However, splicing at A7 depends on activation by its flanking downstream sequences carrying the bipartite ESE3/ESS3 regulatory sequence (cf. Fig. 6A lane 6 with Fig. 6B, lane 7) 17212533. Thus, in this experimental context, the ESE clearly dominates over the ESS function.

Most strikingly, 3'ss A1 extended by its natural exonic sequence turned out to be the most efficient 3'ss of all (Fig. 6B, cf. lane 1 with 2–7). Even in the presence of Rev, only a very small amount of unspliced message was detected comparable to 3'ss A7⁺ (Fig. 6B, cf. lane 1' with 9'). Therefore, from these experimental results we conclude that exon 2 contains a strong splicing regulatory element, which has not been identified so far.

These results combined show that, although all HIV-1 3'ss are predicted to be weak on the basis of their intronic sequences, there are distinct differences in their intrinsic splicing efficiency. To co-ordinate both splicing and Rev function the strength of the individual 3'ss is finally regulated by cis-regulating ESEs and ESSs in their 3' exons.

Quantification of the spliced transcripts from three independent Northern blots in the presence and absence of the downstream flanking exonic sequences revealed that the exon 2 sequence improved splicing at the 3'ss A1 about 11-fold (Fig. 6C, cf. A1 and A1ex). A heptameric motif TGGAAAG occurred twice within this relatively short exon of only 50 nucleotides. Moreover, it is conserved in the different strains of the HIV-1 group M (Fig. 7). Consistent with our observation that at least two SR-binding sites are necessary for supporting U1 snRNA binding at 5'ss D4 34 (Freund and Schaal, unpublished data) we examined whether these heptameric sequences might constitute a bipartite ESE. Therefore, we generated a two-intron minigene construct with exon 2 as the internal exon and mutated either heptamer 1 (M1) or heptamer 2 (M2) (Fig. 8A). RT-PCR analysis of the transcripts following transient transfection of HeLa-T4⁺ cells revealed that mutating either of one of these heptamers totally abolished exon 2 inclusion (Fig. 8B, cf. lane ex2 with Δ M1 and Δ M2). Thus, this heptameric motif most likely constitutes a key element of an ESE in exon 2. Furthermore, it confirms our hypothesis that at least two putative binding sites are necessary to define a functional enhancer. Since GAAAGGA was predicted to bind SF2/ASF by ESEfinder 35 we analyzed SF2/ASF-binding by pull-down and subsequent Western blot analysis using a polyclonal antibody against SF2/ASF. As shown in Fig. 8C immunoblot analysis of proteins from HeLa nuclear extracts pre-incubated with either RNA of in vitro transcribed exon 2 or exon 2 carrying mutations Δ M1 and Δ M2 revealed that the introduced mutations led to reduced reactivity of a polyclonal SF2/ASF antibody with a band of the corresponding molecular mass of SF2/ASF-binding sites. To confirm the SF2/ASF-dependent exon 2 recognition two additional mutations were analyzed which were predicted by ESEfinder to specifically abolish SF2/ASF-binding. As expected these two constructs (Δ M1 SF2^-, Δ M2 SF2^-) led to a complete lack of exon 2 recognition (Fig. 8B) supporting the observation that M1 and M2 represent the SF2/ASF-dependent exon 2 splicing enhancer. Since we were interested in analyzing this newly identified ESE within an infectious molecular clone we set out to test for a silent point mutation predicted to specifically inactivate this enhancer. Based on computer analysis only one mutation was found that fulfilled the desired criterion (Δ M1 – 43). The mutation Δ M1 – 43 resulted in a slightly reduced loss of exon 2 recognition compared to the other mutations (Fig. 8B). The most obvious difference was the appearance of a comparable amount of unspliced transcript. Nevertheless, this result confirmed that the Δ M1 – 43 mutation affected exon 2 recognition.

