Multiple Components of a Complex Androgen-Dependent Enhancer
Adam J. Adler, Arno Scheller, Yetta Hoffman, and Diane M. Robins Department of Human Genetics University of Michigan Medical School Ann Arbor, Michigan 48109-0618
Sex-limited protein (Sip) is expressed in adult male mice. A 160-basepair fragment 2 kilobases upstream of the gene serves as an androgen-dependent enhancer of chloramphenicol acetyltransferase expression in transient transfection assays in cells with endogenous or cotransfected androgen receptor. One element that is necessary, but not sufficient, for induction is a consensus glucocorticoid (or hormone) response element (HRE). This element binds to the mouse androgen receptor in vitro, but with apparent weak affinity. Induction by the HRE is greatly augmented by an accessory sequence within the 160 basepairs, suggesting that cooperative interactions confer strong response to androgen. Additional elements within the enhancer modulate induction, positively or negatively, and exhibit cellspecific behavior. Of particular interest are two degenerate HREs that are adjacent to the consensus sequence; they show no independent activity, but are functionally significant in conjunction with other elements. The complexity of this enhancer may reflect biological mechanisms that ensure specificity of hormonal response and allow gene expression to respond to changes in hormone concentration. (Molecular Endocrinology 5: 1587-1596, 1991)
can all induce a reporter gene via a canonical glucocorticoid response element (GRE) in transfection (3-5), yet their effects in vivo are not synonymous. Specificity in some cases may be determined by which receptors are expressed in the target cell (6). However, many cell types express multiple receptors and still exhibit very precise patterns of gene regulation, suggesting that some component of hormonal response is conferred by additional factors able to distinguish among the receptors. Alternatively, subtle sequence differences distinguish the hormone response elements (HREs) in vivo. Currently, the most studied androgen regulatory sequences are those of mouse mammary tumor virus (MMTV), which responds in vivo and in vitro to glucocorticoid, progesterone, and androgen (3-5, 7, 8). As a model for a more specifically androgen-dependent gene, we have been characterizing the regulation of the mouse sex-limited protein (Sip) gene, which is expressed appreciably only in mature male mice (9). The Sip gene evolved from a duplicated fourth component of complement gene, C4, within the major histocompatibility complex (10). Although greater than 95% homologous to C4 in coding and flanking regions, Sip is inactive in the complement pathway and requires androgen for expression (9, 11-13). We first localized a hormonal regulatory element 2 kilobases (kb) upstream of the Sip gene by virtue of its DNase-l hypersensitivity in liver chromatin in vivo (14). A fragment from this region was able to function as an androgendependent enhancer in transfection (15). This enhancer was found to reside within the 5' long terminal repeat (LTR) of an ancient endogenous provirus, indicating that the hormonal regulation of Sip is a retrotransposoninduced mutation that has been preserved in evolution (16).
INTRODUCTION
In the last several years, the mechanism of steroid hormone action has been clarified in molecular detail, largely due to the cloning and subsequent structural and functional analysis of the specific steroid receptors (1, 2). Hormone-receptor complexes function as transcriptional activators by binding specific DNA sequences that serve as enhancers of gene expression. Several unresolved issues remain, however, and it is not clear that the particulars will be the same for different members of the steroid receptor superfamily. Of particular biological interest is the question of how specific hormonal regulation is exerted, given that several receptors appear to recognize the same DNA sequences. Glucocorticoid, progesterone, and androgen
That the androgen-dependent enhancer derives from proviral sequences simplifies analysis by limiting the hormonal responsiveness to the 500 basepairs (bp) of the LTR. Tissue-specific regulation of Sip is similar to that of C4, in that Sip expresses in the subset of C4expressing tissues that are androgen responsive, such as liver and kidney (17), and, thus, is likely to be determined by sequences common to the two genes. In this report we analyze a 160-bp fragment from the LTR that confers androgen induction in transfection. One element necessary, but not sufficient, for strong response is a consensus HRE that binds androgen
0888-8809/91 /1587-1596$03.00/0 Molecular Endocrinology Copyright © 1991 by The Endocrine Society
1587
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 16 November 2015. at 20:36 For personal use only. No other uses without permission. . All rights reserved.
