Unintegrated HIV-1 DNAs are loaded with core and linker histones and transcriptionally silenced

Significance

Retroviruses are characterized by the reverse transcription of the viral RNA genome into DNA and the integration of that DNA to form the provirus. However, little is known about the nature of unintegrated HIV-1 DNAs early upon delivery into the nucleus. Using chromatin immunoprecipitation assays, we found that both core and H1 linker histones are deposited onto unintegrated HIV-1 DNAs. We also confirmed transcriptional silencing of unintegrated HIV-1 DNAs and determined the presence of posttranslational histone modifications characteristic of inactive chromatin. Our results will help to increase the efficiency of expression from nonintegrating HIV-1–based vectors and after transient transfections with DNA.

Abstract

Upon delivery into the nucleus of the host cell, linear double-stranded retroviral DNAs are either integrated into the host genome to form the provirus or act as a target of the DNA damage response and become circularized. Little is known about the chromatinization status of the unintegrated retroviral DNAs of the human immunodeficiency virus type 1 (HIV-1). In this study, we used chromatin immunoprecipitation to investigate the nature of unintegrated HIV-1 DNAs and discovered that core histones, the histone variant H3.3, and H1 linker histones are all deposited onto extrachromosomal HIV-1 DNA. We performed a time-course analysis and determined that the loading of core and linker histones occurred early after virus application. H3.3 and H1 linker histones were also found to be loaded onto unintegrated DNAs of the Moloney murine leukemia virus. The unintegrated retroviral DNAs are potently silenced, and we provide evidence that the suppression of extrachromosomal HIV-1 DNA is histone-related. Unintegrated DNAs were marked by posttranslational histone modifications characteristic of transcriptionally inactive genes: high levels of H3K9 trimethylation and low levels of H3 acetylation. These findings reveal insights into the nature of unintegrated retroviral DNAs.

Despite extensive research on the human immunodeficiency virus type 1 (HIV-1) life cycle in past years, only the broad outlines of the early steps after infection are known. The detailed events that occur immediately after entry of the retroviral DNA into the nucleus, before the formation of the integrated provirus, are especially unclear. The retroviral life cycle begins with receptor-mediated fusion of the virion envelope with the host-cell membrane and subsequent release of the retroviral capsid into the cytoplasm. The viral RNA genome is reverse transcribed to form a linear double-stranded DNA in a large structure resembling the virion core, termed the preintegration complex (PIC) (1, 2), and this viral DNA complex is delivered into the nucleus by various mechanisms, depending on the virus. The PIC of the simple retroviruses, such as the Moloney murine leukemia virus (MLV), gains access to the nucleus only after nuclear membrane disassembly during cell mitosis (3, 4). In contrast, entry of HIV-1 DNA into the nucleus involves active transport through nuclear pores and uses a variety of pore-associated Nup proteins (5–8). Once inside the nucleus, the linear viral DNA is integrated into the host genome by the viral integrase protein (IN), forming the provirus. A portion of the viral DNA that fails to integrate becomes circularized to yield circles with either 1 copy of the long terminal repeats (1-LTR circles) (9) or 2 tandem copies of the repeats (2-LTR circles) (10). If proviral formation is prevented either by an IN mutation or by an IN inhibitor, the circular DNAs tend to accumulate to higher levels than when integration is allowed (11). The unintegrated DNAs—the linear and the circular forms—do not replicate and gradually disappear as the host cell continues to proliferate.

It has long been known that the unintegrated retroviral DNAs are poorly transcribed and that, after successful proviral formation in permissive cells, the transcriptional silencing is relieved (12, 13). The state of the unintegrated viral DNA, however, only recently is becoming clarified. The unintegrated DNA of MLV is rapidly loaded with core histones and that the histones acquire covalent marks characteristic of silent chromatin (14). The silencing of the unintegrated DNA is relieved by histone deacetylase (HDAC) inhibitors such as trichostatin A, suggesting that histone modifications may mediate the silencing (15, 16). An unbiased genome-wide CRISPR-based knockout screen for genes required for the silencing of unintegrated MLV DNAs revealed the involvement of a large DNA-binding protein (NP220), all 3 subunits of the eponymous HUSH complex, and the H3K9 trimethyl transferase ESET/SETDB1, as well as selected HDACs (17). These various proteins were shown to bind to the unintegrated MLV DNA and to be removed from the DNA after integration, coordinately with the activation of normal transcription from the provirus. There were indications that different viruses utilized different components of the silencing machinery; for example, the HUSH complex was more important in silencing unintegrated MLV DNAs than unintegrated HIV-1 DNAs.

