|Home | About | Journals | Submit | Contact Us | Français|
DNA methylation, by regulating the transcription of genes, is a major modifier of the eukaryotic genome. DNA methyltransferases (DNMTs) are responsible for both maintenance and de novo methylation. We have reported that human immunodeficiency virus type 1 (HIV-1) infection increases DNMT1 expression and de novo methylation of genes such as the gamma interferon gene in CD4+ cells. Here, we examined the mechanism(s) by which HIV-1 infection increases the cellular capacity to methylate genes. While the RNAs and proteins of all three DNMTs (1, 3a, and 3b) were detected in Hut 78 lymphoid cells, only the expression of DNMT1 was significantly increased 3 to 5 days postinfection. This increase was observed with either wild-type HIV-1 or an integrase (IN) mutant, which renders HIV replication defective, due to the inability of the provirus to integrate into the host genome. Unintegrated viral DNA is a common feature of many retroviral infections and is thought to play a role in pathogenesis. These results indicate another mechanism by which unintegrated viral DNA affects the host. In addition to the increase in overall genomic methylation, hypermethylation and reduced expression of the p16INK4A gene, one of the most commonly altered genes in human cancer, were seen in cells infected with both wild-type and IN-defective HIV-1. Thus, infection of lymphoid cells with integration-defective HIV-1 can increase the methylation of CpG islands in the promoters of genes such as the p16INK4A gene, silencing their expression.
One function widely ascribed to DNA methylation is that of a genome defense mechanism against foreign invaders. Retroviruses integrate randomly in the host genome, where they are susceptible to silencing through methylation, which depends on the local DNA environment (6, 26, 33, 87). However, methylation of proviral DNA, which retains the ability to be reactivated, leading to productive infection, allows the virus to survive by escaping immune surveillance. DNA methylation is a major modifier of the eukaryotic genome (8, 85). In mammals, DNA methylation is essential for normal embryonic development, as it plays an important role in the regulation of gene expression, X chromosome inactivation, genomic imprinting, chromatin modification, and silencing of endogenous retroviruses (2, 42, 58, 77). Exactly how DNA methylation recognizes methylated and nonmethylated sequences, transcriptionally represses genes, and stably maintains these patterns during cell replication is not clear. The packaging of the DNA template within chromatin largely controls the transcriptional activity of a gene (85). In 1998, two groups (37, 57) showed that a methyl-binding protein (MeCP2) forms a complex with histone deacetylase (HDAC) and a transcriptional repressor (Sin-3). This linked DNA methylation with histone deacetylation, a universal mechanism for gene silencing (34). To achieve stable repression, the chromatin around the inactive gene becomes more densely methylated and more condensed through histone deacetylation, which can be involved in oncogenic transformation and other pathogenic states (36).
Three enzymes responsible for DNA methylation, known as DNA methyltransferases (DNMTs), have been identified. The carboxy-terminal domain of DNMT1 catalyzes the methylation of DNA containing hemimethylated CpG dinucleotides more efficiently than that of unmethylated DNA in vitro (6). DNMT1 is the enzyme mainly responsible for maintaining DNA methylation patterns in adult mammalian tissues and also participates in de novo methylation on C-type CpG islands in human carcinogenesis (4, 38). It has recently been shown that the noncatalytic N-terminal domain of DNMT1 can act as a transcriptional repressor binding directly to HDAC2 and a new corepressor, DMAP1, to form a complex at the replication foci (73). DNMT1 also forms a complex with Rb, E2F1, and HDAC1 to repress transcription of E2F-responsive promoters (71), suggesting that DNMT1 has activities other than the enzymatic function of a methyltransferase. Two other forms of DNMTs have been isolated in mammals (62, 84). The recently identified DNMT3a and DNMT3b are essential for de novo methylation and for mouse development (64).
During retroviral infection, methylation is increased throughout the viral genome, particularly the viral long terminal repeats (LTR), suggesting that methylation can be a mechanism of suppression of viral expression and latency for human immunodeficiency virus type 1 (HIV-1) and human T-cell leukemia virus type 1 (HTLV-1) (20, 39, 51, 74, 75). Previously, we showed that acute infection of cells with HIV-1 results in an increase in DNMT1 expression and activity, an overall increase in methylated genomic DNA in infected cells, and the de novo methylation of a single CpG dinucleotide in the gamma interferon (IFN-γ) gene promoter, which subsequently down-regulated the expression of interferon (53). However, little is known about the mechanism(s) of the methylation changes seen after HIV-1 infection. As unintegrated circular retroviral DNA has been implicated in the pathogenesis of several retroviral infections, we asked whether replication of HIV-1 is necessary for the increases in DNMT activity and the increased capacity of infected cells to methylate genes de novo. Here, we report that integration-defective HIV-1 could also induce increased DNMT1 expression and activity, resulting in hypermethylation and reduced expression of the tumor suppressor gene p16INK4A following acute HIV-1 infection in lymphoid cell lines. This study further implicates aberrant methylation as a mechanism of pathogenesis during HIV-1 infection and suggests the methylation machinery as a novel target for AIDS therapy.
