ABSTRACT
We have tested the inhibitory potential of peptide nucleic acid (PNA) on in vitro reverse transcription of the HIV-1 gag gene. PNA was designed to target different regions of the HIV-1 gag gene and the effect on reverse transcription by HIV-1, MMLV and AMV reverse transcriptases (RTs) was investigated. We found that a bis-PNA (parallel antisense 10mer linked to antiparallel antisense 10mer) was superior to both the parallel antisense 10mer and antiparallel antisense 10mer in inhibiting reverse transcription of the gene, thus indicating triplex formation at the target sequence. A complete arrest of reverse transcription was obtained at ~6-fold molar excess of the bis-PNA with respect to the gag RNA. At this molar ratio we found no effect on in vitro translation of gag RNA. A 15mer duplex-forming PNA was also found to inhibit reverse transcription at very low molar ratios of PNA/gag RNA. Specificity of the inhibition of reverse transcription by PNA was confirmed by RNA sequencing, which revealed that all tested RTs were stopped by the PNA/RNA complex at the predicted site. We propose that the effect of PNA is exclusively due to steric hindrance, as we found no signs of RNA degradation that would indicate PNA-mediated RNase H activation of the tested RTs. In conclusion, PNA appears to have a potential to become a specific and efficient inhibitor of reverse transcription in vivo, provided sufficient intracellular levels are achievable.
Replication of human immunodeficiency virus (HIV) invariably occurs through reverse transcription of the viral genomic RNA upon entry of the viral core into the host cell, as is typical for all retroviruses. Synthesis of the first cDNA strand, the minus strand, is initiated at a tRNA primer, which binds to the primer binding site located close to the 5'-end of the genomic RNA. The first cDNA transcript ends at the 5'-end of the RNA, which is characterized by a sequence R, also present at the 3'-end of the RNA. Removal of the RNA, by the RNase activity of the HIV-1 reverse transcriptase (RT), allows binding of the cDNA from the 5'-end to the R region at the 3'-end of the RNA and perpetuation of minus strand cDNA synthesis. Initiation of second (plus) strand DNA synthesis takes place at the polypurine tract (PPT) located near the 3'-end of the genomic RNA. The PPT shows some resistance to RNase H during reverse transcription, thereby allowing plus strand cDNA synthesis to initiate from the PPT-cDNA hybrid. Completion of plus strand cDNA synthesis is accomplished after one more step of strand transfer (1 ). Targeting the process of cDNA synthesis appears to be a logical approach for anti-retroviral therapy. Although a broad spectrum of anti-RT compounds extending from nucleoside analogues (azidothymidin and dideoxyinosin) to non-nucleoside inhibitors, such as tetrahydroimidazo[4,5,1-jk][1,4]benzodiazepine-2(1H)-one and tetrahydroimidazo[4,5,1-jk][1,4]benzodiazepine-2(1H)-thione derivatives, nevirapine, pyridinone and bisarylpiperazine, is currently available, emergence of viral strains resistant to these inhibitors has limited their use in the treatment of AIDS (2 ,3 ). Another area of possible anti-RT approaches is represented by antisense DNA oligonucleotides and oligonucleotide analogues. Strategies aimed at blocking HIV-1 RT have so far included competitive inhibition of the formation of RT/primer/template complex by template/primer DNA oligonucleotides (2 ), degradation of the template RNA through antisense DNA oligo-RNA hybrids, triggering the RNase H activity of HIV-1 (4 -6 ), inhibition of plus strand cDNA synthesis by triplex binding of the PPT-cDNA hybrid (1 ), inhibition of HIV-1 U5 long terminal repeat (LTR) integration by triplex binding of a polypurine stretch in the integrase binding site (7 ) and direct inhibition of cDNA extension by antisense template hybrids (8 ). Although some promising results have been obtained in vitro, such obstacles as inefficient in vivo deliverance, short half-lives of unmodified oligonucleotides in serum and cytoplasm and reduced specificity/affinity of modified oligonucleotides preclude a successful clinical application of antisense-based inhibition of HIV RT (9 ). Thus, the introduction of major improvements in molecular design is needed to render antisense oligonucleotides a viable therapeutic alternative.
