| Nucleic Acids Research | Pages |
The internal chondrocyte-specific promoter of the chick type III collagen gene is activated by AP1 and is repressed in fibroblasts by a complex containing an LBP1-related protein
Introduction
Materials And Methods
Construction of substitution mutants of the internal promoter
Functional analysis of promoter constructs
Identification of potential transcription factor binding sites
Electrophoretic mobility shift assays
Results
The internal promoter of the type III collagen gene contains both an activation domain and a fibroblast-specific repressor domain
An AP1 site is required for activation of the internal promoter
Sequences upstream of the AP1 site are required for optimal AP1 binding and promoter function
The repressor element contains a potential binding site for the transcription factor LBP1
Mutagenesis of both the activator and repressor elements confirms that AP1 functions as an activator in fibroblasts
Discussion
An AP1 site is essential for activation of the internal promoter
Repression of the internal promoter in fibroblasts requires an element containing an LBP1 site
The internal promoter of the type III collagen gene differs from other chondrocyte-specific promoters
Acknowledgements
References
The internal chondrocyte-specific promoter of the chick type III collagen gene is activated by AP1 and is repressed in fibroblasts by a complex containing an LBP1-related protein
Received May 12, 1999; Revised August 8, 1999; Accepted August 20, 1999
ABSTRACT The chick type III collagen gene contains an internal promoter in intron 23 in addition to the promoter preceding exon 1. The internal promoter, which is used preferentially in cultured chondrocytes, directs production of an alternative transcript that cannot encode type III collagen. This promoter is used inefficiently in skin fibroblasts, which transcribe the gene from the upstream promoter. We show below that the internal promoter is regulated by an activation element containing a potential activator protein 1 (AP1) site and a repressor element containing a potential binding site for leader binding protein 1 (LBP1). Electrophoretic mobility shift assays indicate that the activation and repressor elements are bound by AP1 and an LBP1-related protein, respectively. Replacement of the AP1 site resulted in substantially decreased promoter activity in both chondrocytes and fibroblasts, indicating that this site is required for promoter function, but the low level of promoter activity in fibroblasts is not due to loss of functional AP1. In contrast, replacement of the LBP1-like site increased activity only in fibroblasts, suggesting that this site is responsible in part for repression of promoter activity in fibroblasts.
INTRODUCTION
Type III collagen, a fibril-forming collagen, is widely distributed in vertebrate connective tissues in heterotypic fibrils with type I collagen (1-5). The human, mouse and chick type III collagen genes each contain 51 exons that are required to encode type III collagen mRNA and protein (6). We recently discovered an additional exon, 23A, within intron 23 of the chick type III collagen gene (7,8). This exon is not present in type III collagen mRNA, but is found at the 5[prime]-end of a unique alternative transcript derived from the type III collagen gene; in the alternative transcript, exon 23A is spliced to exon 24, thus replacing exons 1-23. The alternative transcript cannot encode a normal type III collagen subunit because the substitution of exon 23A for exons 1-23 results in omission from the mRNA of the exons encoding the signal peptide, the N-terminal propeptide and one-third of the triple helical domain. Furthermore, the alternative transcript potentially encodes one or more small non-collagenous proteins (7). The alternative transcript is present as early as 2.5 days of embryonic development; it is subsequently found at low levels in many embryonic tissues, but it appears most abundant in limb mesenchyme and is found at high levels in cultured chondrocytes derived from embryonic growth cartilage (7).
We showed previously that a 300 bp DNA fragment (259 bp of intron 23 immediately preceding exon 23A and 41 bp of exon 23A) functions as an active promoter in chondrocytes which contain the alternative transcript, but is only 20% as active in skin fibroblasts, which do not contain significant amounts of this RNA (8). In the experiments described below, we demonstrate that the activity of this promoter requires an activation element containing a binding site for activator protein 1 (AP1) (9). However, this element does not appear to contribute to chondrocyte-preferential transcriptional activity, since both chondrocytes and fibroblasts contain AP1 and ablation of the AP1 site reduces promoter activity in both cell types. We also define a repressor element that appears to be responsible for the low level of promoter activity in fibroblasts. This element is bound by a leader binding protein 1 (LBP1)-like transcription factor (10); replacement of this element results in dramatically increased promoter activity in fibroblasts, but does not affect activity in chondrocytes.
