The acute promyelocytic leukemia–associated protein, promyelocytic leukemia zinc finger, regulates 1,25-dihydroxyvitamin D₃–induced monocytic differentiation of U937 cells through a physical interaction with vitamin D₃receptor

Steroid and vitamin hormones play an integral part in the maturation and differentiation of hematopoietic cells. Members of the steroid and nuclear receptor superfamily, including the retinoic acid receptor (RAR), have an important role in this process, presumably through the ligand-dependent activation of genes that initiate or facilitate differentiation.1 The hormone 1,25-dihydroxyvitamin D₃(1,25(OH)₂D₃), a metabolite of vitamin D, can induce differentiation of hematopoietic cell lines such as HL-60 and U937 along a macrophage-monocyte lineage, including a decrease or cessation in their proliferation.2-6 The effects of 1,25(OH)₂D₃ on differentiation are mediated through its transcriptional modulation of key target genes by heterodimers consisting of the vitamin D₃ receptor (VDR) and the retinoid X receptor.7-10 During differentiation of hematopoietic cells, 1,25(OH)₂D₃ strongly induces expression of macrophage-monocyte markers, such as CD14, the lipopolysaccharide receptor, and the family of β₂leukocyte integrins including CD11a, CD11b, and CD11c.11 12 

In a search for target genes that might mediate differentiation of myelomonocytic cells by 1,25(OH)₂D₃, our laboratory previously identified the cell-cycle inhibitor p21^WAF1/CIP1 and the homeobox geneHoxA10 as direct transcriptional targets of VDR.7,9 Furthermore, exogenous expression of these genes in U937 cells was sufficient to induce differentiation.8Ligand-dependent activation of such target genes is thought to occur through recruitment of coactivator complexes, including p160–CREB-binding protein (CBP) and VDR interacting protein (DRIP).13 

Repression of VDR-dependent gene activation is not as well understood. Clearly, proper control of ligand-dependent expression of genes that mediate differentiation and cell-cycle arrest is an essential component of the function of normal, noncancerous cells. Aberrations in hormonal control of the differentiation process in hematopoietic cells occur in such cancers as acute promyelocytic leukemia (APL).

In APL, the M3 subtype of acute myelogenous leukemia, differentiation of cells is blocked during an intermediate promyelocyte stage of myeloid differentiation.14-18 These malignant promyelocytes can be induced to differentiate into mature polymorphonuclear leukocytes by providing pharmacologic doses of all-trans retinoic acid (ATRA). In contrast, these cancers are resistant to the effects of 1,25(OH)₂D₃. APL is thought to be caused by one of several reciprocal translocations involving RARα.19 One of these is t(11;17), which joins the promyelocytic leukemia zinc finger (PLZF) gene to RARα to create the PLZF-RARα fusion protein. PLZF–RAR-α expression is sufficient to block differentiation by 1,25(OH)₂D₃ or ATRA in both U937 and HL-60 cells.20 

Several groups21-24 characterized the molecular mechanism of PLZF-RARα resistance to ATRA. However, the mechanism by which VDR-mediated effects are interrupted has not been described. Thus, we investigated the function of the untranslocated PLZF protein with respect to 1,25(OH)₂D₃-dependent transcription and differentiation of the histiocytic lymphoma cell line U937 to gain a more thorough understanding of the pathogenesis of t(11;17)-derived APL.

The PLZF gene encodes a 673–amino acid protein that contains a broad complex, tram-trak, bric-a-brac/pox virus zinc finger (BTB/POZ) domain—regions thought to be critical for transcriptional repression by PLZF25 —and 9 Krüppel-like, Cys₂His₂ zinc fingers at its C-terminal that form the DNA-binding domain (DBD).26 The 118-amino acid BTB/POZ domain was shown previously to mediate protein-protein interaction and repression.21,27,28 PLZF is highly expressed in CD34⁺ bone marrow progenitor cells, but it is expressed only at low levels, if at all, in mature hematopoietic cells such as peripheral blood mononuclear cells or lineage-restricted, partly differentiated hematopoietic cell lines.26,29Accordingly, expression of PLZF is down-regulated in promyelocytic cells on differentiation treatment with retinoic acid.26PLZF has been shown to cause growth arrest and cell-cycle inhibition in myeloid cell lines.30 31 

Because 1,25(OH)₂D₃ is a potent inducer of monocytic differentiation32 and VDR is widely expressed in hematopoietic cells, we wondered whether PLZF might oppose this effect. We found that PLZF and VDR physically interact and that PLZF expression both abrogates the ability of VDR to activate transcription and impairs its ability to mediate differentiation of U937 cells. This study provides the first characterization of a protein that may regulate 1,25(OH)₂D₃-dependent differentiation of hematopoietic cells by blocking transcriptional activation by the receptor.

