Many eukaryotic promoters contain either a TATA box or an initiator (INR), or both, as a core promoter element that serves to direct transcription initiation17. The TATA box functions to direct transcription initiation approximately 30 base pairs (bp) downstream, whereas an INR is a sequence motif that encompasses and orients the transcription start site, generally from a conserved internal C A (+1) 141617. Importantly, an INR can act independently to direct the assembly of transcription complexes that recruit RNA polymerase II. The terminal deoxynucleotidyl transferase (TdT) INR determines the position of the start site of transcription in the absence of other core promoter elements (ibid, above). Extensive mutagenesis and analysis of the function of variant sequence INRs suggested an approximate consensus INR of 5' Py Py A (+1) N T/A Py Py 3' (Figure 1A) (ibid, above). We observed the presence of a sequence in HIV-1 TAR region DNA, 5'CCAGATCTGAG3', that when reversed, 5'

CTCA

TCTGG

CTCA

TCTGG

CA⁺¹ N T/A

T

The HIV-1 LTR contains three putative initiator (INR) elements within the first 50 bp of the R region or TAR DNA (Figure 1A). Two INRs previously described are oriented in the sense direction and are located at the cap site and downstream of the transcription start site (Figure 1A, underlined)19. The HIV-1 TATA box is upstream of these two INRs and would be expected to enhance the functioning of even a weak INR element. The HIVaINR is located inbetween these two sites, but is oriented to start transcription in the opposite direction (Figure 1A, overbar, reverse arrow). We examined whether antisense transcripts were specifically produced off the HIV-1 LTR using an in vitro transcription reaction system (Figure 1, and described in U.S. patent 5,919,677). A variety of truncated fragments of the HIV-1 LTR were used as templates in these in vitro transcription reactions (diagrammed in Figure 1B, and not shown). Drosophila nuclear extracts contain RNA polymerases and transcription factors previously shown to interact effectively with INR elements and were used in the transcription reactions with the LTR templates2021. Primer extension or RNase digestion was used to map the origins of the purified RNAs synthesized in vitro from the designated HIV-1 LTR templates (Figure 1B, 1C). The expected size cDNA following primer extension for an antisense transcript originating from the HIVaINR was clearly identified using two separate biotin-labeled primers, as shown in Figure 1C (lanes 1,2, or 5,6 with the expected size products of 213 and 274 bp, respectively). As a control, the same in vitro synthesized and purified RNA samples were simultaneously assayed for sense RNA transcripts by primer extension and no cDNA product was observed (Figure 1C: lanes 8–10,12,13: expected size product would be 97 bp). Thus the putative sense orientation INR elements were not functioning in this system. However, they can be viewed as mutant internal control INRs within the same LTR DNA with the sequence surrounding the HIV-1 transcription start site, 5'CTGGGTCT 3', and at a site 37 nucleotides from the start site with the sequences 5'CTCTCTGG 3' (Figure 1A, brackets)19. Except for a 3 bp internal deletion, the latter sequence is identical to the reversed HIVaINR (Figure 1A). Additional controls included reactions with no template (NT) and no nuclear extract (lanes 7 and 14) or simply no template (lane 4), in which only the biotinylated primer was detected (Figure 1C). This indicated that the nuclear extract was not contributing any of the product bands observed.

These results suggested a preferential accumulation of the Drosophila transcription factors (and RNA polymerase) at the HIVaINR over the two putative sense INRs and identified for the first time a novel antisense promoter element within HIV-1 TAR region DNA. In addition, these experiments identified a single HIV antisense transcript originating from the region of the HIVaINR. If there had been multiple initiation sites off the R-U3 regions of the LTR, one would have expected to see additional product bands in these experiments. No additional bands suggesting alternative start sites were observed with any of the HIV-1 LTR templates studied. RNase digestion, performed simultaneously with primer extension, confirmed these results (not shown). These experiments also revealed that the 3' end of the antisense transcript in the same samples was not detected by the complementary U3 primer (Figure 1C, lane 3).

