The Cell Cycle Independence of HIV Infections Is Not Determined by Known Karyophilic Viral Elements

Human immunodeficiency virus and other lentiviruses infect cells independent of cell cycle progression, but gammaretroviruses, such as the murine leukemia virus (MLV) require passage of cells through mitosis. This property is thought to be important for the ability of HIV to infect resting CD4+ T cells and terminally differentiated macrophages. Multiple and independent redundant nuclear localization signals encoded by HIV have been hypothesized to facilitate migration of viral genomes into the nucleus. The integrase (IN) protein of HIV is one of the HIV elements that targets to the nucleus; however, its role in nuclear entry of virus genomes has been difficult to describe because mutations in IN are pleiotropic. To investigate the importance of the HIV IN protein for infection of non-dividing cells, and to investigate whether or not IN was redundant with other viral signals for cell cycle-independent nuclear entry, we constructed an HIV-based chimeric virus in which the entire IN protein of HIV was replaced by that of MLV. This chimeric virus with a heterologous IN was infectious at a low level, and was able to integrate in an IN-dependent manner. Furthermore, this virus infected non-dividing cells as well as it infected dividing cells. Moreover, we used the chimeric HIV with MLV IN to further eliminate all of the other described nuclear localization signals from an HIV genome—matrix, IN, Viral Protein R, and the central polypurine tract—and show that no combination of the virally encoded NLS is essential for the ability of HIV to infect non-dividing cells.


Introduction
Human immunodeficiency virus and other lentiviruses have the ability to infect non-dividing cells [1][2][3]. This property allows HIV to integrate into two major types of virus reservoirs in vivo: resting CD4þ T cells and macrophages [4]. However, the ability to productively infect nondividing cells is not shared by all retroviruses [5]. For example, the gamma retroviruses as exemplified by the murine leukemia virus (MLV) requires mitosis for integration [6,7]. Infection and transduction with foamy retroviruses also depends on cell cycle and requires mitosis [8][9][10]. An alpharetrovirus, the avian sarcoma virus, appears to be able to integrate viral genomes in non-dividing cells [11,12], but fails to produce virus particles, indicating that it requires mitosis for a later stage of the viral life-cycle [13,14].
After entry into the cytoplasm, retroviruses undergo an uncoating and reverse transcription process that yields a large nucleoprotein complex called the preintegration complex (PIC) [15]. Nuclear entry of viral DNA is an essential step in the retroviral life cycle since viral genomic DNA in the PIC must enter the nucleus to be integrated into host cell chromosomes. The prevailing model to explain the ability of lentiviruses to infect cells independent of the cell cycle is that lentiviruses can target their viral genomes into the nucleus of non-dividing cells via active nuclear transport, while gammaretroviruses that cannot infect non-dividing cells gain an access to the host chromosomes only when the nuclear membrane breaks down at mitosis [6,7]. Thus, it has been hypothesized that the PIC of lentiviruses contain virally encoded nuclear localization signals (NLS), which allow active nuclear transport independent of the cell cycle, whereas the PIC of gammaretroviruses do not contain virally encoded NLS, and thus can not enter the nucleus until mitosis (reviewed in [16]).
Several lentiviral elements that contain a potential NLS and are present in the PIC have been identified including the matrix (MA) [17], integrase (IN) [18], and Viral Protein R (Vpr) [19] proteins and a cis-acting element called the central polypurine tract (cPPT) [20]. However, the importance of each of these elements is controversial since subsequent studies have shown that HIV lacking one or several mutations in these NLS elements still retains a significant ability to infect non-dividing cells [16,[21][22][23][24][25][26][27]. The IN protein is a particularly attractive candidate to mediate nuclear import of HIV genomes since it is part of the PIC through all steps of infection until viral integration, and IN is necessary for nuclear localization and transposition of the yeast elements Ty1 and Ty3 [28][29][30]. Moreover, HIV IN contains nuclear import activity [18,25,[31][32][33][34], whereas MLV IN lacks such nuclear import activity [35,36]. However, the role of HIV IN within nuclear import of viral genomes has been difficult to definitively address because mutations or deletions within IN often show pleiotropic effects on virus replication, including assembly, and reverse transcription, in addition to integration (reviewed in [37]). Recent work has suggested that HIV IN itself does not contain an NLS, but rather traffics to the nucleus by virtue of binding the lens epithelium-derived growth factor (LEDGF)/p75 protein [36,[38][39].
