HIV-1 integrase integrates a DNA carrying two LTRs into the yeast genome
yeast strain expressing HIV-1 IN was grown under optimal condition for the expression of an active enzyme (72 h, 0.1% glucose as determined previously (10
), confirmed by both western blot and in vitro
To detect the integration, we implemented the strategy shown in . A DNA fragment containing the two HIV-1 U3 and U5 LTRs ends flanking the zeocin-resistance encoding gene under the dependence of the yeast constitutive TEF1 promoter was introduced by electroporation into the cells after IN expression. Yeast cells were grown further for 1–5 h without selection to allow: (i) the interaction between IN and its substrate, (ii) nuclear import of both DNA and IN and (iii) integration of the DNA fragment. Approximately 109 viable yeast cells were then plated on a solid medium containing 400 μg/ml of zeocin and the zeocin-resistant clones were selected 5 days after plating. The first resistant clones were observed after 2 h of culture and the maximum number was attained after 4–5 h (). Zeocin-resistance analysis of the selected clones indicated a 100-fold increase compared to that of the initial yeast strain, thus confirming the acquisition of resistance.
Figure 2 Selection of zeocin-resistant clones from yeast cells expressing wt IN (A), D116A IN (B) or no IN (C). Yeast cells (109) were grown from 0 to 5 h after transformation with the DNA integration substrate containing the two LTRs under non-selective conditions. (more ...)
Using yeast strains expressing the inactive mutant D116A IN or no IN only few clones were observed ( and C), indicating that only a minor portion of the clones obtained with the cells expressing wild type IN (wt IN) was linked to an IN-independent process as the putative acquisition of mutational resistance.
Molecular analysis of the resistant clones
To confirm that the zeocin-resistant clones were representative of genuine integration events, several different molecular assays were performed. The expected 800 bp fragment was detected by PCR on total genomic DNA of the clones using internal primers for all the resistant clones obtained from h.RAD52+ (wt IN), whereas this product was not detected with the total DNA from h.RAD52+ (D116A IN) clones or with the control DNA from non IN-expressing yeast ().
Figure 3 PCR (A), Southern blot (B) and sequence analysis (C) of the selected zeocin-resistant clones. (A). Aliquots from three different clones obtained from cells that expressed wt IN (wt1–3), D116A IN (D116A1–3), or no IN (-IN) were subjected (more ...)
Southern blot analysis using 5′-end (32P) radiolabeled U3-Zeo as probe revealed a band for h.RAD52+(wt IN) clones but not for h.RAD52+ expressing D116A IN or no IN (). The negative result obtained with h.RAD52+ (D116A) clones confirmed that the resistance observed in that case was not due to integration of the substrate into the genome but rather to another cellular mechanism.
To identify the integration loci and analyze the fidelity of integration events, total genomic DNA of 20 previously selected clones were sequenced as described in the Materials and Methods section. For all of these clones, the integrated DNA was found disrupting a distinct open reading frame (ORF) ( shows an e.g. of the genetic structure of the integrated fragments of the three clones analyzed in and B).
Since the sequence of the junctions between integrated LTRs and the target DNA constitutes a specific signature of the IN involved in this process, they were carefully analyzed. The 5 bp repeats characterizing HIV integration were recovered for nine clones, confirming that HIV-1 IN was responsible for the process (). In the remaining clones no repeat was found, whereas a deletion of the ORF sequence was observed (from 5 to 10 bp) and correlated well with the in vitro concerted integration data (). Likewise, nearly 50% of the in vitro integration products present under our conditions contained the specific 5 bp duplications.
IN can thus catalyze, viral DNA integration in yeast despite the lack of other viral factors thereby providing a way to study several parameters of the isolated retroviral integration step, such as the influence of viral DNA ends structure and the involvement of cellular machinery.
Influence of viral LTR structure on integration in yeast
Recently it has been shown that the processing of viral DNA ends by IN could channel their concerted integration (19
). To determine the role of the viral ends structure and their integration in a cellular context, we followed IN activity using different DNA substrates. A DNA substrate that lacked LTR sequences was first used, in which case we recovered a number of resistant clones identical to the background () as observed in yeast cells expressing the inactive D116A IN (). PCR and Southern blot analysis of the corresponding genomic DNA also gave negative results indicating that, in contrast with the results obtained with the substrate containing the two HIV-1 LTRs, no IN-dependent integration was detected in the absence of viral sequences.
Figure 4 Effect of LTRs on integration catalyzed by IN in yeast. Experiments similar to those described in were performed using either the DNA substrate containing the complete LTR sequences (wt LTR), the 3′-processed LTRs (3′-OH LTR) (more ...)
Since the DNA substrate used in our yeast integration assay carried the two non-processed LTR ends, it can be assumed from our results that HIV-1 IN could catalyze both 3′-processing and strand transfer in yeast cells. We thus sought to determine whether integration in yeast could be affected by pre-processing the DNA substrate as in in vitro assays. When the processed DNA was used in our system, a significantly higher number of resistant clones were obtained in comparison with the blunt-ended DNA (). In addition, the maximum number of selected clones was attained earlier (~3 h), suggesting that the global rate of integration was improved. This result suggests that processing is a limiting step of the integration and can be dissociated in time from strand transfer IN activity.