The histiocytic lymphoma cell line U93733 and its stably transfected derivative, PLZF45, were maintained as follows. Cells were grown at 37°C in 5% carbon dioxide in RPMI 1640 tissue culture medium (Gibco, Rockville, MD) supplemented with 10% fetal-calf serum (FCS; Gibco), 1 mM sodium pyruvate, 2 mM L-glutamine, and 100 units/mL penicillin and 100 μg/mL streptomycin. The stable cell line PLZF45 was maintained in this medium supplemented with 0.5 μg/mL puromycin, 0.1 μg/mL tetracycline, and 1.0 mg/mL G418. The U937 cells were used in all transient transfections as described previously.34 After transfection, ligand (1,25(OH)₂D₃ [1 × 10⁻⁸ M], ATRA [1 × 10⁻⁶ M]), or trichostatin A (TSA [150 ng/mL]), or a combination of these, was added for 24 hours. Luciferase assays were done with 10 μg whole-cell extract prepared from harvested cells by using the Promega (Madison, WI) luciferase assay system. Where applicable, luciferase values were normalized to β-galactosidase activity as described previously.35 

U937T cells, a line of U937 cells stably expressing tet-virion protein (VP) 16 under the control of the tet-operator promoter,36 were used to produce a tetracycline-inducible PLZF cell line.37 Briefly, PLZF(B) complementary DNA26 was cloned into the EcoRI site of pTRE (pUHD10-3) to yield pUHD-PLZF. Ten micrograms pUHD-PLZF and 1 μg pIND (Invitrogen, Carlsbad, CA) as a neomycin-selection marker were linearized and cotransfected into 2 × 10⁷ U937T cells by electroporation (Bio-Rad, Hercules, CA) in 400 μL additive-free RPMI at 960 μF and 0.17 kV. After 24 hours, cells were plated into methylcellulose containing 0.1 μg/mL tetracycline, 0.5 μg/mL puromycin, and 10% FCS with 1 mg/mL G418, and resistant clones were isolated.

Nuclear receptor proteins or domains were fused to glutathione-S-transferase (GST) and expressed inEscherichia coli. Putative interacting partners were translated in vitro (TNT kit; Promega) in the presence of sulfur 35 (³⁵S)–methionine (Amersham, Piscataway, NJ). GST fusions and translated proteins were incubated in buffer A (170 mM potassium chloride [KCL], 20 mM Tris [pH 7.9], 20% glycerol, 0.2 mM EDTA, 4 mg/mL bovine serum albumin [BSA], 0.05% NP-40, 0.1 mM phenylmethylsulfonyl fluoride [PMSF], and 1 mM dithiothreitol [DTT]) for 3 hours before 3 washes in buffer B (same as buffer A but without BSA). Beads were resuspended in sodium dodecyl sulfate (SDS) sample buffer, separated by electrophoresis through 7.5% to 10% SDS polyacrylamide gels, dried, and exposed overnight at room temperature to Bio-Max film (Kodak, Rochester, NY).

Nuclear extracts were prepared from U937-PLZF45 cells as described previously.38 Then, 200 μg nuclear extract was resuspended in buffer A.1 (170 mM KCl, 20 mM Tris [pH 7.9], 10% glycerol, 0.2 mM EDTA, 0.05% NP-40, 0.1 mM PMSF, and 1 mM DTT) to a final volume of 445 μL. Extracts were precleared with normal rabbit serum (Santa Cruz Biotechnology, Santa Cruz, CA) and 50 μL protein A/G plus agarose beads (Santa Cruz Biotechnology) for 1 hour and then combined with 50 μL protein A/G plus beads, antibody, and either ethanol or 1,25(OH)₂D₃ (final concentration, 1 × 10⁻⁷ M) per reaction). Reactions were carried out at 4°C for 2 hours and were followed by washing in buffer A.1 to 200 mM KCl. Washed pellets of beads were resuspended in SDS sample buffer, separated by electrophoresis through 7.5% to 10% SDS polyacrylamide gels, dried, and transferred to nitrocellulose filter paper. The filters were subsequently immunoblotted for various proteins by using standard procedures. SKNO-1 cells were cultured in RPMI 1640 with 10% FCS. Cells (1 × 10⁷) were lysed for each immunoprecipitation experiment. A PLZF mouse monoclonal antibody (IgG_2a isotype)30,39 and isotype control preimmune mouse monoclonal IgG_2a (Jackson Immunoresearch, West Grove, PA) were used at a concentration of 1 mg/mL. Subsequent steps in the immunoprecipitation were done as described previously.40 The precipitated proteins were electrophoresed through a 12% SDS-polyacrylamide gel and transferred to an Immobilon P membrane (Millipore, Bedford, MA). Immunoblotting was done by using standard methods.

