Cyclin T1 and CDK9 T-Loop Phosphorylation Are Downregulated during Establishment of HIV-1 Latency in Primary Resting Memory CD4⁺ T Cells

Abstract

P-TEFb, a cellular kinase composed of Cyclin T1 and CDK9, is essential for processive HIV-1 transcription. P-TEFb activity is dependent on phosphorylation of Thr186 in the CDK9 T loop. In resting CD4⁺ T cells which are nonpermissive for HIV-1 replication, the levels of Cyclin T1 and T-loop-phosphorylated CDK9 are very low but increase significantly upon cellular activation. Little is known about how P-TEFb activity and expression are regulated in resting central memory CD4⁺ T cells, one of the main reservoirs of latent HIV-1. We used an in vitro primary cell model of HIV-1 latency to show that P-TEFb availability in resting memory CD4⁺ T cells is governed by the differential expression and phosphorylation of its subunits. This is in contrast to previous observations in dividing cells, where P-TEFb can be regulated by its sequestration in the 7SK RNP complex. We find that resting CD4⁺ T cells, whether naïve or memory and independent of their infection status, have low levels of Cyclin T1 and T-loop-phosphorylated CDK9, which increase upon activation. We also show that the decrease in Cyclin T1 protein upon the acquisition of a memory phenotype is in part due to proteasome-mediated proteolysis and likely also to posttranscriptional downregulation by miR-150. We also found that HEXIM1 levels are very low in ex vivo- and in vitro-generated resting memory CD4⁺ T cells, thus limiting the sequestration of P-TEFb in the 7SK RNP complex, indicating that this mechanism is unlikely to be a driver of viral latency in this cell type.

INTRODUCTION

The development and implementation of HAART (highly active antiretroviral therapy) has drastically reduced patient morbidity and mortality associated with HIV-1 infection. Despite its effectiveness, continuous treatment with HAART is required, as any treatment interruption leads to an inevitable rebound of viral replication (1, 2). The inability of HAART to eradicate HIV-1 is attributed mainly to the persistence of the virus in cellular reservoirs (3). Similar to other lentiviruses, HIV-1 can replicate in cells of the macrophage-monocyte lineage, but HIV-1 shows preferential tropism for replication in activated CD4⁺ T cells. Consequently, when these cells revert to a resting memory phenotype, a stably integrated and transcriptionally silent viral reservoir can be generated. Under this reversible nonproductive state of infection, HIV-1 is refractory to immune surveillance and eradication by HAART. However, upon immune activation of the resting memory cells, this latent viral reservoir can be reactivated, resulting in productive viral replication (4).

The establishment and maintenance of latency in memory CD4⁺ T cells is a complex multifactorial process that is believed to be determined chiefly by the activation status of the cell. Studies have shown that various factors operating at the levels of transcription and posttranscription can restrict the expression of the integrated provirus in resting CD4⁺ T cells (5, 6). Transcriptional blocks to productive HIV-1 replication include epigenetic modifications at the viral long terminal repeat (LTR) and inadequate availability of activation-dependent transcription factors such as P-TEFb, NF-κB, NFAT, Sp1, AP-1, and C/EBP (7–10). Posttranscriptional factors regulating viral latency include inhibition of viral mRNA export to the cytoplasm and regulation of viral gene expression by cellular microRNAs (miRNAs) (11–13). However, upon CD4⁺ T cell activation, these blocks to productive viral replication are removed and transcription of the viral genome occurs through the coordinated actions of host cell transcription factors and the viral transactivator protein Tat.

The HIV-1 protein Tat is crucial for the stimulation of productive RNAPII (RNA polymerase II) transcription from the integrated 5′ viral LTR (14). Although RNAPII initiates transcription from the viral LTR in the absence of Tat, processive elongation of the viral transcript is deficient because of the association of the negative regulatory factors DSIF and NELF bound to RNAPII (15). To overcome these negative blocks to elongation, Tat recruits the host transcription factor P-TEFb to RNAPII at promoter-proximal regions (16). Tat can also recruit other elongation factors, such as ELL2, AFF4, ENL, and AF9, to the HIV-1 LTR in addition to P-TEFb, thereby forming a superelongation complex (17, 18).

Core P-TEFb is a heterodimeric cellular kinase composed of a cyclin subunit, either Cyclin T1 or Cyclin T2, and an invariant CDK9 (cyclin-dependent kinase 9) subunit (19). Tat makes direct protein-protein contact with Cyclin T1 and can therefore target only Cyclin T1-containing P-TEFb complexes (20). P-TEFb kinase activity is dependent on the phosphorylation of Thr-186 in the CDK9 T loop (21). Tat recruits T-loop-phosphorylated P-TEFb to a stem-loop RNA structure called TAR (trans-acting response), which forms at the 5′ end of the nascent viral transcript (22). CDK9 phosphorylates the C-terminal domain (CTD) of RNAPII and the associated negative elongation factors, resulting in the conversion of DSIF to a positive elongation factor and inhibiting NELF from associating with the viral transcripts (19). Therefore, P-TEFb is critical for promoting processive elongation from the viral LTR by antagonizing the negative factors associated with RNAPII.