In eukaryotic cells, histones are organized in an octamer nucleosomal structure, consisting of 2 copies of each the histones H2A, H2B, H3, and H4 (18, 19). Approximately 150 bp of DNA is wrapped twice around each nucleosome (19). In addition to canonical core histones, there are also specific core histone variants of H2 and H3 that may contribute to regulation and control of gene transcription (20). The histone variant H3.3 is a conserved protein which shares structurally very high similarity with the canonical histones H3.1 and H3.2 (21). The function of the H3.3 variant has not been fully elucidated. On one hand, it has been associated with active transcription (22, 23); on the other hand, it has been suggested as accumulating at silent loci in telomeric or pericentric heterochromatin regions (24–26). Interestingly, it has been shown to play a role in transcriptional silencing of certain retroviral elements in mouse embryonic stem cells (27–29). In addition to the core histones, chromosomal DNA is also condensed by the binding of so-called H1 linker histones. They are required to mediate higher-order chromatin structures. There are 5 somatic subtypes described (H1.1 to H1.5) and, in addition, different H1 variants. In addition to their function in stabilizing nucleosomes, H1 linker histones have been shown to play a role in regulation of gene expression (19, 30, 31). In eukaryotic chromosomes, histones are added either onto daughter DNA strands behind the replication fork or are replaced on DNAs at sites of damage or regions of changing transcriptional regulation. The loading of nucleosomes onto unintegrated retroviral DNA occurs in an extraordinary circumstance, in that the DNA has never previously been associated with histones. How histones are loaded on retroviral DNAs is not yet known. The state of unintegrated HIV-1 DNA and the identities of the histones loaded on the DNA have not been reported. In this study, we investigated the nature of unintegrated HIV-1 DNA. We showed that core histones, the H3 variant H3.3, linker H1 histones, and the posttranslational histone 3 lysine 9 trimethylation (H3K9me3) modification are all rapidly loaded onto unintegrated HIV-1 DNA. Furthermore, we found that H1 linker histones and the histone variant H3.3 were also associated with extrachromosomal DNAs of MLV.

Results

Unintegrated HIV-1 DNAs Are Transcriptionally Silenced Unlike Integrated Proviral DNA.

We first confirmed the strong silencing of unintegrated HIV-1 DNAs after infection of HeLa cells. We utilized reporter viruses encoding ZsGreen, based on the pNL4-3.R⁻.E⁻ strain of HIV-1, pseudotyped with the vesicular stomatitis virus glycoprotein (VSVg) envelope, produced after transfection of 293T cells. We chose to employ a commonly used viral construct lacking the Vpr protein that would otherwise target various host proteins for degradation, activate DNA damage responses, and cause cell cycle arrest (32–34). To better monitor unintegrated retroviral DNAs, we utilized an integration-defective mutant (IN-D64A), which harbors a single point mutation at the catalytic site of the viral enzyme integrase (35). This mutant allowed normal viral DNA synthesis and nuclear import of the retroviral DNA but prevented integration of the viral DNAs into the host-cell genome. For comparative studies we used an integration-proficient reporter virus (IN-wt). Producer cells were generated by transfection with the 2 viral constructs in parallel, and infections were carried out with equal volumes of virus preparations harvested from the producer cells. We performed flow cytometry analyses of reporter gene expression 24 h after infection (representative plots are shown in Fig. 1A). The integrase-defective mutant yielded a significantly reduced number of ZsGreen-positive cells as compared to the IN-wt (Fig. 1 A and B), and moreover, the mutant resulted in significantly reduced levels of viral expression compared to IN-wt measured by the mean fluorescence intensity (MFI) of the ZsGreen-positive population (Fig. 1 A and C). To include consideration of both the number of ZsGreen-positive cells and their MFI after infection, we calculated the product of these 2 readouts. This value for gene expression revealed the strong defect in expression from the unintegrated DNAs even more clearly, with an ∼9-fold decrease in the IN-D64A virus as compared to IN-wt (Fig. 1D). Notably, we performed an additional readout to monitor silencing using viral luciferase reporters. We detected a very pronounced silencing of viral expression of IN-D64A–infected cells compared to IN-wt in this readout (SI Appendix, Fig. S1). To confirm that there was no defect in viral DNA synthesis, we isolated DNA and performed a quantitative real-time PCR (qPCR) utilizing ZsGreen-specific primers to detect total viral DNA, which includes linear, circular, and (for the IN-wt virus) integrated viral DNA. Heat-inactivated virus was used as a control for potential plasmid DNA contamination. Importantly, these data exhibited similar levels of total viral DNA of IN-D64A and IN-wt present in the cells (Fig. 1E). We also used qPCR with appropriate primers to assess the levels of 2-LTR circles, specifically reporting the unintegrated viral DNAs formed after nuclear entry. As expected, infection with the IN-D64A virus resulted in higher levels of 2-LTR circles arising as a result of the block to integration (Fig. 1F). The results show a strong block to expression from the unintegrated HIV-1 DNA compared to expression from integrated DNA, as has been seen with MLV (12, 14, 17).