We recently reported that the capsid protein (CA) is a dominant determinant of retroviral infectivity in non-dividing cells since HIV containing MLV CA lost the ability to infect non-dividing cells, even though it still contains proteins with an NLS [40]. Because HIV CA is not nucleophilic [41,42] and is not stably associated with the HIV PIC [43][44][45][46][47][48], these data led us to propose that nuclear entry is not the ratelimiting step in the ability of HIV to infect non-dividing cells [40,49]. However, because we were unable to eliminate all of the proposed NLS in HIV, we could not rule out the possibility that the chimeric HIV containing MLV CA masked a pathway usually used by HIV for entry into the nucleus.
Here, we directly tested the involvement of IN within infection of non-dividing cells by constructing an HIV-MLV chimeric virus in which the HIV IN coding sequence was replaced with the IN coding sequence of MLV. Somewhat remarkably, this chimeric virus was infectious at a low level and was able to integrate in an IN-dependent manner. Furthermore, this virus infected non-dividing cells as well as it infected dividing cells.
While individual NLS-containing proteins, in some cases combinations, have been mutated or deleted from HIV in previous studies, it could be argued that the effect of the different NLS are redundant, and therefore HIV still retained some ability to infect non-dividing cells because of the presence of other NLS on other proteins. The ability to generate an HIVbased chimera with MLV IN allowed us to further eliminate all of the other described NLS (MA, Vpr, and the cPPT) from an HIV infectious clone. We report here that this chimeric virus without any of the previously described NLS is still able to infect non-dividing cells. We discuss the possibility that uncoating of the entering viral particle, rather than nuclear import, is the rate-limiting step that determines the cell cycle dependence/independence of retroviral infections.

Generation of an Infectious Chimeric HIV-1 with MLV IN That Is Integration-Competent
HIV IN localizes to the nucleus when stably expressed in cells, whereas MLV IN does not [36]. Therefore, to determine if the karyophilic property of IN is essential for the infectivity of HIV in non-dividing cells, HIV-1 IN was replaced with MLV IN within an HIV-based provirus, generating the chimeric clone called MHIV-mIN (which encodes MLV IN instead of HIV IN while the rest of the provirus is HIV) (Figure 1). Transfection of this chimeric provirus showed that it produces virus particles as indicated by the presence of virus-specific proteins in culture supernatants of transfected cells ( Figure 2). As expected, MLV IN, and not HIV-1 IN, was detected in virions ( Figure 2). The amount of virions made by MHIV-mIN was between 3-to 30-fold lower than that made by wild-type HIV-1 as measured by p24gag ELISA (unpublished data). Nevertheless, processing of reverse transcriptase (RT) and IN appeared normal in virions produced by MHIV-mIN virus ( Figure 2).
We tested the infectivity of MHIV-mIN together with wildtype HIV-1 in a single-cycle replication assay [50]. While the titer of MHIV-mIN was about 3-logs lower than that of wildtype HIV-1 when normalized by the amount of p24gag ( Figure 3A), it was still well above the background. Real-time PCR data indicated that MHIV-mIN produces 3-to 5-fold less cDNA than wild-type HIV-1 ( Figure 3B). Thus, a decrease of reverse transcription of MHIV-mIN alone cannot explain the reduced infectivity of MHIV-mIN.
The integration reaction requires specific recognition by viral IN of short DNA sequences (;10 bp) at both ends of viral DNA, called the attachment (att) site. A previous report indicated that replacement of HIV att sites with MLV att sites at both ends of the long terminal repeat (LTR) reduced viral titer to 0.5% level of the wild-type level [51], while others have found that the att sequences other than the conserved CA dinucleotide motif are not very important in vivo [52,53]. To test this, we also made a chimeric clone that contains MLV att sequences in both ends of the LTR (Figure 1, called pMHIV-mIN/matt), and examined the infectivity of these chimeras ( Figure 3C). We could obtain titers of up to 1 3 10 6 infectious units per ml after concentration of both viruses, but MHIV-mIN/matt did not show any significant increase in infectivity when compared with MHIV-mIN ( Figure 3C). The infectivity of MHIV-mIN was also sensitive to reverse transcriptase inhibitors ( Figure 3C and Figure S1), and thus depends on de novo genomic DNA synthesis. Therefore, we found that there is not a requirement for an MLV-specific att site in the context of a chimeric HIV with MLV IN.