Since P-TEFb is required for the efficient transcriptional elongation of the majority of cellular genes (23), its activity must be regulated to complement the dynamic transcriptional needs of the cell. In HeLa cells, approximately half of P-TEFb is found in the 7SK RNP complex, which includes 7SK snRNA as the scaffold and protein components, such as HEXIM1 or HEXIM2, MePCE, and LARP7 (24–26). P-TEFb kinase activity is repressed in this complex because of association with the HEXIM1/2 protein. However, cellular and viral factors can release P-TEFb from this complex by altering the cellular conformation of 7SK RNA (27, 28).

In primary CD4⁺ T cells, P-TEFb is highly regulated. While the total levels of CDK9 remain relatively constant through resting versus activated states, the levels of T-loop-phosphorylated CDK9 (pCDK9) are very low in resting CD4⁺ T cells due to the action of the phosphatase PPM1A (29, 30). However, activation of these cells induces phosphorylation of the CDK9 T loop by an activating kinase that has recently been shown to be CDK7 (31). Cyclin T1 is also subject to regulation in CD4⁺ T cells, as it is expressed at low-to-undetectable levels in resting CD4⁺ T cells (32). Activation of these cells results in an increase in Cyclin T1 protein levels, while the levels of Cyclin T1 mRNA do not change significantly (33, 35). The transcription-independent increase in Cyclin T1 protein levels is in part due to regulation by differentially expressed miRNAs (11). However, translation initiation from the Cyclin T1 3′ untranslated region (UTR) has also been shown to be regulated by an RNA binding protein, NF90, independently of miRNAs in certain cell lines (34). In late differentiated macrophages, Cyclin T1 protein expression is regulated by proteasome-mediated proteolysis, indicating that multiple mechanisms regulate Cyclin T1 protein expression (35).

Regulation of P-TEFb and its association in the HEXIM1 complex have not been explored in activated CD4⁺ T cells as they transition to a resting memory phenotype. In this study, we used a central-memory-specific in vitro model of HIV-1 latency to examine the regulation of P-TEFb in central memory CD4⁺ T cells in response to activation signals (36, 37). Our results indicate that Cyclin T1 and pCDK9 levels are low in naïve CD4⁺ T cells, and upon activation their levels increase significantly. Infection of these cells with HIV-1, followed by transition to a central-memory phenotype, resulted in the generation of latently infected memory cells, which are quiescent (36). The low levels of Cyclin T1 and pCDK9 found in resting memory CD4⁺ T cells appeared to be independent of the presence or absence of latent proviruses in these cells. We conclude that the limited availability of P-TEFb in resting memory CD4⁺ T cells is an important determinant of viral latency. Furthermore, we show that the low levels of Cyclin T1 found in naïve CD4⁺ T cells and resting memory CD4⁺ T cells are partly due to proteasome-mediated proteolysis of Cyclin T1 and also likely due to posttranscriptional regulation by miRNAs such as miR-150. We also show that in quiescent memory CD4⁺ T cells, due to low expression of HEXIM1, very little P-TEFb is found associated with the 7SK RNP complex. Reactivation of the cells led to an increase in the HEXIM1 levels and a consequent increase in the association of P-TEFb with the 7SK RNP complex. Our data indicate that the sequestration of P-TEFb in the 7SK RNP complex in resting memory CD4⁺ T cells is unlikely to be a mechanism by which latency is established or maintained. Thus, our data indicate that multiple mechanisms restrict the expression, catalytic activity, and availability of P-TEFb in resting memory CD4⁺ T cells, thereby limiting the transcriptional activity of the HIV-1 provirus.

(M.F. conducted this study as partial fulfillment of her Ph.D. in Molecular Medicine, Section of Basic and Applied Immunology, Vita-Salute San Raffaele University, Milan, Italy.)

MATERIALS AND METHODS

Generation of latently infected memory CD4⁺ T cells.

Naïve CD4⁺ T cells were isolated through negative selection from whole blood obtained from healthy anonymous blood donors by using the EasySep human naïve CD4⁺ T cell enrichment kit (Stem Cell Technology). In vitro-cultured central memory CD4⁺ T cells were generated and infected as previously described, with a few modifications (36). Briefly, naïve CD4⁺ T cells were cultured in medium containing transforming growth factor beta (TGF-β) and, anti-interleukin-12 (anti-IL-12) and anti-IL-4 antibodies (Peprotech) and activated with anti-CD3/anti-CD28-coated Dynabeads (Invitrogen) for 3 days. Cells were cultured for an additional 2 days in medium containing 30 IU/ml IL-2. Five days postactivation, 10⁶ cells were infected with 500 ng of DHIV/X4 (p24 concentration measured by enzyme-linked immunosorbent assay (Zeptometrix) by spinoculation for 2 h at 2,900 rpm at 37°C. Cells were cultured for an additional 9 days postinfection. Latent proviruses were reactivated by activation with anti-CD3/anti-CD28-coated Dynabeads.

DHIV generation.

DHIV is a molecular clone derived from HIV-1_NL4-3 harboring a frameshift mutation in the env gene that renders it capable of only a single round of infection. DHIV was pseudotyped with the HIV-1_LAI envelope glycoprotein and generated by calcium phosphate-mediated transient transfection of HEK 293T cells.

p24 detection by flow cytometry.

