Primitive hematopoietic cells resist HIV-1 infection via p21^{Waf1/Cip1/Sdi1}

p21-restricted HIV-1 replication in primitive hematopoietic cells.

HSCs and progenitor cells are CD34⁺ cells and represent only 0.5–3% of mononucleosis cells in human bone marrow. To assess the role of p21 in HIV-1 replication in these cells, we first knocked down p21 expression with 2 types of siRNA, an in vitro–synthesized siRNA and an in vivo plasmid–transcribed short hairpin RNA (shRNA) (15). Both forms of siRNA suppressed p21 mRNA expression in cells in the range of 62%–98% (Figure 1A), as measured by real-time RT-PCR, which detected as few as 10 copies of target mRNA in a sample of 50 ng total cellular RNA (Supplemental Figure 1; supplemental material available online with this article; doi:10.1172/JCI28971DS1). Western blot analysis confirmed that p21 protein expression was decreased in p21 siRNA–treated cells but not in control cells (Figure 1B).

It has been shown by us and others that HSCs and progenitor cells resist HIV-1 infection (1–3). Treatment of bone marrow CD34⁺ cells with p21 siRNA resulted in a marked increase in productive HIV-1 infection. At 14 days after infection, HIV-1 Gag p24 levels were 50-fold higher in p21 siRNA–treated cells than in cells that were either mock treated or treated with control siRNA (Figure 2A).

To assess the knockdown of p21 in cells more tractable to in-depth study, we examined HIV-1 replication in CMK cells, a p53-deficient human megakaryoblastic cell type with known ability to modulate p21 (16). CMK cells were derived from megakaryoblasts, which are precursor cells of megakaryocytes and platelets (17). Treating these cells with p21 siRNA increased HIV-1 replication up to 14-fold in comparison with that in controls (Figure 2B). Further, expression of p21 in CMK cells can be stimulated by 12-O-tetradecanoylphorbol-13-acetate (TPA) or retinoic acid without affecting the levels of other CKIs, such as p27^Kip1, p16^INK4A, p15^INK4B, and p18 (16). To determine whether stimulation of p21 expression could inhibit HIV-1 replication, we treated CMK cells with TPA for varying intervals prior to infection with a high dose of HIV-1 (10⁴ 50% tissue culture infective dose (TCID₅₀)/10⁶ cells). The expression of p21 mRNA was increased in a dose-responsive manner after TPA treatment, and HIV-1 replication was blocked in all TPA-treated cells. This was noted in cells treated with TPA for as few as 24 hours. In contrast, mock-treated cells, which had low levels of p21 expression, had high levels of HIV-1 replication (Figure 2, C and D).

Further evaluation of the role for p21 in HIV infection was performed using cells that had been engineered to be deficient in p21. Brown et al. generated a normal diploid human embryonic lung fibroblast line that was then rendered p21^–/– by homologous recombination (18). The parent and p21-deficient cells were then transduced with HIV-1_{NLX.Luc(R–)} (a pseudotype of HIV-1 that only replicates a single round in cells) at 25 reverse transcriptase unit (RTU)/cell, and production of HIV-1 p24 was monitored in the cell supernatant. Cells deficient in p21 produced approximately 3-fold greater p24, consistent with an inhibition on HIV-1 production imposed by p21 (Figure 2E).

p21-restricted viral DNA integration into the host cell genome.

To elucidate the mechanism by which p21 affects HIV-1 susceptibility in primary hematopoietic cells, we examined the different replication steps in the HIV-1 life cycle. First, we examined cell-surface expression of the HIV-1 coreceptors CD4 and CXCR4. There was no detectable change in the amount of each receptor between cells treated by p21 siRNA and control siRNA as determined by flow cytometric analysis (Supplemental Figure 2).