The more telling question was whether HIV-1 antisense transcription could be detected in vivo with human cells susceptible to infection by HIV-1. In many previous systems it has been difficult to detect the presence of antisense transcripts in vivo despite apparent biological effects22. In preliminary experiments to determine the appropriate primers for use in RT-PCR, we used transient transfection of HIV-1 LTR-containing plasmids or controls into human Jurkat T cells (see additional File 2 (Figure 2S) from U.S. patent 6,392029). We had to redesign primers several times before finding regions that were specific only for the HIV LTR, and that did not also pick up intrinsic cellular sequences because of the shared transcription factor binding sites (not shown). For instance, primers overlapping the transcription factor binding site SP1 or the HIVaINR within the LTR also detected intrinsic cellular sequences, presumably because of the many human cellular genes also containing SP1 or INR sites (not shown). One of the primer sites that reliably detected only the antisense transcript was the complementary Ava I primer, used during the reverse transcription step (Figure 4). However, this site represents a region of considerable sequence heterogeneity in the HIV LTR and needed to be specifically designed with the HIV-1 strain in mind18. The second primer 441 was biotin-labelled and was added during the polymerase chain reaction (PCR). Biotin-labelled 441 was derived from sequence following the initial 5' terminal hairpin of the HIV-1 antisense transcript (Figure 4). This non-radioactive labelling modification of RT-PCR enhanced the sensitivity and specificity of detection (Figure 2B and Additional File 2). We employed rigorous internal controls for the RT-PCR, and compared the same total RNA samples following DNase treatment or DNase and RNase treatment, in order to ensure that we were studying RNA, and not a DNA contaminant.

Many experiments indicated that the HIV-1 antisense RNA transcript is constitutively produced, with two examples given here (Figure 2B, and additional File 2). A single expected product band of 183 bp was observed following RT-PCR of purified RNA from Jurkat T cells transfected with pHIV-CAT vector containing the HIV-1 LTR (additional File 2, lanes 7a–12a)23. It was only observed with pHIV-CAT transfected cell RNA, but not mock transfected (no template, NT) or control Jurkat T cell RNA (additional File 2, lanes 1a–6a) indicating that intrinsic cellular RNA was not being detected. The expected product band was also only observed when the Jurkat T cells transfected with the HIV-1 LTR vector were amplified with the Ava 1 primer and the biotin-labeled 441 primer, but not with the Ava 1 primer and a biotin-labeled Mae 1 primer (additional File 2, compare lanes 7a–12a with 7b–12b). This indicated that the antisense RNA was initiating in the region of the HIVaINR inbetween the Mae 1 and 441 site as expected (diagrammed in Figure 2A). This also showed that the RNA samples were not contaminated with DNA from the transfected LTR vector (which would contain the Mae 1 site). Thus the HIV-1 antisense RNA originated between these two primer sites, situated 24 nucleotides apart in the HIV-1 TAR region DNA.