Intracellular p24 Gag expression at days 2, 5, and 7 after infection was analyzed by fixing and permeabilizing 2 × 10⁵ cells with Cytofix/Cytoperm (BD Bioscience) for 30 min at 4°C. Following fixing, cells were washed with Perm/Wash buffer (BD Bioscience) and stained for 30 min at 4°C with a 1:50 dilution of anti-p24 antibody (AG3.0, kindly provided by the late Jonathan Allan, Southwest Foundation for Biomedical Research, San Antonio, TX) in 100 μl Perm/Wash buffer. Stained cells were washed with Perm/Wash buffer and incubated for 30 min at 4°C with a 1:1,000 dilution of allophycocyanin (APC)-conjugated goat anti-mouse IgG (H+L) in Perm/Wash buffer, followed by flow cytometry analysis.

Intracellular p24 Gag expression of reactivated cells was analyzed by fixing and permeabilizing cells with Cytofix/Cytoperm for 30 min at 4°C. After washing with Perm/Wash buffer, cells were stained for 30 min at 4°C with 1 μl fluorescein isothiocyanate-conjugated mouse anti-p24 antibody (clone KC57; Beckman Coulter) in 100 μl Perm/Wash buffer. Cells were washed with Perm/Wash buffer and analyzed by flow cytometry.

Total RNA and miRNA extraction and qRT-PCR.

For detection of cyclin T1 mRNA and cyclin T1 targeting miRNAs by quantitative reverse transcription (qRT)-PCR, total RNA enriched for the miRNA fraction was isolated by using the miRVana miRNA isolation kit (Ambion), following the manufacturer's instructions. Cyclin T1-targeting miRNAs were quantified by using TaqMan miRNA assays (ABI). Total RNA was reverse transcribed using the iScript cDNA synthesis kit (Bio-Rad) and amplified using the iQ SYBR green mix (Bio-Rad). The following primer sequences were used for qRT-PCR: GAPDH-F, 5′-CGCCAGCCGAGCCACATC-3′; GAPDH-R, 5′-AAATCCGTTGACTCCGACCTTCAC-3′; Cyclin T1-F, 5′-AACCTTCGCCGCTGCCTTC-3′; Cyclin T1-R, 5′-ACCGTTTGTTGTTGTTCTTCCTCTC-3′.

Immunoblots, immunoprecipitations, and antibodies.

Cell lysates were prepared in EBCD buffer (50 mM Tris-HCl [pH 8.0], 120 mM NaCl, 0.5% Nonidet P-40, 5 mM dithiothreitol, 4 mM MgCl₂ buffer containing protease inhibitor cocktail [Sigma Chemical Co.]). The protein concentration of cell lysates was measured by the Bio-Rad protein assay, which is based on the Bradford method. Proteins were separated by 10% SDS-PAGE, transferred to a nitrocellulose membrane, and probed with the antibodies indicated. Protein detection was carried out by using enhanced-chemiluminescence Western blotting substrate (Pierce).

Immunoprecipitations were carried out by using Dynabeads protein A (Invitrogen) according to the manufacturer's instructions. Briefly, total CDK9 antibody was bound to 40 μl of Dynabeads protein A for 20 min at room temperature, followed by two washes with phosphate-buffered saline (PBS)-Tween (0.2%). Cells were lysed with EBCD buffer and centrifuged at full speed in a microcentrifuge to pellet cellular debris. Supernatants were incubated with Dynabeads protein A bound with CDK9 antibody overnight at 4°C. To remove unbound and nonspecifically bound proteins, beads were washed three times with PBS. The bound proteins were eluted by boiling in 1× Laemmli sample buffer for 7 min at 95°C. Portions of immunoprecipitates were separated by SDS-PAGE and immunoblotted to probes for specific proteins.

CDK9 (sc-484) and cyclin T1 (sc-8127) antibodies were purchased from Santa Cruz Biotechnology. Phosphorylated CDK9 antibody (2549) was purchased from Cell Signaling Technology. Anti-actin antibody (MAB1501) was purchased from Millipore. The HEXIM1 antibody was kindly provided by Jiemin Wong (Baylor College of Medicine).

RESULTS

Expression of P-TEFb in naïve and central memory CD4⁺ T cells.