MOL ENDO-1991 1588
Vol 5 No. 11
receptor (AR) in vitro. The effect of the HRE is greatly augmented by additional sequences, and these effects can be modulated further, both positively and negatively, by elements within the 160 bp. Two degenerate HREs are positioned in tandem with the consensus element; they show no independent function, but are significant in the activity of the entire enhancer. The complex interactions evidenced for this enhancer may reflect the apparently weak affinity of AR for DNA (18, 19). This weak affinity and subsequent reliance on other factors for activation (or repression) may be part of a mechanism to ensure the specificity of the hormonal response.
To test the significance of the putative HREs, the best-fit HRE (HRE 3) was deleted by cleavage with Ddel. The resulting fragment, C A3, lost most androgen sensitivity compared to C'A2. Thus, the HRE consensus sequence was necessary for a strong hormonal response. However, a Haelll fragment with all three HRE-like sequences plus 35 bp up-stream (C'A4) was unresponsive, as was a smaller fragment lacking the third functional HRE (C'A5). The up-stream half of the enhancer, C'A6, showed no response by itself to hormone. Therefore, sequences within C'A6 were required to interact with the HRE(s) for hormonal response or to interact with other sequences within C A 4 to relieve repression of the HRE(s).
RESULTS
HRE 3 Can Function Independently as an Androgen Response Element
A 162-bp Fragment from a Proviral LTR Confers Androgen Induction Previously, a 750-bp Sma\-Xba\ fragment 2 kb from the start site of Sip transcription (Fig. 1A) was shown to function as an androgen-responsive enhancer in transfection (15). This enhancer was complex, in that smaller fragments initially tested did not show activity similar to the whole. The observation of four nuclear proteinbinding sites (footprints I-IV) suggested that multiple domains might be required for function. Discovering that the enhancer encompassed a proviral LTR (16) also intimated that promoter competition or occlusion might be interfering with the experimental definition of hormonal response. A series of restriction fragments was tested upstream of a thymidine kinase chloramphenicol acetyltransferase (tkCAT) reporter gene by transfection into androgen-responsive T47D human mammary carcinoma cells (Fig. 1B). The Sip-proximal Sty\-Xba\ fragment (C) in either orientation conferred greater than 3fold induction with dihydrotestosterone (DHT). However, in the reverse orientation, there was also 12-fold increased expression over that of tkCAT alone in the absence of hormone. This effect on basal expression was reduced to 3-fold by a 5' deletion that removed the proviral CAAT box (C'A1). A 3' deletion to a Haelll site resulted in C'A2, which conferred strong induction, with no effect on basal expression in either orientation. This 162-bp fragment showed a 12-fold response to DHT in the natural orientation, similar to the induction of a transfected full-length Sip gene with 10 kb of 5' flank (15, 20). There are some notable features within C'A2 (Fig. 1C). DNase-l footprints III and IV (FPIII and FPIV) are observed with nuclear extracts from several cell types (15) in a region marked by androgen-responsive hypersensitive sites in liver chromatin (14). FPIII partially overlaps the down-stream copy of a 17-bp direct repeat mismatched at only one base. Most intriguingly, the right end of the fragment contains a good HRE consensus (2), preceded by two similar sequences, one of which has a perfect TGTTCT half-site.
