Cell context-dependent involvement of ATR in early stages of retroviral replication

Abstract

Retroviral DNA integration leaves behind a single-strand DNA discontinuity at each virus:host DNA junction. It has long been proposed that cellular proteins detect and repair the integrated DNA and that failure to do so might lead to apoptotic cell death, but their identity remains unknown. PIKK family members ATM, DNA-PKcs and ATR have all been proposed to be important for HIV-1 replication, but these findings turned out to be very controversial. In order to clarify their role in retroviral replication, we analyzed the effect of pharmacological inhibitors and of a dominant-negative version of ATR on the replication of retroviruses in cell lines relevant to HIV-1 infection. Our data show that ATR and probably other PIKKs as well are involved in retroviral replication in some but not all cell lines and that ATR increases the frequency of retroviral transduction by a mechanism other than the enhancement of infected cell survival.

Introduction

Retroviral integrase proteins catalyze the simultaneous ligation of the two viral DNA ends into the host DNA. Successful retroviral cDNA integration causes two specific types of damage at each extremity of the viral genome: a gap of a few nucleotides adjacent to a two-nucleotide 5′ overhang (Skalka and Katz, 2005). In addition, it is now generally accepted that the retroviral genome itself bears a single-strand discontinuity at its middle, called the “DNA flap” (De Rijck and Debyser, 2006; Royer-Leveau et al., 2002; Zennou et al., 2000), although the importance of this structure for the virus is still disputed. In summary, integrated retroviruses will bear at least three single-strand discontinuities. Because retroviral enzymes have not been found to be able to repair such DNA damage, it is assumed that specialized cellular pathways repair integrated retroviral genomes.

DNA damage is detected and repaired through the action of many proteins that can be tentatively broken down into three families: DNA damage sensors, damage signal transducers and downstream effectors (Bakkenist and Kastan, 2004). Double-strand DNA breaks (such as those caused by γ-irradiation) are especially lethal (van Gent et al., 2001) and are sensed by either the Mre11–Rad50–Nbs1 (MRN) complex, or by the Ku70/Ku80 heterodimer, depending on the cell cycle stage (Czornak et al., 2008). More subtle DNA damage, such as UV-induced crosslinks or base adducts, may not be detected right away but will disrupt the progression of DNA replication forks (Cimprich and Cortez, 2008). This will expose a single-stranded stretch of DNA region next to a double-stranded region with a free 5′ extremity, a structure that will be recognized by other sensors such as replication protein A (RPA) and the Rad9–Rad1–Hus1 (9–1–1) complex (Ellison and Stillman, 2003; Zou and Elledge, 2003).

PIKKs (phosphoinositide-3-kinase-related protein kinases) are a small family of proteins relaying the DNA damage signal by interacting with the damage sensor proteins outlined above and then by phosphorylating dozens of different effector proteins. PIKKs are very large protein kinases that include ATM (Ataxia-telangeictasia mutated), ATR (ATM- and Rad3-related), mTOR (mammalian TOR), DNA-PK_cs (DNA-dependent protein kinase catalytic subunit) and SMG-1 (suppressor of morphogenesis in genitalia-1). Among these, ATM, ATR and DNA-PKcs play an essential role in DNA damage signal transduction. These three proteins recognize an acidic motif present in their interacting partners, leading to their local recruitment at sites of DNA damage (Hiom, 2005).

PIKKs regulate the activity of several proteins involved in various mechanisms of DNA repair. Upon UV irradiation, for instance, ATR promotes the nuclear accumulation of xeroderma pigmentosum group A (XPA), a protein involved in nucleotide excision repair (Wu et al., 2007). ATR also colocalizes with and regulates Werner’s syndrome protein (WRN), which is involved in repair of DNA at stalled replication forks by recombination (Pichierri et al., 2003). ATM phosphorylates many proteins important for homologous recombination, such as histone H2A variant (H2AX) (Burma et al., 2001) and breast cancer 1 (BRCA1) (Cortez et al., 1999). One other major function of ATM and ATR is to prevent cell cycle progression following DNA damage, through the phosphorylation of checkpoint kinases Chk1 (by ATR) and Chk2 (by ATM) and of many other proteins (Eastman, 2004), thus presumably allowing time for the cell to repair its DNA. DNA-PKcs is perhaps even more directly involved in DNA repair, as it is part of the protein scaffold responsible for the NHEJ repair mechanism (Lovejoy and Cortez, 2009).