The HIV antisense transcript also initiated from the region of the HIVaINR in the stably-transfected U38 cells at levels comparable to or greater than HIV-1 sense transcripts in the absence of the HIV-1 Tat protein, as illustrated in Figure 2B. U38 is a human U937 derivative cell line containing stably integrated copies of the HIV-1 LTR promoter linked to the CAT gene24. The expected 183 bp size product band was observed with the HIV antisense RT-PCR (figure 2B, upper panel, arrow labeled as+). As well, HIV sense transcripts were produced in this cell line as indicated by the expected 283 bp product band in U38 RNA samples (HIV sense RT-PCR (figure 2B, lower panel). After stimulation with calcium ionophore and phorbol myristate ester (PMA), some increment in both sense and antisense transcription is observed (Figure 2B, *U38, Stim +). However, along with stimulation (and the production of more HIV-1 sense transcripts) additional antisense RT-PCR product bands are observed indicative of autopriming. One of these additional bands at 312 bp would be expected if the HIV-1 antisense RNA transcript extended from a 5' terminal hairpin to a predicted pseudoknot structure, for instance 25. All primers were functioning effectively, as demonstrated by the RT-PCR Internal Standard RNA (IS+) controls, included in each analysis, as well as by PCR + controls (Figure 2B). The primers were not contaminated with template, as shown by the identical reaction mixtures receiving primer alone (labelled pr, Figure 2B). RNase treatment of the U38 RNA samples eliminated all the RT-PCR product bands, whether following HIV antisense RT-PCR or sense RT-PCR (Figure 2B, labelled RNase +). This confirmed the authenticity of the RNA synthesized in vivo, and analyzed by RT-PCR. The quality of the total RNA was also monitored by simultaneous analysis RT-PCR for intrinsic cellular G3PDH RNA transcripts, and additional controls performed included no reverse transcriptase added, and DNA contamination controls in each RT-PCR analysis (not shown). Similar results were obtained with BF24, a monocyte cell line that had been stably transfected with the HIV-1 LTR linked to the CAT gene, or with a cell line containing an HIV-1 truncated provirus, HL2/3, following transfection with pHIV-CAT 2324.

There were three possible start sites for translation from the intrinsic antisense RNA produced utilizing the HIV-1 LTR as template (Figure 3 and Figure 4). Each translational start was in a separate reading frame and contained at least one presumptive stop codon (diagrammed as a filled box, Figure 3). A short ORF initiating from an internal AUG could encode an 81 aa protein, if a +1 ribosomal frameshift occurred (Figure 3c and discussed in U.S. patent 6,392,029). The initial AUG encountered on the antisense RNA could initiate translation in another reading frame, and with a +1 ribosomal frameshift would generate a 105 aa protein (diagrammed in Figure 3b). Translation could also conceivably initiate at a CUG start codon as has been described for human fibroblast growth factor 2 26. Initiation of translation from non-AUG codons has also been described for HTLV-127. This reading frame would generate a 108 aa protein in the wild-type virus and contained an early TGA stop codon (Figure 3a). The predicted amino acid sequences from each translation are shown in Figures 3 and 4. Translational recoding events such as read-through or ribosomal frameshifting through a stop codon are already utilized by retroviruses, including HIV-1 72829. HIV-1 employs translational recoding in order to translate the gag-pol precursor polyprotein. In addition, given the appropriate downstream RNA structure, the TGA codon may actually code for selenocysteine 30. The HIVaINR antisense RNA, in addition to a 5' terminal hairpin (Figure 4), has very extensive structure by Mfold analysis (not shown). Secondary/tertiary RNA structure of this intrinsically produced antisense RNA was already suggested by the RT-PCR experiments (Figure 2 and not shown). The presence of RNA structures such as hairpins or pseudoknots would suggest an inherent plasticity available for translational decoding 252829.

The HIV-1 antisense gene sequence from +14–368 consisting of 354 bp was introduced, with no modifications, into a recombinant vector encoding an epitope tag sequence, DYKDDDDK, or FLAG (diagrammed in Figure 5A). Expression of FLAG, situated at the carboxyterminal end of the recombinant HIV-1 antisense gene-FLAG fusion protein would indicate RNA and protein product(s) were synthesized. The vector, itself, contains no productive signal for translation initiation upstream of FLAG, such that translation must initiate off RNA containing the inserted HIV-1 antisense gene sequences (Figure 4). Two vector constructs were made: one with the HIV-1 antisense gene linked to the carboxyterminal FLAG sequences (HIV-AS-FLAG), and a negative control vector with the same HIV-1 sequences in the opposite orientation, corresponding to nucleotides 87–441 of the LTR (REV) (Figure 5A). These two constructs, the empty vector as an additional negative control (-), or a control vector containing irrelevant DNA but the same pCMV-FLAG vector (C) used in construction of HIV-AS-FLAG or REV were separately transfected into HL2/3 cells. HL2/3 cells contain a molecular clone of HIV-1, HXB2/3 gpt, and express Gag, Env, Tat, Rev, and Nef, but not reverse transcriptase 31. This cell line used in these experiments did not synthesize antisense RNA by RT-PCR and the HL2/3 proviral LTR was truncated by PCR analysis (data not shown). Because of the requirement for the HIV-1 Tat and Rev proteins for efficient transcriptional elongation and nuclear export for HIV-1 mRNAs, it was presumed these were also necessary to enable production of the HIV-1 antisense proteins. Following simultaneous control and HIV-AS-FLAG transfections, the cells were washed, lysed, and the lysate proteins were immunoaffinity-purified using mouse monoclonal anti-FLAG Ab (M2) affinity columns or resin split between lysate samples. The transfected lysate eluates were then concentrated.