Fig. 1.

Unintegrated HIV-1 DNAs are potently silenced after nuclear entry. HeLa cells were infected with equal amounts of viral supernatants collected after transfection of 293T cells in parallel with either integrase-deficient mutant (IN-D64A) or wild-type IN (IN-wt) DNAs. Cells were monitored 24 h after infection. Data points are the results of independent experiments with 3 independently produced viral supernatants. The error bars indicate mean ± SD. Statistical significance was determined by unpaired Student’s t test. Mock: noninfected control; hi: heat-inactivated viral supernatant as a plasmid contamination control. (A) Flow cytometry analyses were conducted, and representative histograms are shown from IN-D64A and IN-wt infected cells as well as nontransduced control cells (NTD). (B) The percentage of ZsGreen-positive cells is depicted. *P = 0.032. (C) MFI of ZsGreen-positive population is illustrated. **P = 0.004. (D) The percentage of ZsGreen-positive cells and their MFI were multiplied and are represented relative to IN-wt. (E) Total viral DNA levels were evaluated with qPCR using ZsGreen-specific primers. Data were normalized to GAPDH copies. (F) 2-LTR circles were determined by qPCR. Data were normalized to GAPDH levels.

H1 Linker and Core Histones Are Loaded onto HIV-1 Unintegrated DNA.

We used chromatin immunoprecipitation (ChIP) to directly examine the histone content of unintegrated HIV-1 DNAs. HeLa cells were infected with equal amounts of IN-D64A or IN-wt viral supernatants, and at 24 h post infection ChIP was performed followed by qPCR-based analyses of the viral DNAs. ChIPs utilized H1.2- and H1.4-specific antibodies, as members of the linker histone family, and H2B- and H3-specific antibodies as representatives of core histones. Two-LTR circle-specific primers were used for the subsequent qPCR to monitor loading on unintegrated virus DNA, and other primers for the cellular housekeeping gene glyceraldehyde 3-phosphate dehydrogenase (GAPDH) as an active gene control and the beta globin gene as a heterochromatin control. We detected loading of all tested histone members on the unintegrated HIV-1 DNA of the IN-D64A mutant (Fig. 2A) as well as IN-wt (Fig. 2B). H2B and H3 were present on both GAPDH and beta globin as expected, with H1.2 and H1.4 antibodies recovering a higher level of beta globin DNA compared to GAPDH DNA (SI Appendix, Fig. S2 A and B). To test for posttranslational modifications of histones, we used ChIP antibodies to detect H3K9me3, a marker for transcriptionally inactive chromatin, and acetylated H3 (H3ac), a marker for transcriptionally active genes. ChIP with H3K9me3 antibodies recovered high levels of unintegrated HIV-1 DNA, whereas H3ac antibodies yielded reduced levels as compared to GAPDH (Fig. 2 A and B and SI Appendix, Fig. S2 A and B). As expected, DNAs recovered with H3ac antibodies were high in GAPDH and very low in beta globin. In contrast, levels of DNAs associated with H3K9me3 were high in beta globin and low in GAPDH. These data demonstrate that core and H1 histones are loaded onto unintegrated HIV-1 DNAs and modified with marks associated with silencing.

Fig. 2.

Unintegrated HIV-1 DNA is bound to linker and core histones. IN-D64A or IN-wt particles were applied to HeLa cells, and ChIP analyses were conducted after 24 h. Histone-specific antibodies were used for ChIP as indicated, along with a species-specific isotype control. Mouse IgG (mIgG) served as a control for H2B, rabbit IgG (rIgG) for all other antibodies. Data are expressed as the percentage of input DNA and are shown as mean ± SD. Data points represent independent experiments with 3 independently produced viral supernatants. (A) ChIP samples from IN-D64A–infected cells were analyzed by qPCR with 2-LTR–specific primers. (B) ChIP samples from IN-wt–infected cells were analyzed by qPCR with 2-LTR–specific primers.

Histones and H3K9me3 Are Deposited onto HIV-1 Unintegrated DNAs Early After Infection.