Therefore, we inserted the Δ M1 – 43 mutation into the molecular clone NL4/3 and investigated its effect on viral replication in PM1 cells. Unexpectedly, no difference in the replication kinetic was observed throughout the infection period of up to 10 days (data not shown) suggesting that Δ M1 – 43 could not dramatically impair viral gene expression. However, analyzing the viral mRNA of the infected cells at day 5 post-infection by RT-PCR using primers specific for the upstream region of the HIV-1 genome revealed a different splicing pattern for the non-coding leading exons (Fig. 8A). As observed within the splicing reporter, the presence of the Δ M1 – 43 mutation led to a significant loss of exon 2 recognition (cf. band 1.2.4 in lanes NL4/3 and Δ M1-43) confirming that the newly identified ESE within exon 2 supports exon definition within the context of the viral genome. Furthermore, the lack of exon 2 recognition was accompanied by an increase in exon 3 recognition (cf. band 1.3.4 in lanes NL4/3 and Δ M1-43). These results support recent findings that the optional leader exons might not play a direct role in viral gene expression 36. The most abundant detectable spliced isoform corresponded to a 1.4 transcript which was not expected to be affected by the mutation. However, most surprisingly, the amount of vif mRNA was also not altered by the Δ M1 – 43 mutation (cf. band 1.2l in lanes NL4/3 and Δ M1-43), demonstrating that this mutation exclusively impaired exon 2 definition but not intron (D1 – A2) definition. To confirm these results functionally we compared the replication kinetics of NL4/3 and NL4/3 Δ M1-43 in the vif non-permissive and permissive cell lines CEM and CEM-SS respectively. Consistent with the PM1 infection experiments no difference in the viral replication kinetics could be observed between CEM and CEM-SS (data not shown) indicating that the level of vif expression is not impaired by the Δ M1 – 43 mutation during the infectious experiment up to 16 days.

To further investigate the apparent discrepancy between the transient transfection experiment using the splicing reporter (cf. Fig. 8B, lane Δ M1-43) and the infection experiments we specifically analyzed the effect of the Δ M1 mutation on splice site usage of A1 in the splicing reporter by excluding exon definition through deletion of 5'ss D2. Similar to the infection experiments 3'ss A1 usage involving in intron definition still occurs in the presence of Δ M1 mutation although to a somewhat lesser extent (Fig. 9B, D2^- Δ M1). In contrast, mutating both heptameric sequences (Fig. 9B, cf. D2^- Δ M1 with D2^- Δ M1/2) resulted in a total failure of 3'ss A1 recognition. Thus, the intron-containing Rev-dependent vif mRNA is less dependent on the strength of the bipartite ESE (if defined through the number of SR binding sites).

Oligonucleotides were synthesized and purified as previously described 58. Sequences are given in Tab. 1.

SVcrev was constructed by cloning the EcoRI-XhoI fragment from pUHcrev 59 into pSVT7. The SV40 early env expression vector SV E/X tat^- rev^- contains the EcoRI-XhoI fragment of pNLA1 60, a cDNA derivative of pNL4/3. The 5' ss mutations were constructed as previously described 28. To introduce the mutations A7^- and A7⁺ in 3'ss A7, overlapping oligos were PCR amplified (A7^-: #626, #628; A7⁺: #797, #798), the PCR products were restricted with AvaI/AflII (A7^-) and SpeI/XmaI (A7⁺), respectively, and cloned into the appropriate vector backbones.

The one-intron constructs (SV-SD/RRE/SA-pA) and mutations in the 5'ss were constructed as previously described 28. Mutations in the 3'ss were introduced by swapping the BstEII-XmaI fragment except for A7ex which was cloned as a BstEII-BamHI fragment. For A7⁺ overlapping oligos (#796/#797) were amplified. For A7 ^- a fragment was amplified using primer pair #985/#986 and plasmid SV E/X tat ^-rev ^- SA7 ^- as template. HIV 1 3'ss A1 sequence was amplified using pNL-gpt (kindly provided by Valerie Bosch, ATV-DKFZ Heidelberg) as template both with 5' primer #948 and 3' primer either #947 (A1) or #1089 (A1ex). All other HIV 3' ss were amplified using pNLA1 as template and primers #946 and either #945 (A2) or #1091 (A2ex), #936 and #939 (A3) or #1492 (A3ex), #940 and #941 (A4cab) or #943 (A4cab5) or #1088 (A4cab5ex), #942 and #943 (A5) or #1088 (A5ex), #800 and #1065 (A7) or #139 (A7ex). The A5 mutations were constructed by introducing PCR products from overlapping oligos (A5 AG ^-: #1591, #1483; A5 b1^-b2⁺AG ^-: #1586, #1484; A5 b1^-b2⁺AG ^-Py⁺: #1586, #1592; A5 b1^-b2⁺AG ^-Py⁺⁺: #1586, #1587; A5 b1^-AG ^-Py⁺⁺: #1482, #1389; A5 AG ^-Py⁺⁺: #1388, #1590). For A5 Py⁺⁺ (#942, #1481) and A5 Py⁺ (#942, #1486) pNLA1 was used as a PCR template.