IN mutants of HIV that are defective for integration support low levels of infectivity in the multinuclear activation of galactosidase indicate cell (MAGI) assay, probably due to weak expression of the tat gene products from unintegrated DNA [54][55][56]. However stable expression of transduced genes usually requires integration of viral DNA into host chromosome [57,58]. Thus, to genetically test for integration, we made use of a reporter virus system in which the puromycin-resistant gene was put in place of the nef gene, and infected cells were selected for puromycin resistance. Compared with HIV, MHIV-mIN exhibits ;4 log decrease of infectivity in the puromycin-based assay ( Figure  3D), which is about one log lower than the virus titer difference in MAGI assay ( Figure 3A). The difference between the MAGI titer and the puromycin-resistance titer is likely due to expression of Tat from unintegrated DNA [54][55][56]. Nonetheless, these data show that MHIV-mIN is capable of stable transduction.

Synopsis
Human immunodeficiency virus can infect many cells irrespective of whether or not they are dividing, whereas some other retroviruses, such as the murine leukemia virus can only infect cells that are proliferating. This property is important for the ability of HIV to establish infections in critical cell types in infected people. Multiple and redundant signals encoded by HIV have been hypothesized to facilitate migration of viral genomes into the nucleus. However, here the authors eliminated all four described nuclear localizing signals from an HIV genome and show that no combination of these virally encoded signals is essential for the ability of HIV to infect nondividing cells. They suggest that another step of the virus lifecycle, other than nuclear import, is the rate-limiting step that determines the cell cycle dependence/independence of retroviral infections.
To more directly address the question of whether or not the MHIV-mIN virus can carry out bona fide IN-mediated integration, we extracted genomic DNA from the puromycinresistant colonies to amplify and sequence the junction between viral and host sequences. There are two characteristic features of retroviral integration of viral DNA into the host genome. First, two nucleotides are deleted from both ends of viral DNA. Indeed, we observed the deletion of two nucleotides of both ends of all sequenced clones ( Figure 4). The second characteristic of IN-mediated integration is that the target sequence of the host DNA is duplicated after the integration event. The size of duplication differs among retroviruses [59]; for example, HIV integration yields 5-bp duplication of the target sequence, whereas MLV integration creates 4-bp duplication. In each case, the length of the duplicated sequence was 4-bp, which is consistent with integration of the HIV chimeric virus mediated by the MLV IN. Taken together these findings demonstrate that MHIV-mIN is competent for all of the early steps of virus replication including integration.

HIV IN Is Not Essential for Infection of Non-Dividing Cells
HIV efficiently infects non-dividing cells, whereas MLV infection is restricted in non-dividing cells. To determine if IN plays an essential role in this difference, growth-arrested cells prepared by treatment of HeLa cells with aphidicolin were challenged with the chimeric virus MHIV-mIN along with control viruses, and infectivity was judged by measuring the output of the luciferase gene encoded by reporter virus constructs. As expected from previous studies, wild-type HIV was capable of infecting non-dividing cells as efficiently as dividing cells, while transduction of the luciferase gene by MLV was reduced in non-dividing cells compared with in dividing cells ( Figure 5). The phenotype of MHIV-mIN was similar to that with HIV, but not with MLV, in that it was not decreased in non-dividing cells relative to dividing cells ( Figure 5). In fact, we saw a slight increase of infectivity by MHIV-mIN on non-dividing cells relative to dividing cells ( Figure 5). This increase may be due to expression of the  reporter gene from unintegrated DNA in non-dividing cells, as described in the case of infection of non-dividing cells with feline immunodeficiency virus IN mutants [57]. Nonetheless, these results demonstrate that IN is not an essential determinant for the ability of HIV-1 to infect non-dividing cells relative to dividing cells.