Among the three principal PIKKs involved in DNA repair, ATR would be the most likely to be involved in retroviral post-integration DNA damage. However, ATR-independent DNA damage response pathways were found to determine the fate of unintegrated viral cDNAs. Reverse transcription generates a linear double-stranded viral DNA, and the two free extremities are expected to be recognized by specialized sensors ku70/80 or by the MRN complex, activating the NHEJ or HR pathways, respectively. Indeed, once in the nucleus, part of the linear retroviral cDNA is converted to so-called 1-LTR and 2-LTR circles. The 1-LTR forms are generated by homologous recombination between the two retroviral long terminal repeats (LTRs) and this process requires a functional MRN complex (Kilzer et al., 2003). The 2-LTR circles are produced by ligation between the two ends of the linear viral cDNA in a process involving the NHEJ pathway (Jeanson et al., 2002; Li et al., 2001). In fact, Ku70/80 can be co-purified with retroviral replication complexes in acutely infected cells (Li et al., 2001). Formation of these retroviral DNA circles is, at first glance, detrimental to the virus, since circularized DNA cannot be integrated any longer. However, it has been proposed that they are necessary to prevent apoptosis induction caused by the presence of pools of free, unintegrated viral DNA ends (Li et al., 2001). The PIKK proteins ATM and DNA-PK might thus be important for the formation of 1-LTR and 2-LTR circles, respectively.

Therefore, one or more proteins of the PIKK family might be involved in the early stages of retroviral replication. This hypothesis has been explored since 1999, when Daniel et al. (1999) first reported a role for DNA-PK in the survival of retrovirus-infected cells. The same groups later reported that ATR was required for DNA integration, and that ATR-deficient cells died of apoptosis following retroviral infection (Daniel et al., 2003). Evidence supporting these conclusions came from the use of murine cells knocked-out for actors of the cellular DNA repair pathways, from the use of the broad PIKK inhibitors caffeine and wortmannin, and from expression of a kinase-mutant variant of ATR (Daniel et al., 1999, 2001, 2003, 2005; Nunnari et al., 2005), and was reviewed by Skalka and Katz (2005). Several other groups, however, reached opposite conclusions on the role of PIKKs in retroviral replication, even though they were using some of the same investigative tools (Ariumi et al., 2005; Baekelandt et al., 2000; Dehart et al., 2005). Two of these groups, additionally, transfected human cells with siRNAs targeting ATM, ATR and DNA-PKcs and saw no or little effect on retroviral transduction (Ariumi et al., 2005; Dehart et al., 2005). Therefore, the question of the importance of PIKKs in viral replication is among the most controversial ones in retrovirology.

However, upon closer examination, almost all of the experiments described in these papers used cell types that were not relevant to the replication of retroviruses in vivo, i.e. fibroblast cell lines (HeLa mostly), Chinese hamster ovary cells, or mouse embryo fibroblasts. In addition, some of the data published by Daniel and collaborators used experimental systems that did not allow quantification of the number of infected cells, or that required maintaining cells in culture for up to a week before readout. Thus, we sought to revisit the issue of the role of PIKKs in the replication of multiple different retroviruses, using only human cell lines that were representative of HIV-1 natural host cells, and using only infectivity assays that allowed quantitation of infected cells without interference from transcriptional effects. Our results do not support the idea that PIKKs are required for retroviral transduction, but clearly ATR and possibly other PIKKs can modulate the efficiency of early retroviral replication events in at least some cell lines.

Results

There is considerable discrepancy in the literature regarding the effects of PIKK inhibitors on retroviral replication. Caffeine, a methylxanthine derivative, was demonstrated to sensitize cells to radiation-induced DNA damage by inhibiting ATM phosphorylation of downstream substrates (Blasina et al., 1999). However, it is also active against other PIKKs such as ATR, hSMG-1 and mTOR but not DNA-PKcs (Sarkaria et al., 1999). Caffeine competes with ATP for the binding to kinases, but the drug–protein interaction is reversible. Moreover, this drug’s specificity is poor and in addition to kinases, caffeine has been reported to inhibit other proteins such as ATPases. Wortmannin, a fungal metabolite, is a general inhibitor of PI3Ks and PIKKs (Abraham, 2004). It similarly competes with ATP for binding to kinase enzymatic sites which it then permanently disrupts (Walker et al., 2000). Wortmannin efficiently inhibits ATM, DNA-PKsc, mTOR and hSMG-1 but is much less active against ATR (Sarkaria et al., 1998). Compared with caffeine, wortmannin inhibits PIKKs at much lower concentrations and appears to be more specific to PI3Ks and PIKKs (Abraham, 2004).