A GRE consensus element confers glucocorticoid, progesterone, and androgen induction in transfection (5). However, few native androgen-dependent elements have been characterized sufficiently, independent of other enhancer sequences, to determine whether androgen exerts an effect through GREs in vivo (21). The HRE-like sequences in the Sip enhancer were tested to determine whether they represented natural androgen response elements. Oligonucleotides were synthesized to the 15-bp sequences (HREs 1, 2, and 3), with SamHI termini. Single copies of HRE 1 or 2 did not confer androgen induction on tkCAT, whereas one copy of HRE 3 induced androgen 2-fold (Fig. 2). Multimerized HREs were tested for cooperativity that might enhance marginal responses, as is seen for weak GREs (22) and for elements in complex enhancers that only are active when multimerized (23). The activity of HRE 3 was accentuated by multimerization; that is, induction increased dramatically with copy number, to almost 50-fold. In contrast, multimers of HRE 1 or 2 showed virtually no response. Combinations of the sequences were tested for cooperativity. In conjunction with HRE 3, HRE 1, but not 2, increased induction somewhat over HRE 3 alone (3- vs. 2-fold). A 47-bp oligonucleotide containing all three HRE-like sequences as they are in the enhancer was not inducible. Since FPIV overlaps HRE 1 and is protected in vitro by a ubiquitous abundant protein (15), induction could have been repressed by binding of the FPIV factor. However, removal of 8 bp of FPIV sequence, leaving the consensus half-site of 1 (the 39mer), did not restore induction. Thus, HRE 3 is a component of a bona fide response element for androgen. The perfect GRE half-site present in HRE 1 was not sufficient for androgen induction, even when multimerized, although half-sites function in glucocorticoid induction and are largely responsible for the activity of the MMTV LTR. Since T47D cells have abundant progesterone receptor (PR), these constructs were tested for response to progesterone. Results were qualitatively similar; HRE 3 induced well, whereas HRE 1 and 2 were marginal at best (not shown).
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 16 November 2015. at 20:36 For personal use only. No other uses without permission. . All rights reserved.
1589
Androgen-Dependent Sip Enhancer
S
A
X
Provirua
Sip
1 kb
B
a
Ha
D
Ha
I
I
I
NDUCTIO N with DHT
TATA CAAT
100 bp
3.211.1
3.210.5 *
C'A1
2.010.4
8.410.9 *
C'A2
12.613.7
3.411.0
C'A3
1.810.4
1.510.2
CA4
1.110.2
1.310.2
CA5
1.510.2
1.610.2
1.110.1
1.210.2
CA6
FP III C' A6
MTCACAAGGGGGAACTAAAMTC^CMGGGGAMCTCCCTGGTTCTCMGATTTTAATCTMCTTTTGTGGTTGG
C' A4
CX^GGTtDCTGGAACMGGTCACTGU\ACCTGCTTATGTMTTATCTGTTCTGTGGT(>GCCAGTTCTCAGAACAGGCTGTTTCAGGG
FP IV •*—— 1
Ode I
•
Adam J. Adler, Arno Scheller, Yetta Hoffman, and Diane M. Robins Department of Human Genetics University of Michigan Medical School Ann Arbor, Michigan 48109-0618
Sex-limited protein (Sip) is expressed in adult male mice. A 160-basepair fragment 2 kilobases upstream of the gene serves as an androgen-dependent enhancer of chloramphenicol acetyltransferase expression in transient transfection assays in cells with endogenous or cotransfected androgen receptor. One element that is necessary, but not sufficient, for induction is a consensus glucocorticoid (or hormone) response element (HRE). This element binds to the mouse androgen receptor in vitro, but with apparent weak affinity. Induction by the HRE is greatly augmented by an accessory sequence within the 160 basepairs, suggesting that cooperative interactions confer strong response to androgen. Additional elements within the enhancer modulate induction, positively or negatively, and exhibit cellspecific behavior. Of particular interest are two degenerate HREs that are adjacent to the consensus sequence; they show no independent activity, but are functionally significant in conjunction with other elements. The complexity of this enhancer may reflect biological mechanisms that ensure specificity of hormonal response and allow gene expression to respond to changes in hormone concentration. (Molecular Endocrinology 5: 1587-1596, 1991)
can all induce a reporter gene via a canonical glucocorticoid response element (GRE) in transfection (3-5), yet their effects in vivo are not synonymous. Specificity in some cases may be determined by which receptors are expressed in the target cell (6). However, many cell types express multiple receptors and still exhibit very precise patterns of gene regulation, suggesting that some component of hormonal response is conferred by additional factors able to distinguish among the receptors. Alternatively, subtle sequence differences distinguish the hormone response elements (HREs) in vivo. Currently, the most studied androgen regulatory sequences are those of mouse mammary tumor virus (MMTV), which responds in vivo and in vitro to glucocorticoid, progesterone, and androgen (3-5, 7, 8). As a model for a more specifically androgen-dependent gene, we have been characterizing the regulation of the mouse sex-limited protein (Sip) gene, which is expressed appreciably only in mature male mice (9). The Sip gene evolved from a duplicated fourth component of complement gene, C4, within the major histocompatibility complex (10). Although greater than 95% homologous to C4 in coding and flanking regions, Sip is inactive in the complement pathway and requires androgen for expression (9, 11-13). We first localized a hormonal regulatory element 2 kilobases (kb) upstream of the Sip gene by virtue of its DNase-l hypersensitivity in liver chromatin in vivo (14). A fragment from this region was able to function as an androgendependent enhancer in transfection (15). This enhancer was found to reside within the 5' long terminal repeat (LTR) of an ancient endogenous provirus, indicating that the hormonal regulation of Sip is a retrotransposoninduced mutation that has been preserved in evolution (16).