Figure 5 demonstrates HAPs production. Given the finite HIV-1 DNA sequence of 354 bp of the HIV antisense gene inserted inbetween the CMV promoter of the vector and the multiple termination signals following the carboxyterminal FLAG sequences, the larger protein species selected on anti-FLAG Ab affinity resins can only be explained by multimerization. The choice of affinity tag, such as Myc, FLAG, or histidine, can strongly influence the oligomerization state of the resulting recombinant protein 32. Oligomerization of HAPs was confirmed by Western blotting of the concentrated HIV-AS-FLAG eluates (Figure 5D, labeled anti-FLAG). Immunodetection of FLAG epitopes specifically identified proteins migrating as 53–54, 31–32, 27–28, and 24–25 kDa molecular mass bands, as well as fainter species ≈ 49, 44, and < 20 kDa. These were consistent with two of the predicted HAPs that contained the carboxyterminal FLAG epitope (diagrammed in Figure 3a and 3b). The expected molecular mass of the recombinant HIV-1 antisense protein including the FLAG epitope (AUG start) is ≈ 13.3 kDa; a dimer, trimer and tetramer would have a expected molecular mass ≈ 26,6, 39.9 and 53.2 kDa, respectively. The recombinant protein utilizing a CUG start would yield multimers of 13.7 kDa at ≈27.4, 41.1, and 54.8 kDa. The dimer and tetramer-sized molecular masses (migrating at 27–28 and 53–54 kDa) were especially prominent bands on multiple Westerns of HIV-AS-FLAG affinity-purified proteins probed with the directly-labelled mouse monoclonal antibody to the FLAG epitope (DYKDDDDK) (Figure 5D, anti-FLAG). To further add to the complexity, hetero-oligomerization could occur between the two larger HAP proteins (Figure 3a and 3b), or alternatively, between one of the larger HAP proteins (3a or 3b) and the shortest protein diagrammed in Figure 3c. The 81 amino acid protein predicted in the WT sequence of molecular mass 8.5 kDa would be slightly altered at the carboxyterminal end because of the recombinant vector sequences and is not in frame with the FLAG sequences. However, hetero-oligomerization between a and c (Figure 3) would yield dimer and tetramer forms of predicted 22 and 44 kDa molecular mass, respectively, which were bands repeatedly observed along with the 27–28 and 53–54 kDa species, both on silver-stained gels of HIV-AS-FLAG immunoaffinity-purified samples and, more faintly, on the anti-FLAG immunodetection (Figure 5D, anti-FLAG).

Of more particular significance is the prominent immunodetection of these same 44 and 54 kDa molecular mass proteins isolated from HIV-AS-FLAG immunoaffinity-purified samples on Western analysis by immunoglobulin purified from the sera of AIDS patients but not normal sera (Figure 5D, labelled AIDS sera and NORMAL sera).