In order to obtain data as much relevant as possible to natural HIV-1 infection, we chose to infect T cell lines and cells representative of the monocyte/macrophage lineage. These cells were infected with a vector (designated HIV-1_GFP throughout this work) derived from HIV-1 and expressing “enhanced” GFP under control of the cytomegalovirus intermediate-early (CMVie) promoter. Like all viruses used in this study, this vector was pseudotyped with the G protein from the vesicular stomatitis virus, allowing infection of various cell lines with a wide range of viruses and bypassing possible effects of the drugs on entry mediated by retroviral envelope proteins. Because the time at which a putative post-integration defect would negatively affect the number of GFP-expressing cells is difficult to predict, in the first experiment we chose to monitor retroviral transduction at multiple times. In the experiment shown in Fig. 1, cells were challenged with HIV-1_GFP and continuously treated with caffeine or wortmannin. In the four cell lines studied, the percentage of infected (GFP-positive) cells in the absence of drug plateaued at about 7 days post-infection. Wortmannin had no effect on HIV-1_GFP transduction of SupT1 and Jurkat cells. In contrast, it significantly decreased HIV-1_GFP infection of both U937 and THP-1 cells. This drug-dependent inhibition was observed at all time-points tested and was dose-dependent. Caffeine treatments resulted in more complex phenotypes. In both Jurkat and SupT1 cells, caffeine actually increased the % of infected cells in a dose-dependent fashion. In U937 cells, caffeine decreased the % of infected cells at all times after infection, and the magnitude of the effect was similar to what was observed with wortmannin (2- to 3-fold). Caffeine reduced HIV-1_GFP infection of THP-1 cells in a drug-dependent fashion, thus resembling the data obtained in U937 cells.

Fig. 1.

Caffeine and wortmannin inhibit HIV-1 vector transduction of monocytic cell lines. The indicated cell lines were infected with HIV-1_GFP, an HIV-1 vector expressing only GFP. Infections were performed in the absence of drug or in the presence of caffeine or wortmannin at the indicated concentrations. Every 2 or 3 days, between half and two-thirds of each culture was processed for FACS analysis and replaced with medium containing the appropriate drugs. The percentages of GFP-positive cells were determined by FACS and are indicated as % infected cells. Error bars show standard deviations from triplicate infections.

Why did caffeine increase transduction of the two T cell lines is unclear, but it was accompanied by an increase in levels of GFP expression (up to 4-fold) among GFP-positive cells (not shown). Thus, perhaps caffeine somehow promoted the survival of cells expressing high GFP levels or enhanced promoter activity. However, caffeine and wortmannin clearly and consistently inhibited HIV-1_GFP infection of U937 and THP-1 cells (Fig. 1), without affecting levels of GFP expression in GFP-positive cells (not shown).

We next determined whether these drugs would have a similar effect on other retroviruses. U937 and THP-1 cells were infected with vectors derived from human immunodeficiency viruses type 1 and 2, from the mac239 strain of the simian immunodeficiency virus and finally with a murine leukemia virus (MLV)-based vector. The HIV-1_NL-GFP used in this experiment is an HIV-1 based vector in which the coding sequence for GFP replaces that of the viral protein Nef. Thus, GFP here is expressed under control of the viral LTR promoter, and moreover, all viral proteins are expressed with the exception of Nef and the envelope glycoproteins. HIV-2_ROD-GFP and SIV_mac-GFP were constructed in a similar fashion, while MLV_GFP resembles HIV-1_GFP since GFP is expressed from a CMVie promoter and is the only protein expressed from the vector in infected cells. Cells were treated with the drugs only once and then analyzed for GFP expression at days 2 and 5. Caffeine and wortmannin strongly inhibited infection of U937 cells with all four viruses (Fig. 2). The inhibition was more apparent at day 2 than at day 5, perhaps owing to drug instability. The drugs also inhibited infection of the THP-1 cells, but the magnitude of the inhibition was smaller than in U937 cells, and no significant inhibition was seen with HIV-1_NL-GFP. As in the U937 cells, caffeine and wortmannin decreased the % of GFP-expressing cells at day 2 more efficiently than it did at day 5. Neither drug affected the infection of SupT1 cells by any of the four viruses used in Fig. 2 (not shown). Thus, the extent of inhibition of retroviral replication by caffeine and wortmannin varies from cell line to cell line rather than from virus to virus.