Soon after entry, viral reverse transcriptase converts genomic RNA into linear double-strand cDNA, and viral integrase integrates the viral cDNA into a host cell chromosome (19). To determine whether p21 inhibited reverse transcription or integration, we examined the levels of total HIV-1 cDNA, 2–long terminal repeat (2-LTR) circles, and integration by real-time PCR assays (Supplemental Figure 3) in 2 independent infection systems. First, we used vesicular stomatitis virus — glycoprotein (VSV-G)–pseudotyped HIV-1 to infect CMK cells in a single-round infection after siRNA treatment. From the same infectious dose, viral cDNA was detected in p21 siRNA– and control siRNA–treated cells at comparable levels (Figure 3A). However, 2-LTR circles, markers of aborted integration (20, 21), were significantly increased in control siRNA–treated cells in contrast with p21 siRNA–treated cells (Figure 3B). At 24 hours after infection, a high copy number (4.3 ± 0.11 × 10³) of HIV-1 DNA was detected in the genomes of p21 siRNA–treated cells, but only a marginal level (7.9 ± 0.03 × 10) of integrated HIV-1 DNA was detected in control siRNA–treated cells (Figure 3C). Note the inverse relationship between viral 2-LTR circles and integrated DNA in p21 siRNA– and control siRNA–treated cells (Figure 3, B and C). The single-cycle HIV-1 growth experiments demonstrated that p21 prevented viral DNA integration and enhanced 2-LTR circles in cells, which is a marker of aborted integration.

We then examined wild-type HIV-1 integration in primary CD34⁺ cells and CMK cells. At 16 hours after infection, cells treated with control siRNA had significantly higher levels of 2-LTR circles than cells treated with p21 siRNA (2,721.8 ± 142.6 copies versus 21.7 ± 1.46 copies/6 × 10³ cells) (Figure 3D). Examining the HIV-1 provirus at 18 days after infection revealed that cells treated with p21 siRNA had significantly increased levels of viral integrated DNA in the cell genome in contrast with the controls. An average of 2.9 integrated copies per cell was detected in p21 siRNA–treated cells in contrast with no copies detected per cell in control siRNA–treated cells (Figure 3E). In CMK cells, an average of 2.4 integrated copies per cell were detected in cells treated with p21 siRNA in contrast with 0.07 copies per cell in cells treated with control siRNA (data not shown). Taking these data together, knocking down p21 expression dramatically increased the levels of viral integration in both primary primitive CD34⁺ cells and in the CMK cell line. The high level of 2-LTR circles detected in cells that expressed p21 represented the end products of aborted HIV-1 integration. A high level of 2-LTR circles has been previously seen in cells infected with integrase mutant viruses, which specifically lose the ability to integrate, or in cells treated with specific inhibitors that inhibit the function of HIV-1 integrase (20, 21).

HIV-1 preintegration complexes (PICs) are the functional unit required for integration of the retroviral genome into the host genome. To determine whether p21 interacts with the viral integration machinery, PICs were isolated from control and siRNA-treated CMK cells, and interaction of p21 with PIC components was determined by coimmunoprecipitation–Western blot assays. The viral matrix protein is a PIC component (22, 23), and anti-matrix antibodies coimmunoprecipitated integrase from the initial cytoplasmic extracts (Figure 4A) as well as from PICs purified by spin column chromatography (Figure 4B). Similarly, anti-p21 monoclonal antibody coimmunoprecipitated integrase from starting and purified PICs (Figure 4, A and B). In addition, p21 siRNA, but not the control siRNA, aborted coimmunoprecipitation of integrase from purified PIC samples (Figure 4B). Further, p21 was not detected in PICs derived from H9 cells, a T cell line highly susceptible to HIV-1_IIIB (the original HIV-1 clone isolated from an AIDS patient) infection (Figure 4C). Combined, these data demonstrate that p21 in primitive primary cells or CMK cells interacted with the HIV-1 integration machinery and that the presence of p21 in these complexes was associated with an inability of viral DNA to integrate into cellular DNA.

p21-restricted HIV-1 infection before cell cycling.