MATERIALS AND METHODS
Construction of substitution mutants of the internal promoter
Substitution mutagenesis (11,12) was used to define important elements between nt -145 and -56 within the context of pC-259 (8), which includes 259 nt immediately preceding exon 23A and 41 nt of exon 23A inserted into the multiple cloning site of p0CATntLPA (p0CAT; 13); a diagram showing the location of the substitution mutations is provided in Figure 1. We initially substituted 18 nt at a time with the sequence 5[prime]-CATATGCTCGAGGTCGAC containing an XhoI site (bold); this linker was used previously for mutagenesis of the promoter in the human immunodeficiency virus-1 long terminal repeat (HIV LTR) (14), in which these sequences were transcriptionally neutral. We subsequently replaced individual transcription factor binding sites with linkers containing XhoI, SalI or StyI sites. A genomic HindIII fragment containing intron 23, including exon 23A (8; GenBank accession no. U72880), was used as the template for PCR amplification in most cases; the double mutant smAP/LBP was constructed by mutagenesis of the AP1 site using sm-73 as the template. For each mutant, two PCR products were generated, as described below.
Figure 1. Functional analysis of substitution mutants defines both activation and repression domains in the internal promoter. The diagram shows the internal promoter of the type III collagen gene ligated to the CAT gene in p0CAT (13); +1 is the most 5[prime] of the three identified transcription start sites (8). Potential transcription factor binding sites in the region analyzed (-145 to -56) are indicated. The substitution mutants are designated sm followed by the 5[prime]-most nucleotide altered or the transcription factor binding site replaced. In sm5[prime]AP1, six nucleotides upstream of the AP1 site were replaced; in smLBP-L, the left half-site of the potential LBP1 site was replaced; in smAP/LBP, both the AP1 and the potential LBP1 sites were replaced. Data are expressed as mean c.p.m. acetylated [14C]chloramphenicol per amount of cell lysate containing 0.6 U of [beta]-galactosidase activity ± SEM. The mean activity of pC-259 in chondrocytes (11 291 c.p.m.) represents 9.5% chloramphenicol acetylation.
PCR1. The first PCR reaction for each mutant used a common 5[prime] primer containing nt -269 to -252 (5[prime]-ATTTTGACCCGGG-ACGTG), which includes an SmaI site (bold). The 3[prime] primers for the first reaction differed for each mutant, containing 15 nt of linker sequence at the 5[prime]-end and 10-17 nt of promoter sequence (immediately upstream of the sequences to be replaced) at the 3[prime]-end; the sequences and location of these primers are provided in Table 1 (PCR1 3[prime] primers). These PCR products were digested with SmaI, which cuts at -259, and XhoI, SalI or StyI, which cut within the linkers.
Table 1. Primers used for construction of substitution mutants
The mutants are designated by sm (for substitution mutant), followed by the number of the first nucleotide replaced. The linker sequences containing XhoI, SalI or StyI recognition sequences are shown in bold; the recognition sequences used for digestion are underlined. The nucleotide numbers identify the regions of the promoter to which the primers anneal.
PCR2. The second PCR reaction for each mutant used a common 3[prime] primer containing nt +24 to +41 with an SphI site (bold) added to the 5[prime]-end (5[prime]-CAGGCATGCATCCAGTGAGATACCTCT). The 5[prime] primers for the second reaction differed for each mutant, containing 15 nt of linker sequence at the 5[prime]-end and 12-18 nt of promoter sequence (immediately downstream of the sequences to be replaced) at the 3[prime]-end (Table 1, PCR2 5[prime] primers). These PCR products were digested with XhoI, SalI or StyI, which cut within the linkers, and SphI, which cuts at +41.
For each mutant, the vector p0CAT (8,13) was digested with SmaI and SphI, which cut in the multiple cloning site upstream of the CAT gene, and ligated with the two relevant PCR products. All constructs were sequenced to confirm that the appropriate substitution mutations, and no others, had been introduced. The mutants are designated sm (substitution mutant), followed by the number of the 5[prime]-most nucleotide replaced (e.g. sm-145, in which nt -145 to -128 were replaced) or by the name of the transcription factor whose binding site was replaced (e.g. smAP1). Other mutants were: sm5[prime]AP1, with sequences between the helix-loop-helix (HLH) and AP1 sites replaced; smLBP-L, with the left half of the LBP1 site replaced; smAP/LBP, with both the AP1 and LBP1 sites replaced.