We have previously shown that cyclin T1 levels are low in unfractionated resting CD4⁺ T cells, which are a mixture of naïve and memory CD4⁺ T cells. Activation of these cells leads to significant induction of cyclin T1 protein levels, primarily by a posttranscriptional mechanism (33, 38). Similarly, pCDK9 levels are considerably upregulated upon activation (30). In this study, we wished to determine if P-TEFb shows a similar pattern of regulation in naïve and quiescent memory CD4⁺ T cells that were either uninfected or latently infected with HIV-1. To address this issue, we used a central memory CD4⁺ T cell-specific model of HIV-1 latency (36). In this model, naïve CD4⁺ T cells obtained from healthy blood donors are activated with anti-CD3/anti-CD28-coated Dynabeads and cultured in the presence of TGF-β1 and anti-IL-2 and anti-IL-4 antisera to induce the generation of nonpolarized cells, the ex vivo phenotypic and functional counterparts of central memory CD4⁺ T cells. Following activation, cells are infected with a virus derived from HIV-1_NL4-3 with a mutation in the env gene and pseudotyped with HIV-1_LAI env. In a typical experiment shown in Fig. 1, the efficiency of infection was determined by intracellular p24 Gag staining at 2, 5, and 7 days after infection. At day 2 after infection, 3.6% of the total cell population was productively infected (Fig. 1A). There was a progressive decrease in p24 Gag staining at days 5 and 7 after infection. This decrease in productively infected cells was previously shown to be due to cellular apoptosis (39). Infected cultures were maintained in medium containing recombinant IL-2 (rIL-2) for 9 days postinfection, allowing the cells to naturally revert to a quiescent state and acquire a central-memory phenotype. During this transition, latent HIV-1 infection is established in a significant fraction of the cells in the infected culture and the latent virus can be efficiently reactivated at a later time. As shown in Fig. 1B, uninfected resting memory CD4⁺ T cells and latently infected resting memory CD4⁺ T cells, referred to hereafter as URM and LIRM CD4⁺ T cells, respectively, were reactivated at day 9 after infection by incubation with anti-CD3/anti-CD28-coated Dynabeads. Reactivation of the latent reservoir was verified by intracellular p24 staining (Fig. 1B). In the absence of activation signals, only 2.1% of p24 Gag positive cells could be detected. However, in the presence of activating stimuli, the latent subset was efficiently reactivated, as indicated by the progressive increase in the percentage of p24-positive cells, with 78.5% of the cells staining positive for p24 Gag after 72 h of activation.

Fig 1.

Generation of latently infected memory CD4⁺ T cells in vitro. (A) At days 2, 5, and 7 postinfection (P.I.) of activated CD4⁺ T cells with DHIV, intracellular p24 Gag expression in cells was assessed by flow cytometry. Each panel indicates the percentage of p24-positive cells. (B) Uninfected or DHIV-infected cells were cultured in the absence or presence of anti-CD3/anti-CD28-coated Dynabeads at 9 days postinfection for the indicated times. Intracellular p24 Gag expression was assessed by flow cytometry, and the percentages of p24-positive cells are indicated in the panels. SSC, side scatter; FITC, fluorescein isothiocyanate.

We then examined the status of P-TEFb in these cells at various stages of differentiation. Figure 2A shows the analysis for cells obtained from a representative donor. Cell lysates were prepared at different time points, and extracts were probed in immunoblot assays for the expression levels of Cyclin T1, total CDK9, and pCDK9. Cyclin T1 and pCDK9 levels were low to undetectable in naïve CD4⁺ T cells (Fig. 2A, lane 1). In contrast, total CDK9 was always detected at relatively high, constant levels (Fig. 2A, lane 1). Upon the activation of naïve CD4⁺ T cells, Cyclin T1 and CDK9 T-loop phosphorylation levels increased significantly (Fig. 2A, lanes 2 and 3). The high levels of Cyclin T1 and CDK9 T-loop phosphorylation were sustained even at 120 h postactivation (Fig. 2A, lane 4). Activation did not significantly change the total CDK9 levels.

Fig 2.

P-TEFb expression in latently infected CD4⁺ T cells. (A) Activated naïve CD4⁺ T cells obtained from a representative donor were left uninfected or infected with DHIV and allowed to return to a quiescent memory state. The URM and LIRM CD4⁺ T cells were reactivated with anti-CD3/anti-CD28-coated Dynabeads for the times indicated. Cell lysates were prepared at indicated time points and probed in immunoblots for the expression of Cyclin T1, pCDK9, total CDK9, and anti-actin (α-Actin) antibody (loading control). (B) Naïve and memory CD4⁺ T cells were isolated from whole blood from a representative donor and activated for 48 h with anti-CD3/anti-CD28-coated Dynabeads. Cell lysates were probed at the time points indicated for the expression of Cyclin T1, pCDK9, total CDK9, HEXIM1, and anti-actin antibody (loading control). PA, postactivation.

The transition of activated naïve CD4⁺ T cells to a central-memory phenotype by incubating cultures 9 days in the presence of rIL-2 led to a marked reduction in Cyclin T1 and pCDK9 (Fig. 2A, lanes 5 [URM CD4⁺ T cells] and 7 [LIRM CD4⁺ T cells]). Reactivation of the URM and LIRM CD4⁺ T cells, respectively, significantly upregulated cyclin T1 protein levels and T-loop phosphorylation levels (Fig. 2A, lanes 6 and 8). However, there was no change in total CDK9 levels after the reactivation of URM and LIRM CD4⁺ T cells. Our analysis also indicated that HEXIM1 was differentially expressed in resting and activated naïve URM and LIRM CD4⁺ T cells (also see Fig. 5A).

Fig 5.

Coimmunoprecipitation of HEXIM1 with CDK9 in latently infected CD4⁺ T cells. URM and LIRM CD4⁺ T cells were generated as described in Materials and Methods. (A) Whole-cell lysates prepared at the times indicated were probed for the expression of HEXIM1, Cyclin T1, total CDK9, and anti-actin antibody (loading control). CDK9 was immunoprecipitated in cell lysates obtained from naïve and activated naïve CD4⁺ T cells (B), URM and reactivated URM CD4 ⁺T cells (C), and LIRM and reactivated LIRM CD4⁺ T cells (D). Immunoprecipitates were probed for the association of HEXIM1 and Cyclin T1 in immunoblots. IP, immunoprecipitation; WB, Western blot.