In this study we set out to investigate the potential of peptide nucleic acid (PNA) as an inhibitor of HIV-1 RT. PNA is a DNA analogue in which the phosphate backbone has been substituted for an achiral and uncharged pseudopeptide backbone composed of N-(2-aminoethyl)glycine (10 ). This structure is virtually resistant to degradation by nucleases and proteases and ensures that PNA retains its full stability in plasma and cell extracts (11 ). Furthermore, PNA forms exceedingly stable triplexes with homopurine DNA or RNA targets in which one strand is engaged in conventional Watson-Crick pairing while the other PNA strand is bound in the major groove of the PNA-DNA duplex helix by Hoogsteen pairing (10 ). Triplex binding of PNA to a DNA or RNA target has been shown to inhibit the elongation process of the ribosome (12 ,13 ) and DNA polymerase (12 ) and preliminary experiments have also shown inhibition of MMLV reverse transcriptase (12 ) in vitro. In the present paper we have investigated the effect on the function of several RTs, including HIV-1 RT, of PNAs bound by triplex formation or duplex hybridization to selected regions of HIV-1 gag RNA.
To test the inhibitory potential of PNA on reverse transcription, first plasmid pSP64-gag-polyA, enabling in vitro synthesis of HIV-1-specific poly(A)+ RNA, was constructed. This was accomplished by inserting the 1.7 kb BglI fragment, containing ~0.4 kb of non-coding proviral/chromosomal DNA in addition to 1.3 kb of the gag gene, from the HIV-1 full-length clone pNLBT (14 ) into the BamHI polylinker site of plasmid pSP64polyA (Promega). In the resulting construct the insert came under control of the plasmid SP6 promoter and the end was bounded by a 30 base long poly(A)+ tail and a single EcoRI site.
For in vitro transcription the plasmid was linearized with EcoRI and 2 [mu]g were used in a reaction which consisted of 80 [mu]l transcription buffer (Boehringer Mannheim), 0.5 mM of each of the four nucleotides ATP, UTP, CTP and GTP (Pharmacia), 0.25 mM DTT (Boehringer Mannheim), 0.1 [mu]g/ml tRNA (Escherichia coli; Boehringer Mannheim), 0.05 [mu]g/ml heparin (Sigma), 100 U RNA polymerase (Pharmacia) and 25 U human placental RNase inhibitor (RNAguard; Pharmacia). The transcription reaction was kept for 40 min at 37oC, after which 5 U RNase-free DNase I (Boehringer Mannheim) were added and the incubation continued for 15 min at 37oC. For some in vitro translation experiments the RNA was artificially capped by adding 0.2 A250 units m7G(5')ppp(5')G (Boehringer Mannheim) and by lowering the GTP concentration to 0.1 mM. Purification of the gag-poly(A)+ transcript was done using a mRNA purification kit (Dynal) according to the manufacturer's instructions. The recovered RNA was quantified spectrophotometrically and analysed by agarose gel electrophoresis. Single reactions typically yielded 3-4.5 [mu]g full-length 1.7 kb transcript.
Synthesis of the first strand cDNA was primed with a DNA oligonucleotide primer (A318, 5'-ATCACTTCTGG GCTGAAAGC-3'), which is complementary to positions 486-505 in the gag open reading frame (ORF) of the gag-poly(A)+ transcript (Fig. 1 ). After incubation of the template RNA (200-250 ng, giving a final molarity of ~0.05 [mu]M in the assay) for 10 min at 80oC, the primer (10 pmol) as well as PNA or antisense oligonucleotides were allowed to anneal for 5 min at 37oC before the extension reaction was carried out at 37oC for 60 min. The reaction contained 50 mM Tris-HCl, pH 8.3, 75 mM KCl, 3 mM MgCl2,10 mM DTT, 5 U human placental RNase inhibitor, 2 mM dATP, dTTP and dGTP, 0.2 mM dCTP, 0.5 [mu]Ci [[alpha]-32P]dCTP (800 Ci/mmol; Amersham) and 5-8 U HIV RT (Amersham) or AMV RT (Pharmacia) or 100 U MMLV RT (Gibco) in a final volume of 10 [mu]l. After incubation the cDNA was ethanol precipitated and an aliquot was run on a 6% denaturating polyacrylamide gel. Quantitation of the cDNA yield was in an InstantImager (Packard).