To investigate the timing of histone loading during the retroviral life cycle, we exposed HeLa cells to IN-D64A particles and after 6 h removed the virus and washed the cells to prevent further infection. We next isolated DNA and performed qPCR to follow viral DNA synthesis and also conducted ChIP followed by qPCR at 6, 12, 24, and 48 h post infection (Fig. 3A). The level of total viral DNA, determined by qPCR with ZsGreen-specific primers, was maximal within 12 h after virus application and gradually decreased thereafter (Fig. 3B). qPCR analyses with 2-LTR circle-specific primers revealed the formation of 2-LTR circles between 6 and 12 h after virus application (Fig. 3C). To examine the time course of histone loading, we used H2B- and H3-specific antibodies, assaying the prototypical core histones, an H1.4-specific antibody, assaying a linker histone family member, as well as a H3K9me3-specific antibody detecting a major silencing mark. In all cases, the loading of each histone and the formation of the silencing mark on the 2-LTR circles occurred rapidly and peaked within 12 h after virus application (Fig. 3D). Interestingly, the time course of histone loading onto the total viral DNA, based on qPCR with ZsGreen-specific primers after ChIP, showed more gradually increasing levels, with a peak occurring only at 48 h after virus application (Fig. 3E). qPCR analyses with GAPDH or beta globin-specific primers showed comparable results at all time points as well as between sample replicates (SI Appendix, Fig. S3 A and B). As control for nonspecific ChIP of DNA, we always conducted tests with respective isotype control (IgG)-specific antibodies. Rabbit IgG background levels of DNA were uniformly low with primers for 2-LTR circles (Fig. 3D), GAPDH (SI Appendix, Fig. S3A), or beta globin (SI Appendix, Fig. S3B). Throughout our studies, however, the rabbit IgG gave significantly elevated background signals when scoring total viral DNA, especially at earlier time points after virus application (Fig. 3E). The basis for the background ChIP, seen only with the total viral DNA primers, is not clear. The background prevented meaningful readouts of loading on total viral DNA in some cases.

Fig. 3.

Histone loading onto unintegrated HIV-1 DNAs occurs early after infection. (A) Schematic overview of experimental setup. HeLa cells were infected with IN-D64A virus and washed 3 times at 6 h after virus application, and cells were harvested at indicated time points for qPCR or ChIP with subsequent qPCR. Data points represent 2 independent experiments with 2 independently produced viral supernatants and are shown as mean and range. (B) Cells were pelleted, and DNA was isolated to measure total viral DNA levels by PCR with ZsGreen-specific primers. Data were normalized to GAPDH levels. (C) Cells were harvested, and DNA was isolated to assess 2-LTR circles by qPCR. Data are depicted relative to GAPDH levels. (D) ChIP analyses were conducted using histone-specific antibodies as indicated, along with their respective isotype control (H1.4, H3, H3K9me3: rIgG; H2B: mIgG). qPCR was performed using 2-LTR–specific primers. Data are expressed as percentage of input DNA. (E) qPCRs were performed on ChIP samples with ZsGreen-specific primers to detect total viral DNA. Data are shown as percentage of input DNA.

The Histone Variant H3.3 Is Loaded onto HIV-1 Unintegrated DNA.

The H3 histone variant H3.3 has been shown to play a role during silencing of endogenous retroviral elements (27–29). Therefore, we were motivated to determine whether the H3.3 variant is deposited onto unintegrated HIV-1 DNA. To address this question, cells were exposed to either IN-D64A or IN-wt particles, and at 24 h post infection ChIP was performed with a H3.3-specific antibody. qPCR analyses of ChIP samples revealed substantial levels of unintegrated HIV-1 DNA marked by H3.3 deposition (Fig. 4A). The experiments also revealed the presence of H3.3 histones in the housekeeping GAPDH (Fig. 4B) as well as the heterochromatic beta globin gene (Fig. 4C), although at a lower level.

Fig. 4.

The histone variant H3.3 is loaded onto unintegrated HIV-1 DNA. Equal amounts of viral supernatants of IN-D64A or IN-wt particles were applied to HeLa cells, and ChIP analyses with H3.3-specific antibodies and rabbit IgG were performed after 24 h. Data are depicted as percentage of input DNA and are shown as mean ± SD; data points represent independent experiments with 3 independently produced viral supernatants. (A) qPCR reactions were performed on ChIP samples with 2-LTR–specific primers. (B) qPCRs of ChIP samples were performed with GAPDH-specific primers for housekeeping control. (C) ChIP samples were analyzed by qPCR using beta globin-specific primers as a heterochromatin control.

Transient H3.3 Knockdown Has Only a Minor Impact on H3.3 Deposition onto Unintegrated HIV-1 DNA.

The H3 variant H3.3 is encoded by 2 genes, H3F3A and H3F3B. The complete knockout of both genes in mice is described to be semilethal (36). To investigate any impact on H3.3 loading onto unintegrated HIV-1 DNA and potential changes in viral gene expression, we therefore chose a small interfering RNA (siRNA) knockdown (KD) approach, transiently knocking down both genes followed by an infection and subsequent analyses after 24 h. siRNA KD of transcripts of either one of the H3F3A and H3F3B genes alone did not prevent H3.3 protein expression, but KD of both together led to an efficient reduction in H3.3 protein levels as assessed by Western blot (Fig. 5 A, Left). Levels were unaffected in nontreated mock cells. These experiments demonstrate that both genes are expressed and that KD of both transcripts is required to efficiently reduce the levels of H3.3 histones. These experiments also demonstrate the specificity of the H3.3-specific antibody utilized throughout our ChIP analyses, since the H3.3 KD led to strong loss of signal detected with H3.3-specific antibody (Fig. 5 A, Left). No loss of signal was detected with the canonical H3-specific antibody (Fig. 5 A, Right).