We performed a statistical analysis of the n-fold changes in Cyclin T1 and CDK9 T-loop phosphorylation following the transition of naïve CD4⁺ T cells activated for 48 h to a resting state in different donors (Table 1). The statistically significant n-fold reduction in Cyclin T1 levels in URM CD4⁺ T cells were 85% (donor 79; data not shown), 88% (donor 80; Fig. 2A), and 80% (donor 151; see Fig. 5A). Similarly, the n-fold decrease in Cyclin T1 levels in LIRM CD4⁺ T cells was statistically significant, ranging from 70% (donor 79 and donor 80) to ∼80% (donor 151). T-loop phosphorylated CDK9 levels also showed a statistically significant decrease in both URM and LIRM CD4⁺ T cells, ranging from ∼75% to ∼90% in two donors examined (donors 79 and 80). However, the difference in the n-fold reduction of both Cyclin T1 and pCDK9 levels between URM and LIRM CD4⁺ T cells was not statistically significant.

Table 1.

Changes in cyclin T1 and pCDK9 levels in URM and LIRM CD4⁺ T cells^a

Protein and CD4+ T cell type (time point)	Change (n-fold) in:			P value
	Donor 79	Donor 80	Donor 151
Cyclin T1
Naïve (48 h PA)b	1	1	1	0.0007
URM	0.15	0.1	0.2
Naïve (48 h PA)	1	1	1	0.002
LIRM	0.3	0.28	0.18
URM	0.15	0.1	0.2	0.238
LIRM	0.3	0.28	0.18
pCDK9
Naïve (48 h PA)	1	1		0.05
URM	0.12	0.25
Naïve (48 h PA)	1	1		0.019
LIRM	0.21	0.16
URM	0.12	0.25		1
LIRM	0.21	0.16

^aImmunoblots shown in Fig. 2A (donor 80) and 5A (donor 151) and similar immunoblots of cells from donor 79 (data not shown) were quantified by using ImageJ software. After normalization to the α-actin loading control, n-fold changes in Cyclin T1 and pCDK9 levels were calculated by setting their respective levels in naïve CD4⁺ T cells activated for 48 h to 1. P values were calculated by using a paired t test.

^bPA, postactivation.

To rule out the possibility that the differential expression of Cyclin T1 and pCDK9 observed in resting memory CD4⁺ T cells was due to in vitro culturing, we determined their levels in naïve and memory CD4⁺ T cells isolated directly from whole blood obtained from a healthy donor. As shown in Fig. 2B, Cyclin T1, pCDK9, and HEXIM1 levels were reduced in both naïve and memory CD4⁺ T cells (lanes 1 and 3). Activation of these cells for 48 h with anti-CD3/anti-CD28-coated Dynabeads significantly upregulated cyclin T1, pCDK9, and HEXIM1 levels (lanes 2 and 4). In both naïve and memory CD4⁺ T cells, there was a relatively high basal level of total CDK9, which increased moderately upon activation of the memory cells, but not the naïve CD4⁺ T cells. Taken together, these results indicate that the repression of P-TEFb in latently infected memory CD4⁺ T cells due to the regulated expression and posttranslational modifications of its subunits likely contributes to the transcriptional silencing of HIV-1.

Cyclin T1 downregulation in naïve and resting memory CD4⁺ T cells involves proteasome-mediated proteolysis.

We have previously shown that Cyclin T1 levels are highly upregulated following differentiation of primary monocytes to macrophages. However, after 1 to 2 weeks postdifferentiation in vitro, Cyclin T1 protein levels decrease as the result of proteasome-mediated proteolysis (35); IL-10 accelerates this proteolysis of cyclin T1 in macrophages (40). We therefore wanted to investigate whether proteasome-mediated proteolysis is involved in shutting off Cyclin T1 protein levels in naïve, URM, and LIRM CD4⁺ T cells. The three different cell populations were obtained as described above and treated with the proteasome inhibitor MG132 (Fig. 3A). Treatment of naïve CD4⁺ T cells with MG132 for 4 h led to a 2.71-fold increase in cyclin T1 levels. Similarly, treatment of URM CD4⁺ T cells for 2 h and 4 h resulted in 1.9- and 2.0-fold increases in cyclin T1 protein levels, respectively. In LIRM CD4⁺ T cells, 2-h and 4-h MG132 treatments led to 1.65- and 1.82-fold increases in cyclin T1 protein levels, respectively. These results indicate that the low levels of cyclin T1 expression observed in naïve, URM, and LIRM CD4⁺ T cells are due in part to proteasome-mediated proteolysis. It is possible that the PEST sequence at the carboxyl terminus of Cyclin T1 may confer this sensitivity to proteolysis (22).

Fig 3.

Regulation of Cyclin T1 in latently infected memory CD4⁺ T cells. Naïve, URM, and LIRM CD4⁺ T cells generated as described in Materials and Methods were cultured in the absence or presence of the proteasome inhibitor MG132 for 2 h or 4 h. Cell lysates were probed for the expression of Cyclin T1 (A), PPM1A (B), HSP90 (C), and anti-actin (α-Actin) antibody (loading control) in immunoblots. Fold changes in cyclin T1 protein were calculated by normalization to anti-actin antibody and setting the cyclin T1 levels in naïve, URM, and LIRM CD4⁺ T cells, respectively, to 1.