INTRODUCTION
In the last several years, the mechanism of steroid hormone action has been clarified in molecular detail, largely due to the cloning and subsequent structural and functional analysis of the specific steroid receptors (1, 2). Hormone-receptor complexes function as transcriptional activators by binding specific DNA sequences that serve as enhancers of gene expression. Several unresolved issues remain, however, and it is not clear that the particulars will be the same for different members of the steroid receptor superfamily. Of particular biological interest is the question of how specific hormonal regulation is exerted, given that several receptors appear to recognize the same DNA sequences. Glucocorticoid, progesterone, and androgen
That the androgen-dependent enhancer derives from proviral sequences simplifies analysis by limiting the hormonal responsiveness to the 500 basepairs (bp) of the LTR. Tissue-specific regulation of Sip is similar to that of C4, in that Sip expresses in the subset of C4expressing tissues that are androgen responsive, such as liver and kidney (17), and, thus, is likely to be determined by sequences common to the two genes. In this report we analyze a 160-bp fragment from the LTR that confers androgen induction in transfection. One element necessary, but not sufficient, for strong response is a consensus HRE that binds androgen
0888-8809/91 /1587-1596$03.00/0 Molecular Endocrinology Copyright © 1991 by The Endocrine Society
1587
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 16 November 2015. at 20:36 For personal use only. No other uses without permission. . All rights reserved.
MOL ENDO-1991 1588
Vol 5 No. 11
receptor (AR) in vitro. The effect of the HRE is greatly augmented by additional sequences, and these effects can be modulated further, both positively and negatively, by elements within the 160 bp. Two degenerate HREs are positioned in tandem with the consensus element; they show no independent function, but are significant in the activity of the entire enhancer. The complex interactions evidenced for this enhancer may reflect the apparently weak affinity of AR for DNA (18, 19). This weak affinity and subsequent reliance on other factors for activation (or repression) may be part of a mechanism to ensure the specificity of the hormonal response.
To test the significance of the putative HREs, the best-fit HRE (HRE 3) was deleted by cleavage with Ddel. The resulting fragment, C A3, lost most androgen sensitivity compared to C'A2. Thus, the HRE consensus sequence was necessary for a strong hormonal response. However, a Haelll fragment with all three HRE-like sequences plus 35 bp up-stream (C'A4) was unresponsive, as was a smaller fragment lacking the third functional HRE (C'A5). The up-stream half of the enhancer, C'A6, showed no response by itself to hormone. Therefore, sequences within C'A6 were required to interact with the HRE(s) for hormonal response or to interact with other sequences within C A 4 to relieve repression of the HRE(s).