Analysis of the immunoaffinity-purified HIV-AS-FLAG transfected cell lysates and control lysates was performed on a variety of gel types. These included Tris-glycine SDS-PAGE and Tris-Tricine-SDS-PAGE with the inclusion of 8 M urea in an attempt to adequately resolve the molecular masses of the proteins/polypeptides under 60 kDa in size (Figure 5B, mini-gel apparatus and 5C, large gel apparatus) 3334. Even following reduction with dithiothreitol (DTT) and alkylation with iodacetamide (Figure 5C, 5D, labelled D/I), the proteins remained highly oligomerized, possibly secondary to the FLAG tag, although the smaller predicted 13–14 kDa protein band was now observed (faintly) (Figure 5B, HIV-AS-FLAG). Control lysates from HL2/3 cells undergoing mock transfection (-), transfection with an irrelevant DNA/pCMV vector control (C) (Figure 5D), or transfection with the REV vector control (Figure 5C) did not produce any protein bands following immunoaffinity isolation, even with concentration in PEG (Figure 4B, 4D), providing the M2 affinity gel was performing adequately. This indicated that the HL2/3 cell line used in the transfection experiments detailed herein was not intrinsically capable of producing the HIV-AS-FLAG fusion protein(s) isolated on the M2 anti-FLAG antibody resins.

To further identify HIV-1 antisense proteins (HAPs) in HIV-1 infected human cells, which would not have an epitope tag, it was imperative to develop antisera that would recognize HAP(s) internal peptide sequences. We had peptides synthesized, based on the derived amino acid sequences for HAPs, for immunization of rabbits, to enable testing of whether HIV antisense proteins were intrinsically produced by wild-type virus in human cells. Peptides were synthesized from immunogenic regions of each of the proposed translation products indicated in Figure 3 (arrowheads). One peptide, designated 4041 (amino acid sequence underlined in Figure 3 and boxed in Figure 4) was particularly successful in inducing antibodies in rabbits. Polyclonal rabbit antisera from two rabbits immunized with the 4041 peptide was affinity-purified over a 4041 peptide-linked column (Biosource). The affinity-isolated anti-4041 Ab reacted specifically with the 4041 peptide, but not a control peptide on dot blot and Western blot analysis (compare lane 5 (4041 peptide (+)) and lane 6 (4040 peptide (-)) Figure 6, Figure 7D). In addition, on Western blot analysis the anti-4041 Ab recognized protein bands migrating as 28–30 and 53–54 kDa molecular mass species from HIV-AS-FLAG transfected cells that had been affinity-purified on the M2 anti-FLAG Ab column and PEG concentrated but not the pCMVTag4A empty vector control-transfected HL2/3 proteins that had also been affinity-purified and PEG concentrated (compare lanes 1 and 2, Figure 6). Thus the affinity-purified proteins from the recombinant HIV antisense gene transfected cells, but not control transfected cells were recognized three ways by immunodetection on Western blots: the 53–54 kDa species was recognized by AIDS sera but not normal sera, by biotin-labelled mouse monoclonal antibody to the (entire) FLAG epitope, and by affinity-purified rabbit antibody to an internal HAP epitope, 4041. The latter two antibodies also recognized a 27–28 kDa species in affinity-purified HIV-AS-FLAG transfected proteins on these same blots.

