Tat and Trans-activation-responsive (TAR) RNA-independent Induction of HIV-1 Long Terminal Repeat by Human and Murine Cyclin T1 Requires Sp1*

P-TEFb, cyclin T1 (cid:1) CDK9, is needed for the expression of cellular promoters and primate lentiviral long terminal repeats (LTRs). Curiously, cellular and lentiviral promoters differ dramatically in the requirements for positive transcriptional elongation factor (P-TEF) b activity. Lentiviral LTRs, but not cellular promoters, need an RNA-associated P-TEFb/Tat/TAR (trans-activa-tion-responsive) RNA ternary complex. Ternary complex defective murine cycT1 is apparently inactive for lentiviral transcription. Why P-TEFb requires Tat/TAR for LTRs but not for cellular promoters remains un-known. To explore this question, we sought to determine whether DNA targeting of murine and human cyclin T1 can reconstitute a Tat/TAR-independent activity to the HIV-1 LTR. In the absence of Tat and TAR, we found that both HuCycT1 and MuCycT1 can robustly activate the HIV-1 LTR. We further showed that Sp1 is necessary and sufficient for this DNA-targeted activity. Thus, like cellular promoters, HIV-1 LTR can use P-TEFb function without a Tat/TAR RNA complex. This activity

Recent findings suggest that P-TEFb also plays an essential role in the expression of mammalian and D. melanogaster genes (10,13,14,18,19). In these settings, P-TEFb must engage promoter-DNA independently of TAR RNA and must promote processive transcription without Tat. Indeed, it is a mystery why the cellular-promoter operative activity of P-TEFb does not apparently apply to the HIV-1 LTR. Here, we sought to determine whether DNA targeting of murine or human cycT1 to the HIV-1 LTR can bypass a requirement for Tat/TAR. We show that both murine and human cycT1 can interact with the Sp1 "A" domain to reconstitute Tat/TAR independent transcriptional activity and that Sp1 is necessary and sufficient to recruit cycT1 to the HIV-1 LTR. These results support the finding of P-TEFb in HIV-1 preinitiation complexes before the synthesis of TAR RNA (20) and help to explain the normal expression of HIV-1 LTR in rat cells that cannot form an RNA-bound P-TEFb (21,22). Thus, we propose that lentiviral LTRs, like cellular promoters, can use P-TEFb despite absence of Tat/TAR. This proposition concurs with recent suggestions that successful promoter recruitment of cycT1 is fully sufficient for HIV-1 LTR transcription (23).
Cell Culture, Transfection, and Reporter Assays-Cell propagation, transfection, and reporter assays were as described previously (28). All transfections were repeated three or more times and were normalized to ␤-galactosidase activity expressed from a cotransfected pCMV-␤ (Clontech). Average values Ϯ S.D. are shown. In all experiments, we have assayed both the HuCycT1 and the equivalent MuCycT1 plasmids. Because of space limitations, only one or the other set of results is shown.
* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Western Blotting, Immunoprecipitation, and Confocal Imaging-Western blotting and immunoprecipitation were performed as described previously (28). For confocal microscopy, HeLa cells were cultured on 25-mm coverslips (Thomas Scientific) and transfected with plasmid DNA. One day later, cells were fixed with 3.7% formaldehyde, permeabilized with PBS containing 0.1% Triton X-100, and incubated with monoclonal anti-HA antibody followed with anti-mouse conjugated to Alexa Fluor 488 (Molecular Probes, Eugene, OR). Nuclei were stained with 4,6-diamidino-2-phenylindole (Molecular Probes). Coverslips were mounted onto glass slides with ProLong antifade kit (Molecular Probes) and examined with a Leica laser-scanning microscope.

Robust Activation of Primate Lentiviral LTRs by cycT1 in the
Absence of Tat-The great preponderance of HIV-related P-TEFb studies have focused on transcriptional elongation in the context of a ternary P-TEFb/Tat/TAR RNA complex (20,24,29). Recent findings, however, demonstrate that P-TEFb activates numerous cellular promoters in Tat/TAR independent settings (18,30). Moreover, P-TEFb is in HIV-1 preinitiation complexes before the synthesis of TAR RNA (20, 31), suggesting a role distinct from its Tat/TAR RNA-bound form. Prompted by these observations, we sought to investigate the impact of P-TEFb on lentiviral LTR-expression in the absence of Tat/TAR.
