Considerable experimentation has been directed toward understanding the mechanism of HIV-1 strand transfer, a possible therapeutic target. The HIV-1 RT is a key component of the strand transfer reaction, although its specific roles are not fully understood. RT possesses several distinct activities, including DNA- and RNA-template-dependent DNA synthesis, RNase H cleavage, and strand displacement synthesis, all proposed to contribute to successful transfer. Analyses in vitro have led to the proposal of the invasion mechanism of transfer, described earlier. Since this transfer mechanism relies on the catalytic activities of the RT, it was reasonable to expect that the RT also direct the switching of templates from donor to acceptor.
In fact, the earliest analysis of strand transfer mechanism supported this expectation16
. In this system oligo-dT was extended over poly-rA. The reaction was performed in the presence of an RT trapping molecule. Results showed that a single RT molecule could mediate a series of transfers without dissociation producing a transfer product much longer than the average template length. This could only occur if the RT did not dissociate during the primer transfer. Moreover, since each primer had to have transferred many times in the presence of the trap, it appeared that RT dissociation during transfer either never occurred or occurred very rarely.
Consistent with this interpretation, HIV-1 RT was found to be capable of simultaneous binding to the primer, donor and acceptor strands of a transfer intermediate33
. Evidence for a three-strand complex with the RT implies that the active site of the RT is designed to mediate transfer while the templates are exchanging. Indeed, the formation of such complexes has been observed in different organisms as part of reaction intermediates.
Later, trapping experiments with a heteropolymeric set of substrates produced an entirely different result. In this case, the primer could extend on the donor in the presence of trap, but evidence of transfer to the acceptor and subsequent extension was entirely absent. This result was interpreted to mean that either the RT is obligated by requirements of the mechanism to dissociate during transfer, or, even though it could remain bound during transfer, it simply does not. Why were these opposite results obtained? One possibility lies in the structural dynamics of a homopolymeric T/A primer-template. Because the sequence is the same at every position, the primer can shift position on the template with no change in binding free energy. Such changes of position are also favored by the weak hydrogen bonding of AT base pairs, which should result in low transition energy for movement of the primer on the template. This situation could allow the primer to slide on the template into a single stranded 3′ overhang. The overhanging strand could then simply anneal to the next template, and then move back into the RT active site for more synthesis. Whereas strand sliding is not possible on the heteropolymeric template, it must employ a fundamentally different mechanism. Since natural transfer during infection involve heteropolymeric sequences, that mechanism is most likely the one used in vivo.
The detection of three strand complexes with the RT is consistent with RT-mediated transfer, but does not prove that it occurs. One can envision that although a three-strand complex can form, the actual completion of template exchange throughout the active site cleft of the RT may be blocked at some point.
Had measurements with the heteropolymeric substrates indicated that the RT does not dissociate during transfer, the result would have been unambiguous. However, as discussed above, the actual result showing that trapping the RT eliminates transfer had two equally probable interpretations. The first was that the RT almost always dissociates during the strand exchange. The second was that trapping all non-synthesizing RTs prevents polymerization-independent RNase H activity needed to facilitate transfer. We distinguished these possibilities by devising a substrate that allowed transfer in the absence of RNase H. Results showed that the trap still prevented transfer. We interpret this to mean that the RT dissociates during transfer. We emphasize that the result does not provide information on the importance of polymerization-independent RNase H for transfer, since the substrate was purposely designed so that that mode of cleavage was not required.
Does efficient transfer require polymerization-independent RNase H? This question may be even more difficult to address unambiguously than the issue of RT dissociation. We have attempted to perform strand transfer on a heteropolymeric template using progressively lower amounts of RT, to determine whether the transfer efficiency decreased. However, the low amount of synthesis precluded making an accurate measurement. Moreover, considering that the RT dissociates during transfer, the transfer efficiency may appear low at reduced RT levels because after the transfer there is a long delay before an RT comes to re-initiate synthesis on the acceptor template.
Is dissociation of the RT a fundamental requirement made necessary by the structure of the RT active site? Clearly it is not required with the homopolymer substrate. We considered that some natural variants of the RT might also transfer on heteropolymer templates without dissociation. To maximize our chances of seeing such a phenomenon, we tested the AZTr mutant RT, that displays an seven-fold greater substrate binding affinity than the wild type, having an (SG) insertion mutation between amino acids 69 and 70, and TAM. This finding that this mutant RT also dissociated during transfer suggests that common variants all dissociate, and that dissociation is the norm and not the exception.
Interestingly, when the E478Q RT was tested with NC protein, a small fraction of the extended primers transferred successfully in the presence of trap. This result suggests that the RT could have evolved to remain bound to the DNA primer terminus during the transfer reaction if there had been natural selective pressure to do so. The implication of this result is that the final strand exchange that leads to terminus transfer does not need protein-mediated assistance. In fact, the exchange may be slowed in its movement through the active site cleft in the RT. Such a kinetic disadvantage, might have naturally selected for RT variants that dissociate during transfer.
The invasion mechanism of transfer has been proposed to have three steps: acceptor invasion at a short gap in the donor template created by the RT-RNase H activity, propagation of the double strand formed between the primer and the acceptor RNA, and final primer terminus transfer. Based on the results we have obtained here, we propose that the entire process proceeds most efficiently without the direct help of the RT. The initial invasion would occur in a region of the primer-donor that has already cleared the back end of the RT. Propagation of the primer-acceptor hybrid would proceed to the active site cleft and possibly induce RT dissociation. DNA primer terminus transfer would then proceed to completion.