A 55-Kilodalton Accessory Factor Facilitates Vitamin D Receptor DNA Binding
Teruki Sone, Keiichi Ozono, and J. Wesley Pike*
The interaction of the vitamin D receptor with a vitamin D-responsive element (VDRE) derived from the human osteocalcin promoter in vitro has been shown to require a nuclear accessory factor (NAF) derived from monkey kidney cells. In this report we show that this factor is widely distributed in cells and tissues, including those that do not express the vitamin D receptor (VDR). NAF is required for VDR binding to a variety of known VDREs. VDR and NAF independently bind the VDRE weakly, as assessed by elution profiles generated during VDRE affinity chromatography. Together, however, both proteins coelute from this column with a profile that indicates a tighter strength of interaction. Analogous chromatography of the VDR derived from ROS 17/2.8 cells treated with 1,25-dihydroxyvitamin D3 in culture also reveals a dual profile of weak and strong binding, suggesting that in vivo modifications are unlikely to alter receptor DNA binding. NAF is a protein of 55 kDa, as assessed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis, and cross-linking experiments suggest that the VDR and NAF together form a heterodimer on a single VDRE with a mol wt of 103 kDa. These data demonstrate that NAF is required for VDR binding to specific DNA in vitro and suggest the possibility that NAF may be required for the transactivation capability of the VDR in vivo. (Molecular Endocrinology 5: 1578-1586,1991)
INTRODUCTION
Transcriptional regulation of gene expression is mediated by a family of ligand-activated nuclear receptors that include those for the steroid, thyroid, and retinoic acid hormones (1-3). These receptors are DNA-binding proteins that interact in a sequence-specific manner with c/s-acting elements located in or near hormonesensitive promoters (3). The DNA sequence motifs that mediate receptor response are generally comprised of 0888-8809/91/1578-1586$03.00/0 Molecular Endocrinology Copyright © 1991 by The Endocrine Society
1578
direct or inverted repeats, such as those found in the tyrosine aminotransferase (4), vitellogenin (5), laminin B1 (6), GH (7, 8), and osteocalcin (9-12) genes. These elements often mediate the action of a single hormone, although within certain genes they may mediate the action of several hormones whose receptors exhibit related sequence specificity. Thus, for example, the progesterone and glucocorticoid receptors recognize identical responsive elements (13, 14), and thyroid, retinoic acid and vitamin D receptors (VDR) also recognize very similar elements (6-12, 15-17). The hormonal specificity for transactivation demonstrated by certain genes clearly suggests that determinants other than the nucleotide sequence of the receptor-binding site may be required. These determinants may include half-site spacing, chromatin structure, receptor expression, and, perhaps most importantly, expression of additional protein factors that participate in the activation process. The DNA-binding domain of the steroid receptors is comprised of two potential loop structures, each folded about a single zinc atom (18). The three-dimensional solution structure of this domain in the estrogen (19) and glucocorticoid (20) receptors suggests that the carboxyl side of each loop contains an a-helix. The first helix is postulated to lie in the major groove and confer specificity, whereas the second stabilizes the complex (21). Certain of the receptors appear to bind to responsive elements as cooperative dimers (22-24). The highest affinity DNA binding interaction of the receptor, however, requires the DNA-binding domain as well as a separate and distinct domain located within the carboxy-terminal region (25-27). While the structure of this domain remains largely uncharacterized, it is clear that this domain participates not only in homodimer formation (26), but may also mediate formation of dimers comprised of heterologous receptors (25, 27) or dimers comprised of a specific receptor and unknown nuclear proteins (28-34). Examples include retinoic acid-thyroid receptor heterodimers (25, 30), heterodimers composed of the thyroid receptor and a protein designated TRAP (32-34), and heterodimers between the retinoic acid receptor and several different proteins that appear to be expressed in a tissue-specific manner (31). In