Fig. 5.

Transient H3.3 KD has a minor impact on H3.3 deposition onto unintegrated HIV-1 DNA and its expression. (A) Cells were transfected twice on 2 consecutive days either with siRNAs against H3F3A or H3F3B or with siRNAs against both H3F3A and H3F3B. Mock cells served as a nontreated control. Cells were harvested and lysed 24 h after the second siRNA transfection. Western blot was performed using an H3.3-specific antibody (Left) or an H3-specific antibody (Right). GAPDH-specific antibody was used as a housekeeping control. (B) HeLa cells were transfected twice on 2 consecutive days with siRNAs against H3F3A and H3F3B, both encoding H3.3, and with a nontargeting siRNA control (NT). H3.3 KD or NT cells were infected 24 h after the second siRNA transfection with IN-D64A virus. Data are depicted as mean ± SD, and data points represent independent experiments with at least 3 independently produced viral supernatants. ChIP analyses with indicated histone-specific antibodies and their isotype control were performed 24 h after virus application. qPCR data for 2-LTR–specific primers are shown as percentage of input DNA. (C) H3.3 KD (H3F3A+H3F3B) or NT cells were infected 24 h after the second siRNA transfection with various volumes of viral supernatants of IN-D64A or IN-wt virus as indicated. Data are shown as mean and range, and data points represent independent experiments with independently produced viral supernatants. Flow cytometry analyses were conducted, and the percentage of ZsGreen-positive cells is shown. Data are depicted as mean and range from 2 to 3 independently produced viral supernatants as indicated. (D) MFI from the same experiment is shown. (E) The product of the percentage of ZsGreen-positive cells and their MFI is presented relative to control NT cells for each setting.

To monitor H3.3 loading after KD, both H3.3 KD and nontargeting siRNA control cells (NT) were exposed to virus, and ChIP experiments were performed at 24 h after infection. Surprisingly, only a minor decrease in levels of DNA recovered with H3.3-specific antibodies was detected in the H3.3 KD cells compared to NT (Fig. 5 B, Left). Thus, although the levels of H3.3 histones in the cells were strongly reduced as assessed by Western blot, sufficient H3.3 was still loaded to allow for substantial ChIP of the unintegrated HIV-1 DNA. The H3.3 KD also led to a small reduction in the levels of unintegrated DNA recovered after ChIP with the canonical H3-specific antibody (Fig. 5 B, Right). Interestingly, H3K9me3 levels were slightly increased in H3.3 KD in contrast to NT control cells (Fig. 5 B, Right). Of note, qPCR analyses after ChIP with GAPDH- or beta globin-specific primers exhibited no changes in ChIP with H3 or H3K9me3 antibodies (SI Appendix, Fig. S4 A and B, Left) and only a minor decrease in H3.3 antibodies (SI Appendix, Fig. S4 A and B, Right).

To address a possible correlation of H3.3 KD and viral gene expression, we exposed siRNA-treated cells to either IN-D64A or IN-wt particles of indicated volumes of viral supernatants and monitored expression. Flow cytometry analyses were conducted 24 h after virus application. H3.3 KD cells exhibited a very modest increase in ZsGreen-positive cells (Fig. 5C) and in MFI levels (Fig. 5D) as compared to control cells after infection with both IN-D64A and IN-wt virus. Combining the readouts of the ZsGreen-positive cells and the MFI, by multiplying both values, revealed only a minor increase in viral gene expression for H3.3 KD cells treated with IN-D64A compared to NT control cells (Fig. 5E).

Unintegrated DNAs of MLV, like Those of HIV-1, Are Loaded with H1 Linker and H3.3 Histones.

In addition to test histone loading on HIV-1–based reporter viruses, we performed analyses with MLV, another member of the retrovirus family. Our group has previously shown that core histones are loaded onto unintegrated MLV DNA early after nuclear entry (14). We utilized an MLV-based reporter virus harboring an IN-inactive mutation D184A and encoding GFP. We infected HeLa cells and performed ChIP experiments after 24 h. Both the circular unintegrated MLV DNAs, scored using 2-LTR–specific primers (Fig. 6A), and the total unintegrated MLV DNA, scored using GFP-specific primers (Fig. 6B), were decorated with linker histones, shown by H1.4 deposition, as well as with the H3.3 variant. qPCR analyses of the DNAs after ChIP with primers specific for GAPDH (Fig. 6C) or beta globin (Fig. 6D) showed expected levels of H1.4 and H3.3. Thus, like HIV-1 DNAs, unintegrated MLV DNAs are also loaded with both H1 linker and H3.3 variant histones.