To confirm the exact location of the blocking site, RNA sequencing of the gag-poly(A)+ RNA was performed in parallel reactions containing no PNA or PNA 897 at a concentration of 0.3 [mu]M. The procedure consisted of the same steps as the reverse transcriptase assay with the modifications that the A318 primer was 5'-end-labeled and the concentration of all four nucleotides was 1.3 mM. Further, 0.5 mM of either ddGTP, ddATP, ddTTP or ddCTP were present in four separate termination reactions. Radioactive labelling of the primer and subsequent purification were done using polynucleotide kinase (Pharmacia), [[gamma]-32P]ATP (Amersham) and Microspin G-25 columns (Pharmacia) according to the manufacturer's instructions. The sequencing products were resolved by denaturing PAGE (6%) and visualized on X-ray film by autoradiography.
To determine the potential of PNA-RNA complexes to trigger RNase H activity of HIV-1 RT, a radioactively labelled transcript from the pSP64-gag-polyA construct was prepared by performing in vitro transcription as presented above, with the exception that 1 [mu]Ci [[alpha]-32P]CTP (2000 Ci/mmol; Amersham) was added and at the same time the CTP concentration was lowered to 0.05 mM. The RT reaction was done exactly as described for the RT inhibition assay, except that [[alpha]-32P]dCTP was omitted. Any changes in the size distribution of RNA were analysed by fractionation on a 5% denaturating polyacrylamide gel.
For in vitro translation, a transcription/translation vector, pBAB72, enabling in vitro synthesis of full-length gag polyprotein p55 was obtained through the NIH AIDS Research and Reference Reagent Program (15 ). pDAB72 was linearized with BamHI and 2 [mu]g of the linearized template DNA was transcribed as described above for transcription of the pSP64-gag-polyA vector, with the exception that T7 RNA polymerase (Pharmacia) was used instead of SP6 RNA polymerase. Transcription of the pDAB72 vector gave rise to a RNA product of ~1.5 kb, consisting, from the 5'-end, of 11 ribonucleotides of non-coding RNA flanked by the full-length gag and a portion of the pol ORF. The transcript was analysed in and subsequently purified from a 1% low melting temperature agarose gel (NuSieve; FMC BioProducts). For purification, the RNA band was excised and melted at 67oC in 3 vol. sterile water followed by phenol/chloroform extraction (1:1) and standard ethanol/sodium acetate precipitation. Finally, the RNA was resuspended in DEPC-treated water and quantified spectrophotometrically. For in vitro translation, 200-250 ng of this RNA (amounting to a final concentration of ~0.02 [mu]M in the assay) was employed for peptide synthesis by a rabbit reticulocyte translation system (Boehringer Mannheim), according to the manufacturer's instructions. Translation was carried out in the presence of [35S]methionine (Amersham) and translation products were analysed by 10% SDS-PAGE and vizualized by electronic autoradiography in an InstantImager.
PNAs were synthesized as described (16 -18 ). The PNAs were purified by HPLC and their identity was confirmed by MALDI-TOF mass spectrometry. The bis-PNA, PNA 897, was linked by eg1:8-amino-3,6-dioxaoctanoic acid.
Primer and oligophosphorothioates were supplied by DNA Technology (Denmark). The primer was synthesized by phosphoroamidite chemistry and phosphorothioates were synthesized as has been described (19 ) and purified by HPLC.
First, we tested the inhibitory potential on RT activity of 10mer PNAs designed to form triplex hybrids with a highly conserved polypurine region of the HIV-1 gag sequence. [Please note that the purine-rich region of the gag region of the HIV-1 genome is not to be mistaken for the purine-rich region known as the polypurine tract (PPT) which occurs in the coding region of the integrase and also adjacent to the 3'-LTR.]