Downloaded from https://academic.oup.com/mend/article-abstract/5/11/1578/2714228 by guest on 19 January 2019
Departments of Pediatrics (T.S., K.O., J.W.P.) and Cell Biology (J.W.P.) Baylor College of Medicine Houston, Texas 77030
VDR-NAF
VDR Requires a Mammalian Cell Nuclear Factor for VDRE Binding
VDR-
RESULTS Cellular Distribution of NAF We used a bandshift enhancement assay to identify the presence of NAF activity in cellular extacts. As seen in Fig. 1, while cellular extracts containing this activity do not bind to a VDRE probe in the absence of VDR, the addition of this protein, either from crude yeast cytosols or as purified VDR, leads to the clear demonstration of a protein-DNA complex. As with NAF alone, yeast VDR independently does not generate a complex. These experiments indicate that while both proteins together generate a DNA complex of sufficient affinity for identification during bandshift analysis, neither protein independently forms equivalent complexes. We used this assay to determine the presence of NAF or NAF-related activity in the cultured cells and mouse tissues indicated in Table 1. Clearly, this protein or one functionally related to NAF is widely distributed. Although VDR-NAF complexes comigrate independent of cell source, we cannot conclude that NAF is identical, as small differences in mol wt are unlikely to be detected in the bandshift assay. The fact that NAF is expressed in mouse liver and in cells lines such as CV-1, neither of which contains the VDR, suggests that the protein is not coexpressed with the VDR. Requirement for NAF on VDREs VDR derived from tissue nuclear extracts has been demonstrated to bind VDREs in the rat OC and mouse osteopontin genes (10,15). We evaluated the requirement for NAF in VDR binding to these labeled elements by incubating the VDR with or without extracts contain-
12
3 4
Fig. 1. Bandshift Enhancement Assay for VDR and NAF Cellular extracts were incubated with VDR DNA probe and then electrophoresed as described in Materials and Methods. Lane 1, pAVhVDR-transfected COS-1 cell nuclear extract (1 ng protein). Lane 2, COS-1 cell nuclear extract (1 ng protein). Lane 3, VDR-expressing yeast cytosol (0.1 ng protein). Lane 4, VDR-expressing yeast cytosol (0.1 ^g protein) and COS-1 cell nuclear extract (1 ^g protein).
Table 1. Cellular and Tissue Distribution of NAF or NAFRelated Factor Mouse tissues Liver, kidney Cultured cells Kidney fibroblasts (CV-1, COS-1) Hepatoma (HepG2) Cervical carcinoma (HeLa) Breast cancer (T47Dco) Osteosarcoma (ROS 17/2.8) Calvaria(MC3T3-E1) Fibroblasts (human primary) Lymphoblasts (human, Epstein-Barr virus-transformed) Spodoptera fugiperda (insect, Sf9)
ing NAF and resolving the protein-DNA complexes by bandshift assay. As observed in Table 2, NAF was required for binding of the VDR to all of these elements, suggesting a generalized requirement for this protein.
Downloaded from https://academic.oup.com/mend/article-abstract/5/11/1578/2714228 by guest on 19 January 2019
each situation these proteins facilitate DNA binding of their respective receptor partners. While the formation of heterodimers may well increase the diversity and complexity with which cells can respond to hormone, the physiological relevance of these protein factors remains unknown. The VDR is a member of the steroid receptor family and mediates the genomic action of 1,25-dihydroxyvitarnin D3 [1,25-(OH)2D3] (1-3,35). Vitamin D-responsive elements (VDRE) have been identified in the human (9) and rat (10-12) osteocalcin (OC) genes and mouse osteopontin (15) gene. Activation of the human OC promoter requires the presence of a functional receptor (36). The VDR has been shown to bind to each of these response elements in vitro (10,12,15,37,38), although the details of this binding are unknown. Recently, we demonstrated that the interaction of the VDR with the human OC VDRE requires the presence of a mammalian cell protein factor that we termed nuclear accessory factor (NAF) (39). In this report we further characterize the distribution and properties of this protein factor that facilitates strong binding of the VDR to OC VDRE DNA in vitro.