Human and subhuman primate lentiviral LTRs conserve biological function (32) without preserving sequence identity. For example, although three or more copies of Sp1 binding motifs are found in all primate lentiviral LTRs, NF-B-binding sites are sporadically reduced in number or absent altogether in simian immunodeficiency (SIV) viral LTRs (Fig. 1A). Using three different primate LTRs, HIV-1 LTR (HXB2), and SIV LTRs (SIVmac and SIVsykes), we asked how each responds to trans-expression of either human cyclinT1 (HuCycT1) or mu-rine cyclinT1 (MuCycT1). When the appropriate LTR-luciferase reporters were cotransfected into HeLa cells with either HuCycT1-alone ( Previously, the salient difference between HuCycT1 and MuCycT1 was that the latter cannot form a functional P-TEFb/ Tat/TAR ternary complex (20,24,29). Interestingly, the current Tat-independent assays revealed equivalent activities for both cyclin T1s. Dose-dependent titrations further confirmed that MuCycT1 activates expression effectively from NF-Bcontaining and NF-B-absent LTRs (Fig. 1C).
Requirement for Sp1 in cycT1-mediated LTR Activation-The HIV-1 LTR contains several transcription factor binding sites, including AP-1, NFAT, NF-B, Sp1, and TAR. Results in Fig. 1A suggest that NF-B is not needed for cycT1-activation. To address the other motifs in the HIV-1 LTR, we employed several, otherwise isogenic, linker-scanning LTR-mutants ( Fig.  2A). In parallel assays, we found that the Sp1 motifs were strictly required for cycT1 activity, whereas AP-1, NFAT, NF-B, or TAR sequences were dispensable ( Fig. 2A). Hence, although Tat-dependent activity of cycT1 requires TAR, the Tatindependent activity does not.

The Sp1 A Domain Suffices for Functional and Physical
Interaction with cycT1-Sp1 is a 95-to 105-kDa transcription factor that binds DNA through its C-terminal zinc fingers (33,34). Previously, it was shown that Sp1 contributes importantly to the expression of the HIV-1 LTR (35) and that its serine/ threonine-and glutamine-rich "A" activation domain suffices for functional interaction with Tat (27,36). These observations together with above results (Fig. 2) lead us to ask whether the functional interaction between cycT1 and Sp1 is entirely circumscribed in the A domain. To address this, we transfected HeLa cells with a chloramphenicol acetyltransferase reporter (G3(Ϫ)43CAT) composed of three Gal4 binding sites positioned upstream of a TATA box with a downstream chloramphenicol acetyltransferase cDNA (Fig. 3A). To this reporter, we separately added Gal4BD-, Gal4Sp1A-, or Gal4E1a-plasmid expressing respectively the Gal4-DNA-binding-domain, a binding-domain-Sp1A-fusion, or a binding-domain adenovirus E1a fusion. These DNA mixes were cotransfected with or without a MuCycT1 expression vector (Fig. 3B). We found that neither Gal4BD (Fig. 2B, lanes 1 and 2) nor Gal4E1a (Fig. 2B, lanes 5  and 6) supported MuCycT1-activated expression over basal. By contrast Gal4Sp1A plus MuCycT1 mediated a Ͼ70-fold increase in expression (Fig. 2B, lanes 3 and 4), consistent with the Sp1A segment being wholly sufficient for function.
To confirm whether Sp1A-alone was sufficient to bind CycT1 The C Terminus of Cyclin T1 Interacts with Sp1A to Functionally Activate Expression-Human cyclin T1 is a 726-amino acid protein. Its N-terminal 1-290 amino acids contain a "cyclin homology box." This "box" binds cdk9 (39 -41) and mediates physical interaction with HIV-1 Tat. Indeed, a single cysteine at position 261 of HuCycT1 is critical for forming a human P-TEFb/Tat/TAR complex (12,17). Above, our results indicate a cycT1-Sp1 activity distinct from that of P-TEFb/Tat/TAR. We next examined whether the former uses a region of cycT1 different from that of the latter.