Hut 78, a mature CD4+ T-cell line derived from a T-cell lymphoma, was cultured at 37°C and 5% CO2, in RPMI 1640. Dulbecco's modified Eagle Medium (DMEM) was used to culture 293 T cells. Media (Biowhittaker, Walkersville, Md.) were supplemented with 10% heat-inactivated fetal calf serum (HyClone, Logan, Utah), 300 μg of l-glutamine/ml, 100 μg of penicillin/ml, and 100 μg of streptomycin/ml.
All viral stocks were generated by transient transfection of 293 T cells using a calcium phosphate transfection system (Life Technologies, Inc., Gaithersburg, Md.). Cells were plated in 100-mm tissue culture dishes 24 h before transfection. Cells were refed fresh, complete DMEM 3 h before transfection. The precipitate was kept on cells for 24 h prior to replacement with fresh, complete medium and culture for an additional 24 h. Supernatants were filtered through a 0.45-μm-pore-size filter, which collected all of the medium cell-free. The amount of viral p24 antigen was determined using an enzyme-linked immunosorbent assay kit (Cellular Products, Buffalo, N.Y.) with a sensitivity of 10 pg/ml. The strains used included WT NL4–3 (a full-length molecular clone of HIV-1 ), D116N (containing a mutation in the catalytic core domain; kindly provided by Alan Engelmann, Dana-Farber Cancer Institute, Boston, Mass. [18, 19]), and a reverse transcription (RT)-negative mutant (kindly provided by Robert Gorelick, AIDS Vaccine Program, SAIC—Frederick, Frederick, Md. ).
Hut 78 cells were infected at a concentration of 107 cells in 1 ml of total volume in a 50-ml conical centrifuge tube to which 10 to 30 ng of p24 antigen from mutant or wild-type HIV-1 was added. After incubation in a shaking water bath for 2 h at 37°C, cells were washed twice, resuspended in 30 ml of RPMI complete medium in a T75 flask, and incubated at 37°C. At various times postinfection (p.i.), DNA, RNA, and nuclear proteins were isolated. In some experiments, Hut 78 cells were pretreated overnight with the hypomethylating agent 5-azacytidine (5-AzaC) before infection. Cells (5 × 106) were seeded in a 100-mm dish 12 to 24 h before treatment and were then exposed to 1 to 10 μM 5-AzaC or 5-deoxy-AzaC for 24 h. Control cultures (mock treated) were treated with the same volume of phosphate-buffered saline (PBS). Twenty-four hours after addition of 5-AzaC, the culture medium was replaced with drug-free medium, and the cells were infected using D116N or WT NL4–3.
DNA was isolated according to the Hirt method (30). Briefly, cells were first washed in PBS without Ca2+ and Mg2+ and then centrifuged at 750 × g; the pellet was resuspended in 470 μl of 10 mM EDTA, pH 7.5, and transferred to a 1.5-ml Eppendorf style centrifuge tube. Thirty microliters of 10% (wt/vol) sodium dodecyl sulfate was added to this pellet and mixed by gentle inversion of the tube to prevent shearing of chromosomal DNA. Following a 20-min incubation at room temperature, 125 μl of 5 M NaCl was added and mixed again by gentle inversion. After incubation overnight at 4°C, the solution was centrifuged at 17,000 × g for 30 min at 4°C. The supernatant was phenol-chloroform extracted at least three times until no interface was observed. The aqueous layer was precipitated in 70% ethanol and resuspended in 25 to 50 μl of Tris-EDTA (pH 8.0). The viral DNA synthesis status was analyzed by PCR using primers which amplify sequences from the 2-long terminal repeat (2-LTR)-containing circles formed in the nuclei of infected cells. The primer sequences were 5′-GAG ATC CCT CAG ACC CTT TTA G-3′ (sense) and 5′-GTC AGT GGA TAT CTG ATC CCT G-3′ (antisense). Reaction components were mixed at room temperature and heated to 94°C prior to addition of Taq polymerase. We performed 34 cycles of denaturation at 94°C for 45 s, annealing at 60°C for 30 s, and polymerization at 72°C for 2 min, and 1 cycle at 72°C for 5 min. We then electrophoresed 20 μl of each reaction mixture in 2.0% agarose gels (Novex, San Diego, Calif.). Gels were stained with 0.5 μg of ethidium bromide per ml to visualize the DNA. The oligonucleotide primer pair M667 (sense)–AA55 (antisense) was used to determine the 5′ R-U5 region of the LTR in HIV-1 DNA (88). The sequence of M667 was 5′-GGC TAA CTA GGG AAC CCA CTG-3′, and the sequence of AA55 was 5′-CTG CTA GAG ATT TTC CAC ACT GAC-3′. The PCR program was 91°C for 1 min, 65°C for 2 min, 72°C for 1 min, and one cycle at 72°C for 5 min.