1579
Vol 5 No. 11
MOL ENDO-1991 1580
Table 2. Requirement of NAF for VDR Interaction with VDREs Sequence H
NAF Required
GGGGCA-3' AGGACA-3' GGTTCA-3' AGTTCA-3'
+ + + +
Element H
Human OC Rat OC Mouse OP Mouse |8RAR
5'-GGGTGA ACG 5'-GGGTGA ATG 5'-GGTTCA CGA 5'-GGTTCA CCGAA
NAF was probably present in the protein-DNA complexes previously reported (10, 15). Interestingly, VDR also binds to the mouse retinoic acid receptor /3 gene retinoic acid response element (RARE) in the presence of NAF despite the fact that this gene is not activated by vitamin D (16). We do not find this in vitro interaction of the VDR surprising in view of the distinct homology observed between this RARE and that of the VDREs, particularly with the element found within the osteopontin gene (15). In the latter case, the nucleotide spacing between the two GGTTCA half-sites [3 basepairs (bp) in the VDREs and 5 bp in the RARE] represents the only fundamental different between the two elements. Interaction of VDR and NAF with VDRE Affinity Resins We examined the binding properties of the VDR and NAF further through VDRE affinity chromatography. In contrast to the results obtained by bandshift analysis, the yeast VDR alone was capable of weakly binding concatemerized human osteocalcin VDREs, binding under low salt conditions, and eluting during a linear gradient at 0.13 M KCI (Fig. 2A). NAF was equally capable of a similar weak interaction on the VDRE when extracts were chromatographed and aliquots of each fraction were subjected to bandshift analysis in the presence of added VDR (Fig. 2B). In contrast, chromatography of a combination of both yeast VDR and nuclear extracts containing NAF leads to the elution of two peaks of VDR activity, one at 0.13 M and a stronger interaction at 0.26 M (Fig. 2C). Bandshift assay of individual fractions for NAF activity clearly revealed that NAF activity was associated only with the high affinity peak (Fig. 2D). These results suggest that both proteins independently bind VDRE DNA. Together, however, they coelute from the VDRE affinity column, and their desorption requires higher ionic strength.
Physical Properties of NAF Additional chromatographic analyses revealed that NAF bound a variety of ion exchange and affinity resins, including diethylaminoethyl Sephadex, phosphocellulose, calf thymus DNA-cellulose, and heparin-agarose (data not shown). Moreover, the protein was eluted from calf thymus DNA cellulose with p-chloromercuribenzene sulfonate, a property characteristic of the VDR (43) as well as several of the other steroid receptors (44, 45). We partially purified NAF from HeLa cell cytosols and then determined its molecular mass after resolution by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE; see Materials and Methods). As observed in Fig. 4A, bandshift analysis of protein fractions resolved by SDS-PAGE revealed a precise molecular mass of 55 kDa. This size estimate differed from that of the 48-kDa VDR simultaneously determined by Western blot analysis of a parallel electrophoretic lane (Fig. 4B), suggesting that NAF is not simply a modified form of the VDR. Importantly, the capacity of renatured NAF to facilitate DNA binding of exogenously added VDR in the absence of added ATP also suggests that NAF is not a protein kinase. Finally, the observation that the VDR-NAF-DNA complex derived from partially purified NAF comigrated during bandshift analysis with unpurified material (Fig. 4A, lane c) indicates that NAF was not proteolyzed during the enrichment procedure.
Formation of 1,25-(OH)2D3-Receptor Complex in Culture
VDR-NAF Cross-Linking
We treated confluent ROS 17/2.8 cells with radiolabeled 1,25-(OH)2D3 and chromatographed the resulting 1,25-(OH)2D3-receptor complex on the VDRE affinity column to determined whether the formation of the 1,25-(OH)2D3-receptor complex in intact cells might alter
VDR was translated in vitro in the presence of [35S] methionine and then incubated with VDRE DNA endfilled with biotinylated 11-dUTP. Precipitation of the VDRE with immobilized streptavidin followed by SDSPAGE revealed the presence of a VDR protein predom-
Downloaded from https://academic.oup.com/mend/article-abstract/5/11/1578/2714228 by guest on 19 January 2019
Sequence is indicated where the half-sites are designated H and the spacing nucleotides are designated S.
