INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Vitamin A is involved in a wide variety of normal processes, including immunity, differentiation, growth, reproduction, and vision (19, 32). The main dietary sources of vitamin A, retinol and the provitamin -carotene, are converted to a number of bioactive derivatives, including, among others, all-trans retinoic acid (RA) and 9-cis RA (9cRA). Cellular effects of these metabolites are mediated by nuclear receptors, which are ligand-dependent transcription factors (17-19, 37). Two families of retinoid receptors have been identified. The RA receptors (RAR, RAR, and RAR) bind to and are activated by both RA and 9cRA. The retinoid X receptors (RXR, RXR, and RXR) bind only 9cRA. Recent experiments have shown that RAR binds DNA as a heterodimer with RXR and activates transcription following ligand binding (18). Ligand-associated RXRs are thought to bind DNA either as homodimers or as heterodimers in association with RARs and other members of the steroid-thyroid hormone receptor family.

Several properties of human immunodeficiency virus type 1 (HIV-1)-induced disease are inversely correlated with a person's serum vitamin A level. Semba et al. (35) found that the rates of mother-to-child transmission of HIV-1 were increased over fourfold in women who were vitamin A deficient and that vitamin A deficiency was associated with a fourfold increase in risk of death in HIV-1-positive individuals (33, 34). Moreover, levels of dietary vitamin A have been shown to influence progression to AIDS (39), and supplementation can decrease HIV-1-associated morbidity in children (7). The association between vitamin A and HIV-1 disease progression cannot be attributed simply to overall malnutrition. In a study correcting for other factors, including energy intake, Tang et al. (39) still found that vitamin A deficiency held an elevated risk for AIDS mortality. Underscoring the complex nature of the relationship between vitamin A status and disease, these authors also found that a vitamin A intake of >20,000 IU/day increased the relative risk of AIDS mortality (39). This relationship between vitamin A status and disease is further complicated by the observation that AIDS patients are compromised in their ability to maintain vitamin A homeostasis (6, 12, 28).

The effect of vitamin A on HIV-1 disease could reflect either a direct effect of retinoids on virus gene expression or a general effect on immune function (32). Vitamin A has profound effects on hematopoeisis and is required for differentiation of cells of the myeloid lineage (24, 40, 42). Retinoids have also been shown to regulate HIV-1 gene expression in myeloid cells. We and others have shown that vitamin A directly modulates HIV-1 replication in myeloid cells (20, 30, 41, 43, 44). Pretreatment of the promonocytic cell line U937 with RA activated both HIV-1 and simian immunodeficiency virus from rhesus macaques (SIV_mac) long terminal repeat (LTR)-directed transcription approximately threefold by itself and up to 300-fold in synergy with phorbol 12-myristate 13-acetate (PMA) (20). The cis-acting sequences required for this synergistic activation were localized between nucleotides 50 and +1 of the SIV_mac promoter and 83 and +80 of the HIV-1 promoter. Activation required greater than 4 days of RA pretreatment. This lag, along with the demonstration that RA altered the pattern of proteins bound to 50 through +1 of the SIV_mac LTR, indicated that RA was modulating the expression of cellular factors.

Poli et al. (30) have shown that RA treatment stimulated HIV-1 replication in acutely infected U937 cells. Paradoxically, RA treatment of chronically infected U937 cells repressed PMA, interleukin-6, or granulocyte-macrophage colony-stimulating factor activation of HIV-1 (30, 44). The opposite responses of acutely versus chronically infected cells to RA might reflect the differentiation state of these cells. HIV-1 infection induces U937 promonocytic cells to differentiate to a more mature monocyte-like state (27, 29). If the response to RA proved to depend on the differentiation state of the infected cell, then in vivo, vitamin A should have therapeutic potential, since macrophages are the predominant myeloid cell type infected by HIV-1. Consistent with this prediction, RA inhibited HIV-1 replication in alveolar macrophages (44), and in monocyte-derived macrophages (MDMs) (30). In the latter case the RA effects were donor dependent.

