The development of ASV and HIV-1 reconstituted concerted integration systems has permitted a direct comparison of the two reactions as well as a vehicle to study the role of host proteins belonging to the HMG family. All of the ASV integration products sequenced that were stimulated by HMG-1 or HMG-I(Y) and used a wild-type donor DNA arose from a concerted mechanism. This is in contrast to the HIV-1 IN-dependent reaction in the presence of Mg2+, in which less than half of the HMG-2 and approximately 75% of the HMG-I(Y) integrants examined resulted from a concerted reaction. One may speculate that the higher percentage of nonconcerted integration events catalyzed by HIV-1 than by ASV IN is an indication that complexes between this IN and donor DNA ends are not as stable as complexes with ASV IN. Alternatively, this difference could reflect the lower specific activity of the HIV-1 IN or some difference in the multimer structure of the two enzyme preparations.
In the case of ASV IN, the percentage of nonconcerted integration events can be increased by introducing base changes into the LTR sequences (unpublished observations), which presumably alters the binding affinity of IN for the LTR recognition sequences. While HIV-1 IN appears to be less efficient in catalyzing concerted integration, the size of the duplication of the acceptor DNA at the site of integration is more homogeneous. All but one of the HIV-1 concerted integrants showed a 5-bp duplication of the acceptor DNA, characteristic of in vivo integration. In contrast, about 60% of the ASV-dependent integrants exhibited the expected 6-bp duplication (Fig. ). This could reflect differences in stability of protein-protein interactions between ASV or HIV-1 IN protomers with the ends of the donor and with the acceptor DNA. The configuration of the IN protomers could change the spacing of staggered breaks introduced into the acceptor at the site of integration and thereby alter the size of the duplications. The fact that both ASV and HIV-1 IN-dependent in vitro systems produce a high percentage of concerted DNA integration products suggest that they will be very useful for the discovery of new drugs to treat AIDS.
The results from these in vitro studies indicate that multiple HMG proteins can stimulate integration. However, our observation that HMG-I(Y) is the most effective HMG protein to stimulate concerted DNA integration by ASV and HIV-1 IN supports the notion that HMG-I(Y) facilitates integration in vivo. Recently, HMG-I(Y) has been detected in HIV-1 preintegration complexes isolated from infected cells (9
). Moreover, the activity of these complexes was reported to be dependent on the HMG-I(Y) protein. In contrast to the results with the purified system presented here, these investigators could not demonstrate stimulation of preintegration complexes by HMG-2. The reason for this difference is unclear.
The results of coprecipitation and gel shift experiments show that direct protein-protein interactions between IN and HMG proteins vary with the different proteins tested and are probably not required for stimulation. In addition, HMG proteins from several different species, including calf thymus and rat HMG-1, human HMG-I(Y), and human HMG-2, all stimulate integration of a given IN protein. These observations argue that the stimulatory action of HMG proteins does not require species specific protein-protein interactions. However, both stimulation of enzymatic activity and complex formation require HMG proteins that are competent to bind DNA. The truncated HMG-I(Y) protein (Δ50-91) that lacks the N-terminal-most DNA-binding domain motif (A-T hook region) and its acidic C-terminal tail region is as capable of stimulating integration as wild-type HMG-I(Y). The truncated protein also assembles a ternary complex detected by gel shift. In contrast, another mutant protein, HMG-I(Y) (II, III), which has a number of single amino acid substitutions in the last two A-T hooks that prevent protein-DNA interactions while retaining the capacity for specific protein-protein interaction, did not stimulate integration in vitro or gel shift the donor DNA into a ternary complex in the presence of IN. These results clearly indicate the HMG-I(Y) protein needs to associate with the donor DNA both to form a ternary complex with IN and to stimulate integration. In addition, they indicate that only the last two A-T hook DNA-binding regions and the intervening peptide backbone of HMG-I(Y) are necessary for this stimulation.
The DNA-binding domains of the HMG-1/-2 and HMG-I(Y) proteins have markedly different three-dimensional structures and somewhat different DNA-binding properties (6
). This raises the question as to the mechanism(s) by which apparently different proteins stimulate the same integration reaction. For example, the HMG-1/-2 proteins interact in a sequence-independent manner with the minor groove of DNA. The interaction occurs through two DNA-binding domains known as HMG-1 boxes (17
) which are a conserved set of amino acids folded into three alpha helices forming an L-shaped structure (22
). The HMG-I(Y) proteins also bind preferentially to the minor groove of A-T-rich regions of B-form DNA but through DNA binding domains known as A-T hooks (23
). When bound to DNA, these domains assume an extended planar crescent-shaped structure similar to the A-T-minor-groove-binding drugs netropsin and distamycin (13
). While these two classes of HMG proteins possess different folds, they both insert segments directly into the minor groove and can exert similar effects on the DNA to which they bind. This may account for the functional similarities observed with proteins from each class. As noted previously, among these common functional characteristics is the ability to bend and unwind DNA in vitro. For example, DNA ligase-mediated ring closure (i.e., cyclization) assays have demonstrated that both HMG-1 (21
) and HMG-I(Y) (24a
) are capable of bending short, rigid pieces of DNA (i.e., below the persistence length) into closed circles. Therefore, the simple notion has been advanced that both the HMG-1 (1
) and HMG-I(Y) (9
) proteins might stimulate concerted integration by bending the donor DNA to bring the ends into close proximity. This would facilitate concerted recognition of the termini by IN and accelerate the efficiency and/or rate of the in vitro integration reaction.
Alternatively, because HMG-1 (14
) and HMG-I(Y) (20
) proteins are capable of unwinding DNA substrates in vitro, it is possible that they function by modulating the helical twist at the ends of the donor DNA. This could improve the efficiency of nucleophilic attack on the acceptor DNA. Recent experiments (15a
) indicate that ASV IN unwinds and distorts DNA ends. This distortion appears to be required for viral DNA end processing, but the enzyme will also unwind and distort DNA ends that lack viral sequences. Others (26
) have shown that HIV-1 IN preferentially processes frayed DNA ends. Thus, it seems possible that the unwinding activity of HMG proteins could facilitate binding of IN proteins to DNA ends and their subsequent distortion. In earlier experiments (1
), we observed that HMG protein maximally facilitated concerted joining when preincubated with the donor but not the acceptor DNA. Preferential activity of HMG on DNA ends is consistent with the fact that the donor but not the acceptor DNA molecules possess free ends.
It is interesting that there is a striking similarity in the biochemical mechanisms determined for retroviral integration and the initial steps in immunoglobin gene V(D)J recombination, catalyzed by the cellular enzymes RAG1 and RAG2 (28
). Moreover, like retroviral integration, V(D)J recombination is stimulated by HMG protein family members (29