Fig. 2.

Caffeine and wortmannin inhibit multiple retroviruses in U937 and THP-1 cells. Cells were challenged with retroviral vectors derived from HIV-1_NL4-3, HIV-2_ROD and SIV_mac239 which express GFP and also encode all but two viral proteins. Infections were performed in triplicates and in the absence or presence of caffeine or wortmannin at the indicated concentrations. Virus doses were adjusted for each cell type so that about 5% of the cells would be GFP-positive at day 2 post-infection in the absence of drug. Half of each culture was analyzed by FACS at day 2, and the remainder was kept in culture for a second FACS analysis at day 5. Cells were also challenged with an MLV-based vector, but we found monocytic cells to show little permissiveness for this type of vector, and thus only 1% of U937 cells could be infected in the absence of drugs, while no appreciable transduction of THP-1 could be obtained. Error bars show standard deviations.

Wortmannin and especially caffeine are relatively unspecific drugs. Thus, we tested the effect of a third drug, CGK733, on retroviral replication. CGK733 was isolated in a screen for molecules inhibiting cellular senescence, and ATM/ATR were subsequently shown to be targeted by it (Bhattacharya et al., 2009; Crescenzi et al., 2008; Cruet-Hennequart et al., 2008; Goldstein et al., 2008; Won et al., 2006). We infected U937, THP-1 and SupT1 cells with HIV-1_GFP at CGK733 concentrations previously used by others (Cruet-Hennequart et al., 2008; Won et al., 2006). Results show that, like caffeine and wortmannin, CGK733 decreased the infectivity of HIV-1_GFP in both U937 and THP-1 cells but not in SupT1 cells (Fig. 3).

Fig. 3.

A specific ATM/ATR inhibitor reduces HIV-1 transduction of monocytic cell lines. U937, THP-1 and SupT1 cells were challenged with HIV-1_GFP in the presence or not of CGK733. The experiment was conducted as in Fig. 2. Error bars show standard deviations from triplicate infections.

Taken together, the results in Figs. 1–3 show that targeting PIKKs using pharmacological agents decreases the efficiency of early steps of retroviral replication in monocytic cell lines. Because ATR is the PIKK member most likely to be involved in the repair of single-strand DNA damage, we turned our attention to the negative-dominant D2475A mutant of ATR, called ATR-kd (for “kinase-dead”). Expression of this ATR point mutant, which suppresses its kinase activity, inhibits the endogenous wild-type ATR (Cliby et al., 1998). We cloned wild-type and mutant ATR into an MLV-based retroviral vector also expressing a puromycine resistance marker. U937 and SupT1 cells were transduced with the vectors. Probably due to the large size of the ATR cDNA, the efficiency of transduction was low (1% or less), yet puromycine-resistant cells were obtained and grown. Upon analysis of transgene expression by western blotting using an ATR antibody, we found that expression of the mutated form was more efficient than that of the wild-type protein (Fig. 4A). It is unlikely that this would result from multiple copies of the mutant being transduced, since transduction efficiency was low for both wild-type and mutant. Thus, the D2475A mutation of ATR probably increased steady-state levels of the protein in the cell lines used. To confirm that ATR-kd had a negative-dominant effect on endogenous ATR activity, U937 cells expressing ATR-wt or ATR-kd or mock-transduced control cells were treated with aphidicolin, a DNA replication inhibitor that activates the ATR kinase pathway (Hekmat-Nejad et al., 2000; Tibbetts et al., 2000). Upon treatment, the G2/G1 ratio was increased 3 times in the control cells and in the cells over-expressing ATR-wt, but only had a small effect (1.4-fold) on the cells transduced with ATR-kd. Thus, ATR-kd expression inhibited the whole ATR kinase-dependent pathway as expected.

Fig. 4.