The three PNAs employed for this purpose were: PNA 283, H-T6CT3-NH2, which is complementary to position 74-83 of the HIV gag ORF (primary target for the tested triplex-forming PNAs; PTTP) and is also predicted to hybridize to position 332-341 of the HIV gag ORF (secondary target for the tested triplex-forming PNAs; STTP) with only one terminal mismatch (a U instead of an A at the 5'-end) when compared with positions 74-83; PNA 284, H-T3CT6-NH2,which is antiparallel to PNA 283; PNA 897, H-T6CT3(eg1)3T3JT6-LysNH2 (J is pseudoisocytosine), which is a bis-PNA consisting of both PNA 283 and PNA 284 motifs covalently linked.
The position of the A318 primer binding site (RT initiation site; RTIS) and the relative positions of PTTP and STTP for the triplex-forming PNAs are summarized in Figure 1 A. As is obvious from Figure 1 B, the predicted size of the unblocked RT product is 900 bp (corresponding to cDNA synthesis of 500 bases of the gag coding region and ~400 bases of the non-coding proviral/plasmid sequences), whereas RT product blocked at PTTP would have a size of 400 bp and blockage at STTP is expected to give a product of 160 bp.
The effect on reverse transcription was examined with PNAs 283, 284 and 897 individually and with PNAs 283 and 284 in combination. All PNAs, or combinations thereof, were tested in overall concentrations ranging from 0.03 to 3 [mu]M, i.e. at molar ratios of PNA/RNA ranging from 0.6 to 60. The inhibitory effect on HIV-1 RT activity is shown in Figure 2 . When the PNAs were used at 3.0 [mu]M concentration, almost complete and therefore presumably non-specific inhibition of the reaction was observed in all cases. At 0.3 [mu]M concentration, however, PNAs 283 and 284, and the combination thereof, resulted in three bands. These bands were consistent with the predicted size of the full-length product (900 bases), an RT product blocked at PTTP (of ~400 bases) and an RT product blocked at STTP (of ~160 bases). PNA 897 resulted only in appearance of the latter band, indicating a total block at STTP. At a concentration of 0.03 [mu]M, which represents a molar ratio of 0.6 with respect to the template RNA, PNAs 283, 284 and the combination of 283 and 284 resulted mainly in the full-length RT product, while blocking with PNA 897 resulted in the full-length cDNA as well as truncated cDNAs, predominantly of the size (400 bp) expected for blockage at PTTP. A similar pattern of inhibition was obtained when MMLV RT was used instead of HIV RT (data not shown). The above results strongly indicate that the tested RTs may become blocked through triplex binding of the PNAs at the two predicted sites in the gag RNA. The fact that antiparallel antisense PNA 284 is as efficient as the parallel antisense PNA 283 and that their combination slightly increases the efficiency of inhibition implies formation of triplex structures at the target sequences. Moreover, the notion that triplex clamping at the targeted sites may represent a paramount obstacle to reverse transcription was corroborated by the bis-PNA (PNA 897), which was found to be the most potent inhibitor.
To confirm the sequence specificity of the studied PNAs and to investigate the exact position at which reverse transcription is stalled by these antisense oligomers we performed sequencing of the RNA template using HIV-1, MMLV and AMV RTs. The representative data obtained with AMV RT and HIV RT are shown in Figure 3 . Similar results were also produced with MMLV RTs and show in accordance that DNA synthesis is blocked at 2-4 nt before the target sequence (in this case STTP). Thus, inhibition of reverse transcription appears to be accomplished in a site-specific manner through steric hindrance by the annealed PNAs. Moreover, the sequencing experiments also revealed an interesting difference in the ability of the tested RTs to act as polymerases for RNA sequencing of gag RNA. While AMV RT resulted, over a wide range, in the most readable sequence (probably due to equally efficient incorporation of the four ddNTPs), HIV RT was superior to both AMV RT and MMLV RT in resolving secondary structures of the RNA that otherwise resulted in compression of the sequencing products. Such compression could be seen at two places above the site of blockage when sequenced with AMV RT, while no or only slight compression was seen when the reaction was driven by HIV-1 RT (Fig. 3 ).