We previously showed that replacement of part of the gag gene of HIV with that of MLV would convert HIV into a virus that had lost the ability to infect non-dividing cells [40]. Similarly, we found that we can change the phenotype of the MHIV-mIN by replacing the MA and CA proteins of HIV with the MA, p12, and CA proteins of MLV (MHIV-mMA12CA-mIN in Figure 1). Indeed, addition of Gag proteins of MLV into the HIV provirus that already contains MLV IN increased the infectivity in dividing cells, but specifically lost the ability to infect non-dividing cells ( Figure 5: Compare MHIV-mMA12-CA/mIN with MHIV-mIN). These data demonstrate that Gag, rather than IN, is the dominant determinant for the ability of HIV to infect cells independent of cell cycle progression.

Normal Levels of Nuclear Import by MHIV-mIN
A recent report showed that efficient nuclear entry of HIV can occur independently of mitotic nuclear disassembly in cycling cells [60]. Thus, one interpretation of our results is that elimination of an NLS from HIV would result in lack of infectivity both in dividing and non-dividing cell populations. Indeed, the new chimeric virus created in the present study, MHIV-mIN, infects dividing cells and non-dividing cells with an equal efficiency, but the overall infectivity by MHIV-mIN is severely reduced from that of wild-type HIV-1 ( Figure 3A). Thus, to directly determine whether or not MHIV-mIN is restricted at nuclear import of viral DNA, infected cells were separated into cytoplasmic and nuclear fractions and realtime PCR was used to measure late reverse transcription products. The results indicate that there is little apparent difference of viral DNA associated with nuclear fractions between MHIV-mIN and HIV ( Figure 6A). Although higher  amounts of viral DNA were associated with the nuclear fractions of HIV (;75%) than MHIV-mIN (;50%), this level of difference cannot explain the decrease of infectivity by MHIV-mIN (3-log reduction compared with wild-type HIV). Control experiments using a cytoplasmic protein (LDH I) as a maker for the cytoplasmic fraction indicated that contamination of cytoplasm into the nuclear fraction is less than 1% ( Figure 6B. compare the 125-fold dilution of the cytoplasmic fraction in lane 2 with the nuclear fraction in lane 6). These data indicate that nuclear entry of MHIV-mIN is essentially not inhibited and that reduced infectivity of the chimeric virus is due to a post-nuclear entry event (most likely integration). We also examined 2-LTR circles, which are often used as a surrogate marker for nuclear entry. We found that the ratio of 2-LTR circles to total viral DNA in cells infected with MHIV-mIN is roughly equivalent (or even slightly higher) to the ratio in cells infected with parental HIV-1 ( Figure 6C). The slight increase in the average number of 2-LTR circles per total viral DNA for MHIV-mIN relative to wild-type virus likely reflects the fact that mutants that integrate inefficiently often accumulate 2-LTR circles [56]. Nonetheless, in sum, these data further support the idea that HIV IN is not essential for the nuclear transport of viral DNA and infectivity in non-dividing cells.