The directly biotin-labelled, affinity purified antibody to the internal HIV-1 antisense protein peptide 4041 was then used to probe human PBL cell lysates following 4–20% tris-glycine SDS-PAGE gel electrophoresis and transfer to membranes (Figure 7A, labelled B-4041 Ab). As a control, biotin-labelled rabbit anti-keyhole limpet hemocyanin (KLH) antibody was used in tandem to probe the same cell lysates (Figure 7A, labelled B-KLH Ab). This was a necessary control because the human lymphocyte cell lysates undoubtedly contained immunoglobulins (Ig) and other Ig-binding proteins. These would be capable of interacting with the immunoblotting antibody detection system following the renaturation that occurs upon transfer to a blot and washing. This capacity of antibody to recognize its antigen even following transfer to a blot from a denaturing gel is demonstrated with Figure 7D, where the rabbit anti-4041 antibodies (lane 1), but not control antibodies (lanes 2,3) bind specifically to the biotin-labelled 4041 peptide, but not a control biotin-labelled 4040 peptide (Figure 7D, compare lane 1 probed with 4041 peptide as opposed to 4040 peptide). Four HIV+ individual's PBL cell lysates, as well as a HIV negative PBL cell lysate control were analyzed by gel electrophoresis and then immunoblotted with either rabbit anti-KLH or rabbit anti-4041 antibody (Figure 7A). A protein band migrating at 8.5 kDa, appeared to react with both biotin-labelled antibodies and the interaction was judged to be non-specific (open arrowhead, Figure 7A). However, only immunodetection with the biotin-labelled anti-4041 Ab revealed two protein bands migrating at a molecular mass of 11.3 and 11.5 kDa from HIV positive PBL lysates, which were at the predicted molecular mass for the two larger HIV-1 antisense proteins of 105 and 108 aa in the wild-type virus (Figure 7A, 7C, labelled HAPs). Further analysis of this individual's HIV+ cell lysate proteins, run in replicate on the same gel and then simultaneously immunoblotted, showed that the 11.3/11.5 kDa species were recognized with the biotin-labelled rabbit anti-4041 antibody or AIDs antisera, but not normal sera, nor by control rabbit p24 antiserum (Figure 7B). Control HIV-negative human cell lysates (PBL 1) were also probed with the same antibody preparations and demonstrated no reactive bands at 11.3 or 11.5 kDa molecular mass (Figure 7E). However, antibody detection using both a primary and secondary labelled antibody universally had faint bands at 8–9 kDa and 14 kDa (Figure 7E).

Reducing disulfide bonds in the HIV+ cell lysate sample significantly removed the contribution of "non-specific" bands and allowed prominent detection of the HAPs 11.3 and 11.5 kDa molecular mass species with the affinity-purified antibody to the internal HIV antisense protein epitope (Figure 7C, immunoblot with biotin-labelled 4041 Ab). As an additional control, we used the affinity-purified antibody to the internal HAP epitope 4041 to perform affinity-isolation of the same human HIV+ PBL cell lysate. Once again the expected molecular mass bands of 11.3 and 11.5 kDa were prominent on the silver-stained gel, particularly when the eluate fractions were reduced and alkylated (D/I, Figure 7F).

HIV + and HIV – blood donors were recruited from an Infectious Disease Clinic at Buffalo General Hospital, after informed consent, as approved by the institutional review board (State University of New York at Buffalo). Blood was diluted in Hank's balanced salt solution (HBSS) (Gibco BRL), layered on top of Histopaque 1077 (Sigma), and then centrifuged at 425 × g for 30 minutes at room termperature. The peripheral blood mononuclear cell (PBMC) layer was removed, washed three times with HBSS, and then resuspended in serum-free X-vivo 15 media (BioWhittaker) for incubation in a plate at 37°C in 5% CO₂. After two hours of incubation, non-adherent cells were removed, washed in PBS, and then kept frozen in liquid nitrogen until experimental use, as described below. These cells were predominantly peripheral blood lymphocytes (PBLs).

The following reagents were obtained through the AIDS Research and Reference Reagent Program, Division of AIDS, NIAID, NIH: pHIV-CAT plasmid containing HIV-1 LTR U3 and R sequences 5' to the chloramphenicol acetyl transferase (CAT) gene from Drs. Gary Nabel and Neil Perkins23; the U38 and BF24 cell lines containing stably integrated and silent copies of the HIV-1 LTR-CAT sequences; the HL2/3 cell line containing stably integrated copies of the HIV-1 molecular clone HXB2/3gpt from Drs. Barbara K. Felber and George N. Pavlakis2431; human HIV immunoglobulin (HIV-Ig), and rabbit p24 antiserum.