A kinase-dead mutant of ATR inhibits retroviral transduction of U937 cells. cDNAs for wild-type ATR or for ATR bearing a mutation in the kinase active site were cloned into a retroviral vector and then transduced into U937 or SupT1 cells. (A) characterization of ATR-wt and ATR-kd transduction in SupT1 and U937 cells. Left panel, western blot analysis of ATR expression in transduced cells. Actin is used as a loading control. Right panel, aphidicolin-induced cell cycle arrest in G2. The indicated U937 cell lines were treated or not with 0.25 μM of aphidicolin and stained with propidium iodide 2 days later. The ratio between cells in G2 and cells in G1 is indicated in each panel. (B) cells over-expressing wild-type or kinase-dead ATR, or transduced with the empty LPCX vector as a control, were challenged with the indicated retroviral vectors in triplicates. Percentages of GFP-expressing cells were determined at days 2 and 5 as explained in Fig. 2, with error bars showing standard deviations. (C) U937 cells over-expressing wild-type of kinase-dead ATR were challenged with HIV-1_GFP in the absence or presence of caffeine (4 mM), wortmannin (3 μM) or CGK733 (6 μM). The percentages of infected cells were analyzed by FACS. Error bars show standard deviations from triplicate infections.

We thus analyzed the effect of over-expressing wild-type or mutated ATR on the infection of U937 or SupT1 cells by a variety of retroviral vectors. Again, the percentages of cells expressing GFP were monitored at 2 and 5 days after infection with a single dose of each virus (Fig. 4B). We found that most SupT1 transduced by HIV-1_NL-GFP (as seen by GFP expression at day 2) had disappeared by day 5, obviously due to the toxicity of some HIV-1 proteins in these cells (since the HIV-1_GFP vector, which does not express any HIV-1 protein, did not affect SupT1 survival). In U937 cells, over-expression of wild-type ATR did not significantly affect the efficiency of the infection, compared with cells transduced with the empty vector (LPCX). Over-expression of mutated ATR, however, reduced GFP transduction efficiency by 1.5- to 4-times. This inhibition phenotype was more apparent at day 2 for some viruses and at day 5 for others. The only exception was U937 cells infected with HIV-1_GFP at day 2, where there was no significant difference between ATR-wt and ATR-kd. In SupT1 cells, we obtained very different data: over-expression of the wild-type version of ATR itself caused a significant decrease in GFP transduction efficiency (about 2-fold for most vectors, but about 8-fold for the MLV-based vector). Unlike what was seen in U937 cells, however, ATR-kd behaved similarly to ATR-wt in SupT1 cells (Fig. 4B). Therefore, and consistent with our pharmacological data, ATR kinase activity seems to be important for retroviral infections in U937 cells but not in SupT1 cells. The observation that wild-type ATR over-expression in SupT1 cells decreases levels of infection by all vectors tested is intriguing, and so is the finding that this effect is more pronounced for the MLV-based vector. However, it is unlikely to be related to the role of ATR in DNA repair, since the same phenotype was found with ATR-kd.

Treatment with PIKK-inhibiting drugs and expression of the ATR mutant both had cell-specific, virus-nonspecific effect on retroviral transduction. It was thus expectable that both interventions would inhibit the same step of retroviral replication. To test that hypothesis, we infected U937 cells expressing either wild-type or mutant ATR with HIV-1_GFP in the presence or absence of caffeine, wortmannin, or CGK733. As expected, over-expression of ATR-kd decreased infection of U937 by 2-fold, and this effect was seen at days 2 and 5. In addition, and as found before, treatment with PIKK drugs decreased HIV-1_GFP infection of ATR-wt-expressing cells by 4-times (caffeine) to 10-times (wortmannin and CGK733) at day 2. At day 5, the inhibition mediated by the drugs was less acute (2-fold to 4-fold decrease in infectivity). We found that ATR-kd over-expression and treatment with caffeine or wortmannin had additive effects on HIV-1 infection. In other words, over-expression of ATR-kd decreased HIV-1_GFP infection of U937 cells by 2-fold or more, whether these cells were untreated or treated with caffeine or wortmannin. In contrast, over-expression of ATR-kd had no effect on cells that were treated with CGK733, and this was true at day 5 as well as at day 2. This result strongly suggests that ATR-kd and CGK733 interfered with the same ATR function. On the other hand, caffeine and wortmannin seemed to inhibit retroviral replication through a mechanism independent of ATR and thus likely to involve other PIKKs.