receptor elution. 1,25-(OH)2D3 is capable of regulating the phosphorylation state of the VDR when added to intact cultured cells (40, 41), and this modification as well as the capacity of the hormone to up-regulate the receptor (42) might feasibly affect DNA binding. As observed in Fig. 3, both weak and strong interactions were evident upon chromatographic analysis, and their elution profiles were not significantly different from those identified when the VDR was labeled in vitro. The relative distribution of both peaks was somewhat similar, and bandshift analysis indicated that the high affinity complex comigrates with the known VDR-NAF complex (data not shown). The results of this experiment suggest that formation of the VDR-hormone complex in culture does not substantially alter its capacity to interact with the VDRE independently, and that the VDRNAF complex does not demonstrate an interaction different from that observed after in vitro VDR labeling. These data also suggest that NAF activity is insufficent in these experiments to saturate endogenous levels of the VDR. Thus, the protein probably exists in concentrations that correspond to those of normal VDR.
VDR-NAF
1581
0.13 0.13
0.26
D.
0.13
20 30 40 50 Fraction Number
60
0.26
10 20 30 40 50 60 Relative Fraction Number
0.5 0.4 0.3 0.2 0.1 0
Fig,. 2. VDRE Affinity Chromatography of the VDR and NAF Cellular extracts from COS-1 cells or yeast were incubated for 3 h at 4 C with or without 1,25-(OH)2-[3H]D3 (4 nM) and then chromatographed on a VDRE-Sepharose column (5 ml) containing concatemerized VDRE oligonucleotide (25 Mg/ml). Samples were applied in TD buffer containing 0.05 M KCI, washed, and then eluted with TD buffer containing a linear gradient of KCI. Aliquots (0.5 ml) were quantitated for tritium by liquid scintillation spectrophotometry or asssessed for NAF activity by bandshift assay in the presence of added yeast VDR. A, Chromatography of VDR-containing yeast cytosol (140 ng cytosol protein; 1 pmol VDR) prelabeled for 3 h at 4 C with 1,25-(OH)2D3. B, Chromatography of COS-1 cell extracts (3 mg cytosol protein) and assessment of relative NAF activity. The NAF-VDR-VDRE complex was excised and quantitated. C, Chromatography of prelabeled VDR-containing yeast cytosol sample, as in A, premixed with nuclear extract (140 ^g protein) derived from nontransfected COS-1 cells, as in B. D, Assessment of NAF activity in aliquots of the chromatography in C. VDR was added in excess to each aliquot.
0.14
0.26
20 30 40 Fraction Number
60
Fig. 3. Formation of 1,25-(OH)2D3-VDR Complexes in Culture Does not Alter the VDR-NAF Profile ROS 17/2.8 cells were incubated with 1,25-(OH)2D3 in culture, as described in Materials and Methods. Cytosol was prepared, and 0.86 mg protein containing 0.94 pmol labeled VDR was chromatographed on a VDRE affinity column, as described in Fig. 2.
inantly when HeLa cell extracts were included in the incubation (Fig. 5). Incubation with unprogrammed lysates did not result in precipitable 35S-labeled protein. Most importantly, when these precipitates were sub-
jected to the cross-linking reagent bismaleimidohexane (BMH), a VDR-related cross-linked band at 103 kDa was evident (Fig. 5). This size is significantly greater than that of a VDR homodimer at 96 kDa. Larger crosslinked species are also evident (J 20-135 kDa), which may reflect either nonspecific interactions or additional heterogeneous protein-protein interactions with the VDR/NAF complex. This 103-kDa cross-linked complex was similarly observed when VDR was incubated with extracts of COS-1, ROS 17/2.8, and mouse fibroblasts (data not shown). These observations are, therefore, consistent with a molecular mass for NAF of approximately 55 kDa. More importantly, they suggest that the interaction of the VDR on the OC VDRE probably occurs as a heterodimer with NAF.
DISCUSSION
We have previously reported the existence of a trypsinand heat-sensitive component in mammalian cells which facilitates the binding of the VDR to OC VDRE DNA (39). In this report we describe several additional prop-
Downloaded from https://academic.oup.com/mend/article-abstract/5/11/1578/2714228 by guest on 19 January 2019
10 20 30 40 50 60 Relative Fraction Number
Vol 5 No. 11
MOL ENDO-1991 1582
A.