Functional analysis of promoter constructs
Chondrocytes from lower sternal cartilage of 18-day-old chick embryos (15,16) and fibroblasts from skin of 12-day-old embryos (17,18) were isolated, cultured and transfected with 12.4 µg of plasmid DNA in the presence of lipofectamine (Gibco BRL) as described previously (8,13). Cells were co-transfected with 0.5 µg of pCh110 (19), which expresses [beta]-galactosidase under control of the SV40 early promoter; we showed previously that chondrocytes and fibroblasts are transfected with comparable efficiency (8). CAT activity of cell extracts containing 0.6 U of [beta]-galactosidase activity was determined using xylene extraction (20). Each construct was analyzed in 2-8 replicate plates in 3-34 independent experiments using cells isolated from different batches of embryos; at least two preparations of each plasmid were analyzed. Activities of constructs within a single cell type were compared using a paired t-test, and activities of individual constructs in chondrocytes and fibroblasts were compared using an unpaired t-test; values of P < 0.05 were considered statistically significant.
Identification of potential transcription factor binding sites
Relevant regions of the DNA sequence were analyzed using the FindPatterns program (GCG Package v.9.1, Madison, WI) with the transcription factor database maintained by David Ghosh (21).
Electrophoretic mobility shift assays
Nuclear extracts were prepared from cultured lower sternal chondrocytes (15,16) and skin fibroblasts (17,18) using established procedures (22). Oligonucleotides containing consensus binding sites for AP1 and AP2 were obtained from Promega; all other oligonucleotides were synthesized by Biosynthesis Inc. (Lewisville, TX); the upper strand of each oligonucleotide is shown in Table 2. Double-stranded oligonucleotides were labeled with T4 polynucleotide kinase in the presence of [[gamma]-32P]ATP. The labeled oligonucleotides (1 × 104 c.p.m.) were incubated with 10 µg of nuclear extract with 2 µg of the non-specific competitor poly[d(I-C)] (Boehringer Mannheim) for 20 min at room temperature in 4% glycerol, 10 mM Tris, pH 7.5, 50 mM NaCl, 1 mM MgCl2, 0.5 mM EDTA, 0.5 mM DTT, 0.4 mM PMSF; protein-DNA complexes were analyzed by electrophoresis on non-denaturing 5% or 4-10% gradient polyacrylamide gels. Specificity of the binding reactions was determined by preincubation of nuclear extracts for 5 min with a 100-fold molar excess of the indicated competitor oligonucleotides. Components of AP1 in protein-DNA complexes were identified by addition of 1 µg of purified polyclonal antibodies (Santa Cruz Biotechnology) to the reactions. The antibody c-Fos (K-25)-G, raised against a highly conserved domain of human c-Fos that is identical in the chicken protein, is broadly reactive with all Fos family members. The antibody Fra-2 (Q-20), raised against a highly conserved domain of human Fra-2, cross-reacts with chicken Fra-2 and is specific for Fra-2. The antibody c-Jun/AP-1 (N)-G, raised against a conserved epitope of mouse c-Jun, cross-reacts with chicken c-Jun and is specific for c-Jun. A polyclonal antiserum against a recombinant form of the N-terminal half of human LBP1a (23) was generously provided by Dr Robert Roeder (Rockefeller University).
Table 2. Oligonucleotides used for electrophoretic mobility shift assays
Potential transcription factor binding sites are shown in bold; mutated nucleotides are underlined.
RESULTS
The internal promoter of the type III collagen gene contains both an activation domain and a fibroblast-specific repressor domain
We showed previously that a construct containing only 259 nt of the internal promoter was highly active in chondrocytes and significantly less active in fibroblasts (8); thus this region appears to contain most of the sequences necessary for chondrocyte-preferential activity. Analysis of 5[prime]-end deletion mutants indicated that nt -145 to -65 contain a domain that represses activity in fibroblasts (8), suggesting that this region may be responsible for chondrocyte-preferential promoter function.
The regulatory elements within this region were defined more precisely using linker substitution mutagenesis, systematically replacing 18 nt at a time between -145 and -56; this strategy allowed us to identify regulatory elements in their natural context, without removal of upstream sequences. The parent for the linker substitution mutants was pC-259 (Fig. 1), which was significantly more active in chondrocytes (n = 34) than in fibroblasts (n = 22) (P < 0.0004), and thus displayed the pre-viously observed differential activity between cell types.