Electrophoretic mobility shift assay (EMSA) was done as described previously41 by using 5 μg nuclear extract as described above. Cold specific and nonspecific competitor oligonucleotides were added to the reactions in 10 times the concentration of specific radioactive oligonucleotide. Antibody supershifts were done by adding anti-VDR monoclonal antibody or anti-PLZF polyclonal serum to the reactions. The vitamin D₃ response element (VDRE) from the mouse osteopontin gene (OPN; also called Spp1) was generated as complementary oligonucleotides of the sequence 5′-GATCCACAAGGTTCACGAGGTTCACG-3′ (top strand). The oligonucleotide used as the nonspecific control was a glucocorticoid response element of the top strand sequence 5′-GATCCGACCGAGAACAAGATGTTCTGTCGAG-3′.

PLZF45 cells were washed in phosphate-buffered saline (PBS), transferred to medium without tetracycline, and allowed to grow for 16 hours before addition of 1,25(OH)₂D₃(10⁻⁸ M), ATRA (10⁻⁶ M), or ethanol and dimethyl sulfoxide. At various times, 1 × 10⁶ cells were obtained for flow cytometric analysis. Each sample was washed twice in ice-cold PBS and 1% BSA and incubated in 150 μL cold PBS and 1% BSA with 1.5 μL of either IgG1–fluorescein isothiocyanate conjugated (FITC) and IgG1– R-phycoerythrin or IgG1-FITC anti-CD14 or anti-CD11c (Caltag, Burlingame, CA) for 45 minutes. After incubation, the cells were washed in PBS and 1% BSA 3 times; resuspended in 400 μL PBS, 1% BSA, and 0.5% paraformaldehyde; and stored at 4°C. Flow cytometry was done on a fluorescence-activated cell-sorter scanner (FACS Calibur; Becton Dickinson, Mountain View, CA).

At various times during the PLZF-1,25(OH)₂D₃-ATRA induction experiment, 2 × 10⁵ cells were removed from each growth condition, cytospun onto microscope slides at 600 rpm for 10 minutes, air dried, and fixed in 100% methanol for 5 minutes. They were then stained with Giemsa staining (Hema-tek 2000; Bayer Corp, Pittsburgh, PA). Several fields on each slide were examined, and the cell dimensions were measured from digitized images by using National Institutes of Health Image 1.62 software. Total-cell and nuclear areas were determined, and the percentage of each cell occupied by the nucleus was derived for each cell. Ten cells were measured, and the average nuclear-area percentage was calculated for each condition.

PLZF-RARα expression can block 1,25(OH)₂D₃-mediated monocyte differentiation.20 To determine the possible molecular mechanism for this activity, we used PLZF-RARα, PLZF, and VDR to transfect U937 cells, along with a reporter gene containing a consensus VDRE, a 6–base pair direct repeat with a spacing of 3 nucleotides between the half sites. As shown in Figure1A, exogenous expression of either PLZF or PLZF-RARα inhibited VDR-mediated activation of a luciferase reporter (lane 1 versus lanes 2 and 3), regardless of the presence of the histone deacetylase (HDAC) inhibitor, TSA (lane 4 versus lanes 5 and 6). PLZF-RARα and PLZF were expressed at similar levels (data not shown).