Cells were cultured at 106/ml for 6 h in 10 μg of brefeldin A (Sigma, St. Louis, Mo.)/ml. After three washes in PBS, the cells were resuspended in 1 ml of freshly prepared 2% paraformaldehyde in PBS (pH 7.2) and incubated for 2 h at 4°C. The cells were then washed three times in PBS, resuspended in 1 ml of PBS containing 0.1% saponin (Sigma), and incubated for 10 to 30 min at 4°C for efficient permeabilization, which was tested using an anti-actin antibody (Sigma). A phycoerythrin-labeled anti-HIV core antibody (KC-57; Beckman-Coulter, Brea, Calif.) was added, and the cells were incubated for 45 min at 4°C in the dark. After being washed, the cells were left undisturbed for 10 min, resuspended, and immediately analyzed by flow cytometry using a FASCAN (Becton Dickinson, Mountain View, Calif.). Data were analyzed using Flow Jo software (Tri Star, Inc., San Carlos, Calif.).
Total cellular RNA from Hut 78 cells infected with HIV-1 or mock infected was extracted and purified with TRIzol Reagent (Life Technologies). RNA was resuspended in diethyl pyrocarbonate-treated water and quantitated by the optical density at 260 or 280 nm. An agarose gel using ethidium staining also verified the quantitation. The samples were treated with DNase I (Roche, Indianapolis, Ind.). RT reactions using 2.5 μg of total RNA in a total reaction volume of 20 μl were performed using Superscript II reverse transcriptase (Life Technologies). PCR mixtures containing 2.5 mM MgCl2, 0.2 mM each deoxynucleoside triphosphate (Boehringer Mannheim), 1 μM each primer, 2 μl of cDNA from the RT reaction, and 2.5 U of Taq DNA polymerase (Sigma) in a volume of 50 μl were amplified using the following PCR conditions taken from the work of Robertson et al. (DNMT3a and β-actin) (69) and Mizuno et al. (DNMT1 and DNMT3b) (54). For DNMT3a and β-actin, 35 and 25 cycles, respectively, of 94°C for 2 min, 94°C for 0.5 min, the transcript-specific annealing temperature (65°C for DNMT3a and 60°C for β-actin) for 1 min, and 72°C for 1 min (with one 322-bp fragment of β-actin cDNA used as a control) were carried out. For DNMT1 and DNMT3b, the PCR programs were 94°C for 30 s, 55°C for 30 s, and 72°C for 1 min. Primer sequences (Life Technologies) used were as follows: for DNMT1 (GenBank accession number XM017218), 5′-ACC GCT TCT ACT TCC TCG AGG CCT A-3′ (sense) and 5′-CCA CAG TGT TCA CAG AGG ACT GCA AC-3′ (antisense); for DNMT3a (GenBank accession number NM022552), 5′-GGG GAC GTC CGC AGC GTC ACA C-3′ (sense) and 5′-CAG GGT TGG ACT CGA GAA ATC GC-3′ (antisense); for DNMT3b (GenBank accession number XM009449), 5′-AAT GTG AAT CCA GCC AGG AAA GGC-3′ (sense) and 5′-ACT GGA TTA CAC TCC AGG AAC CGT-3′ (antisense); and for β-actin (GenBank accession number XM004814), 5′-GGA GTC CTG TGG CAT CCA CG-3′ (sense) and 5′-CTA GAA GCA TTT GCG GTG GA-3′ (antisense). The density of each band obtained by RT-PCR in each lane was normalized to the amount of total RNA as determined by the density of the band obtained by RT-PCR for β-actin; i.e., if the β-actin control value was 30,000 U (pixels of brightness), then the calculation used to normalize DNMTs to β-actin can be expressed as [30,000/(density of β-actin)] × (density of DNMTs). The RT-PCR analysis was done at least three times.
Quantitative RT-PCR (QRT-PCR) was performed on a Light Cycler (Roche Molecular Systems) by using syber green, which fluoresces upon binding to double-stranded DNA, according to the manufacturer's instructions, and results were normalized to the β-actin control as described above. To discriminate between specific products and nonspecific products such as primer dimers, DNA melting curves were generated (68). Fluorescence data were converted to melting peaks using Roche data analysis software, then plotted as the negative derivative of fluorescence with respect to temperature (−dF/dT versus T, where F is fluorescence and T is temperature). The area of the specific melting peak is directly proportional to the amount of intended product (65, 68). The area under the melting peak was determined using Gaussian Fit software (Roche Molecular Systems).