HIV Lacking All of the Known Types of NLS Still Infects Non-Dividing Cells as Efficiently as Dividing Cells As mentioned above, IN is not the sole candidate that potentially encodes a viral NLS. MA, Vpr, and the cPPT have all been described as elements that are important for entry of HIV-1 PIC into the nucleus. To formally address the argument that other described NLSs in HIV as well as the cPPT are redundant for nuclear import with the NLS in HIV IN, a mutant HIV-1 lacking all the NLS candidates was generated. This mutant (HIV-DNLS), carrying MLV MA and IN instead of HIV counterparts, lacks a functional vpr gene, and has a mutated cPPT (Figure 1). We found that HIV-DNLS had reduced infectivity relative to wild-type HIV (Figure 7), but the infectivity of HIV-DNLS is sensitive to reverse transcriptase inhibitors ( Figure S1), and thus is not an artifact of the virus concentration. Importantly, the infectivity of HIV-DNLS is independent of cell cycle conditions (Figure 7). It should be noted that our luciferase system can detect reduction even when the activity is low (See reduction of MLV infectivity in 0.08 ll in Figure 7, for example). The phenotype of HIV-DNLS in non-dividing cells is in marked contrast to that of MLV which is dependent on the cell cycle, and in contrast to a previously described chimeric HIV virus containing MLV MA, p12, and CA (MHIV-mMA12CA: Figure  1) [40], which has specifically lost the ability to efficiently infect non-dividing cells (Figure 7). Therefore, these data to cytoplasmic (C) and nuclear (N) fractions. Viral DNA was extracted and subject to real-time PCR to measure late products of reverse transcription. These data represent one of two independent experiments. Control viruses, with the presence of reverse transcription inhibitors or without VSV-G protein, were also used in the experiments to monitor retrovirus-dependent DNA synthesis, and showed that contamination of plasmid DNA used for transfection is less than 1% of the total DNA. (B) Western blot analysis of total cell lysates (To), cytoplasmic extract (Cy) and nuclear lysates (Nu). Contamination of cytoplasmic extract and the presence of intact cells in nuclear fractions were tested by checking for the presence of a cytoplasmic protein, LDH-I, in each fraction (upper lanes). Five-fold dilutions of the cytoplasmic extract (5-, 25-, and 125-fold dilutions, lanes 4, 3, and 2, respectively) were made to assess the degree of contamination of the nuclear fraction. Presence of proteins was confirmed by antibody against a nuclear pore complex protein (mAb414; lower lane). (C) Nuclear import was monitored by measuring late reverse transcription products and 2-LTR circles. The ratio of total viral DNA and 2-LTR circles was obtained by dividing the copy number of late RT products by the copy number of 2-LTR circle. The parental wild-type strain of HIV-1 (shown here as wt) was compared with the chimeric virus MHIV-mIN (shown as mIN). Control infections with reverse transcription inhibitors (AZT and 3TC: 50 lM each) yielded viral copy numbers that are less than 10% of copy numbers of the samples without reverse transcription inhibitors, indicating that contamination of plasmid DNA used to produce virus stocks does not affect the final results. Two independent experiments gave substantially identical data. DOI: 10.1371/journal.ppat.0010018.g006 demonstrate that HIV without any of the previously described NLS elements is fully capable of infecting nondividing cells as well as it infects dividing cells, and suggest that the virally encoded NLS elements are not rate-limiting for this process.

Discussion
In the present study, we created a chimeric HIV [16,24,25,61]. Indeed, a recent report by Lu et al demonstrated that mutations in a putative NLS of HIV IN results in class II mutations, which are defective at a postnuclear entry step rather than at nuclear import [61]. In contrast, Ikeda et al claimed that nuclear import of viral DNA is affected by reduced binding of IN to viral cDNA [62]. However, the reduction of nuclear import of such IN mutants (with reduced binding ability to viral cDNA) was at most 40% of the wild-type level as judged by nuclear DNA and that amount of reduction does not seem to explain severely reduced infectivity of their mutants (less than 1% of the wild-type level) [62]. Moreover, other studies have shown that HIV IN localizes to the nucleus by virtue of binding LEDGF/p75 [36,38,39]. However, reduction of LEDGF levels by small interfering RNA (siRNA) affected the nuclear localization of HIV IN, but did not affect the ability of HIV to infect non-dividing cells [36]. Therefore, although we cannot completely rule out the possibility that HIV IN is involved in nuclear migration of viral DNA, we believe that its role is minor.
Although other studies have ruled out a role for individual and some combinations of putative karyophilic viral elements in the HIV PIC, it has not been possible up to now to eliminate all of the identified elements at once in order to test the hypothesis that infection of non-dividing cells is reliant on multiple redundant NLS. However, we were able to create an HIV mutant lacking all of the known NLS-encoding elements, and demonstrated that not only IN, but also none of the other NLS-encoding elements have any effect on the ability of HIV-1 to infect non-dividing cells. Thus, our data are not consistent with a previous suggestion that mutation of single (or double) NLS-encoding elements had little phenotypic change because of redundant NLS-encoding elements that are responsible for nuclear transport of HIV PIC and for infection in non-dividing cells. One possible interpretation of our results is that we have not yet found the most important NLS encoded by HIV. While this is still formally possible, the present results along with our previous results that found that CA is a dominant determinant for retrovirus infectivity in non-dividing cells [40], suggest that these virally encoded karyophilic elements are not the major determinants for the infectivity of HIV in non-dividing cells. Rather, we consider that our data lend support to the alternative hypothesis that nuclear entry is not the rate-limiting step for infection of non-dividing cells. Our hypothesis is also consistent with the findings that the addition of NLS encoding sequences to MLV does not render it infectious to non-dividing cells [35,63].