To examine the specificity of proteasomal inhibition for Cyclin T1 protein expression, we examined the effects of MG132 treatment in naïve, URM, and LIRM CD4⁺ T cells on the expression of HSP90 and PPM1A, a Mg⁺²/Mn⁺²-dependent phosphatase. Treatment of naïve, URM, and LIRM CD4⁺ T cells had no effect on the expression of HSP90 or PPM1A compared to that in untreated cells (Fig. 3B and C). This suggests that the proteasome-mediated proteolysis of Cyclin T1 is likely a specific posttranslational mechanism that regulates its protein levels in naïve and resting memory CD4⁺ T cells.

We also examined the expression pattern of miRNAs previously shown to target Cyclin T1 for repression in resting CD4⁺ T cells (11). Naïve, URM, LIRM CD4⁺ T cells were generated, and total RNA was extracted for qRT-PCR assays to determine the expression of miR-27b, miR-29b, and miR-150, three miRNAs that we have previously shown to repress Cyclin T1 protein expression posttranscriptionally in unfractionated resting CD4⁺ T cells (11). As indicated in Fig. 4A, Cyclin T1 mRNA levels did not change significantly upon the activation of naïve, URM, and LIRM CD4⁺ T cells, indicating that the increase in Cyclin T1 protein expression upon cellular activation (Fig. 2A), in addition to being regulated by proteasome-mediated proteolysis, is also posttranscriptional in nature. Of the miRNAs assayed, only miR-150 showed a reproducible differential pattern of expression in the donors examined; data from a representative donor are shown in Fig. 4B. Our previous study demonstrated that an antagomir to miR-150 increased cyclin T1 protein levels in resting CD4⁺ T cells from several donors. However, this study suggested that miR-150 may represses Cyclin T1 through an indirect mechanism, as we were unable to identify miR-150 target sequences in the Cyclin T1 3′ UTR (11). As shown in Fig. 4B, miR-150 levels were high in naïve, URM, and LIRM CD4⁺ T cells, suggesting that miR-150 plays a role in repressing Cyclin T1 protein expression in these cells. Activation of naïve CD4⁺ T cells and reactivation of the URM and LIRM CD4⁺ T cells led to a marked decrease in miR-150 levels, and this correlated with an increase in Cyclin T1 protein levels (see Fig. 2). Taken together, our results suggest that Cyclin T1 protein levels are repressed in naïve, URM, and LIRM CD4⁺ T cells by both proteasome-mediated proteolysis and likely the action of miR-150. Cellular activation appears to alleviate these repressive mechanisms, resulting in increased cyclin T1 protein expression.

Fig 4.

Cyclin T1 mRNA and miR-150 expression in latently infected memory CD4⁺ T cells (A) Cyclin T1 mRNA was detected by qRT-PCR, and relative abundance was calculated by normalizing to glyceraldehyde 3-phosphate dehydrogenase (GAPDH) mRNA. (B) miR-150 levels were examined in CD4⁺ T cell subsets as indicated by TaqMan miRNA RT-PCR assays. Fold changes were calculated by normalization to U6 snRNA and setting the miR-150 level in naïve CD4⁺ T cells to 1. The data shown in panels A and B are representative of a single donor.

HEXIM1 expression and its association with P-TEFb are reduced in resting memory CD4⁺ T cells, independent of their latent infection status.

The association of catalytically active P-TEFb with the 7SK RNP complex in transformed cell lines is well established and thought to be a mechanism by which cellular gene expression is regulated (27). It has been suggested on the basis of studies done with transformed cell lines that sequestration of P-TEFb in the 7SK RNP complex could be a mechanism involved in latent HIV-1 infection by limiting the availability of P-TEFb (25, 41). We therefore wished to examine the expression of HEXIM1 and its association with CDK9 in naïve, URM, and LIRM CD4⁺ T cells. Cells were cultured and infected as described above, and cell lysates were prepared for immunoblot analysis and coimmunoprecipitation using a CDK9 antiserum (Fig. 5). The analysis of cell lysates revealed that HEXIM1 levels are low to undetectable in all three quiescent populations analyzed, namely, naïve, URM, and LIRM CD4⁺ T cells (Fig. 5A, lanes 1, 4, and 8). Activation of the naïve CD4⁺ T cells for 24 h and 48 h upregulated HEXIM1 levels (Fig. 5A, lanes 2 and 3). Reactivation of URM CD4⁺ T cells and LIRM CD4⁺ T cells also resulted in a significant increase in HEXIM1 levels at 24, 48, and 72 h postactivation (Fig. 5A, lanes 5 to 7 and 9 to 11).

We further wanted to determine if the low levels of HEXIM1 in URM and LIRM CD4⁺ T cells would affect the partitioning of P-TEFb in the 7SK RNP complex. Using the cell lysates examined in Fig. 5A, we immunoprecipitated CDK9 and measured the association of HEXIM1 and cyclin T1 with CDK9 in immunoblots. In naïve CD4⁺ T cells, the association of CDK9 with HEXIM1 was below the level of detection (Fig. 5B, lane 1), whereas activation of these cells resulted in a readily detectable association between CDK9 and HEXIM1 at both 24 and 48 h postactivation (Fig. 5B, lanes 2 and 3). As expected, there was reduced association of cyclin T1 with CDK9 in naïve CD4⁺ T cells (Fig. 5B, lane 1) and activation strongly increased this association (Fig. 5B, lanes 2 and 3). Similarly in URM CD4⁺ T cells, there was a strong diminution of the amount of HEXIM1 associated with CDK9, and this association markedly increased upon reactivation at the 24- and 48-h time points examined (Fig. 5C, lanes 1 to 3).