VDR/NAF VDR
Teruki Sone, Keiichi Ozono, and J. Wesley Pike*
The interaction of the vitamin D receptor with a vitamin D-responsive element (VDRE) derived from the human osteocalcin promoter in vitro has been shown to require a nuclear accessory factor (NAF) derived from monkey kidney cells. In this report we show that this factor is widely distributed in cells and tissues, including those that do not express the vitamin D receptor (VDR). NAF is required for VDR binding to a variety of known VDREs. VDR and NAF independently bind the VDRE weakly, as assessed by elution profiles generated during VDRE affinity chromatography. Together, however, both proteins coelute from this column with a profile that indicates a tighter strength of interaction. Analogous chromatography of the VDR derived from ROS 17/2.8 cells treated with 1,25-dihydroxyvitamin D3 in culture also reveals a dual profile of weak and strong binding, suggesting that in vivo modifications are unlikely to alter receptor DNA binding. NAF is a protein of 55 kDa, as assessed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis, and cross-linking experiments suggest that the VDR and NAF together form a heterodimer on a single VDRE with a mol wt of 103 kDa. These data demonstrate that NAF is required for VDR binding to specific DNA in vitro and suggest the possibility that NAF may be required for the transactivation capability of the VDR in vivo. (Molecular Endocrinology 5: 1578-1586,1991)
INTRODUCTION
Transcriptional regulation of gene expression is mediated by a family of ligand-activated nuclear receptors that include those for the steroid, thyroid, and retinoic acid hormones (1-3). These receptors are DNA-binding proteins that interact in a sequence-specific manner with c/s-acting elements located in or near hormonesensitive promoters (3). The DNA sequence motifs that mediate receptor response are generally comprised of 0888-8809/91/1578-1586$03.00/0 Molecular Endocrinology Copyright © 1991 by The Endocrine Society
1578
direct or inverted repeats, such as those found in the tyrosine aminotransferase (4), vitellogenin (5), laminin B1 (6), GH (7, 8), and osteocalcin (9-12) genes. These elements often mediate the action of a single hormone, although within certain genes they may mediate the action of several hormones whose receptors exhibit related sequence specificity. Thus, for example, the progesterone and glucocorticoid receptors recognize identical responsive elements (13, 14), and thyroid, retinoic acid and vitamin D receptors (VDR) also recognize very similar elements (6-12, 15-17). The hormonal specificity for transactivation demonstrated by certain genes clearly suggests that determinants other than the nucleotide sequence of the receptor-binding site may be required. These determinants may include half-site spacing, chromatin structure, receptor expression, and, perhaps most importantly, expression of additional protein factors that participate in the activation process. The DNA-binding domain of the steroid receptors is comprised of two potential loop structures, each folded about a single zinc atom (18). The three-dimensional solution structure of this domain in the estrogen (19) and glucocorticoid (20) receptors suggests that the carboxyl side of each loop contains an a-helix. The first helix is postulated to lie in the major groove and confer specificity, whereas the second stabilizes the complex (21). Certain of the receptors appear to bind to responsive elements as cooperative dimers (22-24). The highest affinity DNA binding interaction of the receptor, however, requires the DNA-binding domain as well as a separate and distinct domain located within the carboxy-terminal region (25-27). While the structure of this domain remains largely uncharacterized, it is clear that this domain participates not only in homodimer formation (26), but may also mediate formation of dimers comprised of heterologous receptors (25, 27) or dimers comprised of a specific receptor and unknown nuclear proteins (28-34). Examples include retinoic acid-thyroid receptor heterodimers (25, 30), heterodimers composed of the thyroid receptor and a protein designated TRAP (32-34), and heterodimers between the retinoic acid receptor and several different proteins that appear to be expressed in a tissue-specific manner (31). In