Instead, we propose that the difference in CA between HIV and MLV affects the progress of uncoating, thereby influencing downstream events such as nuclear import and integration. In this model, uncoating of HIV progresses normally in non-dividing cells and functional PIC enter the nucleus where they integrate viral DNA. In contrast, uncoating of MLV is impaired in non-dividing cells, which results in the failure of subsequent steps of the replication cycle. In this scenario, gammaretroviruses may need mitosis to complete uncoating. In fact, in the case of HIV, CA is dissociated from viral nucleoprotein complexes [43][44][45][46][47][48], while larger amounts of CA are associated with MLV PICs [15,[64][65], suggesting that uncoating of MLV may not be as efficient as that of HIV. An optimal stability of the HIV core appears to be essential for infectivity [66], and complete uncoating may be a prerequisite for nuclear import of PIC. In this hypothesis, the tight association of MLV CA with PIC prevents cellular machinery from interacting with a putative NLS on MLV PICs, thereby retaining PICs within the cytoplasm of interphase cells. On the other hand, the HIV PIC can migrate into the nucleus of interphase cells by using cellular transport machinery. Thus, we are not arguing that nuclear import of the HIV PIC is not essential. Rather, that it is not the ratelimiting step and that cellular rather than viral components of the PIC might play the major role in viral nuclear import after uncoating.  [20] was introduced into pLai by PCR mutagenesis. Construction of a Vpr mutant (pLai-DVpr) [69] and MHIV having the MLV MA [40] was reported previously. All of these mutations and replacements were combined together with the env-deficient provirus clone pLai-DEnv to create the NLS-minus mutant HIV-DNLS.

Materials and Methods
The reporter virus constructs encoding the luciferase gene were made by introducing the luciferase gene from the wild-type Envminus HIV-1 (pLai-DEnv-luc2) [40] into the new HIV-based constructs. The puromycin resistant gene is cloned into the nef gene of molecular clones of HIV-1, MHIV-mIN, and MHIV-Mma12CA/mIN. The HIV puromycin resistant constructs were created in the same way as the luciferase constructs as described above. The vpr mutant pLai-DVpr was used to introduce an insertional mutation in the vpr gene of the puromycin virus constructs.
Western blot analysis. Western blots were probed with the following antibodies: rabbit anti-HIV-1 RT antibody (through the AIDS Research and Reference Reagent Program, Division of AIDS, NIAID); mouse monoclonal anti-HIV-1 IN antibody (Michael Malim, King's College, London); rabbit anti-MLV IN (Frederick Bushman, University of Pennsylvania, [65]); sheep antibody against LDH I (Cortex Biochem, San Leandro, CA); and mouse monoclonal antibody against nuclear pore complex proteins MAb414 ( Covance, Devner, Pennsylvania) [70]. The membranes were washed for 30 min in wash buffer (PBS containing 0.2% Tween 20) and then incubated with a 1:10,000 dilution of horseradish peroxidase-conjugated antibodies that match with the primary antibody for 60 min at room temperature. The membranes were washed three times for 30 min, and the bound antibody was detected with ECL Plus Western blotting detection reagents (Amersham Biosciences, Little Chalfont, United Kingdom). In some cases, membranes were stripped and reprobed with another primary antibody.