One obvious mechanism for a putative role of PIKKs in retroviral replication would be the prevention of apoptosis resulting from integration-related DNA damage. Testing this hypothesis by performing infections at multiplicities of infection (MOI) similar to that used in previous experiments (1–10% infected cells) was complicated by low levels of virus-related apoptosis at these MOIs together with background apoptosis. Thus, we opted for infecting U937 cells expressing wild-type or mutated ATR with large doses of LV-luc, an HIV-1 vector expressing the gene of resistance to puromycine but not encoding any viral protein (which might cause apoptosis themselves). In a preliminary experiment, treatment with puromycine revealed that nearly all parental U937 cells (not expressing exogenous ATR) were infected (puromycine-resistant) at the viral doses used. In an annexin V-binding assay, no significant apoptosis was detected 1 day after adding the virus onto the cells (Fig. 5A). At day 2, the number of annexin V-positive cells was about 2.5-fold higher in the infected cells than it was in control uninfected cells, but no significant differences were observed between the cells expressing wild-type or mutant ATR. We then looked at levels of cleaved poly (ADP-ribose) polymerase-1 (PARP-1) and cleaved caspase-3, two additional markers of apoptosis induction. Activated (cleaved) caspase-3 is an important effector of most apoptotic pathways, and targets many substrates, including PARP-1 (Nicholson et al., 1995; Tewari et al., 1995). Infection with the HIV-1 vector greatly increased levels of both activated PARP-1 and activated caspase-3 (Fig. 5B), and this effect was more marked at 24 h than at 48 h. However, we found no significant difference between cells expressing wild-type and mutated ATR. Intriguingly, cells made to express ATR seemed less prone to apoptosis induction compared with cells transduced with the parent construct (LPCX), regardless of whether ATR was wild-type or mutant. The explanation for this effect is not clear but is independent of ATR kinase activity and thus is probably independent of ATR function in DNA repair.

Fig. 5.

Apoptosis does not explain decreased levels of retroviral transduction of U937 cells over-expressing ATR-kd. U937 cells over-expressing wild-type of kinase-dead ATR, or transduced with the empty vector LPCX as a control, were challenged with high doses (0.5 or 1 ml) of the HIV-1-derived vector LV-luc. (A) the percentage of cells undergoing apoptosis was determined at 24 or 48 hours post-infection by using an annexin V-binding assay. (B) induction of apoptosis was monitored by western blot detection of cleaved PARP-1 and cleaved caspase-3. Actin is used as a loading control.

Discussion

Over-expression of a kinase-dead mutant of ATR and treatment with PIKK pharmacological inhibitors decreased retroviral replication in some cell lines (U937, THP-1) but not in others (Sup-T1). It would be interesting to investigate the involvement of PIKKs in early retroviral replication in mature, differentiated macrophages. Strikingly, in cell lines in which ATR-kd over-expression and PIKK drugs did inhibit replication, it affected all retroviruses used, including three different lentiviruses (HIV-1, HIV-2, SIVmac) and a more simple oncoretrovirus (MLV). This strongly suggests that ATR is involved in a retroviral replication step which is not virus-specific. Post-integration DNA repair is a likely candidate, since the single-strand discontinuities generated by DNA integration are very similar across retroviruses. However, we have not characterized the exact replication step inhibited by ATR-kd expression or by treatment with the PIKK drugs. The relatively small magnitude of the effects seen here prevented a quantitative analysis of integrated versus unintegrated viral cDNA. Therefore, we cannot exclude that the various drugs used, and over-expression of ATR-kd, interfere with steps other than post-integration repair. In addition, more than one replication stage might be affected.

In our hands, over-expression of an ATR negative-dominant mutant did not increase apoptotic cell death. In fact, apoptosis was found to affect much fewer cells than would be expected to account for the decrease in retroviral transduction in U937 cells over-expressing ATR-kd (Fig. 5A). In addition, it is not even clear that apoptosis was related to DNA damage in our experiment. Several HIV-1 proteins are known to be pro-apoptotic (Shedlock et al., 2008) and it is possible that incoming retroviral particles themselves might directly cause cell death. We did not analyze cell death in infections done in the presence of PIKK drugs, as these drugs are clearly toxic even in the absence of virus, which would likely lead to misleading data. While additional experiments would be required to fully assess the role of PIKKs in cell survival following retroviral infections, our data so far do not support the idea that retroviral transduction of ATR-deficient cells causes apoptosis, as proposed by others (Daniel et al., 2003).