RESULTS
HRE 3 Can Function Independently as an Androgen Response Element
A 162-bp Fragment from a Proviral LTR Confers Androgen Induction Previously, a 750-bp Sma\-Xba\ fragment 2 kb from the start site of Sip transcription (Fig. 1A) was shown to function as an androgen-responsive enhancer in transfection (15). This enhancer was complex, in that smaller fragments initially tested did not show activity similar to the whole. The observation of four nuclear proteinbinding sites (footprints I-IV) suggested that multiple domains might be required for function. Discovering that the enhancer encompassed a proviral LTR (16) also intimated that promoter competition or occlusion might be interfering with the experimental definition of hormonal response. A series of restriction fragments was tested upstream of a thymidine kinase chloramphenicol acetyltransferase (tkCAT) reporter gene by transfection into androgen-responsive T47D human mammary carcinoma cells (Fig. 1B). The Sip-proximal Sty\-Xba\ fragment (C) in either orientation conferred greater than 3fold induction with dihydrotestosterone (DHT). However, in the reverse orientation, there was also 12-fold increased expression over that of tkCAT alone in the absence of hormone. This effect on basal expression was reduced to 3-fold by a 5' deletion that removed the proviral CAAT box (C'A1). A 3' deletion to a Haelll site resulted in C'A2, which conferred strong induction, with no effect on basal expression in either orientation. This 162-bp fragment showed a 12-fold response to DHT in the natural orientation, similar to the induction of a transfected full-length Sip gene with 10 kb of 5' flank (15, 20). There are some notable features within C'A2 (Fig. 1C). DNase-l footprints III and IV (FPIII and FPIV) are observed with nuclear extracts from several cell types (15) in a region marked by androgen-responsive hypersensitive sites in liver chromatin (14). FPIII partially overlaps the down-stream copy of a 17-bp direct repeat mismatched at only one base. Most intriguingly, the right end of the fragment contains a good HRE consensus (2), preceded by two similar sequences, one of which has a perfect TGTTCT half-site.
A GRE consensus element confers glucocorticoid, progesterone, and androgen induction in transfection (5). However, few native androgen-dependent elements have been characterized sufficiently, independent of other enhancer sequences, to determine whether androgen exerts an effect through GREs in vivo (21). The HRE-like sequences in the Sip enhancer were tested to determine whether they represented natural androgen response elements. Oligonucleotides were synthesized to the 15-bp sequences (HREs 1, 2, and 3), with SamHI termini. Single copies of HRE 1 or 2 did not confer androgen induction on tkCAT, whereas one copy of HRE 3 induced androgen 2-fold (Fig. 2). Multimerized HREs were tested for cooperativity that might enhance marginal responses, as is seen for weak GREs (22) and for elements in complex enhancers that only are active when multimerized (23). The activity of HRE 3 was accentuated by multimerization; that is, induction increased dramatically with copy number, to almost 50-fold. In contrast, multimers of HRE 1 or 2 showed virtually no response. Combinations of the sequences were tested for cooperativity. In conjunction with HRE 3, HRE 1, but not 2, increased induction somewhat over HRE 3 alone (3- vs. 2-fold). A 47-bp oligonucleotide containing all three HRE-like sequences as they are in the enhancer was not inducible. Since FPIV overlaps HRE 1 and is protected in vitro by a ubiquitous abundant protein (15), induction could have been repressed by binding of the FPIV factor. However, removal of 8 bp of FPIV sequence, leaving the consensus half-site of 1 (the 39mer), did not restore induction. Thus, HRE 3 is a component of a bona fide response element for androgen. The perfect GRE half-site present in HRE 1 was not sufficient for androgen induction, even when multimerized, although half-sites function in glucocorticoid induction and are largely responsible for the activity of the MMTV LTR. Since T47D cells have abundant progesterone receptor (PR), these constructs were tested for response to progesterone. Results were qualitatively similar; HRE 3 induced well, whereas HRE 1 and 2 were marginal at best (not shown).
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 16 November 2015. at 20:36 For personal use only. No other uses without permission. . All rights reserved.
1589
Androgen-Dependent Sip Enhancer
S
A
X
Provirua
Sip
1 kb
B
a
Ha
D
Ha
I
I
I
NDUCTIO N with DHT
TATA CAAT
100 bp
3.211.1
3.210.5 *
C'A1
2.010.4
8.410.9 *
C'A2
12.613.7
3.411.0
C'A3
1.810.4
1.510.2
CA4
1.110.2
1.310.2
CA5
1.510.2
1.610.2
1.110.1
1.210.2
CA6
FP III C' A6
MTCACAAGGGGGAACTAAAMTC^CMGGGGAMCTCCCTGGTTCTCMGATTTTAATCTMCTTTTGTGGTTGG
C' A4
CX^GGTtDCTGGAACMGGTCACTGU\ACCTGCTTATGTMTTATCTGTTCTGTGGT(>GCCAGTTCTCAGAACAGGCTGTTTCAGGG
FP IV •*—— 1
Ode I
•