Surprisingly, the first two substitutions, replacing nt -145 to -128 (sm-145) and -127 to -110 (sm-127), significantly decreased activity in chondrocytes compared with pC-259 (P < 0.002, n = 15, and P < 0.04, n = 4, respectively) (Fig. 1). Thus these mutations identify a previously unrecognized activation domain within nt -145 to -110 that appears to be required for the high level of promoter activity in chondrocytes; further characterization of this domain will be described below.
The next three substitutions, replacing nt -109 to -92 (sm-109), -91 to -74 (sm-91), and -73 to -56 (sm-73), had no effect on CAT activity in chondrocytes compared with pC-259 (Fig. 1). However, sm-73 significantly increased promoter activity in fibroblasts (P < 0.01, n = 14); the activity of sm-73 was similar in fibroblasts and chondrocytes and similar to the activity of the wild-type promoter pC-259 in chondrocytes. Thus this mutation identifies a domain within nt -73 to -56 that appears to repress transcription in fibroblasts; further characterization of this domain will be described below.
An AP1 site is required for activation of the internal promoter
The activation domain identified by the linker substitution mutants sm-145 and sm-127 (nt -145 to -110) contains potential binding sites for an HLH transcription factor (-140 to -135) and for AP1 (-122 to -116), as well as part of an AP2 site (-114 to -106) (Figs 1 and 2). To determine which of these sites are important for transcriptional activation, additional mutants were made in which only the nucleotides comprising the potential transcription factor binding sites were replaced.
Figure 2. Electrophoretic mobility shift assays identify AP1 as the major transcription factor binding to the activation element. (A) A point mutation in the AP1 site abolishes protein binding to the activation element. End-labeled oligonucleotides extending from -130 to -106 were incubated with chondrocyte (Chond.) or fibroblast (Fib.) nuclear extract and protein-DNA complexes were analyzed by electrophoresis. The oligonucleotides were: lanes 1 and 3, -130AP1, the wild-type sequence containing a consensus AP1 site (TGAGTCA); lanes 2 and 4, -130mAP1, containing a single nucleotide change in the AP1 site (AGAGTCA) that abrogates AP1 binding (24). (B) Antibodies against AP1 components supershift the protein-DNA complex formed on the activation element. The -130AP1 oligonucleotide was incubated with chondrocyte nuclear extract; identification of AP1 components in the protein-DNA complexes was accomplished by incubation with antisera against Fra-2 (lane 2), c-Fos (lane 3) or c-Jun (lane 4). (C) No significant protein binding was detected to an oligonucleotide containing the HLH site. Chondrocyte protein binding to an oligonucleotide extending from -153 to -128 (-153HLH), containing the HLH site, was compared with binding to -130AP1. (D) Nucleotides -128 to -130 are required for optimal AP1 binding. A nested set of oligonucleotides containing the AP1 site, with a common 3[prime]-end at -106 and 5[prime]-ends increasing from -127 to -133, were incubated with chondrocyte nuclear extract. The oligonucleotides used were: lane 1, -133AP1 (nt -133 to -106); lane 2, -130AP1 (-130 to -106); lane 3, -127AP1 (-127 to -106). Arrows indicate the major protein-DNA complex; the bracket denotes the antibody-shifted protein-DNA complexes. The diagram shows the sequence of the activation domain with the positions of the substitution mutants (mutants) and oligonucleotides used in gel shift assays (oligos) indicated below; × indicates a point mutation in the AP1 site. The shaded box identifies 3 nt (-130 to -128) that appear to be required for optimal AP1 binding.
Replacement of the HLH and AP2 sites had no significant effect on promoter activity compared with pC-259 (Fig. 1). However, replacement of 6 of the 7 nt constituting the AP1 site essentially abolished activity in chondrocytes (P < 0.0003, n = 22) (Fig. 1, compare smAP1 with pC-259), indicating that this site is essential for promoter function. Furthermore, smAP1 displayed decreased activity in fibroblasts as well (P < 0.008, n = 12), suggesting that, despite the low level of promoter activity in these cells, fibroblasts have the necessary AP1 components for transcriptional activation mediated through this site.