Because PLZF alone was capable of significantly blocking VDR-dependent transactivation, we subsequently focused on PLZF. To test whether the repression of VDR activity by PLZF was a result of general transcriptional repression, PLZF was cotransfected with GAL-VP16 and a GAL operator–containing responsive reporter in U937 cells. As shown in Figure 1B, activation of the GAL reporter by VP16 was not affected by exogenous expression of PLZF, indicating that PLZF does not generally squelch transcription. Additionally, coexpression of PLZF with VDR did not affect VDR expression levels (data not shown). For comparison, PLZF was also cotransfected with RARα and a reporter containing the RARβ₂ RARE. Figure 1C shows that exogenous expression of PLZF did not affect ATRA-dependent activation of this reporter. Thus, we found that PLZF repression of transactivation is selective for VDR.

We addressed the question of whether the PLZF-mediated repression of VDR function was due to a physical interaction between the proteins by conducting in vitro pull-down assays to determine whether³⁵S-labeled PLZF could bind to GST-VDR or GST-RARα in the presence and absence of 1,25(OH)₂D₃ or ATRA, respectively (Figure 1D). In vitro, PLZF associated with VDR independent of ligand (lanes 3 and 4) but did not associate with GST-RARα in either the presence or absence of ATRA (lanes 5 and 6). These findings show that the ability of VDR and PLZF to interact is correlated with the selective loss of VDR-mediated transactivation in the presence of PLZF.

The results of the in vitro interaction assay were confirmed by coimmunoprecipitation studies. Stable U937 cells capable of induced expression of PLZF on removal of tetracycline were used to assess an interaction between endogenous VDR and PLZF in a nuclear extract. Figure 2A shows the induction of PLZF expression after 36 hours of culture in medium without tetracycline. Nuclear extracts were prepared from cells in the presence and absence of tetracycline (the absence and presence of PLZF, respectively) and used in a coimmunoprecipitation assay with antibodies to both VDR and PLZF. The extracts (with and without PLZF) were precipitated with either anti-VDR or anti-PLZF antibodies and immunoblotted with the opposite antibody. As shown in Figure 2B, in the absence of PLZF, neither antibody precipitated the reciprocal protein. However, in the presence of PLZF, anti-VDR and anti-PLZF coprecipitated PLZF and VDR, respectively. As in the in vitro pull-down assays, the interaction between PLZF and VDR occurred independently of 1,25(OH)₂D₃.

The physiologic nature of the PLZF-VDR interaction was explored further in the M2 leukemia cell line SKNO,42 in which both PLZF and VDR are expressed endogenously. VDR could not be detected by direct immunoblotting from SKNO lysates (Figure 2D, lane 2). However, coimmunoprecipitates with the PLZF monoclonal antibody reacted positively (6.1-fold increase over lysate alone with the IgG signal subtracted) when probed with an anti-VDR antibody (Figure 2D, lane 4). These results show that antibodies to PLZF and VDR can reciprocally immunoprecipitate both proteins and that this interaction occurs even at the low endogenous protein levels in SKNO cells.

Because of the apparent ligand-independent nature of the interaction between PLZF and VDR, and given that PLZF can interact with the corepressors SMRT–N-CoR, HDAC-1, and Sin3A,21,28,43 44 we wondered whether VDR associates with components of a corepressor complex through its association with PLZF. As shown in Figure 2C, endogenous Sin3A was clearly immunoprecipitated by anti-PLZF antibodies in the presence of PLZF (lanes 4 and 5) but not by anti-VDR antibodies (lanes 1 and 2). Similar results were observed when VDR coimmunoprecipitations were blotted for SMRT and HDAC-1 (data not shown). Thus, although VDR interacted with PLZF and PLZF interacted with Sin3A and other repressors, a 3-way interaction was not detected.

To determine how PLZF might block transcriptional activation by VDR, an in vitro protein affinity assay was used to identify the domains in PLZF and VDR responsible for the interaction. Figure3A depicts a series of GST fusions containing all or part of VDR. The ³⁵S-labeled, in vitro–translated PLZF was incubated with the individual VDR derivatives (Figure 3B). In vitro–translated PLZF interacted to a significant extent with full-length VDR (Figure 3B, lane 3) but also with the GST-VDR DBD (GST-VDR^DBD; Figure 3B, lane 6). It did not interact with GST alone, the GST-VDR ligand-binding domain (GST-VDR^LBD, or the hinge region (GST-VDR^hinge) (lanes 2, 4, and 5). Consistent with the inability of PLZF to interact with GST-VDR^LBD, a GAL-VDR^LBD construct that strongly activated transcription from a GAL-UAS reporter in response to 1,25(OH)₂D₃ was not affected by coexpression of PLZF in U937 cells (data not shown).