Fig. 6.

Histone variant H3.3 and linker histones are deposited onto unintegrated MLV DNAs. HeLa cells were infected with MLV integrase-deficient mutant IN-D184A virus encoding GFP, and ChIP analyses were carried out with histone-specific antibodies and their isotype control 24 h after virus application. Data shown are percentage of input DNA, as mean ± SD. Data points represent independent experiments with 3 independently produced viral supernatants. (A) ChIP samples were used to perform qPCR with 2-LTR–specific primers. (B) qPCR reactions were performed on ChIP samples with GFP-specific primers to detect total viral DNA. (C) qPCR reactions were performed with GAPDH-specific primers as a housekeeping control. (D) qPCR reactions were performed on ChIP samples with beta globin-specific primers as a heterochromatin control.

Discussion

The suppression of incoming DNA by transcriptional silencing is a major aspect of cellular innate immunity, constituting a block to the initiation of infection by invading viruses. The system is known to act against many incoming viruses, including MLV (17), and many DNA viruses, such as herpes viruses (37, 38), adenoviruses (39), and adeno-associated viruses (40). In all these cases the mechanism involves the loading of histones onto the viral DNA and the posttranslational modification of these histones to induce silencing and regulate gene expression. The importance of the antiviral system is highlighted by the fact that many viruses have evolved genes that function to inactivate the silencing machinery and allow viral expression, such as Vpr of HIV-1 and Vpx of simian immunodeficiency virus type 1 (SIV-1) (41–44). The system is also active in suppression of expression of nonviral DNAs introduced by many transfection methods, and inhibition of the silencing machinery can result in dramatic enhancement of the level of expression of such DNAs (45, 46).

In this study, we investigated the nature of the chromatin formed on unintegrated HIV-1 DNAs, its histone composition, and the time course of histone loading. We made use of IN-deficient virus particles to assure analysis of purely unintegrated DNAs. The circumstances of histone loading on these DNAs are unusual, in that nucleosome formation needs to take place on “virgin” DNA, which has never previously been associated with histones. As we previously have reported with MLV (14), we detected rapid loading of core histones, such as H3 and H2B, on unintegrated HIV-1 DNA. HIV-1 and MLV enter the nucleus through very different mechanisms: MLV requires nuclear membrane disassembly, while HIV-1 engages with nuclear pore proteins. Thus our studies show that histone loading onto the unintegrated retroviral DNAs occurs independently of the route of nuclear entry. Using specific antisera, we also demonstrate loading of the histone variant H3.3 on unintegrated DNAs of both MLV and HIV-1. The significance of this variant for viral DNA processing or expression remains unclear, but may contribute to the limited expression of these DNAs. In addition, and unexpectedly, we detected loading of the linker histone H1 on extrachromosomal HIV-1 and also MLV DNAs. These histones are traditionally associated with higher-order condensation of chromatin and thus may induce a compaction of the DNAs. Linker histones are also described in the literature to play a role in gene regulation, and numerous posttranslational modifications are known, but their function is poorly understood (30). It would be of interest to analyze potential correlations between linker histone H1 and regulation of retroviral gene expression.

It has been previously shown that the unintegrated DNAs of both HIV-1 and MLV are heavily silenced (12–14, 17). Here we provide confirmation that the silencing is pronounced; delivery of a ZsGreen reporter gene by IN-defective HIV-1–based vector induced significantly fewer fluorescence-positive cells with a lower MFI compared to those induced by the IN-wt vector, even though the total viral DNA levels were comparable. We detected high levels of the posttranslational histone modification H3K9me3, a silencing mark, and low levels of H3ac modification, an activation mark. Both modifications correlate with the observed silencing phenotype of unintegrated HIV-1 DNAs.

We showed by time-course analyses that histone deposition and H3K9me3 modification onto unintegrated HIV-1 DNA appears rapidly after nuclear entry. Notably, the kinetics of loading onto circles, judged by ChIP and subsequently qPCR utilizing 2-LTR circle-specific primers, exhibited a peak at 12 h after infection, while the loading on total DNA, utilizing primers for detecting total virus, showed a more gradual increase. It should be noted that the readout of total viral DNA detects many species, including DNA intermediates in the cytoplasm, DNAs in PICs that have not uncoated, as well as nuclear DNAs. In contrast, the readout of circular DNA only detects nuclear forms that are likely fully uncoated and therefore are more accessible both for the host DNA damage repair machinery and for histone loading. These aspects of the assay likely account for the difference in the timing of loading, as measured by the recovered DNA after ChIP relative to the input DNA.