Why, then, is retroviral transduction less efficient in U937-kd cells? We speculate that in the absence of functional ATR in these cells, single-strand discontinuities become double-strand breaks during DNA replication, which are then repaired by NHEJ or by HR. While NHEJ will preserve the integrated retrovirus, interchromosomal HR (Richardson et al., 1998) will likely lead to complete removal of the foreign retroviral cDNA inserted in only one of the two chromosomes, thus accounting for the decrease in % infected cells.

In our view, a strong evidence for the implication of ATR in retroviral transduction of U937 cells comes from the observation that over-expression of ATR-kd and treatment with the specific ATR/ATM inhibitor CGK733 have partly overlapping effects on HIV-1: when cells are treated with CGK733, then over-expression of ATR-kd does not affect HIV-1 vector transduction (Fig. 4C). In contrast to CGK733, wortmannin treatment did not alleviate the effect of over-expressing ATR-kd. This was not surprising since wortmannin has little inhibitory potential against ATR (Sarkaria et al., 1998). This observation suggests that, in addition to ATR, ATM and/or DNA-PKcs are important for retroviral transduction of U937 cells. More puzzling was the observation that ATR-kd retained its inhibitory potential in the presence of caffeine, which itself is believed to inhibit ATR (Sarkaria et al., 1999). Perhaps caffeine does not target ATR as efficiently in live cells as it does the purified protein.

ATR activation results in cell cycle arrest, and the HIV-1 protein Vpr has been proposed to prevent progression from the G2 phase of the cell cycle by activating a pathway that includes ATR (Andersen et al., 2005; Belzile et al., 2007; Dehart and Planelles, 2008; Zimmerman et al., 2006). However, no significant differences in phenotypes were found upon comparison of retroviral vectors bearing a Vpr protein (such as HIV-1_NL-GFP) and those devoid of Vpr (such as HIV-1_GFP or the vector derived from MLV). Therefore, the effects on retroviral replication seen here are clearly independent of Vpr.

At least two groups have shown that knocking down ATR, ATM or DNA-PKcs had no effect on retroviral transduction (Ariumi et al., 2005; Dehart et al., 2005). Accordingly, we found that decreasing the expression of each of these three proteins in U937 cells by the use of small hairpin RNAs had little effect on infection by HIV-1 vectors (not shown). However, negative RNAi data should be interpreted with great care, as they could simply mean that only minute amounts of the targeted proteins are required during viral replication. Both ATM and ATR are very efficiently recruited to sites of DNA damage (Abraham, 2004; Bakkenist and Kastan, 2004; Hiom, 2005), and in our experimental conditions most cells will undergo only one retroviral integration event. Thus, it is conceivable that just a few copies of either ATR or ATM proteins would be sufficient to accomplish their function.

Why are PIKKs seemingly dispensable for retroviral infection of lymphocytic cell lines? Minor single-strand DNA breaks, which are the most common types of DNA damage in cells, are sensed by PARP-1 and repaired by a variety of proteins that collectively form the base excision repair (BER) pathway (Caldecott, 2007; Dianov and Parsons, 2007). BER, which is thought to be independent of PIKKs (Audebert et al., 2004; Cappelli et al., 2000), could possibly repair the single-strand discontinuities generated by retroviral DNA integration in lymphocytes and perhaps in most other cell lines as well. Regardless of what the precise mechanism of post-integration repair in lymphocytic cell lines is, it remains that PIKKs do not seem to be involved. Thus, considering that CD4+ lymphocytes are the major HIV-1 host cells in vivo, PIKKs do not appear to be very valuable targets in the development of antiretroviral compounds for AIDS therapy.

Materials and methods

Cells, plasmids and reagents

293T cells were maintained in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% fetal calf serum and antibiotics at 37 °C. The human monocytic cells U937, THP-1 (a gift from Carlos Reyes-Moreno, UQTR) and the human T lymphocytes SupT1 and Jurkat were cultured in RPMI 1640 supplemented with 10% fetal calf serum and antibiotics at 37 °C.