In an effort to resolve these discrepancies and establish the anti-HIV-1 therapeutic potential of vitamin A, we have started to delineate the molecular mechanisms responsible for retinoid-mediated effects on HIV-1 gene expression. We show that physiological concentrations of either retinol, RA, or 9cRA repress HIV-1 LTR-directed expression in the THP-1 monocytic cell line. These retinoids prevented transactivation of the HIV-1 LTR by both NF-B and the viral protein Tat, reducing LTR-directed expression up to 200-fold. The cis-acting sequences required for repression are contained within the viral core promoter. Repression required retinoid pretreatment of the cells and was associated with a change in the pattern of cellular factors which bound to the core promoter.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Monocyte isolation and culture. Peripheral blood mononuclear cells (PBMCs) were purified from healthy volunteers by Ficoll gradient centrifugation. PBMCs were suspended in RPMI 1640 medium containing 25 mM HEPES (pH 7.4), 0.29 mg of L-glutamine/ml, 50 U of penicillin/ml, 50 µg of streptomycin/ml, 20% fetal bovine serum (FBS), and 10% pooled human AB serum. The serum stocks contained less than 0.3 endotoxin units per ml. All other reagents were endotoxin free. The cells were cultured overnight, after which nonadherent cells were removed, and adherent cells were cultured for an additional 4 days before virus infection. The cultures were infected with HIV-1_Ba-L (3,000 ng of p24/5 × 10⁵ cells) and maintained in medium without human serum. Virus replication was quantified either by measuring extracellular reverse transcriptase activity (9) or by p24 antigen enzyme-linked immunosorbent assay (ELISA) (Coulter). When indicated, retinoids were added from concentrated stocks (in ethanol) daily, beginning at the time of isolation of adherent PBMCs and continuing throughout the experiment. Cells not treated with retinoids received 0.01% ethanol to control for solvent effects.

Plasmids. PCR was used to isolate different regions of the HIV-1 LTR from HIV-1_LAV-1. These fragments were inserted upstream from the chloramphenicol acetyltransferase (CAT) gene and simian virus 40 polyadenylation sequences and cloned into pSP72 (Promega). p(139/+84)CAT includes the binding sites for the cellular transcription factors NF-B and Sp1 as well as the general transcription factor, TFIID. It also encodes TAR, the binding site for Tat. p(86/+84)CAT includes the binding sites for Sp1 and TFIID and encodes TAR. p(51/+84)CAT includes the binding site for TFIID and encodes TAR. p(86/+12)CAT includes the binding sites for Sp1 and TFIID but does not encode TAR.

Transfections and CAT assays. Mid-log-phase THP-1 monocytes were washed, suspended in RPMI 1640 (room temperature), and transfected by electroporation at settings of 300 V and 1,160 µF. Transfections included 8 µg of either an LTR-CAT or CMV-CAT reporter per 10⁷ cells. Some transfections also included 4 µg of pCMVcTat. Those transfections not including pCMVcTat contained pCMV5 instead in order to maintain a constant amount of DNA in each transfection. In some experiments, one-half of the transfected cells were treated with 50 nM PMA. Retinoid treatments were continued in those cultures that were initially pretreated with retinoids. Cellular extracts were prepared from the transfected cells after 20 h and assayed for CAT activity as previously described (20).

Flow cytometric analysis. THP-1 monocytes were washed two times with phosphate-buffered saline (PBS) and resuspended in RPMI 1640 containing 10% FBS (10⁷ cells/ml). MDMs were detached from tissue culture plates by incubating them in RPMI 1640 containing 5% FBS and 10 mM EDTA. The detached MDMs were washed two times with PBS and resuspended in RPMI 1640 containing 10% FBS (5 × 10⁶ cells/ml.) The washed cells were stained by incubation for 30 min at 4°C with either anti-CD11b (fluorescein isothiocyanate [FITC]), anti-CD11c (FITC), anti-CD14 (FITC), or anti-CD33 (phycoerythrin). The cells were washed two times with PBS and fixed in 1% paraformaldehyde. A minimum of 10,000 cells were analyzed for each sample with a Becton Dickinson fluorescence-activated cell sorter. The percentage of positive cells was calculated in comparison to isotype control stained cells.

Electrophoretic mobility shift and protein-DNA cross-linking assays. Cellular extracts and radiolabeled probes were prepared as described previously (20). Mobility shifts were performed with 15 µg of cellular extract and 0.35 ng of probe (approximately 30,000 cpm).