Infectivity assays. Vesicular stomatitis virus G protein (VSV-G)pseudotyped viruses were prepared by transient transfections of 293T cells performed with the FuGene 6 reagent. HIV and MHIV expression plasmids were co-transfected with a VSV-G-expression vector (pL-VSV-G [71]) in addition to pCMV-tat to express the VSV-G for pL-VSV-G. For the production of VSV-G-pseudotyped MLV, the MLV Gag-Pol expression vector (pCS2-mGP) [40] were used along with the murine retrovirus-based vectors [72] encoding the luciferase gene (pLNCluc) [40] as well as the VSV-G construct. To enhance the infectivity, MHIV-mIN and HIV-DNLS were concentrated by ultracentrifugation. Briefly, 25-35 ml of culture supernatant of transfected 293T cells were centrifuged at 500 3 g for 5 min to remove cell debris and then filtered through a 0.22 lm filter. The supernatant were transferred into ultracentrifuge tubes and centrifuged at 64,000 3 g for 90 min within a SW28 rotor (Beckman Instruments, Fullerton, California, United States). The supernatants were carefully removed and 250-350 ll of culture medium was added at 4 8C for 1 hr and freshly used for infection.
Single-cycle infectivity of HIV and MHIV was measured by challenging MAGI cells with serial dilution of virus and staining for b-galactosidase expression as basically described previously [50]. HeLa cells were used for infections with the luciferase reporter virus stocks. Luciferase titer was assayed with the luciferase assay kit (Promega, Madison, Wisconsin, United States) and read on a luminometer. Growth-arrested cells were prepared by treatment with 2 lg per ml of aphidicolin (Sigma, St. Louis, Missouri, United States). Virus binding was enhanced by spinoculation [73] and by addition of 20 lg per ml of DEAE/dextran.
Quantification of p24gag and viral cDNA. The p24gag content of the viral supernatants was determined by an enzyme-linked immunosorbent assay (ELISA; Beckman Coulter, Hialeah, Florida, United States.). Late products of reverse transcription and 2-LTR circles of HIV-1 were measured by using real-time PCR based on a previously published protocol [74] as described previously [40].
Subcellular fractionation. One day before infection, approximately 5 million HeLa cells were seeded onto four 75 cm 2 flasks. The cells were challenged either by the VSV-G-pseudotyped HIV-1 or MHIV-mIN. Cells were infected with virus stocks that can synthesize equivalent amount of viral DNA in target cells. Virus stock of MHIV-mIN was concentrated by ultracentrifugation. Both virus stocks were treated with 50 units of Turbo DNase (Ambion, Austin, Texas, United States) per ml at 37 8C for an hour. Infections were performed with the presence of DEAE/dextran (20 lg per ml).
Subcellular fractionations were carried out based on the method described by Yuan et al [75] with minor modifications. One day after infection, cells were washed, tripsinized, and washed once again with phosphate-buffered saline. In order to extract cell lysates and DNA from intact cells, 20% of the infected cells were kept for further experiments. All the manipulations after this step were carried out at 4 8C. The remaining 80% cells were resuspended in 3 volumes of hypotonic buffer (10 mM HEPES, [pH 7.9]; 1.5 mM MgCl 2 ; 10 mM KCl; 2 mM dithiothereitol ; 20 lg of aprotinin per ml). Resuspended cells were centrifuged at 2,300 3 g for 5 min. The cell pellet was resuspended in 3 volumes of hypotonic buffer and kept on ice for 10 min. The cells were homogenized with 30 strokes in a Dounce homogenizer. Nuclei and cell debris were pelleted by centrifugation at 3,300 3 g for 15 min. The supernatant of this centrifugation was directly used to extract viral DNA or additionally spun down at 13,400 3 g for 20 min. The nuclear pellet were washed with 3 volumes of hypotonic buffer containing 0.005% digitonin once and then washed with hypotonic buffer twice. DNA was extracted from half of each fraction using the QIAamp DNA Mini Kit (Qiagen, Valencia, California, United States), and the other half was used for Western blotting to assess the integrity of the fractionation procedure using antibodies to a cytoplasmic protein or a nuclear pore antigen.
To assess the integrity of the fractionation procedure, we examined the contamination of cytoplasmic fraction into nuclear fraction by monitoring the presence of LDH I. Intact cells (10% of the infected cells) and nuclei (half of the purified nuclei) were first resuspended with 50 ll and 100 ll of NTE buffer (10 mM Tris-HCl, [pH 8.0]; 1 mM EDTA; 50 mM NaCl; 2 mM DTT; 20 lg of aprotinin per ml), respectively. After incubation on ice for 5 min, equivalent amount of NP40-doc buffer (1% NP40; 0.2% sodium deoxycholate; 0.12 M NaCl; 20 mM Tris-HCl, [pH 8.0]) were added to the samples and kept on ice for 10 min. The samples were mixed by vortex and spun down at 9,300 3 g for 5 min. Twenty lg of protein samples was used in SDS-PAGE and western blotting analysis.