Downloaded from https://academic.oup.com/mend/article-abstract/5/11/1578/2714228 by guest on 19 January 2019
Departments of Pediatrics (T.S., K.O., J.W.P.) and Cell Biology (J.W.P.) Baylor College of Medicine Houston, Texas 77030
VDR-NAF
VDR Requires a Mammalian Cell Nuclear Factor for VDRE Binding
VDR-
RESULTS Cellular Distribution of NAF We used a bandshift enhancement assay to identify the presence of NAF activity in cellular extacts. As seen in Fig. 1, while cellular extracts containing this activity do not bind to a VDRE probe in the absence of VDR, the addition of this protein, either from crude yeast cytosols or as purified VDR, leads to the clear demonstration of a protein-DNA complex. As with NAF alone, yeast VDR independently does not generate a complex. These experiments indicate that while both proteins together generate a DNA complex of sufficient affinity for identification during bandshift analysis, neither protein independently forms equivalent complexes. We used this assay to determine the presence of NAF or NAF-related activity in the cultured cells and mouse tissues indicated in Table 1. Clearly, this protein or one functionally related to NAF is widely distributed. Although VDR-NAF complexes comigrate independent of cell source, we cannot conclude that NAF is identical, as small differences in mol wt are unlikely to be detected in the bandshift assay. The fact that NAF is expressed in mouse liver and in cells lines such as CV-1, neither of which contains the VDR, suggests that the protein is not coexpressed with the VDR. Requirement for NAF on VDREs VDR derived from tissue nuclear extracts has been demonstrated to bind VDREs in the rat OC and mouse osteopontin genes (10,15). We evaluated the requirement for NAF in VDR binding to these labeled elements by incubating the VDR with or without extracts contain-
12
3 4
Fig. 1. Bandshift Enhancement Assay for VDR and NAF Cellular extracts were incubated with VDR DNA probe and then electrophoresed as described in Materials and Methods. Lane 1, pAVhVDR-transfected COS-1 cell nuclear extract (1 ng protein). Lane 2, COS-1 cell nuclear extract (1 ng protein). Lane 3, VDR-expressing yeast cytosol (0.1 ng protein). Lane 4, VDR-expressing yeast cytosol (0.1 ^g protein) and COS-1 cell nuclear extract (1 ^g protein).
Table 1. Cellular and Tissue Distribution of NAF or NAFRelated Factor Mouse tissues Liver, kidney Cultured cells Kidney fibroblasts (CV-1, COS-1) Hepatoma (HepG2) Cervical carcinoma (HeLa) Breast cancer (T47Dco) Osteosarcoma (ROS 17/2.8) Calvaria(MC3T3-E1) Fibroblasts (human primary) Lymphoblasts (human, Epstein-Barr virus-transformed) Spodoptera fugiperda (insect, Sf9)
ing NAF and resolving the protein-DNA complexes by bandshift assay. As observed in Table 2, NAF was required for binding of the VDR to all of these elements, suggesting a generalized requirement for this protein.
Downloaded from https://academic.oup.com/mend/article-abstract/5/11/1578/2714228 by guest on 19 January 2019
each situation these proteins facilitate DNA binding of their respective receptor partners. While the formation of heterodimers may well increase the diversity and complexity with which cells can respond to hormone, the physiological relevance of these protein factors remains unknown. The VDR is a member of the steroid receptor family and mediates the genomic action of 1,25-dihydroxyvitarnin D3 [1,25-(OH)2D3] (1-3,35). Vitamin D-responsive elements (VDRE) have been identified in the human (9) and rat (10-12) osteocalcin (OC) genes and mouse osteopontin (15) gene. Activation of the human OC promoter requires the presence of a functional receptor (36). The VDR has been shown to bind to each of these response elements in vitro (10,12,15,37,38), although the details of this binding are unknown. Recently, we demonstrated that the interaction of the VDR with the human OC VDRE requires the presence of a mammalian cell protein factor that we termed nuclear accessory factor (NAF) (39). In this report we further characterize the distribution and properties of this protein factor that facilitates strong binding of the VDR to OC VDRE DNA in vitro.