To examine DNMT protein levels following acute HIV-1 infection, Western blotting was carried out. Nuclear extracts were prepared from both infected and uninfected Hut 78 cells by washing cells with PBS and lysing in buffer A (10 mM HEPES [pH 7.4], 10 mM KCl, 1.5 mM MgCl2, 0.5 mM dithiothreitol [DTT], 0.2 mM phenylmethylsulfonyl fluoride [PMSF], 1 μg of protease inhibitors/ml, 0.025% NP-40) for 15 min with rotation at 4°C. The nuclear pellet was resuspended in buffer B (20 mM HEPES [pH 7.4], 420 mM NaCl, 1.5 mM MgCl2, 0.2 mM EDTA, 25% glycerol, 0.5 mM DTT, 0.2 mM PMSF, 1 μg of protease inhibitors/ml) for 30 min. The soluble nuclear protein was collected by centrifugation. A 200-μg portion of each nuclear extract was boiled in loading buffer for 5 min and then loaded onto a sodium dodecyl sulfate–14% polyacryamide gel. After electrophoresis, proteins were electroeluted onto a polyvinylidene difluoride membrane. A rabbit polyclonal antibody against DNMT1 (diluted 1:500) was purchased from New England Biolabs (Newton, Mass.). The rabbit polyclonal antibody against DNMT3a was a gift from Keith Robertson (National Cancer Institute [NCI], Bethesda, Md.). A mouse monoclonal antibody against DNMT3b (diluted 1:200) was obtained from Alexis Biochemicals (San Diego, Calif.). An antibody against β-actin (Sigma) was used as a control for protein input.
A modified methyl-accepting assay (86) was used to determine the methylation status of DNA isolated from infected and uninfected Hut 78 cells. DNA (200 ng) was incubated with 4 U of SssI CpG methylase (New England Biolabs) in the presence of 1.5 μM S-adenosyl-l-[methyl-3H]methionine (60 to 85 Ci/mmol; TRK 581; Amersham) and 1.5 μM nonradioactive S-adenosylmethionine (SAM) (New England Biolabs). The reaction mixtures (20 μl) were incubated at 37°C for 4 h in a buffer containing 10 mM Tris-HCl (pH 7.9), 50 mM NaCl, 10 mM MgCl2, and 1 mM DTT. The reactions were stopped by addition of 5 μl of 2.5 mM nonradioactive SAM, and reaction products were spotted on 2.4-cm2 Whatman GF/C filter disks, which were air dried for 15 min and washed with 6 ml of 5% (wt/vol) trichloroacetic acid and 70% (vol/vol) ethanol. Disks were counted in Econofluor in a Beckman liquid scintillation counter. Control reactions without DNA or enzyme added were included as background, and these results never exceeded 5% of those in the test samples. All samples were done in triplicate, and values were obtained as disintegrations per minute per nanogram of DNA.
We used the bisulfite treatment method of Clark et al. (15) with some modifications as follows. Two micrograms of total genomic DNA (from at least two independent infections corresponding to RT-PCR experiments) was isolated with the QIAamp DNA Blood Mini Kit (Qiagen Inc.) and then denatured with NaOH and modified with a freshly prepared sodium bisulfite solution (2.35 M) containing hydroquinone (0.04 M). The bisulfite-treated DNA was desalted using the Wizard DNA Clean Up kit (Promega). To amplify the p16INK4A promoter, we used a 0.1-μg aliquot of the converted DNA. Methylation of the 5′ CpG island in the p16INK4A gene was also determined in samples from Hut 78 cells infected with wild-type or mutant HIV-1, as well as those treated with 5-AzaC. Bisulfite-treated DNA was amplified by PCR using primers specific (GenBank accession number X94154) for the methylated (sense, 5′-TTA TTA GAG GGT GGG GCG GAT CGC-3′; antisense, 5′-GAC CCC GAA CCG CGA CCG TAA-3′) or unmethylated (sense, 5′-TTA TTA GAG GGT GGG GTG GAT TGT-3′; antisense, 5′-CAA CCC CAA ACC ACA ACC ATA A-3′ [28, 29]) CpG island. The thermocycler program was 95°C for 5 min, 5 cycles of 95°C for 1 min, 65°C (methylated) or 60°C (unmethylated) for 2 min, and 72°C for 3 min, and then 35 cycles of 95°C for 30 s, 65 or 60°C for 30 s, and 72°C for 30 s in a 50-μl volume containing 100 ng of bisulfite-treated DNA, 0.1 mM deoxynucleoside triphosphates, 2.0 mM MgCl2, and 0.5 μM primers. The PCR product was directly loaded onto 3% agarose gels and electrophoresed. The gel was stained with ethidium bromide and directly visualized under UV illumination. Oligonucleotide primers for a stretch of the MYOD1 gene completely devoid of CpG dinucleotides were used for a control reaction for equal loading and amplification of bisulfite-treated DNA (17). The sense primer was 5′-CCA ACT CCA AAT CCC CTC TCT AT-3′, and the antisense primer was 5′-TGA TTA ATT TAG ATT GGG TTT AGA GAA GGA-3′. The PCR program for MYOD1 was 30 s at 94 °C, 1 min at 55°C, and 72°C for 1 min. Futhermore, wild-type p16INK4A primers were used to ensure that complete conversion of DNA was obtained in the bisulfite reaction. A positive control for complete methylation was also amplified. In this control, DNA isolated from Hut 78 cells was treated with SssI methylase (New England Biolabs) prior to bisulfite treatment. This enzyme methylates all CpG dinucleotides.