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Retinoic acid represses HIV-1 expression in THP-1 monocytes. We previously reported that all-trans retinoic acid and PMA synergistically activated HIV-1 expression in the promonocytic cell line U937. In agreement with these results, Poli et al. (30) reported that RA activated HIV-1 in acutely infected U937 cells but repressed HIV-1 replication in chronically infected U937 cells. Because HIV-1 induces U937 cells to differentiate to a more monocyte-like phenotype (27, 29) we reasoned that the different effects of RA might be related to the differentiation state of the cell. To test this, we transfected either untreated or RA-pretreated THP-1 monocytes (10⁷ M; 4 days) with a biologically active provirus clone, pHIV-1_NL4-3 (1). Untreated, transfected cells transiently produced HIV-1, as determined by p24 antigen capture ELISA. After 3 days cell-free supernatants contained over 300 pg of p24/ml. Treatment of THP-1 cells with RA for 4 days prior to transfection repressed virus expression to the extent that no p24 was detected (Table 1).

Physiological concentrations of retinoids repress HIV-1 replication in MDMs. We next determined whether retinoids could repress HIV-1 replication in primary MDMs. MDMs were prepared from PBMCs by overnight adherence to plastic and were grown either with or without RA for four additional days prior to infection with HIV-1_Ba-L. At 3-day intervals virus levels were quantified by assaying reverse transcriptase activity in the culture medium. In untreated cultures, reverse transcriptase was detected after 3 days, increased to peak levels at 15 days, and thereafter decreased. The decrease was associated with extensive cell death. RA pretreatment delayed the appearance of virus typically until days 6 to 9 and repressed total virus production at all time points (Fig. 1a). Moreover, there were no signs of cell death in the RA-treated cultures. In additional experiments we found that both RA and its precursor, retinol, showed antiviral activity over a wide range of concentrations, including those expected to correspond to plasma levels found in healthy individuals (Fig. 1b). However, in all cases retinoids had antiviral effects only when cells were pretreated. When added postinfection, retinoids failed to repress virus replication and in some instances slightly augmented replication.

View larger version (19K): [in this window] [in a new window]   FIG. 1.   Retinol and RA repress HIV-1 replication in MDMs. MDMs were grown either with or without retinoids for 5 days and then infected with HIV-1Ba-L, a macrophage-tropic virus isolate. After infection, retinoid treatments were continued in those cultures that were pretreated and begun in other cultures that were previously untreated (post-infection). Some cultures were left untreated and served as controls. (a) Virus production in untreated and RA-pretreated (109 M) MDMs was measured as cell-free reverse transcriptase activity. Each time point represents the average (± standard deviation) of three independent infections. (b) Levels of virus production 15 days after infection. Each sample represents the average (± standard deviation) of three independent infections. untr., untreated.

Retinoids repress HIV-1 LTR-directed expression in THP-1 monocytes. To determine if the antiviral activity of retinoids resulted from repressed transcription from the HIV-1 LTR, we examined the transient expression of HIV-1 LTR-CAT reporter plasmids in transfected THP-1 cells. When THP-1 cells were pretreated with RA for 4 days prior to transfection, Tat-activated expression was repressed more than 200-fold (Fig. 2a). Repression of LTR-directed expression was also seen when cells were pretreated for 4 days with either retinol or 9cRA (Fig. 2b).

View larger version (13K): [in this window] [in a new window]   FIG. 2.   Retinol and its metabolic derivatives RA and 9cRA repress HIV-1 LTR-directed expression in THP-1 monocytes. (a) Untreated or RA-treated (107 M; 4 days) THP-1 monocytes were transfected with p(139/+84)CAT, a plasmid containing a portion of the HIV-1 LTR (nucleotides 139 through +84) directing the expression of CAT. Some transfections also contained pCMVcTat (Tat), a plasmid expressing Tat. Levels of CAT activity were measured 16 to 18 h after transfection. The data are the averages (± standard deviations) of four independent transfections. Fold repression was calculated as the ratio of CAT activities measured from untreated cells to those in retinoid-treated cells. +, treated; , untreated. (b) Untreated () or retinoid-treated (retinol [ROL], 106 M, 4 days; RA [AT], 107 M, 4 days; 9cRA [9c], 107 M, 4 days) THP-1 cells were cotransfected with p(139/+84)CAT and pCMVcTat. Levels of CAT activity were measured 16 to 18 h after transfection. The data are the averages (± standard deviations) of three independent transfections. trt, treatment.

View larger version (20K): [in this window] [in a new window]   FIG. 3.   Time course of RA-mediated repression. (a) THP-1 monocytic cells were treated for the indicated times with 107 M RA. Both treated and untreated cells were then cotransfected with p(139/+84)CAT and pCMVcTat. Levels of CAT activity were measured 16 to 18 h after transfection. The data obtained from two independent experiments are shown. (b) THP-1 monocytes were treated as described above and then transfected with pCMVCAT containing the CMV immediate-early promoter directing CAT expression. Levels of CAT activity were measured 16 to 18 h after transfection. The data obtained from two independent experiments are shown.