In all our ChIP experiments, we always performed controls with nonspecific isotype-matched Ig to determine the baseline levels of DNA recovered nonspecifically in the immunoprecipitates. These background levels of DNA are almost always extremely low. We experienced difficulties, however, with high backgrounds throughout our studies when using rabbit IgG antibodies for ChIP and qPCR primers detecting total HIV-1 viral DNA. In some cases, these high backgrounds prevented meaningful readouts of histone loading on total HIV-1 DNAs. We found these high backgrounds only with HIV-1, not MLV; only with total viral DNA and not 2-LTR; and only with rabbit IgG and not mouse IgG. The data suggest that there may be some component of the HIV-1 PICs that binds rabbit IgG nonspecifically. A possible candidate for this is the host protein TRIM21, a cytosolic Fc receptor capable of binding IgGs (47) and known to be specifically packaged into HIV-1 virions (48). Investigators using ChIP to probe HIV-1 PICs should be alerted to this potentially confounding IgG-binding activity. Importantly, the isotype controls for the other target DNAs such as 2-LTR circles, GAPDH, or beta globin DNAs uniformly gave the usual very low, negligible backgrounds.

We were interested in exploring the potential role of the H3.3 variant histone in the chromatinization of the viral DNAs and their transcriptional silencing. Because the complete knockout in the mouse germ line is semilethal (36), we used siRNA-mediated knockdowns to specifically reduce H3.3 expression and monitored the H3.3 loading onto viral DNA. It proved almost impossible to prevent H3.3 loading onto unintegrated HIV-1 DNA, even though the overall levels of the histone variant could be dramatically reduced as assessed by Western blots. The residual pool of H3.3 in KD cells apparently provided adequate supplies for loading on the incoming virus. It should be noted that the quantities of viral DNA are tiny in comparison to the host genomic DNA: at multiplicities of infection of ∼1, they represent roughly one or a few one-hundred thousandths of the host DNA by weight. There was only a very slight reduction in the levels of H3.3 loaded and very little change in the levels of expression from the viral reporter gene. It thus remains unclear if complete loss of H3.3 would have a larger effect on silencing.

The state of condensation of the unintegrated retroviral DNAs may well contribute to their transcriptional silencing. In this regard, the presence of the H1 linker histones on the viral DNA is intriguing. It may be that the H1 organizes the viral chromatin into a highly condensed state and thereby reduces access to the DNA by the transcriptional machinery. Analysis of the density and spacing of nucleosomes on the PICs, as assessed by micrococcal nuclease digestion or assay for transposase-accessible chromatin-sequencing (ATAC-seq) methods, could be informative. Tests for the role of H1 histones in determining such structures might be of great importance. Another important question for the future is the identification of the host machinery involved in histone loading. Manipulation of specific histone chaperones will reveal the major loaders and could determine the importance of histone variants such as H3.3.

To summarize, we described here the chromatinization of unintegrated HIV-1 DNA, its histone composition, and the timing of loading. These events, occurring very early after exposure to virus, can have a major impact on the outcome of the infection and the efficiency of virus transmission. Moreover, the potent silencing of unintegrated HIV-1 DNAs constitutes a major limitation for the use of nonintegrating retroviral vectors for gene therapy applications. Modulation of posttranslational histone modifications through specific inhibitors has been shown to improve gene expression in transient DNA transfection applications (49, 50), and controlling histone-based silencing may similarly improve the utility of nonintegrating viral vectors.

Materials and Methods

Retrovirus Reporter Plasmids.

HIV-1–based reporter virus construct pNL4-3.ZsGreen.R⁻.E⁻, described previously (51), was used throughout the study and is here referred to as IN-wt. We generated the IN-deficient mutant pNL4-3.ZsGreen.R⁻.E⁻ (IND64A), which is here named IN-D64A, by transferring the XhoI–NotI fragment from pNL4-3.Luc.R⁻.E⁻ (IND64A) (17). The mutant harbors a mutation at the catalytic site of the retroviral enzyme integrase and allows nuclear entry but prevents integration (35). The Moloney MLV-based expression constructs pCMV-intron.IND184A (Gag-Pol from NB-tropic MLV) and pNCA-GFP (MLV expressing GFP) were as previously published (14). All reporter virus vectors had a replication-defective single-round design and were pseudotyped with the VSV G envelope encoded by pMD2.G (a gift from Didier Trono, Ecole Polytechnique Federale de Lausanne, Lausanne, Switzerland; Addgene plasmid, 12259), referred to as VSVg here.

Cells and Cultivation.

Human HeLa cell line (American Type Culture Collection, CCL-2) and human Lenti-X 293T cell line (Clontech, 632180), were cultured in Dulbecco’s Modified Eagle Media (DMEM) supplemented with 10% heat-inactivated fetal bovine serum (FBS), 100 U/mL penicillin, and 100 µg/mL streptomycin.