mRNA expression of the p16INK4A gene was determined by RT-PCR of RNA (using the same samples with which the RT-PCR analysis of DNMTs was done) from Hut 78 cells infected with D116N or WT NL4–3 with or without 5-AzaC treatment. The sequences (56) of the primers (GenBank accession number L27211) used were 5′-CCC GCT TTC GTA GTT TTC AT-3′ (sense) and 5′-TTA TTT GAG CTT TGG TTC TG-3′ (antisense). In PCRs, the concentration of each primer was 0.5 μM in 50 μl. PCR amplification was run for 35 cycles, with each amplification cycle consisting of 94°C for 1 min, 58°C for 1 min, and 72°C for 1 min. The size of the PCR product was 355 bp. The primer sequence and PCR program for β-actin cDNA amplification were as described for the DNMT RT-PCR. The PCR product was visualized on 2% agarose gels.
We previously reported that acute HIV infection of primary T cells and lymphoid cell lines results in up-regulation of the expression and activity of DNMT1 and that, consequently, infected cells have an increased capacity to methylate genes (53). In considering mechanisms for this increased DNMT1 expression, the presence of large amounts of unintegrated viral DNA during HIV infection (5, 50, 78) suggested that viral integration was not needed. Therefore, DNMT expression was examined in lymphoid cells at various time points following acute infection with the wild type HIV-1 strain NL4–3 (1), a mutant incapable of integration (D116N [18, 19]), and a mutant incapable of RT (RT negative [23, 24]). Since these mutant proviral clones produced virus particles at lower levels than the wild type when transfected into 293 T cells, in our studies, infections were normalized to the p24 core antigen (CA) content of the virus stock and carried out at a multiplicity of infection which would limit the spread of the wild-type virus. The presence of unintegrated 2-LTR circles demonstrates the nuclear presence of viral intermediates. D116N and WT NL4–3 showed similar levels of 2-LTR circles, while the RT-negative mutant showed no 2-LTR circle formation (Fig. (Fig.1).1). To compare the number of proviral DNA molecules in infected-cell populations, we performed PCR with primers specific for HIV-1 LTR. To estimate viral DNA amounts in this assay, we used DNA isolated from the ACH2 cell line as the positive control (Fig. (Fig.2,2, lane 5). This cell line contains 1 to 2 copies per cell. Similar levels of viral DNA were seen for the D116N and WT NL4–3 viruses, while the RT-negative mutant was essentially negative for viral DNA formation (Fig. (Fig.2).2). β-Actin was used to ensure equal loading. Despite the fact that the D116N mutant completely abolished detectable integrase (IN) activity (reference 18; also data not shown), unintegrated D116N DNA scored 30% positive in MAGI infectivity assays. The lack of integration is further supported by the absence of viral production as determined by cocultivation in Hut 78 cells (data not shown). Furthermore, we examined viral production at the single-cell level. In cultures infected with WT NL4–3, 43 to 52% of the cells had detectable intracellular HIV cores by day 5, while in cultures infected with D116N, no positive cells could be detected.
Next, the expression of DNMT1 and the recently discovered DNMT3a and DNMT3b (63) was examined in the human T-cell line Hut 78 at various time points following infection with wild-type and mutant viruses using RT-PCR, QRT-PCR, and Western analysis. Both cells infected with WT NL43 and cells infected with D116N showed increased levels of DNMT1 RNA (Fig. (Fig.3A),3A), while DNMT3a and -3b showed little or no significant increase in expression in infected cells. The increased expression of DNMT1 was also observed at the protein level, as Western analysis showed increased DNMT1 protein levels in infected cells. (Fig. (Fig.3B,3B, lanes 2 and 4). Western blotting for β-actin demonstrates that equivalent amounts of nuclear extracts were used in this experiment. Consistent with the RT-PCR results, no significant difference between DNMT3a and DNMT3b protein levels was seen after HIV-1 infection (data not shown).
QRT-PCR has the advantage that PCR amplification and product analysis can occur simultaneously, conferring a higher level of specificity on quantitation. To discriminate between specific products and nonspecific products such as primer dimers, DNA melting curves were generated. Fluorescence data were converted to melting peaks using the manufacturer's software, then plotted as the negative derivative of fluorescence with respect to temperature (−dF/dT versus T) (65, 68). The area of the specific melting peak is directly proportional to the amount of specific product (Table (Table1).1). Using QRT-PCR to quantify the changes in DNMT expression that we observed by semiquantitative RT-PCR and Western analysis, we showed that by day 5 p.i., the amounts of β-actin RNA present were remarkably similar in control and virus-infected cultures. Both DNMT3a (Table (Table1)1) and DNMT3b (data not shown) had less RNA in infected than in control cultures. In contrast, by day 5, DNMT1 RNA amounts were 22% greater in D116N-infected cultures and 44% greater in WT NL4–3-infected cultures (Table (Table1).1).