To address the relationship between cell cycle and viral replication in p21 siRNA–treated cells, we analyzed cell cycle status by flow cytometry. Cells were stained with Hoechst 33342 for DNA and pyronin Y (PY) for RNA content at 10 hours and 42 hours after siRNA treatment. Up to the time of HIV-1 infection (42 hours after siRNA treatment), there was no significant change in cycle status between cells treated with p21 siRNA and with control siRNA in either primitive CD34⁺ cells (Figure 5A) or in CMK cells (Figure 5B). Since this analysis provides only a snapshot of cell cycle status, we conducted a BrdU incorporation assay, which provided a more sensitive and more temporally integrated analysis. At 42 hours after silencing, there was no difference in BrdU incorporation detected between cells treated with p21 siRNA and with control siRNA (Figure 5, C and D). It was only after 68 hours following siRNA treatment that we did note more cells incorporating BrdU in p21 siRNA–treated cells than in control cells (Supplemental Figure 4). In our system, this was 26 hours after viral infection. This late-occurring change in BrdU incorporation may well be derived from the virus interaction with host cells, either by synthesis of viral DNA in cells or virus-induced further downregulation of p21. Studies have shown that HIV-1 DNA integration occurs before cell cycling, such as in HIV-1 infection of quiescent T lymphocytes and nondividing cells (22–26). We found that HIV-1 infection ultimately downregulated p21 expression in infected cells that were without RNA interference (RNAi) treatment (Supplemental Figure 5). Others report that active HIV-1 infection of T cells is associated with a reduced expression of p21 (27, 28).

To determine whether other CKIs affect HIV-1 infection, we examined the effects of silencing p27^Kip1 (p27, another member of the CIP/KIP sub-family) and p18 (a member of the INK4 subfamily) expression on HIV-1 replication in CD34⁺ cells. Neither p27 nor p18 silencing increased HIV-1 replication in CD34⁺ cells (Figure 6, A and B).

Inhibition of HIV-1 infection is specific and independent of other restriction factors.

To determine whether expression of other host genes known to inhibit HIV-1 could account for the differences seen in susceptibility after silencing p21, we examined expression of Trim5α, PML, Murr1, and IFN-α by using real-time RT-PCR assays. In cells with or without silencing of p21, we could detect no changes in mRNA expression of any of these cellular virus inhibitors (Table 1). The role of p21 in inhibiting HIV-1 replication in primitive cells, therefore, is not mediated by increased expression of 4 other known modulators of innate cell resistance to HIV-1.

To assess whether the effect of p21 on HIV-1 replication was virus specific, we studied the infectivity of SIVmac-251, a dual tropic (T- and M-tropic) lentivirus related to HIV-1 (29–31), in both p21 siRNA–treated primary cells and in CMK cells. A low level of viral antigen p27 was detected (0.103 ± 0.001 ng/ml) in culture supernatant of CD34⁺ cells at day 7, but this dropped to background levels by day 10 (<0.05 ng/ml; Figure 7A). In contrast, the more mature bone marrow CD34^– primary cells from the same donors were actively infected by SIVmac-251 (SIV capsid protein p27 level >3.92 ng/ml at day 7; Figure 7B). Silencing of p21 expression, therefore, did not induce SIV to replicate in primary CD34⁺ cells and CMK cells, in marked contrast to what was observed with HIV-1.

Cells and viruses.

Human CD34⁺ primary cells were obtained from bone marrow of healthy adult volunteers via a protocol approved by the Institutional Review Board of Massachusetts General Hospital. Mononuclear cells from bone marrow were selected with Ficoll/Paque PLUS (Amersham) density gradient centrifugation. CD34⁺ cells were enriched by immunomagnetic selection according to the manufacturer’s instructions (Miltenyi Biotec) with 96% purity. Human megakaryoblastic leukemia cells (CMKs) were grown in RPMI with 20% FBS, and cells in log growth were used in experiments (16). p21^+/+ cells were normal diploid human lung embryonic fibroblasts (LF-1 strain), and the p21^–/– (HO 7.2-1; HO, homozygous derivative) subclone was obtained by targeted homologous recombination (18). Both cell lines were generous gifts from J. Sedivy (Brown University, Providence, Rhode Island, USA). The viral pool of HIV-1_IIIB was harvested from H9/HTLV-IIIB cell culture (NIH AIDS Research & Reference Reagent Program), and viral stocks were titrated with the TCID₅₀ assay in H9 and in human peripheral blood mononuclear cells (45). HIV-1_{NLX.Luc(R–)} pseudotyped with the VSV-G envelope was prepared as previously described (15, 46). In brief, HIV-1_{NLX.Luc(R–)} has the backbone of pHIvec2.GFP but with a luciferase gene inserted upstream of GFP (15). Three package plasmids, pCMV.G, pCMV.E, and pCMV.R (15), were used to make the pseudotyped virus, and the viral titer was determined by the reverse transcription assay.