View larger version (22K): [in this window] [in a new window]   FIG. 4.   Concentration dependence of retinoid-mediated repression. THP-1 monocytic cells were treated for 4 days with the indicated concentration of either RA (solid symbols) or retinol (open symbols) and then cotransfected with p(139/+84)CAT and pCMVcTat. Levels of CAT activity were measured 16 to 18 h after transfection. The data obtained from two independent experiments are shown. Circles, experiment 1; squares, experiment 2.

Repression of HIV-1 gene expression is not obligatorily coupled with cellular differentiation. Retinoids have been shown to modulate the differentiation of myeloid cells (24, 40, 42). Thus, the repressive effects of retinoids on HIV-1 gene expression might be due to changes in cellular differentiation that could be induced by other agents. We therefore examined whether PMA, which induces macrophage differentiation of THP-1 monocytes, would also repress HIV-1 gene expression. Both untreated and RA-treated (4 days; 10⁷ M) THP-1 monocytes were cotransfected with p(139/+84)CAT and pCMVcTat. After transfection, one-half of the cells were treated for 24 h with 50 nM PMA. RA treatment repressed HIV-1 LTR-directed CAT activity 90-fold (Fig. 5). In contrast, PMA, which induced these cells to become adherent (data not shown), repressed HIV-1 LTR-CAT activity 1.7-fold (Fig. 5). Combined treatment with RA and PMA repressed HIV-1 LTR expression to the same extent as RA treatment alone. These results indicate that repression is not a general property of differentiation but rather is associated with retinoid-specific changes in the cell. Because RA increased the expression of some macrophage markers on THP-1 monocytes, including CD11b and CD14, but decreased the expression of others, including CD11c and CD33 (Table 2), it is possible that repression was associated with a distinct retinoid-differentiated macrophage phenotype.

View larger version (20K): [in this window] [in a new window]   FIG. 5.   RA-mediated repression and THP-1 differentiation are not synonymous. Untreated or RA-treated (107 M; 4 days) THP-1 monocytes were cotransfected with p(139/+84)CAT and pCMVcTat. One-half of the transfected cells were also treated with 50 nM PMA. Levels of CAT activity were measured 24 h after transfection. The data are the averages (± standard deviations) of four independent experiments. +, treated; , untreated.

Retinoid repression requires sequences contained within the HIV-1 core promoter. As a first step in identifying monocyte/macrophage factors responsible for retinoid-mediated repression of HIV-1 gene expression, we constructed LTR-CAT reporter constructs containing various portions of the HIV-1 LTR. These included p(86/+84)CAT (including Sp1 binding sites and the Tat-responsive element, TAR), p(51/+84)CAT (Tat responsive but lacking Sp1 binding sites), and p(86/+12)CAT (including Sp1 binding sites but lacking TAR). RA repressed Tat-induced expression from both p(86/+84)CAT and p(51/+84)CAT and basal expression from p(86/+12)CAT (Fig. 6). RA repression of p(51/+84)CAT and p(86/+12)CAT indicates that the RA-sensitive element(s) lies between nucleotides 51 and +12 and that RA repressed activation by both upstream cellular (NF-B and Sp1) and downstream viral (Tat) activators.

View larger version (26K): [in this window] [in a new window]   FIG. 6.   RA represses HIV-1 expression from the core promoter. Untreated (; shaded bars) or RA-treated (107 M; 4 days) (+; solid bars) THP-1 cells were transfected with HIV-1 LTR-CAT reporter plasmids containing different portions of the HIV-1 promoter. p(86/+84)CAT includes the binding sites for Sp1 and TFIID and encodes TAR, the binding site for Tat. p(51/+84)CAT includes the binding site for TFIID and encodes TAR. p(86/+12)CAT includes the binding sites for Sp1 and TFIID but does not encode TAR. All transfections also contained pCMVcTat. Levels of CAT activity were measured 16 to 18 h after transfection. The data are the averages (± standard deviations) of at least three to four independent transfections.