To determine the functional consequences of increased DNMT1 expression in the lymphoid cell line, we examined the overall methylation status of genomic DNA using a methyl acceptor assay as described in Materials and Methods. The assay takes advantage of the ability of bacterial SssI methylase to methylate all unmethylated CpG dinucleotides. By using radiolabeled SAM as a substrate, a quantitative measure of overall genomic methylation was obtained. If the DNA is more methylated in infected cell lines, then less SAM will be incorporated. Data were expressed relative to SAM uptake in uninfected Hut 78 cells. As shown in Table Table2,2, the uptake of radiolabeled SAM in WT NL-43-infected cells was 55% of that for the control at day 3 and 80% at day 7, demonstrating significantly more methylation of CpG dinucleotides in the genomic DNA of infected cells than in that of uninfected cells. Viral infection of Hut 78 cells with the D116N mutant showed smaller but still significant increases in overall genomic methylation (20 to 25% [Table 2]), while infection with the RT-negative virus showed no significant change. These results correlated with the increases seen in DNMT expression. Since knowledge of the genes that either are regulated by methylation or contain CpG islands in their promoters has been steadily increasing, particularly with sequencing of the human genome and array technology using CpG islands (32), it seems likely that numerous genes that have altered methylation status following HIV-1 infection will be identified.
The cyclin-dependent kinase inhibitor p16 normally inhibits the phosphorylation of RB by cyclin D and cyclin-dependent kinases 4 and 6. This gene is frequently altered in neoplasia, including hematological malignancies, which often result from homozygous deletion or promoter region hypermethylation (29). Since hypermethylation of p16INK4A has recently been demonstrated in HTLV-1-infected cell lines (79) and since p16INK4A is frequently methylated in non-Hodgkin's lymphoma (29), commonly seen in AIDS patients, we examined the methylation status of p16INK4A following acute infection with wild-type HIV-1 and mutant viruses using MSP-PCR (28). Bisulfite treatment converts the cytosine residues in the genomic DNA to uracil, which are amplified as thymine during subsequent PCR.
As shown in Fig. Fig.4,4, Hut 78 cells showed positive 150- and 151-bp bands for methylated and unmethylated specific primer sets for p16INK4A, respectively, indicating that the p16INK4A gene is partially methylated in this cell line. The methylated bands for the p16INK4A gene in D116N- and WT NL4–3-infected Hut 78 cells were consistently stronger than the products of uninfected Hut 78 cells or of Hut 78 cells infected with RT-negative mutant virus (Fig. (Fig.44 and Table Table3).3). As a control for equal loading, a sequence from the MYOD1 gene lacking methylatable CpG dinucleotides (17) was used for normalization instead of β-actin as described in Materials and Methods. Thus, unmethylated product levels were significantly lower, and methylated product levels were correspondingly higher, in HIV-infected cells. Further, bisulfite genomic sequencing of eight CpG islands in the p16INK4A gene promoter revealed 80 to 90% methylation in Hut 78 cells infected with WT NL4–3 or D116N. Partial (30 to 40%) but not complete methylation of the p16INK4A gene promoter was detected in uninfected Hut 78 cells and those infected with the RT-negative mutant (data not shown).
These results suggest that the p16INK4A gene is a target of the increased DNMT activity in HIV-1-infected Hut 78 cells. To further examine this, we decreased methylation of the p16INK4A gene using 5-AzaC. Three days after treatment of uninfected Hut 78 cells with 1 μM 5-AzaC, MSP-PCR revealed a significant increase in the amount of unmethylated product (Fig. (Fig.4,4, lane 9), while a significantly more intense methylated 150-bp band was seen for DNA treated with 5-AzaC and infected with either D116N or WT NL4–3 (Fig. (Fig.4,4, lanes 10 and 11). By day 5, there was partial remethylation of this gene (Fig. (Fig.4,4, lane 12) that was markedly accelerated by HIV-1 infection (Fig. (Fig.4,4, lanes 13 and 14).
To correlate increased methylation of the p16INK4A gene promoter with expression of the gene, we examined the expression of p16INK4A RNA in Hut 78 cells, using semiquantitative RT-PCR (Fig. (Fig.5;5; Table Table4).4). Decreased levels of p16INK4A expression were seen in Hut 78 cells infected with D116N or WT NL4–3 (Fig. (Fig.5,5, lanes 2 and 4; Table Table4)4) but not in cells infected with the RT-negative mutant (Fig. (Fig.5,5, lane 3; Table Table4).4). Hut 78 cells treated with 5-AzaC had a four- to fivefold increase in p16INK4A expression (Fig. (Fig.5,5, lanes 9 and 12; Table Table4).4). Furthermore, infection of Hut 78 cells previously treated with 5-AzaC for 24 h, using either D116N or WT NL4–3, markedly decreased expression (50 and 90%, respectively) at 5 days p.i. (Fig. (Fig.5,5, lanes 12 to 14; Table Table4).4). In addition, we found that p16INK4A mRNA levels were decreased in Hut 78 cells chronically infected with HIV-1 strain 3B, and hypermethylation in the p16INK4A promoter was present (data not shown). These data suggest that methylation of the p16INK4A gene is one of the mechanisms for silencing of p16INK4A expression in Hut 78 cells and that HIV-1 infection can modulate the methylation status of the CpG island in the promoter of p16INK4A, leading to decreased transcription.