An interaction with VDR by means of its DBD could explain why repression by PLZF is ligand independent. To address whether the interaction with the VDR DBD was specific, we conducted a pull-down study using the VDR DBD and another nuclear receptor DBD, that from the orphan receptor HNF4. PLZF interacted with GST VDR^DBD but not with GST-HNF4^DBD (Figure 3B, lanes 6 and 7), thereby indicating that the interaction is specific for VDR and not the result of promiscuous interaction with a hormone receptor zinc finger region.

We used a similar analysis to determine the domains of PLZF protein that facilitate VDR interaction. Figure 3C depicts a series of PLZF mutants containing deletions covering the BTB/POZ domain (PLZFΔBTB/POZ]), a second repression domain (PLZFΔRD2]), and both the RD2 and zinc finger regions (BTB/POZ). The PLZF BTB/POZ domain was implicated as being responsible for protein-protein interactions in several other systems.27 We translated PLZF and derivatives and allowed them to interact with GST-VDR. Deletion of either the RD2 domain or both the RD2 and zinc finger domains of PLZF did not affect binding to VDR (Figure 3D). In contrast, deletion of the BTB/POZ domain completely abrogated binding of PLZF to VDR. Thus, it appears that the major surface responsible for VDR interaction lies within the BTB/POZ domain of PLZF.

We correlated the ability of various PLZF mutants to interact with VDR with their ability to repress VDR-dependent transactivation of cotransfected reporter constructs in U937 cells. Deletion of the BTB/POZ domain abrogated the ability of PLZF to repress VDR transactivation (Figure 4A, lanes 4 and 5) relative to full-length PLZF (lanes 2 and 3). Thus, we found that, in vivo, the BTB/POZ domain plays an important role in the functional repression of VDR, presumably through its ability to physically associate with the receptor. Interestingly, an epitope-tagged version of the BTB/POZ domain did not repress VDR-dependent transcription (Figure 4B). Furthermore, deletion of the zinc finger region of PLZF partly relieved the repression of VDR transactivation (Figure 4C), whereas a deletion of the RD2 domain retained the ability to repress VDR (Figure 4D). Therefore, we found that although the BTB/POZ domain is necessary and sufficient for the interaction of PLZF with VDR, it cannot on its own repress transcription, a result that suggests the existence of additional contributions from other moieties of PLZF.

Our observation of a physical interaction between the VDR DBD and PLZF led us to consider that PLZF may affect the ability of VDR to bind its hormone response element or affect the nature of the DNA-bound complex so that it cannot activate transcription. To explore this issue, nuclear extracts prepared from U937-PLZF cells, grown in either the presence or absence of tetracycline to regulate expression of PLZF, were used in an EMSA (Figure 5). The probe used in this assay was a direct repeat with a spacing of 3 nucleotides (DR-3) derived from the VDRE of the mouse OPNgene promoter and was identical to the canonical DR-3 in the reporter construct used in the transfection experiments. As shown in Figure 5A, addition of extract yielded at least 4 shifted species (lane 2), which showed specific competition with the addition of cold excess VDRE but not a nonspecific competitor (lanes 3 and 4). In extract containing PLZF, there was a significant change in shift mobility (lane 5). This shift showed competition with the specific oligo (lane 6) and was not affected by the nonspecific oligo (lane 7). Addition of an anti-VDR antibody confirmed that these shifted species contained VDR. Anti-VDR antibodies could compromise binding (open arrow) or alter the mobility of several of the major species on the gel and induce a supershift (asterisk in lane 9). In the presence of PLZF, addition of anti-VDR antibody disrupted the large lower shift (solid arrow) and a high-molecular-weight shift appeared (2 asterisks in lane 11). The disappearance of a significant specific shift in the presence of the anti-VDR antibody in lanes 9 and 11 indicates that, in both cases, VDR was present in the complex but the complex was altered by the presence of PLZF. The most supershifted complexes in lanes 9 and 11 appear to have different mobilities, with lane 11 showing slightly more delay, a finding consistent with the presence of PLZF in the DNA-protein complex.