The most serious problem for our experiments was potential contamination of cytoplasmic viral DNA into purified nuclei. Viral DNA is present in a nucleoprotein complex or a free-DNA form, and those viral DNA may behave differently than cytoplasmic proteins such as LDH I in the process of fractionation. To address this possibility, we used a control to determine if quantity of contamination of viral DNA from the cytoplasmic fraction into the nuclear fraction during the washing steps. To this end, we mixed cytoplasmic extracts of infected cells with nuclei of uninfected cells. In these experiments, we used MLV instead of HIV because of the ease of manipulation. Virus stocks of MLV were prepared by harvesting culture supernatant of ecotropic MLV-producing NIH/3T3 cells. Cytoplasmic extracts of acutely infected cells and nuclei of uninfected NIH/3T3 cells were prepared, mixed on ice, and washed as described above. Nuclear-associated DNA was extracted and subject to real-time PCR to measure the copy number of late reverse transcription products as described above. We found that there was less than 1% introduction of cytoplasmic viral DNA into the nuclear fraction during the washing steps (unpublished data).
Sequencing of junctions between host DNA and integrated viral DNA. Junction sequences between host DNA and viral DNA were determined by using an inverse PCR strategy as described before [76]. HeLa cells were infected with VSV-G-pseudotyped MHIV-mIN-DVpr-Puro. Puromycin-resistant cell clones (;130 colonies) were selected for 2-3 weeks with the presence of puromycin (0.7 lg per ml) and used to extract genomic DNA. Genomic DNA (2 lg) from infected cells was digested with 20 U of PstI at 37 8C for 12 h. After heat inactivation at 65 8C for 40 min, 200 ng of digested DNA were taken out for ligation reaction. The ligation reaction was carried out at 16 8C for 12 h. Ligase was then heat inactivated at 65 8C for 15 min. The region of the junction between cellular DNA and the 59 end of the integrated proviral DNA was amplified by nested inverse PCR. The first PCR primers were U3RRG2, 59-GGCAAGCTTTATTGAGGC-39and Gag716, 59-GGTCAGCCAAAATTACCCTATAGTG-39. The second PCR primers were MH536, 59-TCCACAGATCAAGGA TATCTTGTC-39, and Gag934, 59-TGTTAAAAGAGACCATCAAT GAGGAAG-39. The PCR was carried out in 50 ll solution, which contains 1 3 PCR buffer, dNTPs (0.2 mM), primers (1 lM), 10 units of Taq polymerase (Roche, Basel, Switzerland), and 200 ng of ligated DNA. PCR products were purified and used for cloning by using pGEM-T Vector System (Promega). Positive clones were sequenced by using the T7 primer.
Junction sequences between 39 ends of viral DNA and host DNA were determined for three of the clones by nested PCR with 59sense primers matching with 39 LTR of proviral DNA and with 39 anti-sense primers matching with host DNA downstream of viral DNA. The information obtained from 59 junction sequences between host DNA and viral DNA allowed us to map integration sites of these three clones in the human genome sequence deposited in GenBank. Based on this information, 39 primers were designed. Amplified products were cloned into T-vector and sequenced. Figure S1. Single-Cycle Infectivity Assay of MHIV-mIN and HIV-DNLS with Reverse Transcriptase Inhibitors Infections with viruses that encode the luciferase gene in place of nef were performed with the presence (shown in black) or absence (shown in gray) of a reverse transcriptase inhibitor (RTI) (AZT and 3TC: 50 lM each). For details, see the legend to Figure 5. In both cases, the luciferase activity is decreased by RTI which indicates that expression of luciferase relies on de novo RT activity. Also, the presence of aphidicolin does not change the dependence on de novo reverse transcriptase activity for luciferase activity. This is a representative experiment done with two different virus stocks with virtually identical results.