Insertion of foreign invaders into the eukaryotic genome can occur naturally, e.g., during viral infections or under experimental conditions such as microinjection or transfection of DNA for immunization purposes. It has long been suggested that de novo methylation is one mechanism by which the cell or genome is protected from expression of foreign DNA such as viruses (7, 33). Retroviruses, which are essentially movable genetic elements, integrate randomly in the host genome and, depending on the local DNA environment, are susceptible to DNA methylation. A relationship among DNA methylation, retroviral replication, and pathophysiology was first shown for murine leukemia virus (MuLV) (26). The MuLV LTRs became hypermethylated, silencing viral expression that could be reactivated to produce active virus by several means. Methylation has been shown to be a mechanism of suppression of viral expression and latency for both disease-causing human retroviruses, HIV-1 and HTLV-1 (52, 53, 74, 75). The ability of the provirus to become latent through methylation and escape the immune response is a two-edged sword for the host that is clinically relevant to human disease and therapy, e.g., HIV latency after highly active antiretroviral therapy (14, 20). It is possible that primary retroviral virulence (direct viral replicative pathogenesis) is inversely related to the number of proviral CpG dinucleotides available for nuclear methylation and silencing by the vertebrate host. If true, the most pathogenic retroviruses would be those with the lowest frequency of CpGs (39). Interestingly, as in the human genome, CpG dinucleotides are underrepresented in retroviral genomes. To counteract the host defense system of methylation and silencing, HIV-1 evolved or is evolving to decrease the number of methylatable CpGs in its genome, thereby escaping the host's capacity to inactivate viral DNA through methylation. The methylatable-CpG content in the HIV-1 genome is lower than that encountered in the many viruses elsewhere in the animal kingdom (61).
In addition to de novo methylation of foreign viral DNA, the methylation patterns of cellular genes and DNA structures can be profoundly altered (27, 65) by viral infection. These virus-induced changes are likely a reflection of more general alterations in chromatin structure. Compelling evidence for the role of methylation in chromatin structure and vice versa has been published (42, 48, 67). Exactly how DNA methylation silences gene expression has been further elucidated by the observation that methyl binding proteins form complexes with other proteins, such as HDAC, that affect chromatin structure and gene regulation, and complexes such as Sin-3, which are transcriptional repressors (37, 57, 58, 82, 83). There seem to be several levels of stable gene silencing. The chromatin of an inactive gene can be densely methylated by de novo methylation, and additional chromatin condensation can occur through histone deacetylation (46).
Previous studies from our laboratory showed that acute infection of cells with HIV-1 results in increased DNMT1 expression and activity, an overall increase in methylation DNA in infected cells, and de novo methylation of a CpG dinucleotide in the IFN-γ gene promoter, resulting in the subsequent down-regulation of expression of this cytokine (53). However, little is known about the mechanism(s) of increased methylation and DNMT1 activity during HIV infection. DNMT1 is the chief enzyme responsible for maintaining methylation patterns in adult mammalian cells. Disruption of the Dnmt1 gene results in embryonic lethality (45). DNMT1 is a large enzyme (193.5 kDa) composed of a C-terminal catalytic domain, which transfers methyl groups from SAM to cytosines in CpG nucleotides, and a large N-terminal regulatory domain with several functions, including targeting to replication foci (13, 44). Forced overexpression of DNMT1 or cleavage between the N-terminal regulatory domain and the C-terminal catalytic domain has been shown to result in increased de novo methylation activity (6, 80) and cellular transformation (86). Other forms of DNMTs have been isolated in mammals (62, 84), and two recently identified DNA methyltransferases, DNMT3a and DNMT3b, are thought to be essential for de novo methylation (31, 64, 69). Moreover, it has recently been shown that in addition to its capacity to methylate CpG sites, DNMT1 can play other roles in transcriptional regulation. DNMT1 can bind HDAC2 and novel corepressors to form a complex at replication foci (73). DNMT1 also forms a complex with Rb, E2F1, and HDAC1, repressing transcription from E2F-responsive promoters (71).