To determine whether PLZF was present in the DNA-bound complex, we conducted similar antibody additions using a monoclonal antibody to PLZF (Figure 5B). Addition of the anti-PLZF antibody did not affect the shift pattern observed when PLZF was not present (lanes 12 and 13). In the presence of PLZF, increasing amounts of anti-PLZF antibody abrogated both the upper and lower prominent shifts observed in the absence of the antibody (lanes 16 and 17 versus lane 14). The data indicate that PLZF is present in the VDR-VDRE complexes. When coupled with the data on in vitro interaction and transfection described earlier, these findings suggest that PLZF and VDR from nuclear extracts can form complexes on a VDRE and that these complexes could be responsible for the inability of VDR to transactivate in the presence of PLZF.

The stable, tetracycline-inducible U937 cell line PLZF45 was used to assay the effects of PLZF expression on monocytic differentiation induced by 1,25(OH)₂D₃ and, in comparison, ATRA. Both 1,25(OH)₂D₃ and ATRA induce differentiation of hematopoietic precursor cells as measured by surface-antigen expression and changes in cellular morphologic features. PLZF45 cells were cultured in medium without tetracycline, inducing PLZF expression; treated with 1,25(OH)₂D₃ or ATRA for 48 hours; and harvested for analysis of cell-surface antigens by flow cytometry (Figure6). The fold change in mean fluorescence of the monocytic differentiation marker CD14 was measured as a function of PLZF expression and treatment with 1,25(OH)₂D₃ or ATRA (Figure 6A). PLZF alone minimally induced expression of CD14 over background levels (lane 2). ATRA did not induce expression of CD14 (lane 3). CD14 expression was induced by 1,25(OH)₂D₃, to 20 times the levels in uninduced cells, and this induction was substantially reduced by PLZF (Figure 6A, lane 5 versus lane 6). In contrast, PLZF expression did not block ATRA-induced differentiation, as measured by the monocyte-granulocyte marker CD11c after 48 hours of treatment with ATRA (Figure 6B, lane 3 versus lane 4) or alone (lane 1 versus lanes 2 and 6); instead, it appeared to contribute to induction with ATRA. PLZF also did not block induction of CD11c family members CD11b and CD11a or of CD18 (data not shown). Thus, PLZF appears, possibly through a specific block of VDR function, to alter CD14 but not CD11c surface-marker expression. These results suggest that PLZF may alter very specific 1,25(OH)₂D₃-induced monocytic differentiation pathways.

As multipotent precursor cells progress through monocytic differentiation, they undergo morphologic changes. One morphologic indicator of differentiation is a reduction in the volume of the nucleus volume relative to that of the cytoplasm. We used the PLZF45 cell line to assess whether PLZF expression would alter 1,25(OH)₂D₃-dependent changes in cell morphologic features (Figure 7A). For U937T-PLZF45 cells without either 1,25(OH)₂D₃treatment or PLZF expression, nuclei occupied 71.5% ± 2.0% of total-cell area (Figure 7B, lane 1). Expression of PLZF (Figure 7A, panel ii) did not cause a significant decrease in nuclear size. In contrast, treatment with 1,25(OH)₂D₃resulted in a striking decrease in nuclear area, to 45% ± 3.0%, indicative of a differentiating cell (Figure 7A, panel iii). Expression of PLZF during treatment with 1,25(OH)₂D₃blocked the shrinking of nuclear volume (Figure 7A, panel iv, versus Figure 7A panel iii, and Figure 7B; 61.0% ± 3.0% versus 45.0 ± 3.0%, respectively). Unlike results observed with 1,25(OH)₂D₃ treatment, PLZF expression had little effect on the morphologic changes resulting from ATRA treatment (Figure 8A). Treatment with ATRA resulted in a 13% reduction in nucleus-to-cytoplasm area (Figure 8B, lanes 2 and 3), a change that was not affected by the presence of PLZF (Figure 8B, lanes 3 and 4). These results, along with the observed effects of PLZF expression on 1,25(OH)₂D₃-induced CD14 surface expression, strongly suggest that PLZF is a potent and specific regulator of 1,25(OH)₂D₃-dependent monocytic differentiation.