Retinoid treatment is associated with a change in cellular factors which bind to the HIV-1 core promoter. Based on the above results, we predicted that RA modulated the binding of cellular factors to sequences between 51 and +12 of the LTR. To test this prediction, we compared protein-DNA complexes formed between THP-1 and RA-treated THP-1 extracts with end-labeled DNA fragments corresponding to either 51 to +84 or 51 to +12 of the HIV-1 LTR (Fig. 7a and b). With THP-1 extracts, a specific complex, B2, was detected following nondenaturing polyacrylamide gel electrophoresis. When we used RA-differentiated THP-1 cell extracts, an additional complex, B2RA, was seen, which migrated more slowly and replaced the B2 complex seen with untreated extracts. Slower-migrating B2RA complexes were also seen with extracts prepared from either retinol (10⁶ M; 4 days)- or 9cRA (10⁷ M; 4 days)-differentiated THP-1 cells (Fig. 7c). Therefore, the three retinoids we tested which repress HIV-1 LTR-directed expression induce similar changes in the binding of cellular factors to the LTR.

View larger version (26K): [in this window] [in a new window]   FIG. 7.   Retinoid treatment of THP-1 monocytes alters the pattern of factors which bind to the HIV-1 promoter. (a and b) Cellular extracts were prepared from both untreated THP-1 monocytes () and cells that were treated for 4 days with 107 M RA (+). Extracts were incubated with radioactive DNA probes corresponding to either nucleotides 51 through +84 (a) or 51 through +12 (b) of the HIV-1 promoter. Protein-DNA complexes were resolved on a nondenaturing polyacrylamide gel. Extracts from untreated cells formed a complex, B2, with the 51-through-+84 probe. A complex with a slower electrophoretic mobility, B2RA, was formed with extracts from RA-treated cells. Both complexes were specifically competed for by a 100-fold molar excess of unlabeled self DNA (comp) but not by a fragment containing the CMV immediate-early promoter (data not shown). Similar complexes were formed when the 51-through-+12 probe was used. (c) Cellular extracts were prepared from either untreated THP-1 monocytes () or cells treated for 4 days with either 107 M RA, 107 M 9cRA (9c), or 106 M retinol (Rol). Electrophoretic mobility shift assays were performed with a radioactive DNA probe corresponding to nucleotides 51 through +84 of the HIV-1 LTR. (d) Cellular extracts from either untreated () or RA-treated (107 M; 4 days) (+) cells were incubated with radioactive, bromodeoxyuridine-substituted DNA probes corresponding to either nucleotides 51 through +84 or 51 through +12 of the HIV-1 promoter. Bound proteins were cross-linked to DNA by UV light and then separated on either 10% (left) or 6% (right) SDS polyacrylamide gels.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

Freshly isolated PBMCs are refractory to HIV-1 infection but become permissive after in vitro differentiation into macrophages. Addition of physiological concentrations of retinoids, either retinol or retinoic acid, during differentiation results in MDMs that are nonpermissive for HIV-1 replication. However, retinoid treatment after this critical period has no effect on virus replication. Therefore, the presence of retinoids during differentiation induces or maintains expression of cellular factors that inhibit the replication of HIV-1. The inability of retinoids to block virus replication after this critical period argues against a direct effect of retinoids on HIV-1 gene expression.

This indirect effect of retinoids does not result from altered differentiation of either primary monocytes/macrophages or THP-1 cells per se. RA had no effect on differentiation marker expression on primary macrophages and activated some but repressed other differentiation markers on THP-1 cells (Table 2). Moreover, treating THP-1 cells with PMA, a potent inducer of macrophage differentiation, reduced HIV-1 gene expression only modestly (less than twofold, compared to 100- to 300-fold reduction by RA). Although both RA and PMA can induce THP-1 differentiation, the inductions do not result in equivalent macrophage phenotypes. For example, whereas PMA strongly induces THP-1 adherence, RA does so only weakly (data not shown) (22). Moreover, PMA induces interleukin-1 mRNA expression while RA does not (23). In contrast, RA induces lyn mRNA expression while PMA does not (22). It is therefore likely that RA affects macrophage differentiation in subtle ways not revealed by changes in traditional cell surface markers. Although RA is unable to induce complete differentiation of THP-1 cells, RA treatment does increase a number of properties typically associated with macrophages. In vitro, RA treatment of promonocytic cells favors granulocyte differentiation at the expense of macrophage differentiation (40, 42). However, RA does enhance monocyte/macrophage formation induced by other agents, including PMA and vitamin D₃ (24, 31, 38), raising the possibility that macrophages that differentiate and/or mature in the presence of retinoids have properties distinct from those of macrophages that differentiate in the absence of vitamin A. One consequence of these proposed effects of retinoids would be inhibition of HIV-1 gene expression and replication.