Finally, in LIRM CD4⁺ T cells, little HEXIM1 was associated with CDK9 (Fig. 5D, lane 1) and reactivation strongly increased the association at 24 and 48 h (Fig. 5D, lanes 2 and 3). These results demonstrate that the reduced expression of HEXIM1 in naïve CD4⁺ T cells and uninfected and latently infected memory CD4⁺ T cells that have entered a quiescent state greatly reduces the sequestration of P-TEFb in the 7SK RNP complex. Upon activation of the cells, both HEXIM1 and cyclin T1 levels increase and the level of P-TEFb sequestered in the 7SK RNP complex greatly increases. Thus, it is unlikely that the sequestration of P-TEFb in 7SK RNP complex is a dominant mechanism by which P-TEFb availability is limited in quiescent memory CD4⁺ T cells.

DISCUSSION

The HIV-1 latent reservoir in memory CD4⁺ T cells is believed to be generated when activated CD4⁺ T cells are infected as they are making the transition to a resting memory phenotype (4), although infection of resting cells can also occur (42, 43). As the infected cells revert to a metabolically quiescent state, productive transcription of the integrated provirus is shut down (44). Transcription elongation is a major checkpoint in the HIV-1 life cycle, and it has been demonstrated that in order for the latent reservoir to be purged or reactivated, it will likely be essential to activate P-TEFb (10, 44). Therefore, it is necessary to understand the mechanisms by which P-TEFb activity and expression are regulated in latently infected cells. In this study, we used an in vitro model of HIV-1 latency (36) to show that a major mechanism by which P-TEFb availability is limited in the latently infected and quiescent memory CD4⁺ T cell population is the differential expression of Cyclin T1 and phosphorylation of the CDK9 T loop. Furthermore, contrary to previous suggestions, P-TEFb availability in uninfected and latently infected resting memory CD4⁺ T cells in this model system does not appear to be dictated by its sequestration in the 7SK RNP complex, as shown in the model presented in Fig. 6.

Fig 6.

In resting CD4⁺ T cells, P-TEFb availability is dictated predominantly by the low levels of Cyclin T1 protein and T-loop-dephosphorylated CDK9 and is not due to its sequestration in the 7SK RNP complex, which is limited due to low levels of HEXIM1. Cellular activation upregulates HEXIM1, Cyclin T1 protein levels, and CDK9 T-loop phosphorylation, leading to increased association of P-TEFb in the 7SK RNP complex, which also includes MePCE and LARP7. Tat competitively recruits P-TEFb from this complex to the stalled RNAPII at the viral LTR. CDK9 phosphorylates the CTD of RNAPII and the negative elongation factors associated with it (not shown) to promote processive elongation of the viral transcripts. (Adapted from references 6 and 14 with permission of the publishers.)

We have shown in this study that Cyclin T1 and pCDK9 levels are low in a highly purified population of naïve CD4⁺ T cells, indicating that P-TEFb is likely limiting in these cells for HIV-1 replication. Upon the activation of these naïve cells, both Cyclin T1 and pCDK9 levels increase significantly. This is likely in response to extracellular cues such as Ca⁺² signaling, which, as we have shown previously, plays a part in CDK9 T-loop phosphorylation following the activation of resting CD4⁺ T cells (45). Besides the increased availability of catalytically active P-TEFb, activation of CD4⁺ T cells also increases the availability of other host cofactors, such as NF-κB, thereby creating a more favorable environment for HIV-1 transcription (46). In the in vitro system used in this study, the culturing of activated mock-infected and HIV-infected cells following the removal of activating stimuli allows for their natural progression to memory quiescent CD4⁺ T cells (36). In both the uninfected and infected quiescent memory CD4⁺ T cells from different donors, Cyclin T1 and pCDK9 levels show a statistically significant decrease in expression, a state which is reversed upon reactivation through TCR engagement. Similar results were also obtained by the Karn laboratory in another in vitro model of HIV-1 latency, where CDK9 and Cyclin T1 levels were reduced in memory CD4⁺ T cells latently infected with HIV-1 (10). Importantly, the upregulation of Cyclin T1 protein expression was found to be necessary for reactivation of latent virus in this model system of memory CD4⁺ T cells. These data suggest that reversion to a resting memory state and downregulation of P-TEFb reestablish the restrictive environment for HIV-1 transcription found in naïve CD4⁺ T cells and thus likely play a role in the establishment and/or maintenance of HIV-1 latency.