To clone the three-exon-two-intron minigene reporter construct LTR ds ex2 two PCR fragments containing a 5'ss (#377, #918; BssHII-SalI) and a 3'ss (#931, #932; PstI-SalI) were inserted into LTR 1.4tatCAT, a vector coding for transcripts with the native HIV-1 tat 1.4 mRNA leader sequence 61, to generate LTR SD SA tatCAT.

Exon 2 including flanking intronic sequences of pNL-gpt was PCR-amplified with primers #1183 and #1913 and cloned into LTR SD SA tatCAT via EcoRI and PstI. Similarly, to insert the mutations of heptamer 1 and 2 the 3' PCR primer was substituted for #1913a (LTR ds ex2 hept.1) and #1913b (LTR ds ex2 hept.2) respectively.

All plasmid sequences can be obtained on request.

HeLa-T4⁺ cells 62 were transfected with FuGENE™ 6 (Roche Molecular Biochemicals) and total RNA was prepared 30 h after transfection by a modified guanidinium isothiocyanate protocol 63 using RNA-clean (Hybaid-AGS, Heidelberg). The poly(A)⁺ RNA from 80–100 μg total RNA was isolated with Dynabeads^® oligo(dT)₂₅ (Dynal, Oslo), subjected to electrophoresis on a 1.2% agarose-1% formaldehyde-gel, and blotted onto a positively charged nylon membrane (Roche Molecular Biochemicals). After UV crosslinking (0.5 J/cm²), the membrane was hybridized with digoxigenin-(DIG) labeled antisense RNA probes in a buffer containing 50% (v/v) formamide, 5 × SSC, 50 mM sodium phosphate (pH 7), 0.1% (w/v) N-lauroylsarcosine, 7% (w/v) SDS, 2% (w/v) blocking reagent (Roche Molecular Biochemicals), 50 mg yeast RNA/L at 68°C. To monitor transfection efficiency and RNA loading, the membrane was hybridized with a DIG-labeled antisense RNA probe specific for exon 5 of human growth hormone (hGH, expressed from cotransfected plasmid pXGH5 as a transfection control). HIV-specific RNA was detected by a DIG-labeled antisense RNA probe specific for the 3' end of env (LTRcenvpA^-, nt 8648–8887) and detection with anti-Digoxigenin-AP-Fab fragments (50 mU/ml; Roche Molecular Biochemicals) and chemiluminescence substrate (250 μM CDP-Star™; Roche Molecular Biochemicals) as previously described 28. Quantification was done with the Lumi-Imager F1 (Roche Molecular Biochemicals) and the LumiAnalyst™ 3.1 software.

Isolation of total RNA was performed using a modified guanidinium isothiocyanate protocol 63. Cells were washed twice with 2 ml of PBS and cell lysis was performed with 500 μl of buffer D [4 M guanidinium-isothiocyanat, 25 mM Na-Citrat pH 7, 0,5% N-Laurylsarkosin]; 7.6 μl of 2-mercaptoethanol, 50 μl of 3 M sodium acetate (pH 4), 500 μl of phenol and 100 μl of a chloroform-isoamyl alcohol mixture (24:1) were added and mixed for 15 s. After incubation on ice for 15 min, phases were seperated by centrifugation (10,600 × g, 4°C, 20 min). RNA was precipitated in 1 volume of isopropanol overnight. After centrifugation (10,600 × g, 4°C, 20 min) the RNA pellet was washed twice with 70% ethanol and dissolved in 10 μl of DMDC-ddH2O. Prior to reverse transcription, 4 μl of RNA samples were subjected to DNase I digestion using 10U DNase I (Roche Molecular Biochemicals) with 50 mM Tris (pH 7.5) and 10 mM MgCl₂ in a total volume of 10 μl at room temperature for 1 h. After DNase I inactivation at 80°C for 10 min, 4.5 μl of the DNase digested RNA samples were reversed transcribed with 200U SuperScript III RNase H-Reverse Transcriptase (Invitrogen) according to the manufacturer's protocol using 0.375 mM oligo(dT)₁₅ (Roche Molecular Biocemicals) or 0.02 μM sequence-specific oligo #1542 as primer. As a negative control for the remaining plasmid DNA contamination of each sample, a second assay was performed as described above but replacing reverse transcriptase with ddH₂O.