1579
Vol 5 No. 11
MOL ENDO-1991 1580
Table 2. Requirement of NAF for VDR Interaction with VDREs Sequence H
NAF Required
GGGGCA-3' AGGACA-3' GGTTCA-3' AGTTCA-3'
+ + + +
Element H
Human OC Rat OC Mouse OP Mouse |8RAR
5'-GGGTGA ACG 5'-GGGTGA ATG 5'-GGTTCA CGA 5'-GGTTCA CCGAA
NAF was probably present in the protein-DNA complexes previously reported (10, 15). Interestingly, VDR also binds to the mouse retinoic acid receptor /3 gene retinoic acid response element (RARE) in the presence of NAF despite the fact that this gene is not activated by vitamin D (16). We do not find this in vitro interaction of the VDR surprising in view of the distinct homology observed between this RARE and that of the VDREs, particularly with the element found within the osteopontin gene (15). In the latter case, the nucleotide spacing between the two GGTTCA half-sites [3 basepairs (bp) in the VDREs and 5 bp in the RARE] represents the only fundamental different between the two elements. Interaction of VDR and NAF with VDRE Affinity Resins We examined the binding properties of the VDR and NAF further through VDRE affinity chromatography. In contrast to the results obtained by bandshift analysis, the yeast VDR alone was capable of weakly binding concatemerized human osteocalcin VDREs, binding under low salt conditions, and eluting during a linear gradient at 0.13 M KCI (Fig. 2A). NAF was equally capable of a similar weak interaction on the VDRE when extracts were chromatographed and aliquots of each fraction were subjected to bandshift analysis in the presence of added VDR (Fig. 2B). In contrast, chromatography of a combination of both yeast VDR and nuclear extracts containing NAF leads to the elution of two peaks of VDR activity, one at 0.13 M and a stronger interaction at 0.26 M (Fig. 2C). Bandshift assay of individual fractions for NAF activity clearly revealed that NAF activity was associated only with the high affinity peak (Fig. 2D). These results suggest that both proteins independently bind VDRE DNA. Together, however, they coelute from the VDRE affinity column, and their desorption requires higher ionic strength.
Physical Properties of NAF Additional chromatographic analyses revealed that NAF bound a variety of ion exchange and affinity resins, including diethylaminoethyl Sephadex, phosphocellulose, calf thymus DNA-cellulose, and heparin-agarose (data not shown). Moreover, the protein was eluted from calf thymus DNA cellulose with p-chloromercuribenzene sulfonate, a property characteristic of the VDR (43) as well as several of the other steroid receptors (44, 45). We partially purified NAF from HeLa cell cytosols and then determined its molecular mass after resolution by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE; see Materials and Methods). As observed in Fig. 4A, bandshift analysis of protein fractions resolved by SDS-PAGE revealed a precise molecular mass of 55 kDa. This size estimate differed from that of the 48-kDa VDR simultaneously determined by Western blot analysis of a parallel electrophoretic lane (Fig. 4B), suggesting that NAF is not simply a modified form of the VDR. Importantly, the capacity of renatured NAF to facilitate DNA binding of exogenously added VDR in the absence of added ATP also suggests that NAF is not a protein kinase. Finally, the observation that the VDR-NAF-DNA complex derived from partially purified NAF comigrated during bandshift analysis with unpurified material (Fig. 4A, lane c) indicates that NAF was not proteolyzed during the enrichment procedure.
Formation of 1,25-(OH)2D3-Receptor Complex in Culture
VDR-NAF Cross-Linking
We treated confluent ROS 17/2.8 cells with radiolabeled 1,25-(OH)2D3 and chromatographed the resulting 1,25-(OH)2D3-receptor complex on the VDRE affinity column to determined whether the formation of the 1,25-(OH)2D3-receptor complex in intact cells might alter
VDR was translated in vitro in the presence of [35S] methionine and then incubated with VDRE DNA endfilled with biotinylated 11-dUTP. Precipitation of the VDRE with immobilized streptavidin followed by SDSPAGE revealed the presence of a VDR protein predom-
Downloaded from https://academic.oup.com/mend/article-abstract/5/11/1578/2714228 by guest on 19 January 2019
Sequence is indicated where the half-sites are designated H and the spacing nucleotides are designated S.