In vitro transcription reactions were performed with the indicated HIV-1 LTR DNA templates (Figure 1B) and Drosophila nuclear extracts as a source of eukaryotic RNA polymerases, as described by the manufacturer (Drosophila Embryo Nuclear Extract Transcription System II, Promega). The pooled and gel-purified DNA templates were generated by PCR using primers containing specific HIV-1 sequences (Figure 2A), as previously described 63. Each in vitro transcription reaction was performed with 5 u Drosophila nuclear extract and the indicated DNA template, in 7.5 mM HEPES, pH 7.6, 60 mM K glutamate, 3.75 mM MgCl₂, 1.5 mM DTT, 3% glycerol, 0.5 mM each rNTP, 2.5 μg/ml creatine kinase, 0.2 mM creatine phosphate, 6.25 μg/ml polyI poly C, and 10 u RNase inhibitor. Following incubation for 1 hr at 22°C, the reaction mixtures were DNase I treated, phenol-chloroform extracted and ethanol precipitated to purify the RNA synthesized in vitro. Primer extension reactions using biotinylated HIV-1 specific primers (Figure 1B) and 3 u AMV reverse transcriptase (RT) in a reaction buffer consisting of 50 mM Tris pH 8.3, 50 mM KCl, 10 mM MgCl₂, 10 mM DTT, 1 mM each dNTP, 0.5 mM spermidine, and 2.2 mM Na-pyrophosphate were incubated for 60 min at 42°C64. A nuclear extract control, in which in vitro transcription was performed without added DNA template (no template; NT), was also analyzed by primer extension with either a sense primer (5'U3, Figure 1C, lane 4) or antisense primer specific for HIV sequences (3'510–530, Figure 1C, lane 11), and verified the lack of contaminating template contributed by the Drosophila nuclear extract or reaction components. Biotinylated primers utlized in primer extension reactions had the following sequences and are diagrammed in Figure 1B: 5' Ava I:

TAATACGACTCACTATAGGG

TAATACGACTCACTATAGGG

TAATACGACTCACTATAGGG

Human cell lines were expanded in culture with or without stimulation for 2 hours with calcium ionophore (A23187), 50 ng/ml, and phorbol myristate ester (PMA), 50 ng/ml. Total RNA was extracted from cells using TRIzol (Gibco BRL Life Technologies), and equally divided for separate treatments consisting of either two DNase I treatments (75 u, Gibco), or two DNase I treatments plus digestion with RNase A (70 μg/ml) and RNase TI (20 u), as described (Boehringer Mannheim Biochemicals: RNase Protection Kit). A proteinase K treatment (0.4 mg/ml) was essential following the initial DNase treatment. Nucleic acids were then purified by phenol-chloroform extraction and ethanol precipitation. For antisense HIV-1 RNA detection: the Ava1 primer complementary to antisense RNA was utilized (5'TTTCGTCACATGGCCCGAGAGC 3') during the reverse transcription step, then the Ava1 primer and a biotinylated 441 primer (5'CCAGAGAGACCCAGTACAGGCAAAA3') were utilized during PCR. These primers are illustrated in the context of the HIV antisense gene in Figure 4. For sense HIV-1 RNA detection: an antisense 3'CAT primer (5'CAAGAATGTGAATAAAGGC 3') (or alternatively 3'510–530 above) were used during RT, and 3'CAT (or 3'510–530) and biotinylated 5'472 primer (5'CAGATCTGAGCCTGGGA 3') were used during PCR. Reverse transcription was performed with 200 u MuLV-RT and the indicated primer in a reaction mixture containing 10 mM Tris, pH 8.3, 50 mM KCL, 2.5 mM MgCl₂, 1 mM each dNTP, 40 u. RNase inhibitor, and 3 μl of the treated RNA. The reaction mixture was then incubated for 35 min at 42°C. PCR was performed, after adding the appropriate biotinylated primer, 0.2 mM each dNTP, and 2.5 U Taq DNA polymerase, for 30 cycles of denaturing, reannealing and extension at 94°, 45 sec; 60°C, 45 sec; 72°C, 2 min., as described (Perkin Elmer, RNA PCR Core Kit). The RT-PCR products were analyzed by 3% agarose gel electrophoresis, followed by transfer to Biodyne B membrane and colorimetric detection, as previously described 63.