In an attempt to provide additional evidence that steric hindrance is the major (or only) mechanism mediating the effect of PNAs, we tested susceptibility of the inhibitory complexes to RNase H activity of the HIV-1 RT. The results shown in Figure 4 demonstrate integrity of the template RNA in the presence of PNA 897 and HIV-1 RT. Similar results were obtained with MMLV RT (data not shown). These findings rule out the possibility that RNase H activity could participate in the inhibitory mechanism of PNA 897 and are consistent with earlier reports showing that RNA/PNA hybrids are not substrates for RNase H (13 ). Notably, it was found that both the tested RTs bound all the employed RNAs resulting in complete retardation of the RNA when analysed by gel electrophoresis.
It has recently been shown that only triplex-forming PNAs and not duplex-forming PNAs are capable of producing full inhibition of in vitro translation elongation (13 ). To test if triplex-forming PNAs are also required in order to fully inhibit RT activity, a 15mer PNA of sequence H-TGGCCTTAACCGAAT-LysNH2 targetted to positions 54-69 (TDP in Fig. 1 ) in the HIV-1 gag ORF was analysed for its potential to inhibit reverse transcription of the gag template RNA. As demonstrated by the data in Figure 5 , we found that the duplex-forming PNA gave complete inhibition at molar ratios of PNA/RNA of 10 or more, while a ratio of 1 resulted in inhibition of ~70%. These results clearly show that triplex clamping of the RNA by the PNA is not necessary to obtain complete inhibition of RT polymerase activity and further that a 15mer duplex-forming PNA oligomer can inhibit cDNA synthesis at molar ratios of PNA/RNA comparable with the best results obtainable with triplex-forming 10mer PNAs.
With the aim of comparing the anti-RT performance of PNAs with other relevant DNA analogues, we tested the effect on RT activity of two phosphorothioate olignucleotides which were analogous in their base composition to PNAs 283 and 284. The two phosphorothioates were tested separately and in combination in the RT activity assay under identical conditions as were used for the PNAs (3.0-0.03 [mu]M). When tested at concentrations of 0.3 [mu]M (an oligonucleotide/RNA ratio of 6) or less the phosphorothioates failed to demonstrate any effect on HIV-1 RT activity. When tested at a concentration of 3.0 [mu]M, only the combination of the two phophorothioates was able to inhibit the RT reaction. There was a slight reduction in the full-length RT product but no truncated product, as would be expected for specific inhibition (results not shown).
Finally, we set out to investigate the effect of bis-PNA on in vitro translation of the gag-poly(A)+ RNA. Our data show that PNA 897 at a concentration of 0.1 [mu]M (molar ratio of PNA/RNA of ~5) or less had little or no effect on in vitro translation. Concentrations of PNA 897 in excess of 1 [mu]M resulted in complete inhibition of the translational process (Fig. 6 ). The complete absence of translational product is to be expected, as one of the targets for the PNA is situated only 50 bases from the start codon (Fig. 1 ) and a truncated translational product of only 18 amino acids is below the limit of detection in the employed reticulocyte translation system. Collectively, our data suggest that over a specific range of concentrations PNA can be utilized to selectively block reverse transcription without having an adverse, even specific, effect on translation.