receptor elution. 1,25-(OH)2D3 is capable of regulating the phosphorylation state of the VDR when added to intact cultured cells (40, 41), and this modification as well as the capacity of the hormone to up-regulate the receptor (42) might feasibly affect DNA binding. As observed in Fig. 3, both weak and strong interactions were evident upon chromatographic analysis, and their elution profiles were not significantly different from those identified when the VDR was labeled in vitro. The relative distribution of both peaks was somewhat similar, and bandshift analysis indicated that the high affinity complex comigrates with the known VDR-NAF complex (data not shown). The results of this experiment suggest that formation of the VDR-hormone complex in culture does not substantially alter its capacity to interact with the VDRE independently, and that the VDRNAF complex does not demonstrate an interaction different from that observed after in vitro VDR labeling. These data also suggest that NAF activity is insufficent in these experiments to saturate endogenous levels of the VDR. Thus, the protein probably exists in concentrations that correspond to those of normal VDR.
VDR-NAF
1581
0.13 0.13
0.26
D.
0.13
20 30 40 50 Fraction Number
60
0.26
10 20 30 40 50 60 Relative Fraction Number
0.5 0.4 0.3 0.2 0.1 0
Fig,. 2. VDRE Affinity Chromatography of the VDR and NAF Cellular extracts from COS-1 cells or yeast were incubated for 3 h at 4 C with or without 1,25-(OH)2-[3H]D3 (4 nM) and then chromatographed on a VDRE-Sepharose column (5 ml) containing concatemerized VDRE oligonucleotide (25 Mg/ml). Samples were applied in TD buffer containing 0.05 M KCI, washed, and then eluted with TD buffer containing a linear gradient of KCI. Aliquots (0.5 ml) were quantitated for tritium by liquid scintillation spectrophotometry or asssessed for NAF activity by bandshift assay in the presence of added yeast VDR. A, Chromatography of VDR-containing yeast cytosol (140 ng cytosol protein; 1 pmol VDR) prelabeled for 3 h at 4 C with 1,25-(OH)2D3. B, Chromatography of COS-1 cell extracts (3 mg cytosol protein) and assessment of relative NAF activity. The NAF-VDR-VDRE complex was excised and quantitated. C, Chromatography of prelabeled VDR-containing yeast cytosol sample, as in A, premixed with nuclear extract (140 ^g protein) derived from nontransfected COS-1 cells, as in B. D, Assessment of NAF activity in aliquots of the chromatography in C. VDR was added in excess to each aliquot.
0.14
0.26
20 30 40 Fraction Number
60
Fig. 3. Formation of 1,25-(OH)2D3-VDR Complexes in Culture Does not Alter the VDR-NAF Profile ROS 17/2.8 cells were incubated with 1,25-(OH)2D3 in culture, as described in Materials and Methods. Cytosol was prepared, and 0.86 mg protein containing 0.94 pmol labeled VDR was chromatographed on a VDRE affinity column, as described in Fig. 2.
inantly when HeLa cell extracts were included in the incubation (Fig. 5). Incubation with unprogrammed lysates did not result in precipitable 35S-labeled protein. Most importantly, when these precipitates were sub-
jected to the cross-linking reagent bismaleimidohexane (BMH), a VDR-related cross-linked band at 103 kDa was evident (Fig. 5). This size is significantly greater than that of a VDR homodimer at 96 kDa. Larger crosslinked species are also evident (J 20-135 kDa), which may reflect either nonspecific interactions or additional heterogeneous protein-protein interactions with the VDR/NAF complex. This 103-kDa cross-linked complex was similarly observed when VDR was incubated with extracts of COS-1, ROS 17/2.8, and mouse fibroblasts (data not shown). These observations are, therefore, consistent with a molecular mass for NAF of approximately 55 kDa. More importantly, they suggest that the interaction of the VDR on the OC VDRE probably occurs as a heterodimer with NAF.
DISCUSSION
We have previously reported the existence of a trypsinand heat-sensitive component in mammalian cells which facilitates the binding of the VDR to OC VDRE DNA (39). In this report we describe several additional prop-
Downloaded from https://academic.oup.com/mend/article-abstract/5/11/1578/2714228 by guest on 19 January 2019
10 20 30 40 50 60 Relative Fraction Number
Vol 5 No. 11
MOL ENDO-1991 1582
A.
VDR/NAF VDR