To determine whether the presumptive AP1 site was actually bound by AP1, an oligonucleotide extending from -130 to -106, encompassing the AP1 site (Fig. 2, -130AP1), was incubated with chondrocyte nuclear extract; because the AP1 and AP2 sites are separated by only 1 nt, it was necessary to include the AP2 site in this oligonucleotide as well. A single major protein-DNA complex was formed on this oligonucleotide (arrow in Fig. 2A, lane 1). Protein binding was abolished by a single nucleotide change in the AP1 site (TGAGTCA->AGAGTCA) in the mutant oligonucleotide -130mAP1 (Fig. 2A, lane 2); this mutation abrogates AP1 binding to its cognate site in the human osteocalcin promoter (24). Protein binding to the wild-type oligonucleotide was largely abolished by competition with an excess of unlabeled oligonucleotide containing a consensus AP1 site, but was unaffected by competition with an oligonucleotide containing a consensus AP2 site (not shown).
Confirmation that AP1 is likely to be the major factor binding to the -130AP1 oligonucleotide was obtained using several antibodies against AP1 components (Fig. 2B). AP1 is a family of dimeric proteins most often containing one Fos and one Jun family member (25). While mammals contain four Fos and four Jun family members, only four AP1 components, c-Fos, Fra-2, c-Jun and JunD, have been identified in chickens (26-30); all four have been identified in both chondrocytes (31) and fibroblasts (29,32-35).
An antibody against c-Fos that cross-reacts with all Fos family members supershifted or abolished nearly all the protein-DNA complexes formed with chondrocyte nuclear extract (Fig. 2B, lane 3); thus all the complexes formed on this oligonucleotide contain Fos family members. Antibodies specific for Fra-2 and c-Jun each supershifted a portion of the protein-DNA complex (Fig. 2B, lanes 2 and 4), indicating that at least some of the complexes contain these proteins as well. These results, together with the transfection data shown in Figure 1, indicate that AP1 containing both Fos and Jun family members is likely to be an important activator of the internal promoter in chondrocytes.
Fibroblasts also contain functional AP1, as indicated by formation of a similar major protein-DNA complex with the -130AP1 oligonucleotide (Fig. 2A, lane 3). Binding of fibroblast nuclear extract was abolished by a point mutation in the AP1 site in the oligonucleotide -130mAP1 (Fig. 2A, lane 4) and the protein-DNA complexes formed with fibroblast extract were supershifted by antibodies to c-Fos, Fra-2 and c-Jun (not shown). The presence of functional AP1 in fibroblasts is consistent with our observation that the promoter construct smAP1, which replaced the AP1 site with unrelated sequences, displayed reduced activity in fibroblasts compared with the wild-type promoter pC-259 (Fig. 1). Thus the low level of activity of the internal promoter in skin fibroblasts is not due to lack of functional AP1.
Sequences upstream of the AP1 site are required for optimal AP1 binding and promoter function
The substitution mutant sm-145 decreased promoter activity in chondrocytes (Fig. 1), indicating that at least a portion of nt -145 to -128 is important for activation of the internal promoter. The only potential transcription factor binding site identified in this region is an HLH site at nt -140 to -135. Since mutagenesis of the HLH site did not alter promoter function (smHLH in Fig. 1), an HLH transcription factor is unlikely to be required for promoter function. However, sm5[prime]AP1, in which 6 nt downstream of the HLH site (-134 to -129) were replaced, decreased promoter activity significantly in chondrocytes compared with pC-259 (P < 0.05, n = 7) (Fig. 1), indicating that these sequences are required for optimal promoter function.
No known transcription factor binding sites were found in this region, and no discrete protein-DNA complexes were formed with chondrocyte nuclear extract on an oligonucleotide including these sequences (-153HLH, extending from -153 to -125) (Fig. 2C, lane 1). This film was overexposed to demonstrate the absence of protein binding under conditions in which a large amount of AP1 bound to the adjacent oligonucleotide -130AP1 (Fig. 2C, lane 2).
We next used gel shift assays to determine whether sequences within nt -134 to -129, which were replaced by sm5[prime]AP1, contribute to optimal functioning of the AP1 site. We prepared a nested set of oligonucleotides containing the AP1 site with a common 3[prime]-end at -106 and 5[prime]-ends varying from -127 to -133. Oligonucleotides -133AP1 and -130AP1 displayed similar amounts of the expected protein-DNA complex (Fig. 2D, lanes 1 and 2). However, truncation of the oligonucleotide at the 5[prime]-end to -127 abolished AP1 binding (Fig. 2D, lane 3), suggesting that nt -130 to -128 stabilize binding of AP1 to its cognate site at -122 to -116. It is likely that the decreased activity of substitution mutants sm-145 and sm5[prime]AP1 (replacing nt -145 to -128 and -134 to -129, respectively) is due to replacement of these nucleotides upstream of the AP1 site that are required for optimal AP1 binding.