Is HIV-1 replication necessary for increases in DNMTs and genomic methylation? The integration process is the keystone of retroviral replication. Once integrated into the chromosome, the provirus will remain stable throughout the life span of the target cell (12). Integration of retroviral DNA into the host cell genome is required for virus replication and is mediated by viral IN (11). IN function is essential for HIV-1 replication in T-cell lines (9, 43, 76). Mutational analyses of HIV-1 IN indicate that the protein consists of three functional domains: the N-terminal, core (catalytic domain), and C-terminal domains (18, 19). A mutation in the catalytic domain (present in the D116N mutant used in this study) completely abolishes 3′ processing, DNA strand transfer, and disintegration in vitro (18). Although the D116N mutant shows a significant titer in a CD4+ indicator cell assay, it is clearly integration and replication defective (19). As reported for other tissues (69), DNMT1 was expressed at the highest level of the three DNMTs in lymphoid cells. QRT-PCR and Western analysis also showed that acute infection of Hut 78 cells with wild-type or integration-negative HIV-1 markedly up-regulates DNMT1 mRNA and protein expression, respectively. The RT-negative mutant showed no significant ability to regulate DNMT1 expression and no detectable effect on methylation. In contrast to DNMT1, DNMT3a and -3b showed no significant increase in expression following acute HIV-1 infection.
In addition to our previous demonstration that HIV infection can result in the aberrant methylation of single CpG dinucleotides in the IFN-γ promoter (53), we show here that genes which have CpG-rich regions of 1 kb of DNA, termed “CpG islands,” and are usually hypomethylated can be aberrantly methylated in HIV-1-infected cells. In malignant cells, these CpG island regions become methylated and expression of the associated gene is silenced (16, 35). The p16INK4A gene, which encodes a specific inhibitor of cyclin-dependent kinases 4 and 6, is located at the 9p2/1 chromosomal region. Loss of p16INK4A gene expression is a frequent molecular alteration involved in tumorigenesis. Recently, changes of expression and methylation status of the p16INK4A gene have been demonstrated in tumor cell lines and a variety of cancers (3, 21, 22, 25, 40, 49, 54, 60, 81). The hypothesis that expression of the p16INK4A gene may be regulated in part by changes in the methylation status of this CpG island has been substantiated in several tumor models (21, 47, 59). Partial or complete promoter methylation rather than deletion of the p16INK4A gene has been observed in some HTLV-1-infected T-cell lines (79) and here in Hut 78 cells. However, by using completely methylated and unmethylated controls, increased methylation of the p16INK4A gene in HIV-1-infected cells was shown. The p16INK4A gene is frequently methylated in non-Hodgkin's leukemia, a malignancy common in AIDS patients (29), suggesting that HIV-induced aberrant methylation could play a role in disease development.
What is the mechanism of increased DNMT1 expression and hypermethylation which results from infection with integration-defective HIV-1? Infection with the HIV-1 mutant D116N results in accumulation of unintegrated 2-LTR circles in the nucleus, a sensitive indicator of a recent infection. In contrast to wild-type HIV-1, infection with D116N resulted in more 2-LTR circle formation in Hut 78 cells, as previously reported (18, 19). This increased DNMT1 expression is not caused by the mere presence of foreign DNA (RT-negative HIV had no effect). Unintegrated DNA can serve as a template for HIV tat expression (18, 19), a transactivator of many genes, which could stimulate DNMT1 expression in trans. Moreover, it has been shown that 2-LTR circles produce large amounts of tat, which stimulates the β-galactosidase readout in the MAGI assay and leads to p24 production (10). On the other hand, since the N terminus of DNMT1 has recently been shown to form complexes with HDAC, transcription factors, and corepressors to silence transcription of specific genes (70, 71, 73), the effect of HIV-1 on DNMT1 may be due to an as yet unknown effect of viral proteins on a non-DNA methyltransferase function of DNMT1. For example viral proteins, such as tat, could interrupt the normal complex formation. We are currently studying this and other hypotheses to determine how HIV infection might alter these DNMT1 functions.
As nonintegrated circular DNA has long been implicated in the pathogenesis of retroviral infections, the ability to aberrantly methylate genes and/or alter chromatin structure could play a role in such pathogenesis. Examples of such pathogenesis are the association of cell killing with nonintegrated spleen necrosis virus (41), disease-specific production of unintegrated feline leukemia virus DNA in feline AIDS (55), and avian leukosis-virus induced-osteoporosis (72). In AIDS, circular forms of unintegrated HIV have been associated with dementia and giant cell formation (78) and may play a role in neuropathogenesis. Although unintegrated viral DNA has been linked to cell killing during HIV infection, it is not always associated with cytopathology (5, 50). Regardless of the mechanism, this study shows that integration-defective HIV-1 can alter the DNA methylation patterns of infected cells, further implicating aberrant methylation as a mechanism of pathogenesis in AIDS and AIDS-associated malignancies and suggesting that the methylation machinery can be a novel target for AIDS therapy.
J.-Y.F. and J.A.M. contributed equally to this report.
We thank Weimin Zhu, Paula Roberts, Angela Brennan, and Dorjbal Dorjsuren for technical assistance. We also thank Robert Gorelick for providing HIV-1 wild-type and RT mutant plasmids and for useful discussions and Howard Young for review of the manuscript.
The NIH Intramural AIDS Targeted Antiviral Program provided support for this study. Portions of this work were also supported by funds from the NCI under contract NO1-CO-56000.