Motivated by an urgent need to find methods for effectively suppressing expression of pathogenic genes such as oncogenes and retroviral genes, intensive research into the gene suppressing capacity of synthetic oligonucleotides and oligonucleotide analogues has been carried out, especially in the last decade. Although promising results have been obtained in a variety of cell-based or cell-free in vitro systems, no successful clinical application of the antisense technology has been reported so far (20 ). Remarkably, however, a recent report from Monia et al. showed that almost complete and specific inhibition of C-raf-1 kinase gene expression can be achieved both in vitro and in vivo using a carefully selected antisense phosphorothioate applied in nanomolar concentrations (21 ). The strong effect on C-raf-1 kinase gene expression is believed to depend on RNase H cleavage of the target mRNA triggered by binding of the antisense phosphorothioate to an optimal hybridization site within the structured mRNA molecule. Whether expression of HIV proteins can be inhibited in vivo with similar specificity and potency at similar low concentrations of phosphorothioates upon extensive screening of antisense candidates remains to be shown. So far, efficient suppression of HIV-1 in culture has mostly been reported at concentrations of antisense oligomers above 1 [mu]m and many reports have described non-specific inhibition of gene expression and toxic effects of antisense oligomers at these concentrations (9 ,20 ,22 ). Further, it has recently become evident that efficient inhibition of de novo infection with HIV-1 obtained in vitro with a variety of antisense phosphorothioates is to a large extent due to a completely non-specific process in which the phosphorothioates interfere with binding of HIV-1 to its target cells (20 ). However, even if it becomes evident that expression of HIV-1 proteins can be efficiently suppressed in vivo by one or more carefully selected antisense phosphorothioates at non-toxic and obtainable concentrations, the risk of HIV-1 developing escape mutants is high. Thus, the search for a more generally applicable antisense strategy seems necessary. Facing this challenge, we sought to examine the inhibitory effect of PNA on in vitro reverse transcription of the gag gene of HIV-1. Bearing in mind practical implications, we intended to specifically maximize inhibition of HIV-1 RT-driven cDNA synthesis, nonetheless, for comparative purposes we also employed some of the best characterized RTs of animal retroviruses. Our results clearly demonstrated that complete inhibition of RT-driven cDNA synthesis is attainable even at a RNA/PNA molar ratio of only 6, which is at least an order of magnitude less than that accomplished with phophorothioates of similar base composition. Similar superiority of PNAs over analogous phosphorothioates has recently been demonstrated in a study comparing their ability to inhibit telomerase activity in cell extracts and permeabilized cells (23 ). We have also confirmed previous reports that PNA/RNA hybrids are not a substrate for RNase H (13 ), thus indicating steric blockage of the RT as the direct cause of the observed inhibition. The most efficient PNA employed in our study was a bis-PNA capable of clamping the target in a triplex complex. Importantly, however, a 15mer PNA designed only to engage in duplex formation with the target was also found to be very efficient in blocking RT-driven cDNA synthesis. Furthermore, we found that PNA at molar PNA/RNA ratios sufficient to completely block cDNA synthesis had no effect on in vitro translation. This indicates that if used at such concentrations, RT activity can be targeted without risking non-specific and toxic interference with translation. Considering the high mutation rate of HIV-1, allowing the virus to escape most tested nucleotide analogues and virus enzyme inhibitors, it is interesting to mention that in addition to the completely complementary primary target motif, the triplex-forming PNAs also bound with high affinity to a secondary region which exhibited a single mismatch. This observation indicates that triplex-forming PNAs might retain high targeting potential even in the face of the well-recognized rapid HIV-1 mutation rate. From the therapeutic point of view, such a property, along with the previously demonstrated low intracellular toxicity and extraordinary stability (10 ,12 ), makes PNA a very promising candidate for inhibition of RT activity in vivo. It is important to emphasize that our study has been exclusively concerned with inhibition by PNA of the RT elongation process. Our finding that a duplex-forming PNA is very efficient in inhibiting RT elongation opens the possibility that PNAs could also be used as an inhibitor of the initial priming process of HIV-1 cDNA synthesis. A study of the potential of PNA as an inhibitor of HIV-1 RT initiation would need a system in which the RT is primed by tRNA, as is the case in vivo, and further the influence of nucleocapsid protein (NC) would have to be investigated, as this viral protein is known to be important for efficient and specific priming of HIV-1 RT in vivo (24 ,25 ). The influence of factors such as NC and topoisomerase I on HIV-1 RT elongation is currently being investigated (26 -28 ) and a more comprehensive study on the effect of PNA on RT processivity in different parts of the HIV-1 genome would also have to consider the possible implications of such factors.
This work was supported by the Danish Cancer Society and by the Danish Research Council. The skilled technical assistance of Elin La Coste was greatly appreciated.
REFERENCES