The results described above and shown in Figures 1 and 2 indicate that the activation element required for function of the internal promoter of the type III collagen gene contains an essential AP1 site at nt -122 to -116. The element extends from nt -130 (the 5[prime]-most nucleotide required for optimal AP1 binding) to -115 (the last nucleotide before the AP2 site, since replacement of the AP2 site had no effect on promoter activity).
The repressor element contains a potential binding site for the transcription factor LBP1
The substitution mutant sm-73 displayed increased activity in fibroblasts (Fig. 1), indicating that nt -73 to -56 contain an element that represses promoter activity in these cells; this mutation had no effect on activity in chondrocytes. Computer analysis of this sequence identified a potential inverted half-site for the transcription factor LBP1 (10), CCAGA, at nt -71 to -67 (Fig. 3). LBP1 most often recognizes two direct repeat half-sites with the consensus (A/T)CTG(G/C) separated by 5 nt (23); this consensus, and the sequence of the well-characterized binding site in the HIV LTR (10,23), are shown in Figure 3. While computer analysis did not identify a second LBP1 half-site, visual inspection identified a potential half-site in the opposite orientation 5 nt downstream (nt -61 to -57). Fortuitously, the substitution mutant sm-73, replacing nt -73 to -56, precisely replaced this potential LBP1 site (nt -75 to -57).
Figure 3. The repressor element contains a binding site for an LBP1-related protein. (A) An LBP1-related protein in fibroblasts binds to the repressor element. An oligonucleotide extending from -76 to -52 (-76LBP1) was incubated with fibroblast (Fib.) nuclear extract with the following additions: lane 1, none; lane 2, 100-fold molar excess of unlabeled competitor -76LBP; lane 3, 100-fold excess of mutant competitor -76mLBP-L; lane 4, 100-fold excess of mutant competitor -76mLBP-R; lane 5, 1 µl anti-LBP1 antiserum (23). (B) Chondrocytes also contain an LBP1-like factor. The oligonucleotide -76LBP was incubated with fibroblast (lanes 1-4) or chondrocyte (lanes 5-9) nuclear extract, with the following additions: lanes 1 and 6, wild-type -76LBP as competitor; lanes 2 and 7, mutant -76mLBP-LR as competitor; lanes 3 and 8, 0.5 µl anti-LBP1; lanes 4 and 9, 1 µl anti-LBP1; lane 5, none. The diagram shows the consensus LBP1 binding site consisting of two 5 bp direct repeats separated by 5 bp (where N is any nucleotide) and the binding site in the HIV LTR (10,23). The potential LBP1 site in the internal promoter of the type III collagen gene consists of two 5 bp inverted repeats separated by 5 bp. The half-site orientations are indicated by arrows; the positions of the substitution mutants and oligonucleotides used in gel shift assays are shown below.
To determine whether this DNA sequence functions as a binding site for an LBP1-like protein, an oligonucleotide extending from -76 to -52 (-76LBP) was subjected to electrophoretic mobility shift analysis. One major and several minor bands were observed with fibroblast nuclear extract (Fig. 3A, lane 1). The specificity of these complexes was demonstrated by competition with an excess of cold -76LBP oligonucleotide (Fig. 3A, lane 2), which abolished the major protein-DNA complex, as well as two of the minor complexes (large and small arrows).
Confirmation that an LBP1-like protein binds to the -76LBP oligonucleotide was obtained using an antiserum prepared against the highly conserved N-terminal half of human LBP1a (23). This antiserum supershifted or abolished all of the major and minor protein-DNA complexes formed with fibroblast nuclear extract in one experiment (Fig. 3A, lane 5) and a portion of the complexes in another (Fig. 3B, lanes 3 and 4). Thus fibroblasts contain an LBP1-like protein that binds to the repressor element identified by substitution mutagenesis (sm-73 in Fig. 1). These results suggest that an LBP1-like protein may play a role in repressing activity of the internal promoter of the type III collagen gene in fibroblasts.