We were concerned that the well-documented ability of PLZF to cause cell-cycle arrest might affect the differentiation results shown in Figures 6 to 8. Table 1 shows the cell-cycle profile of PLZF45 cells during the 1,25(OH)₂D₃-dependent treatments illustrated in those figures. At the times described, PLZF induced a G₁arrest similar to that induced by 1,25(OH)₂D₃alone (71.7% versus 72.6%). However, PLZF did not induce CD14, nor did it change the nucleus-to-cytoplasm ratio. PLZF expression also did not inhibit ATRA induction of CD11c or ATRA-mediated morphologic changes. These results strongly suggest that, although PLZF does induce a G₁ arrest, it neither blocks all hormone-induced differentiation nor induces general differentiation on its own.

Our study demonstrated that PLZF blocks transcriptional activation of a canonical VDRE reporter construct, revealed a physical interaction between VDR and PLZF, and showed physiologic effects on VDR-dependent processes by PLZF. To more clearly define the role of PLZF in VDR-mediated differentiation, we examined its effect on a direct VDR target gene. Our laboratory previously identified the cyclin-dependent kinase inhibitor p21^WAF1/CIP1 gene as a direct transcriptional target of VDR.7 Expression of p21 was also shown to be sufficient to begin differentiation of U937 cells along a macrophage-monocyte lineage. Using a 2.4-kb fragment of the p21 promoter cloned upstream of the luciferase gene (Figure9A), we here observed that coexpression of VDR and PLZF resulted in a reproducible attenuation of 1,25(OH)₂D₃-mediated activation of p21 transcription (Figure 9B). RARα also can directly transactivate through the p21 promoter.45 Expression of PLZF did not significantly affect the activation of the p21 promoter in response to ATRA and RARα. These data support the results of earlier transfection experiments and protein-protein interaction assays, thereby indicating that PLZF has a selective role in VDR-dependent activation. The findings show a strong correlation between p21 activation and possibly other VDR-dependent target genes and their attenuation by PLZF that results in an alteration of the differentiation course of a cell.

In this study, we found that PLZF is a potent repressor of VDR-dependent transcriptional activation. At the molecular level, this repression appears to be due to a selective physical interaction between VDR and PLZF. We showed that PLZF and VDR can be coimmunoprecipitated from a hematopoietic precursor cell line and that this interaction appears to occur through the DBD of VDR. Consistent with these results, PLZF was detected in a VDR-VDRE–bound complex in vitro, suggesting that the PLZF-mediated block of VDR transactivation might be due to a PLZF-induced alteration in the nature of the receptor-DNA complex that is unfavorable for VDR-dependent transcriptional activation.

The BTB/POZ domain of PLZF plays an essential role in both the interaction of PLZF and VDR and subsequent repression of transcription.46 In studies in Drosophila, proteins containing BTB/POZ were shown to regulate gene expression through local control of nucleosome positioning and remodeling.47 The BTB/POZ domain of PLZF-RARα APL fusion protein was also implicated in transcriptional repression of ATRA-induced genes.22-24,44 PLZF is highly expressed in immature but not differentiated hematopoietic cells and may play a role in the maintenance of uncommitted precursor cells.19,29-31VDR is widely expressed in hematopoietic cells,32 and we showed that endogenous PLZF and VDR can interact in an undifferentiated leukemic cell line. This suggests that the interaction of PLZF and VDR may reflect a facet of normal cellular physiologic characteristics. Therefore, we propose that, in undifferentiated cells, PLZF could repress VDR function through the BTB/POZ domain.

The BTB/POZ domain is able to associate with several corepressors, notably Sin3A, HDAC1, SMRT, and N-CoR.21-24 28 Many researchers have found that constitutive association of these repressors with PLZF-RARα is responsible for the resistance to ATRA therapy in patients with t(11;17)-derived APL. Surprisingly, our data suggest that repression of VDR is not due to the association of Sin3A, HDAC1, or SMRT (Figure 2 and data not shown). In supporting of this idea is our finding that repression of VDR by PLZF was not sensitive to the addition of TSA (Figure 1A). We could not coimmunoprecipitate Sin3A with anti-VDR antibodies by means of a bridging PLZF molecule, despite the fact that anti-PLZF antibodies coimmunoprecipitated Sin3A effectively (Figure 2C). This result indicates that the fraction of PLZF associated with VDR may not be able to also associate with Sin3A. Our data indicate that deletion of the BTB/POZ domain abolishes the interaction of PLZF with VDR and that the BTB/POZ domain alone can interact with VDR. It is conceivable that VDR interacts with the same or an overlapping region of the BTB/POZ domain as the corepressors and that is why we could not show a 3-way interaction among VDR, PLZF, and the corepressors. We conclude that the mechanism of inhibition of VDR by PLZF does not involve corepressors that work though HDAC mechanisms.