To verify that the C-terminal region of cycT1 is discretely sufficient for transcription, we fused a sub-portion, amino acids 419 to 726, to Gal4BD. Gal4HuCycT1 (419 -726) when targeted to promoter upstream Gal4-binding sites was sufficient for activating promoter expression (Fig. 5). Surprisingly, in paral-lel assays, Gal4HuCycT1 (1-444) and Gal4HuCycT1 (1-426), previously shown by others to be competent for cdk9-binding and transcriptional elongation when bound to Tat/TAR RNA, were inactive when targeted to DNA via Gal4-binding sites (Fig. 5, left). Because all Gal4HuCycT1 chimeras were expressed comparably (Fig. 5, right), one interpretation of these results is that simple recruitment of cdk9 insufficiently supports transcription from the HIV-1 promoter.
Sp1A-active HuCycT1s Localize to Nuclear Speckles- Fig. 4 defined forms of HuCycT1 that are highly active, moderately active, and inactive for cooperation with Sp1A. Previously, transcriptionally intact full-length HuCycT1 was found in nuclear dots/speckles (42). To check how Sp1A-competent forms of HuCycT1 might behave, we visualized using confocal microscopy to identify the subcellular location of seven HA-tagged mutants. Intriguingly, both highly active, (HuCycT1 (1-726),

Activation of HIV-1 LTR through an Sp1-dependent Mechanism
HuCycT1 (1-633), HuCycT1(240 -726, not shown), and HuCycT1 (300 -726)) and moderately active (HuCycT1 (419 -726) and HuCycT1 (500 -726)) cyclin T1s stained identically in nuclear speckles (Fig. 6). On the other hand, Sp1A-inactive HuCycT1 (1-444) and HuCycT1 (1-419), shown elsewhere to support P-TEFb/Tat/TAR function (24,39), were not in speckles and stained predominantly cytoplasmic. To the extent that nuclear speckling might be a requisite for transcriptional activity, the divergent patterns suggest that Sp1A-active portion of cycT1 dictates such physical localization. DISCUSSION HIV-1 expression occurs in two phases. After integration of the provirus into cellular chromosome, basal transcription from the LTR must initially drive the synthesis, however inefficiently, of a small amount of full-length 9.6-kb mRNA that provides for the synthesis of Tat. Tat then activates the LTR through TAR RNA (2) to further increase processive transcription. Over the past few years, P-TEFb (reviewed in Ref. 43) has emerged as an important cofactor for Tat/TAR-activated phase of LTR transcription. Thus, we understand the P-TEFb/Tat/ TAR ternary complex modulates transcriptional elongation by hyperphosphorylation of RNAP II CTD (12,14). Interestingly, although P-TEFb is present in elongation complexes, before the synthesis of TAR RNA, it is also found in HIV-1 LTR-engaged preinitiation complexes (20). Currently, despite much learned about Tat/TAR interactions with elongation-associated P-TEFb, little is known regarding how P-TEFb enters preinitiation complexes and the role that it might play in such moieties.
Here, we present evidence that P-TEFb may be recruited into preinitiation complexes through physical association of its cycT1 subunit with DNA-bound Sp1. Sp1 is a glutamine-rich promoter-proximal activator that interacts with TFIID to stabilize preinitiation complexes (44). We found that the 179amino acid Sp1A domain that contains its glutamine-rich se-quence is sufficient for physical (Fig. 3C) and functional (Figs. 3B and 4A) interaction with cycT1. Surprisingly, the portion of cycT1 delineated to cooperate with Sp1A mapped to a short, ϳ419 -633-amino acid C-terminal stretch (Figs. 4A and 6). This Sp1A-active region resides outside of the cycT1 cyclin homology box (i.e. amino acids 1-290) and associates with neither cdk9 nor Tat. Thus, unlike TAR RNA-bound P-TEFb, DNA-bound cycT1-Sp1A activity apparently does not use cdk9. Although provocative, this unexpected conclusion is consistent, in part, with findings that Sp1-driven transcription can occur through a CTD-kinase independent mechanism (45,46) and with evidence that productive and processive transcriptional elongation from the HIV-1 LTR can proceed without cdk9 (47). Currently, we do not rule out that in the absence of cdk9, other kinases such as CDK2 (48) may be involved.