HIV-1 infection and detection, lentivector transduction.

At 42 hours after siRNA treatment, 1 × 10⁶ cells were infected with 1 × 10³ TCID₅₀ of HIV-1_IIIB. Cell number and viability after siRNA treatment were determined by trypan blue exclusion. Viruses and cells were incubated at 37°C for 2 hours, and the unabsorbed viruses were washed out with PBS. Cells were cultured at 37°C, and the culture supernatant or cells were harvested at indicated time points. Beckman Coulter p24 antigen kinetics assay was used to detect HIV-1 levels in culture supernatant. Cells were transduced with HIV-1_{NLX.Luc(R–)} at 45 RTU/cell in integration experiments and 25 RTU/cell in p24 assays. Transduction was conducted 12–14 hours after cell seeding, with a spin transduction method (15).

SIV infection and detection.

SIVmac-251 was a gift from K. Reimann (Beth Israel Deaconess Medical Center, Boston, Massachusetts, USA) (29). Infection was performed as described previously (29). Beckman Coulter SIV p27 antigen ELISA assay was used to detect SIV replication in the supernatant of cell cultures.

Alu-LTR PCR assay.

The Alu-LTR (with Alu denoting the Alu repeat in the human genome) PCR assay (47–49) was used to determine levels of HIV-1 DNA integration (Supplemental Figure 3). We adapted this assay with some modification. In brief, a nested PCR assay was carried out with 2 sets of Alu-LTR primers and a probe. The sequences of the primers for first step PCR were the same as that of Alu forward (MH535) and Alu reverse (SB704) (47). The PCR was performed with Taq polymerase (Invitrogen) and a DNA thermal cycler (PerkinElmer). Viral DNA was detected by using 500 ng of cellular genomic DNA and subjected to denaturation at 94°C for 4 minutes, then 20 cycles of denaturation at 94°C for 0.5 minutes, annealing at 55°C for 0.5 minutes, and extension at 72°C for 3 minutes, with a last cycle extension at 72°C for 7 minutes. Following the initial PCR, a real-time PCR was performed by using an aliquot equivalent to 1/20 of the 20-cycle PCR product with the second-round Alu-LTR primers and TaqMan probe. The sequences of the second set of primer and probe are as follows: Alu forward (3′ of LTR), 5′GGTAACTAGAGATCCCTCAGACCCT-3′; Alu reverse (5′ of Alu), 5′-gcgtgagccaccgc-3′; and Alu probe, 5′-TTAGTCAGTGTGGAAAATCTCTAGCAGGCCG-3′. The copy numbers of integrated viral DNA in samples were derived from IS cell line (a gift from U. O’Doherty, University of Pennsylvania, Philadelphia, Pennsylvania, USA) (49) adjusted with the known copy number of SM2 plasmid spiked with H9 cellular DNA (47, 48). The gel-purified genomic DNA of OM10.1 cells was used as a reference standard (48). Each sample was tested in triplicate. The copy number of viral DNA in samples was calculated directly from the result spreadsheet, which was converted by the threshold cycle numbers (C_T value ≤ 36 with FAM, 6-carboxyfluorescein) (47, 48) versus the standard curves with Sequence Detector version 1.7 software (Applied Biosystems).

2-LTR circle PCR assays.

The 2-LTR circle PCR assays were performed as previously described (47, 48). The 2-LTR standard was a generous gift from F. Bushman (University of Pennsylvania, Philadelphia, Pennsylvania, USA). PCR assay was performed as described, and the primers and probes were the same as described (47, 48). Viral 2-LTR circles were detected from 500 ng cellular genomic DNA in quintuplicate.

PIC isolation.