Retroviral Particle Production and Infection.

One day before transfection, 5 × 10⁶ 293T cells per 10-cm dish were seeded. Retroviral particle production was achieved by the calcium phosphate precipitation method using CaCl₂ and 2x Hepes-buffered saline (pH 7.08). Culture medium was supplemented with 10 mM Hepes during particle production. To package HIV-1–based reporter virus, 10 µg IN-wt or IN-D64A plasmid DNA was cotransfected with 2 µg VSVg DNA in a 10-cm dish. To produce Moloney MLV-based reporter virus, 10 µg pNCA.GFP plasmid was transfected along with 15 µg pCMV-intron (encoding NB-tropic MLV Gag-Pol) and 2 µg VSVg per 10-cm dish. Cells were washed ∼5 h after transfection. Viral supernatants were harvested 36 and 48 h after transfection, filtered through 0.45-µm pore-size filters and concentrated (100×) by ultracentrifugation (2 h, 25,000 rpm, 4 °C). Viral pellets were resuspended in culture medium and stored in aliquots at −80 °C until further usage. HeLa cells were infected with viral supernatants and incubated for at least 5 h if not indicated otherwise. The concentrated IN-wt virus preparations typically exhibited titers on HeLa cells of ∼2.5 × 10⁸ transducing units per milliliter. IN-wt and IN-D64A viruses were simultaneously produced, and equal amounts of supernatants were applied to cells within an experiment for comparative studies. We typically utilized a multiplicity of infection of ≤0.5, which leads to ∼30 to 50% fluorescent-positive cells.

Flow Cytometry.

Cells were harvested at respective time points after infection with 0.25% trypsin- ethylenediaminetetraacetic acid (EDTA) and resuspended in phosphate-buffered saline (PBS) supplemented with 3% heat-inactivated FBS. The BD LSRII flow cytometer (BD Biosciences) was used, and analyses were conducted using FlowJo software. Data were first gated for viable cells through forward scatter height and side scatter height. ZsGreen-positive cells were then scored by gating from the previously defined viable population. MFI was determined based on the gated ZsGreen-positive population.

ChIP.

One day before infection, 2 × 10⁶ HeLa cells per 10-cm dish were seeded. Before infection, viral supernatants were pretreated with 5 U/mL DNase I (Promega, M6101) for 1 h at 37 °C to remove any residual plasmid DNA. Media during infection were supplemented with 8 µg/mL Polybrene. Cells were rinsed twice with 1 × PBS and cross-linked with 1% formaldehyde for 10 min at 37 °C, quenched in 0.125 M glycine for 5 min, and lysed in 700 µL of ChIP lysis buffer (50 mM Tris⋅HCl, pH 8.0, 1% sodium dodecyl sulfate, 10 mM EDTA). Cell lysates were sonicated, immunoprecipitated, and reverse–cross-linked. After DNA isolation, qPCR was conducted with ChIP samples. Extended information for ChIP protocol is provided in SI Appendix, Supplementary Materials and Methods.

Detection of 2-LTR Circles and Total Viral DNA.

HeLa cells were pelleted at indicated time points, and DNA was isolated with the QIAamp DNA blood mini kit following the manufacturer’s instructions (Qiagen). Approximately 50 to 100 ng DNA per sample was used for subsequent qPCR analyses. The cycle threshold (Ct) values were normalized to endogenous GAPDH levels based on the 2^ΔCt method. Extended information for qPCR detection of viral DNAs is provided in SI Appendix, Supplementary Materials and Methods.

siRNA Transfection and Western Blot.

The following siRNAs were purchased from Dharmacon with ON-TARGET plus SMART pool design: H3F3A (L-011684-01-0005); H3F3B (L-012051-00-0005); and nontargeting pool (D-001810-10-20). For siRNA transfection, 4 × 10⁵ HeLa cells per 10-cm dish were seeded, and the next day, transfection was performed with 300 pmol siRNA per dish and Lipofectamine RNAiMAX (Life Technologies, 137785000) for at least 5 h and according to the manufacturer’s advice. Transfection was carried out twice on 2 consecutive days, cells were seeded for infection at the end of the day of the second siRNA transfection, and infection was accomplished 24 h later. Cells were harvested 24 h after infection for ChIP or 30 h after infection for flow cytometry analysis, and cells were pelleted for subsequent Western blot after the last harvest of cells to confirm KD during all experimental procedures. Detailed Western blot protocol can be found in SI Appendix, Supplementary Materials and Methods.

Statistical Analysis.

Data were shown as mean ± SD or mean and range as indicated. We used unpaired t test for comparison of 2 groups, and in the case of high variances between the two groups, Welch’s correction was applied. P values ≤ 0.05 were considered significant (*) and ≤ 0.01 very significant (**).