Our results also suggest that the reduced expression of Cyclin T1 in naïve, quiescent, and latently infected memory CD4⁺ T cells is due to multiple mechanisms which include proteasome-mediated proteolysis and posttranscriptional regulation by miRNAs such as miR-150. In naïve, uninfected, and latently infected resting memory CD4⁺ T cells, Cyclin T1 protein expression is subject to proteasome-mediated proteolysis, similar to Cyclin T1 shutdown in late differentiated macrophages (35). The proteasome-mediated turnover of Cyclin T1 protein levels is likely a specific regulatory mechanism, as the levels of other proteins that we examined, PPM1A and Hsp90 were not affected by MG132 treatment. Upon T-cell activation, the repression of Cyclin T1 protein expression is relieved, resulting in increased accumulation of Cyclin T1. In resting CD4⁺ T cells, which are a mixture of naïve and memory CD4⁺ T cells, repression of Cyclin T1 protein expression appears to be largely posttranscriptional (32, 38). In these cells, we have previously shown that several differentially expressed miRNAs target Cyclin T1 expression directly and indirectly (11). Utilizing a central memory model of HIV-1 latency, we examined several miRNAs that have been shown to repress Cyclin T1 in resting CD4⁺ T cells—miR-27b, mir-29b, and miR-150. However, miR-150 was found to be the only miRNA examined that showed a consistent differential pattern of expression in the donors tested. MiR-150 levels were elevated in naïve, resting, and latently infected memory CD4⁺ T cells and decreased markedly upon cellular activation and reactivation. In our previous study, inhibition of miR-150 by transfection of an antagomir in resting CD4⁺ T cells led to an increase in Cyclin T1 protein expression (11) and it is therefore likely that the increased level of miR-150 in naïve and quiescent mock-infected and latently infected memory CD4⁺ T cells observed here contributes to a repressive environment for transcription of the latent provirus.

We observed here that CDK9 T-loop phosphorylation is low in naïve CD4⁺ T cells and quiescent uninfected and latently infected memory CD4⁺ T cells, whereas activation of these cells leads to a large increase in T-loop phosphorylation. The stability of total CDK9 protein is also dependent on T-loop phosphorylation, as we have previously shown that perturbation of T-loop phosphorylation affects total CDK9 levels (45). However, CDK9 T-loop phosphorylation does not affect Cyclin T1 protein stability. We recently reported that the cellular Mg⁺²/Mn⁺²-dependent phosphatase PPM1A dephosphorylates the CDK9 T loop in resting CD4⁺ T cells (29). It is probable that PPM1A is involved in dephosphorylating the T loop in naïve cells and quiescent memory CD4⁺ T cells.

Treatment of transformed cell lines such as HeLa cells or Jurkat T cells with hexamethylene bisacetamide (HMBA) releases P-TEFb from the inhibitory 7SK RNP complex, and can thus enhance HIV-1 transcriptional elongation (47). Therefore, it has been suggested that this could be a mechanism by which P-TEFb availability and activity are limited in resting memory CD4⁺ T cells (48). Our data challenge this idea, as little P-TEFb can be found in the 7SK RNP complex in naïve CD4⁺ T and uninfected and latently infected resting memory CD4⁺ T cells. The low expression level of HEXIM1 in these cells appears to play a role in the low level of P-TEFb found associated in the 7SK RNP. Our results are in agreement with previous studies where we reported that little P-TEFb can be found associated in the 7SK snRNA-HEXIM1 complex in resting PBLs (peripheral blood lymphocytes) and resting CD4⁺ T cells; upon activation of the PBLs or resting CD4⁺ T cells, we observed a large increase in the level of P-TEFb associated with 7SK snRNA and HEXIM1 (49). It is therefore unlikely that association of P-TEFb in the 7SK RNP complex is a mechanism that drives the establishment and maintenance of latency in resting memory CD4⁺ T cells. On the contrary, it is likely that the sequestration of P-TEFb with the 7SK RNP acts as a reservoir of preactivated P-TEFb with a phosphorylated T loop from which cellular and viral effectors can extract it to stimulate transcription elongation. It was recently reported that P-TEFb can be competitively removed from the 7SK RNP complex by transcriptional activators such as the HIV-1 protein Tat and Brd4 (27, 50). Given that cellular gene expression is controlled at the level of transcriptional elongation, it is unsurprising that P-TEFb activity is tightly regulated. As we have proposed previously, sequestration of P-TEFb in the 7SK RNP complex may be a mechanism through which the cell maintains a pool of excess of P-TEFb so that this elongation factor is not limiting for productive transcription and is readily and quickly available (51). In the case of resting naïve CD4⁺ T cells and quiescent uninfected and latently infected memory CD4⁺ T cells examined in this study, the low level of P-TEFb in the 7SK RNP likely reflects the low transcriptional requirements of the cells.

The observations presented here also have implications for screening efforts geared toward the discovery of antilatency drugs (52–54). In particular, screening methods that use cells in division may favor the discovery of substances that elicit dissociation of P-TEFb from the 7SK RNP complex, for example, HMBA. In contrast, we predict that latency systems using resting memory cells will favor the discovery of antilatency substances that trigger increases in Cyclin T1 expression and/or CDK9 T-loop phosphorylation. We also predict that reagents favoring dissociation of P-TEFb from the 7SK RNP complex will have low or no activity in resting CD4⁺ T cells. Recent experiments in our laboratory have indeed corroborated this expectation, as HMBA had little or no effect in reactivating HIV-1 in our model of central memory T cells (A. Bosque and V. Planelles, unpublished findings).