CMK and H9 cells in log growth were infected by HIV-1_IIIB at 8,000 TCID₅₀/10⁶ (CMK) or 2,000 TCID₅₀/10⁶ cells (H9), respectively. At 7 hours after infection, cells were harvested and washed twice in buffer K (20 mM HEPES, pH 7.6, 5 mM MgCl₂, 150 mM KCl, 1 mM dithiothreitol, and 20 μg/ml aprotinin) and lysed in 1 ml of buffer K-0.025% (w/v) digitonin. Crude cytoplasmic extract cleared by brief centrifugation was treated with RNase A (0.1 mg/ml) for 30 minutes at room temperature. To partially purify PICs, 2 ml of lysate was passed through a BSA-coated Sepharose CL-4B (Amersham Biosciences) spin column (12 ml) equilibrated in buffer K-0.025% digitonin. Both the crude cytoplasmic extract and the column eluate were used for immunoprecipitation.

Coimmunoprecipitation–Western blot assay.

p21 protein in PIC samples was detected following immunoprecipitation with anti-matrix antibody (mAb 3H7, a gift from A. Engelman, Dana-Farber Cancer Institute, Boston, Massachusetts, USA) or anti-p21 antibody (mAb SC-187; Santa Cruz Biotechnology Inc.) (15). HIV-1 Gag protein matrix (MA) is a PIC component, and mAb 3H7 has been previously used to immunoprecipitate PICs (50–52). Antibody (10 μl) was prebound to 15 μl protein A/G-Agarose (Santa Cruz Biotechnology Inc.) in 100 μl of buffer K for 1 hour at 4°C, and 200 μl of CL-4B column purified cytoplasmic PICs was added. The mixture was rocked for 4 hours at 4°C and washed, and the immune complexes were examined in 12% Tris-Glycine gel (Invitrogen). Integrase and p21 protein were detected by using mAb 8E5 (anti-integrase) and SC-187, respectively.

Western blot analysis of p21 after RNAi treatment was performed as previously described (15). In brief, each sample containing the same amount of lysate protein, determined with the BCA Protein Assay (Pierce Biotechnology), was separated on 16% SDS-PAGE (Invitrogen). Following transfer, the membrane was blocked with 5% milk and detected with mAb SC-187, a mouse anti-human p21 monoclonal antibody. The second antibody was an HRP-labeled rabbit anti-mouse IgG (Jackson ImmunoResearch Laboratories Inc.). The image was obtained by immediately exposing the membrane to film for from 30 seconds to 10 minutes. Hewlett-Packard ScanJet IIcx was used to scan the gel photographs, and density of each band was measured and analyzed with ImageQuant (version 1.3; Molecular Dynamics) software.

Cell culture and TPA assay.

CD34⁺ primary cells were cultured in StemSpan medium (StemCell Technologies Inc.) supplemented with 300 ng/ml SCF, Flt3 ligand, and 60 ng/ml IL-3 (R&D Systems) at 37°C under 5% CO₂ (15, 53). The cell culture supernatants after viral infection were collected at indicated time points and replaced with the same amount of fresh medium. CMK cells after HIV-1 infection were cultured in RPMI 1640 medium containing 10% FBS, 2 mM l-glutamine, 100 units/ml penicillin, and 100 μg/ml streptomycin. Cells were cultured at 37°C under 5% CO₂. In TPA experiments, cells were cultured in the presence or absence of 10 nM of TPA for 1, 2, and 3 days. Cell number and viability were determined with the trypan blue exclusion assay, and cells were washed with PBS before being infected with HIV-1_IIIB at 1 × 10⁴ TCID₅₀/10⁶ cells.

Real-time quantitative RT-PCR.