PCR was carried out with 1.25U AmpliTaq (Applied Biosystems) in a total volume of 50 μl according to the manufacturer' protocol in a Robocycler Gradient 96 Temperature Cycler (Stratagene). All primers were used at a final concentration of 0.2 μM. Spliced and skipped RNA was detected with the primer pair #1544/#1542 and hGH mRNA with primer pair #1225/#1224. Prior to PCR the cDNA reaction mixture was denatured at 94°C for 3 min. To determine the linear PCR-amplification range allowing a semi-quantitative estimation of the relative abundance of pSV-1-env and hGH mRNA, a preliminary PCR test series was carried out using the same cDNA sample but varying the PCR cycle numbers between 15 and 30 [94°C, 0.5 min; 52°C (pSV-1-env) and 56°C (hGH), respectively, 1 min; 72°C, 1 min]. The reactions were completed with a final elongation step of premature amplified products at 72°C for 10 min. Accordingly to the obtained results, PCR analysis was performed with 26 cycles for pSV-1-env as well as hGH PCR amplification.

PCR products were separated on 6% non-denaturating polyacrylamide gels, stained with ethidium bromide (10 min) and visualized with the Lumi-Imager F1 (Roche Molecular Biochemicals).

Periodat-labeled in vitro transcripts (1 nmol) of Eco47lII-linearized T7 Ex2 and T7 Ex2 Δ M1/Δ M2 plasmids carrying mutations of the heptameric motive were prepared using T7-MEGAshortscript™ (Ambion) and 0.1 M sodium m-periodat. The binding reaction to the adipic acid dihydrazide agarose (Sigma) was performed over night in 0.1 M NaOAc pH 5.0. Complex formation was performed in a 650 μl reaction volume containing 500 μl of HeLa nuclear extract (Cell Culture Center, Belgium) and 150 μl buffer D (20 mM HEPES-KOH pH 7.9, 100 mM KCl, 20% glycerol, 0.2 mM EDTA, 0.5 mM DTT) by incubation at 30°C for 30 min. After washing five times in 1 ml of buffer D the complexed RNAs were eluted by incubating the beads in 60 μl of 2 × sample buffer, heating at 90°C for 10 min, followed by fractionation on an 7% denaturing polyacrylamide gel. The proteins were blotted onto a PVDF membrane (Immobilon™ P, pore size 0.45 μm; Millipore) by electroblotting with 70V in transfer buffer (200 mM glycine, 25 mM Tris, 20% methanol) for 1 h. Blots were blocked o/n in PBS with 10% bovine serum albumin (BSA), 10% Tween^®-20. Protein detection was performed in PBS, 1% BSA, 1% Tween^®-20, with a goat polyclonal antibody raised against a peptide mapping near the C-terminus of SF2/ASF of human origin (Santa Cruz Biotechnology, Inc. C-19:sc-10255) for 1 h, washed three times, incubated with a horseradish peroxidase conjugated AffiniPure donkey anti-goat IgG (Jackson ImmunoResearch Laboratories, Inc. #705-035-147), washed four times, rinsed with water and visualized by a chemiluminescence detection system (ECL™-system and ECL™ hyperfilm, Amersham; Super Signal^® ultra, Pierce).

Cell cultures were maintained in RPMI 1640 medium containing 10% fetal calf serum (Pansystems GmbH) and antibiotics (penicillin and streptomycin). Viral stocks were prepared by transfecting 293T cells with provirus expression vectors, followed by ELISA (Innotest HIV p24 Antigen mAb; Innogenetics N. V.) of culture supernatants for p24 content. For HIV-1 infection, 5 × 10⁶ PM1, CEM or CEM-SS cells were resuspended in 500 μl culture medium and incubated at 37°C for 3 hours with 100 ng of each HIV-1 viral stock (HIV-1 NL4/3 wild type and ΔM1-43). After infection, cells were washed twice with PBS without Ca²⁺ and Mg²⁺ and further cultured in 5 ml medium for another 5 days. Subsequently, total RNA was isolated using Trizol reagent according to the protocol of the manufacturer (Invitrogen).