It has been suggested elsewhere that cycT1 might be recruited to the HIV-1 preinitiation complexes through binding to NF-B and contact with RNAP II CTD (49,50). Although we have not directly examined interactions between NF-B and cycT1, several observations seemingly disfavor this interpretation. First, the natural phylogeny of primate lentiviral LTRs is such that many simian LTRs have either single or no NF-B binding sites (Fig. 1A). Thus, recruitment of cycT1 by NF-B cannot be a generally conserved mechanism for all primate lentiviruses. Second, our linker-scanning analyses ( Fig. 2A) showed that Sp1 is required, whereas NF-B is not, for DNAtargeted cycT1 activity. Other studies (40,51) have also independently suggested this Sp1-requirement. Hence, although we do not exclude that NF-B might recruit cycT1 to HIV-1 DNA, our current results show that NF-B is unnecessary, whereas Sp1 is necessary and sufficient. Intriguingly, in the two studies that proposed NF-B recruitment of cycT1, presence of Sp1-binding sites in the reporter constructs were required for such activity (49,50). Because NF-B binds the N-terminal cyclin homology domain of cycT1 (49) and Sp1 binds the C terminus of cycT1, it is possible, however, that in the natural context of the HIV-1 LTR, both factors cooperate in high affinity capture of cycT1. DNA-tethered cycT1 may then subsequently contact RNAP II CTD directly (50).
In addition to binding Sp1, the C terminus of cycT1 has another important activity. It is well known that transcription factors and actively transcribing RNAP II physically localize into nuclear foci with speckled patterns (52,53). Accordingly, a prerequisite for P-TEFb-activity is its entry into nuclear speckles (42,54). A priori, one might suppose that P-TEFb is directed to such a locale through the N-terminal Tat/TAR associative domain of its cycT1 component. Surprisingly, our data show that the Sp1A-interactive C terminus (500 -726 amino acids) of cycT1 dictates its localization into nuclear dots (Fig. 6). The N terminus (e.g. 1-444 amino acids) of cycT1, which binds cdk9, has no such function (Fig. 6). Hence, the first step of P-TEFb function, its constitutive (or Sp1-associated) migration to nu-clear speckles, seems to be determined through its C terminus. DNA recruitment of cycT1 by Sp1 serves also to address the long-standing observation that Tat function and HIV-1 transcription are apparently defective in rodent cells (29,55,56). Two recent studies have unexpectedly shown that HIV-1 is both transcriptionally intact and replication competent in rat cells (21,22), despite the fact that rat cycT1 lacks the human specific cysteine 261 residue (29,(55)(56)(57) necessary for forming a P-TEFb/Tat/TAR complex. Accordingly, in rat cells, the prototypic elongation-associated P-TEFb/Tat/TAR complex cannot form. How then can HIV-1 transcription occur successfully in rat cells? Although we have not examined rat cycT1 directly, we find in human cells that MuCycT1 and HuCycT1 are equally active in their Sp1-dependent activity. Thus, DNA-targeted cycT1 does not need cysteine 261 for LTR expression. Intriguingly in rat cells, we have observed that Tat-independent activity of MuCycT1 is actually considerably higher than that of HuCycT1. 2 Conceivably, a rat-specific cell factor may enhance DNA-based Sp1-cycT1 interaction to sufficiently bypass the need for RNA-tethered P-TEFb/Tat/TAR. Indeed active P-TEFb at cellular promoters functions effectively without Tat or TAR (43). We have recently observed that, consistent with the idea that Tat/TAR may not be absolutely necessary for lentiviral LTR and consistent with the proposal of Bieniasz and Cullen (58), overexpression of either HuCycT1 or MuCycT1 fully activates the transcription and synthesis of viral proteins from a Tat(Ϫ) HIV-1 provirus (data not shown).