Real-time quantitative RT-PCR was performed as previously described (10, 15). In brief, the sequences of the primer set for detection of human p27 and p18 mRNA were as follows: p27 forward, 5′-CGGTGGACCACGAAGAGTTAA-3′; p27 reverse, 5′-ggctcgcctcttccatgtc-3′; and p27 probe, 5′-(FAM)CCGGGACTTGGAGAAGCACTGCA-3′(black hole quenchers [BHQ]; Integrated DNA Technologies); p18 forward, 5′-CCCCACAAAACCGGGAA-3′; p18 reverse, 5′-GCTGTAGGCACTCATTGAGCTG-3′; and p18 probe, 5′-(FAM)AAGGAAAGGACAGCGGCGGCA-3′(BHQ). The sequences of the primer set for detection of human Trim5α, PML, Murr1, and INF-α mRNA were as follows: Trim5α forward, 5′-ACCGTGGTCACCACAGGTT-3′; Trim5α reverse, 5′-TGCCTGGAGCTTCACTTGGTA-3′; and Trim5α probe, 5′-(FAM)CCCACAGAGGAGGTTGCCCAGGAG-3′(BHQ); PML forward, 5′-CCGCCCTGGATAACGTCTTT-3′; PML reverse, 5′-CCA CAATCTGCCGGTACACC-3′; and PML probe, 5′-(FAM)TCGAGAGTCTGCAGCGGCGC-3′(BHQ); Murr1 forward, 5′-CAAAAATCCGTGAGAGCCTCA-3′; Murr1 reverse, 5′-CAGCTCAGGCCCCGAAG-3′; and Murr1 probe, 5′-(FAM)AACCAGAGCCGCTGGAATAGCGG-3′(BHQ); IFN-α forward, 5′-CCTCGCCCTTTGCTTTACTG-3′; IFN-α reverse, 5′-GCCCAGAGAGCAGCTTGACT-3′; and IFN-α probe, 5′-(FAM)TGGTCCTGGTGGTGCTCAGCTGC-3′(BHQ). Human cellular GAPDH mRNA was used as the endogenous control in each RT-PCR reaction, and the primers and probes to detect GAPDH mRNA were from the TaqMan GAPDH Control Reagent kit (Applied Biosystems). Total cellular RNA from cells was isolated with the RNeasy kit (QIAGEN) and purified with DNase following the RNeasy clean-up protocol (QIAGEN). An equal amount of purified RNA (50 ng or 100 ng, dependent on each assay) in quadruplicate from each sample was examined in each experiment. A standard curve of the amplicon being measured was run in duplicate with 1 to 1 × 10¹⁰ copies in each assay, with 2 no-template controls. The amplicons from each reaction were analyzed using Sequence Detector version 1.7 software. The RNA copy number in each sample was converted according to the standard curve and the C_T value of the amplicons in each sample (Supplemental Figure 1).

siRNAs and transfection of siRNA into cells.

Both synthetic siRNA duplexes and plasmid-derived shRNA were used in this study, and the sequences of siRNA and shRNA were as previously described (15). The transfection of siRNA and shRNA into cells was as previously described. In brief, siRNA was transfected into CD34⁺ cells by the siPORT method following manufacturer’s instructions (Ambion). In our assay system, siRNA can be efficiently transfected into cells by this method, but shRNA cannot. Electroporation was used to transfect shRNA into cells when sufficient numbers of cells were available (≥ 1.5 × 10⁶ cells/assay condition).

Cell cycle analysis.

The siRNA-treated CD34⁺ cells and CMK cells were stained with CD34⁺ FITC (BD); this was followed by incubation with a DNA dye, Hoechst 33342 (Sigma-Aldrich), and an RNA dye, PY. The proportion of cells in G0 (PY^lowHoechst^low) and in activated status (PY^highHoechst^high) was measured from the CD34⁺ or CMK cells, representing quiescent and activated cells, at 10 hours and 42 hours after siRNA treatment, respectively.

BrdU incorporation assay.

The siRNA-treated CD34⁺ cells were incubated with BrdU in a final concentration of 10 μM. BrdU-pulsed cells were fixed in 70% ethanol at 20°C overnight and denatured by 2N HCl, 0.5% Triton X-100, for 30 minutes at room temperature followed by neutralization with borate buffer (pH 8.5). Treated cells were subjected to dual-color staining with anti-BrdU MoAb (BD) followed by goat anti-mouse FITC-conjugated and 5 μg/ml propidium iodide. Analyses were performed using the FACSCalibur flow cytometer and CellQuest (version 3.3) and ModFit (version 3.0) software (BD).