We also conclude that (i) IN is the minimal retroviral protein needed for DNA integration with two LTRs into eukaryotic cellular DNA and (ii) some or all of the counterparts involved in HIV-1 integration in human infected cells are also present in yeast. These findings led us to further study the cellular mechanisms involved with the retroviral integration step.
Viral DNA integration in yeast is down regulated by RAD51
The last step of proviral integration is the repair of the DNA gaps flanking the integrated product. Several viral or cellular proteins have been proposed to be involved in this mechanism. New data reported here suggest that, in addition to cellular factors, no viral proteins other than IN play a major role in this repair activity.
Cellular repair systems have been previously proposed to be important for retroviral infection, such as the proteins belonging to the RAD52 epistasis group (20
), but no direct evidence of their role in the integration step have been shown. Some of the proteins belonging to the RAD repair system, as RAD18 or RAD51, have been proposed to function either in the retroviral integration step in human cells (RAD18), or in the integration of retrovirus-like elements in yeast (RAD51) (21
). In contrast to the case in human cells, the deletion of RAD51 and RAD18 in yeast does not lead to deleterious effects. We thus took into advantage the possibility of using our yeast system to study the role of those factors specifically on the isolated integration step by using the corresponding DNA repair-deficient mutant yeast stains.
The previously described lethal phenotype observed in a rad52−
deficient yeast strain (10
) further complicated the analysis of the role of RAD52 in the cellular integration mechanism but, as an alternative, we performed the integration assay in haploid yeast strains deficient for either yeast RAD18 or RAD51 encoding genes (respectively h.rad51−
). No difference was observed in the number of zeocin-resistant clones obtained with h.RAD52+
cells but, in contrast, a significantly higher number of zeocin-resistant clones were obtained in h.rad51−
in an extremely reproducible way (P
-value = 0.03 from five independent experiments, ). In addition, no such increase was observed in the yeast cells expressing either the D116A mutant or no IN, indicating that the increase of the integration events observed in the presence of IN was probably not due to a global stimulation of DNA mobility.
Figure 5 Effect of recombination factors on yeast integration catalyzed by IN. The integration assay described in was performed using the h.RAD52+ yeast strain (h.RAD52+), or yeast cells disrupted for RAD18 (h.rad18−) and RAD51 (h.rad51− (more ...)
PCR and Southern blot analysis of the genomic DNA of the clones confirmed that the higher number of clones obtained with h.rad51− was due to an increase in integration events catalyzed by IN (data not shown). Sequencing of several selected clones in the rad51 deleted strain revealed a similar ratio of loci presenting the 5 bp duplication compared to the results obtained with the wild type yeast strain (see and ). This indicates that the absence of RAD51 increased the number of integration events catalyzed by IN without significantly affecting the quality of the integration reaction products. Taken together, these results suggest a possible negative regulation of integration by the RAD51 DNA repair system in yeast. To determine whether this effect was due to a direct inhibition of IN activity by RAD51 we further analyzed the effect of the human RAD51 on in vitro integration activity catalyzed by IN.
Human Rad51 interacts with IN and inhibits its in vitro concerted integration activity
Physical hRAD51-IN interaction was first checked by ELISA experiments using highly purified enzymes. As reported in IN was found to bind hRAD51 with notable efficiency (close to the auto-association level of the enzyme). The interaction was also specific since no binding of IN was observed with non coated wells or with the bacterial class 1 integron recombinase. Nevertheless, as HIV-1 IN is known for its low solubility, the fixed format of ELISA could lead to biased results. In addition, we performed in vitro pull down assays using hRAD51 protein which had been previously coupled to Affi-gel beads. As shown in HIV-1 IN was found to associate with Affi-hRAD51 beads specifically, since no co-elution was observed in presence of Affi-BSA beads. It is noteworthy that the Affi-hRAD51 does not bind all the free IN protein. This could possibly be due to the fact that Affi-beads contain RAD51 ring structures in which not all the surfaces are accessible for additional interactions.
Figure 6 In vitro interaction between human RAD51 and HIV-1 IN (A) ELISA assay. Experiments were performed with a plate either uncoated or coated with the reported protein (30 pmol) and revealed with anti-IN antibodies after IN loading (30 pmol of proteins). The (more ...)
Our results indicate that IN and hRAD51 interact in vitro
. We wondered whether this interaction could be responsible of the IN activity inhibition observed in yeast cells. For this purpose we analyzed the effects of hRAD51 on in vitro
IN integration activity. Increasing amounts of hRAD51 were added in a standard concerted integration assay. Under our conditions hRAD51 caused a strong inhibition of the IN activity (). Quantification of the inhibitory effect of hRAD51 allowed us to determine an IC50
of 0.5 μM (). hRAD51 is known to interact with DNA in presence of ATP (18
). Since some ATP may still be present in the integration labeling reaction it remains to determine the part of inhibition due to the DNA binding property of RAD51 and/or to its the direct interaction with IN.
Figure 7 In vitro effect of human RAD51 proteins on integration catalyzed by IN (A) Concerted integration assays performed with 250 nM of IN, 100 ng of receptor plasmid and 10 ng of donor DNA in absence (lane 2) or in presence of 0.1, 1 or 10 μM of purified (more ...)