The internal standard (

IS+

IS+

IS+

IS+

IS+

IS+

IS+

HL2/3 cells 31 were grown to near confluency in 100 mm plates (~8 × 10⁶ cells) and transfected with either the recombinant HIV-1 antisense gene-FLAG vector (HIV-AS-FLAG); the opposite orientation-HIV-1 sequences within the same vector (REV-FLAG) or empty vector (-) as a negative control; or a luciferase-FLAG vector as an additional control using Transfectam, as described by the manufacturer (Promega). Following transfection and incubation for 2 hr, complete media consisting of DMEM/F12 (Gibco BRL), 10% fetal calf serum, penicillin-streptomycin, 1 × nonessential amino acids and anti-PPLO (Gibco BRL) was added, along with Ca ionophore (50 ng/ml) and phorbol myristate acetate (PMA, 50 ng/ml), and the transfected cells were incubated overnight at 37°C in 5%CO2. Cells were washed in Tris-buffered saline supplemented with 1 mM CaCl₂, and lysed with 1 ml lysis buffer per dish. Lysis buffer contained 25 mM Tris pH 7.4, 150 mM NaCl, 1 mM CaCl₂, 1% Triton X-100, 27 μg/mL aprotinin, 10 μg/mL leupeptin, and 100 mM PMSF). Following incubation on ice for 30 min, cells were scraped, transferred to microtubes and spun at 15,400 × g to pellet cell debris. The cleared lysates from each set of transfections were purified over individual mouse monoclonal anti-FLAG M2 affinity columns, and eluted with 0.1 M glycine (pH 3.5), per the manufacturer (Sigma). Pooled eluates were concentrated in dialysis tubing (MWCO = 500) by packing in PEG 8000, and were then analyzed on Tris-tricine-urea PAGE gels with or without treatment with dithiothreitol (DTT) or DTT and iodoacetamide (D/I) 33. Gels were electroblotted onto nitrocellulose membranes using an Xcell II blot module (Novex). The blots were blocked in either 1% BSA (for anti-FLAG Ab detection), 3% BSA/0.05% Tween-20 in TBS (for biotin-labeled anti-peptide antibody detection), or 5% non-fat dried milk/1% BSA, then directly probed with either biotinylated anti-FLAG M2 antibody (Stratagene), biotin-labeled rabbit anti-peptide antibody (Biosource), or indirectly probed with HIV-Ig (antisera) or normal sera, followed by biotinylated goat anti-human immunoglobulin (Fab₂) (Biosource). Normal mouse sera was added to the immunodetection steps with human sera. Colorimetric detection of the blots was then performed, as described 636465.

HIV-1 antisense gene sequences derived from pHIV-CAT23 were generated by PCR, gel-purified, and cloned into pTarget, as described by the manufacturer (Promega). The inserted sequences were digested with BamH1 (site contributed by pTarget) and EcoRV, gel purified, and directionally subcloned into the pCMV-Tag4A vector (Stratagene) that had also been digested with BamH1 and EcoRV and gel-purified, using standard methods 64. The recombinant HIV antisense gene-FLAG vector contains the CMV immediate early promoter upstream of the HIV antisense gene sequence, but no translation start signal except that intrinsic to the HIV antisense gene. In addition, the HIV antisense gene is linked to 9 amino acid sequences contributed by the vector in frame with the carboxy-terminal FLAG (DYKDDDDK) sequences and termination signals. REV-FLAG contains the identical HIV antisense gene insert sequences, but in the reverse orientation, within the pCMV-Tag 4A vector. Recombinant vector sequences were confirmed by sequencing.