The general PIKK family inhibitors caffeine and wortmannin and the ATM/ATR specific inhibitor CGK733 (Won et al., 2006) were purchased from Sigma. Antibodies for the detection of ATR, actin and cleaved PARP-1 were from Abcam (ab2905), Millipore (murine clone C4) and Cell Signalling (9541), respectively. The antibody specific for cleaved caspase 3 was a gift from Éric Asselin.

pLPCX_AB is a home-made version of the retroviral vector pLPCX (Clontech), which includes additional unique cloning sites (L.B., unpublished). The coding sequences of wild-type and kinase-dead mutants of human ATR were cut out from pBJ5.1-ATR-wt and pBJ5.1-ATR-kd (Cliby et al., 1998), respectively, using Not1, and cloned into pLPCX_AB, thus generating pLPCX-ATR-wt and pLPCX-ATR-kd. pLV-luc was constructed by transferring the shRNA luciferase fragment in pSUPER-LUCi (Sayah et al., 2004) to pLV-PARP1 (Ariumi et al., 2005) cut with BamH I and Sal I. pMD-G, pΔR8.9, pTRIP-CMV-GFP, pCL-Eco, pNL4-3_Nef-GFP, pROD_Nef-GFP and pMAC239_Nef-GFP have all been extensively described before (Berthoux et al., 2003, 2004, 2005; Reuter et al., 2005; Zufferey et al., 1997).

Production of viral vectors

To produce wild-type and kinase-dead ATR-expressing retroviral vectors, 293T cells were transfected using polyethylenimine (Poly-sciences) with 10 μg of pCL-Eco, 5 μg of pMD-G, and 10 μg of the appropriate pLPCX_AB construct as described before (Berube et al., 2007). To produce the MLV vector expressing GFP, 293T cells were similarly transfected with 10 μg of pCL-Eco, 5 μg of pMD-G, and 10 μg of pCNCG. To produce the HIV-1_GFP vector and LV-Luc retroviral vector, 293T cells were transfected with 10 μg of pΔR8.9, 5 μg of pMD-G, and 10 μg of pTRIP_CMV-GFP or 10 μg of pLV-luc. HIV-1_NL-GFP, HIV-2_ROD-GFP and SIV_mac-GFP vectors were produced by transfection of 10 μg of pNL4-3_Nef-GFP, pROD_Nef-GFP or pSIVmac239_Nef-GFP, respectively, along with 5 μg of pMD-G. All virus-containing supernatants were collected 2 days after transfection, clarified by low-speed centrifugation and stored in 1-ml aliquots at − 80 °C.

Creation of cell lines stably expressing ATR

U937 and SupT1 cells were plated at 500,000 cells per well in 12-well plates. LPCX_AB retroviral vectors expressing ATR-wt and ATR-kd were added onto the cells (1 ml per well, or half of the total volume). Two days later, the cell supernatants were replaced with medium containing 8 μg/ml of puromycine (EMD Biosciences). Puromycine selection was allowed to proceed for 1 week. ATR expression was analyzed by standard western blotting. For the analysis of ATR function, cells were treated or not with 0.25 μM of aphidicolin (Sigma). At 48 h post-treatment, cells were stained with the Krishan buffer (0.1% sodium citrate, 0.3% NP-40, 0.05 mg/ml propidium iodide and 0.02 mg/ml RNase) and the cell cycle profile of treated and not treated cells was analyzed by flow cytometry. Acquisition was performed on a FACScan flow cytometer. The mathematical model MODFIT was used to calculate the proportions of cells in the G2/M and G1 phases.

Viral challenges and flow cytometry analysis

A total of 25,000 cells were plated in 1 ml medium per well of 24-well plates. Cells were infected with HIV-1_GFP, HIV-1_NL-GFP, HIV-2_ROD-GFP, SIV_mac-GFP and MLV_GFP vectors. When wortmannin, caffeine or CGK733 were used, they were added 15 min prior to the virus. Cells were analyzed at various days after infection by flow cytometry on a FC500 MPL instrument (Beckman Coulter) using the MXP/CXP software. Intact cells were identified based on light scatter profiles, and only those cells were included in the analysis. Ten thousand cells per sample were processed, and cells positive for GFP expression were gated and counted as a percentage of total intact cells.

Analysis of apoptosis

Cells (LPCX, ATR-wt and ATR-kd) were plated at 50,000 cells per well in 12-well plates. Cells were infected with a large amount (0.5 or 1 ml per well) of the LV-Luc retrovirus vector. Twenty-four hours or 48 h after being infected, apoptotic cells were labeled with annexin V-FITC (clontech) according to the manufacturer’s instructions. The percentages of annexin positive cells were analyzed by FACS as described before (Berthoux et al., 2003). Alternatively, cells were lysed and processed for western blotting detection of cleaved caspase and cleaved PARP-1.