These conclusions are supported by the results of experiments utilizing the THP-1 monocytic cell line. Transient expression from the HIV-1 LTR is repressed in retinoid-pretreated THP-1 cells. High basal (Tat-independent) HIV-1 expression in THP-1 cells results from constitutively active NF-B (10). In the presence of retinoids, this high basal expression was reduced to levels below those seen in untreated THP-1 cells transfected with mutant LTRs lacking NF-B sites [p(86/+84)CAT)] (data not shown). In addition to reducing basal expression, retinoids also prevented transactivation of the LTR by the viral protein Tat. Therefore, retinoids repress both NF-B- and Tat-activated expression from the HIV-1 LTR, an effect relatively specific for the viral LTR, since similar repression of either the CMV immediate-early promoter or an NF-B-collagen promoter was not seen in control experiments (data not shown).

Three properties of retinoid-mediated repression of HIV-1 implicate the general transcription machinery as a primary target. First, the cis-acting sequences required for repression are located within the region of the LTR that includes the binding sites for components of the general transcription machinery. Retinoid-mediated repression mapped to sequences between 51 and +12. This region does not contain a sequence that is similar to the consensus for known retinoic acid response elements (PuGGTCA), nor do RARs and RXRs bind to this region in vitro (data not shown). Therefore, it is unlikely that repression results from the binding of ligand-bound retinoid receptors to the HIV-1 promoter. Second, retinoids block activation by multiple independent factors, including Sp1, NF-B, and Tat. Finally, retinoids virtually shut down HIV-1 expression. The magnitude of this repression (>200-fold) is consistent with retinoids interfering with the general transcription machinery.

Our data are most consistent with two models of repression. In one model, retinoids induce the expression of a negative factor that binds the HIV-1 core promoter. The binding of such a factor to the HIV-1 LTR should be detectable in biochemical assays. Consistent with this mechanism, we found that retinoid treatment altered the mobility of protein-DNA complexes and led to the presence of four new proteins that formed covalent protein-DNA structures following UV cross-linking. The identities and functions of these proteins remain to be determined. Several candidate proteins that bind to the HIV-1 core promoter and repress transcription have been described, but in most cases the magnitude of the repression of Tat activation by these factors was small or ineffective (8, 13, 21, 26, 45). One exception to this is the factor AP-4. By binding to E boxes that flank the HIV-1 TATA element, AP-4 can prevent TATA element binding protein from binding DNA (25). However, since no similar E boxes are present in the SIV_mac LTR, which is repressed by retinoids in an analogous fashion (20a), it is unlikely that AP-4 is responsible for the phenomena we describe here.

Alternatively, the retinoid-induced repressor could interfere with functional interactions between TFIID and transactivators, such as NF-B and Tat, without directly binding to the core promoter. These interactions are likely to be mediated by gene-specific TBP-associated factors (TAFs) and coactivator proteins. Evidence is accumulating that different promoters are bound by particular combinations of TBP and TAFs. In the case of HIV-1, switching the normal TATA element for one resembling that of the simian virus 40 early promoter inhibits Tat transactivation without affecting basal transcription levels (4). Therefore, RA might modulate the pattern of gene-specific TAFs or coactivator proteins that couple the HIV-1 core promoter to transactivators, such as Tat. Alterations in the functions of these factors could result from their altered expression, physical competition between them and the RARs, or interference with their functions by altered posttranslational modifications. There is precedence for retinoids modulating the expression of coactivator proteins which interact directly with both gene-specific activators and components of the basal transcription machinery, such as TFIID. Retinoic acid stimulates the expression of a protein in embryonal carcinoma cells which can complement the activity of the adenovirus coactivator, E1a (16). This protein, E1a-LA, is required for RA-regulated transcription from the RAR2 promoter and functionally interacts with both RAR and TFIID (3, 14). In addition, retinoic acid-mediated repression of AP-1 activity is thought to result in part from competition between RARs and AP-1 for the limiting coactivator CBP (11). Retinoids might also exert their effects through changes in the expression of extracellular signal-responsive protein kinases, including members of the protein kinase C family (5, 15). It is conceivable that the retinoid-mediated repression reported here involves altered phosphorylation of coactivators. Further experiments are needed to distinguish between these models.