The BTB/POZ domain of PLZF is an obligate homodimer,25,46and we found that when bound to its cognate site, PLZF forms a high-molecular-weight complex that may contain as many as 4 PLZF molecules.48 This information suggests that several molecules of PLZF could form a complex with the VDR when the 2 proteins interact. Hence, PLZF may not block VDR action by recruitment of corepressors but may instead work by steric blockade of the VDR.

PLZF, like many of members of its family, contains not only a BTB/POZ domain but also 9 Krüppel-like zinc fingers (Cys₂His₂). We showed previously that PLZF can bind a specific DNA site and repress transcription of reporter genes.48,49 Our data suggest that PLZF is present in a DNA-bound complex with VDR. Although PLZF did not bind the VDRE probe used in the assay (data not shown), the transfection data suggest that the zinc fingers are partly responsible for repression of VDR transactivation (Figure 4C). In a previous study,50 the more N-terminal zinc fingers of PLZF-RAR were found to be required for the dominant negative effects of the fusion protein on RARα, thereby implicating these zinc fingers in protein-protein interactions. We conclude that the combined actions of the BTB/POZ domain and zinc finger region of PLZF are responsible for repression of VDR-mediated transcriptional activation.

Through DNA binding to target promoters, VDR activates target genes, affecting a variety of biologic processes, including differentiation of hematopoietic precursors. Many studies have examined this phenomenon by assessing induction of surface-marker antigens known to indicate a particular cell type. CD14 is a marker of 1,25(OH)₂D₃-mediated monocytic differentiation. We found that contemporaneous expression of PLZF dramatically reduced the number of CD14 molecules expressed per cell, reflecting an alteration in monocytic differentiation. CD14 and CD11c have traditionally been regarded as having similar function (inflammatory and phagocytic response to lipopolysaccharide stimulation) and cell-type expression patterns. However, studies have shown that CD14, but not CD11c, plays a significant role in noninflammatory phagocytosis of apoptotic cells.51 Our data suggest that PLZF selectively regulates the ability of VDR to induce CD14 compared with CD11c (and its family members). Although this idea is unproved, we speculate that PLZF may function to regulate highly specific lineage commitments in precursor hematopoietic cells in response to hormonal signals.

We also showed that morphologic changes associated with monocytic differentiation that occurred when U937 cells were treated with 1,25(OH)₂D₃ failed to occur when PLZF was present. PLZF may inhibit essential VDR target-gene activation, resulting in an inability to differentiate. Indeed, we also observed that PLZF attenuated 1,25(OH)₂D₃-induced activation of p21, a factor with an important role during initiation of differentiation in hematopoietic cells. Interruption of transcription of p21 and other VDR target genes9 could result in a strong block of the ability of VDR to mediate differentiation of precursor hematopoietic cells. In contrast, PLZF did not interact with RARα, and it did not inhibit RARα transactivation or the ability of ATRA to induce differentiation of U937 cells.

PLZF was shown previously to be growth suppressive in tumor cell models and may participate in the establishment and maintenance of a state of quiescence in developing mammalian tissues, including maintaining immature hematopoietic precursors.29,30,49,52,53 PLZF has been thought to show this growth-suppressive effect by repressing specific growth-promoting target genes that contain a specific DNA-binding site.48,49 54 We found that PLZF physically interacts with the nuclear-localized transcription factor VDR and that PLZF expression inhibits the ability of 1,25(OH)₂D₃ to induce specific hematopoietic differentiation. We suggest that, through a combination of these mechanisms, PLZF participates in the establishment of distinct populations of monocytic hematopoietic cells in response to the presence of external endocrine signals. We hope that additional studies will determine whether PLZF-mediated regulation of VDR, or an aberration thereof, participates in APL leukemogenesis.

We thank C. Bromleigh, C. Rachez, G. Farmer, and T. Staeva for helpful discussions; A. Ward for manuscript preparation; the members of the Flow Cytometry Core Facility of Memorial Sloan-Kettering Cancer Center and the Mount Sinai Flow Cytometry Shared Research Facility for their assistance; and V. Richon and F. Sladek for antibodies against mSin3A and the HNF-4 DBD, respectively.