Several observations indicate that the fibroblast LBP1-like protein bound to the repressor element differs from the LBP1 family members that regulate transcription from the HIV LTR. Previous studies showed that LBP1 requires two intact half-sites for effective DNA binding (23). However, oligonucleotides with mutations introduced into each half-site individually (-76mLBP-L and -76mLBP-R, for mutations in the left and right half-sites) competed effectively for binding to the wild-type -76LBP oligonucleotide (Fig. 3A, lanes 3 and 4). Oligonucleotides containing these half-site mutations did not compete for LBP1 binding to sites in the HIV LTR (23). Furthermore, although replacement of the entire LBP1 site relieved the repression of promoter activity in fibroblasts (sm-73 in Fig. 1), ablation of a single half-site did not (smLBP-L in Fig. 1). These results suggest that the LBP1-related protein in chick fibroblasts does not require two intact half-sites for DNA binding. However, an oligonucleotide with mutations in both half-sites (-76mLBP-LR) failed to compete for binding to the wild-type sequence (Fig. 3B, lane 2), indicating that at least one intact half-site is required for effective binding.
Surprisingly, incubation of chondrocyte nuclear extract with the -76 LBP oligonucleotide resulted in formation of a major protein-DNA complex similar to that observed with fibroblast extract (Fig. 3B, lane 5); the apparent quantitative difference between binding of fibroblast and chondrocyte nuclear proteins in this experiment was not reproducibly observed. There appear to be differences in the minor protein-DNA complexes formed with chondrocyte and fibroblast extracts (e.g. compare lanes 2 and 5 in Fig. 3B); the basis for these differences is unknown at present. Both major and minor chondrocyte protein-DNA complexes were partially supershifted or abolished by the LBP antiserum (Fig. 3B, lanes 8 and 9), indicating that an LBP1-related protein is present in chondrocytes as well.
Mutagenesis of both the activator and repressor elements confirms that AP1 functions as an activator in fibroblasts
We constructed the double mutant smAP/LBP by mutagenizing the AP1 site in the context of sm-73, in which the LBP1-like site was already replaced (Fig. 1). The activity of this double mutant was similar to that of smAP1, in which only the AP1 site was mutagenized. Thus the promoter is non-functional in the absence of the activation element whether the repressor element is present (in smAP1) or absent (in smAP/LBP). Furthermore, comparison of smAP/LBP with sm-73 confirms that AP1 functions as an activator for this promoter in fibroblasts. The activity of sm-73 (containing the activation element, but not the repressor element) is high in fibroblasts. Mutagenesis of the AP1 site in the context of sm-73 (in smAP/LBP) resulted in a 95% decrease in promoter activity (P < 0.03, n = 6).
DISCUSSION
The chick type III collagen gene is regulated by two promoters separated by at least 20 kb of DNA (8). Use of the two promoters is cell type-specific and gives rise to transcripts with significantly different structures (7) and, of necessity, different functions. The upstream promoter directs production of type III collagen mRNA, and thus is active in all cells that produce type III collagen (7). Although type III collagen is found in many connective tissues (1-5), it has not been observed in embryonic cartilage formed during endochondral ossification (36,37), and we have observed little detectable type III collagen mRNA in cultured chondrocytes derived from this cartilage (7,8,17). Furthermore, expression constructs including various portions of the upstream promoter of the chick type III collagen gene are active in skin fibroblasts, but inactive in chondrocytes (8), suggesting that chondrocytes lack a factor that is required for activation of this promoter or contain a repressor that prevents activation.
In contrast, the internal promoter of the type III collagen gene directs production of a unique 4 kb alternative transcript that is abundant in cultured chondrocytes, but is essentially undetectable in cultured skin fibroblasts (7,8,17). While the function of the alternative transcript is unknown (7), preliminary in vitro transcription/translation data (S.L.Adams, unpublished observations) suggest that the most likely protein product is a 20 kDa protein bearing some resemblance to a brain-specific protein of unknown function (38). We previously identified one major and two minor start sites for the alternative transcript (8) (shown schematically in Fig. 1). There is a potential imperfect TATA box, ATTTAA, 24 nt upstream from the major transcription start site and potential initiator elements (39) adjacent to all three start sites. The internal promoter is much more active in cultured chondrocytes than in skin fibroblasts (8; see also Fig. 1). In the experiments described above, we have shown that the chondrocyte specificity of this promoter is determined by a fibroblast-specific repressor, and have defined an activation element whose integrity is essential for promoter function.
An AP1 site is essential for activation of the internal promoter
Linker substitution mutagenesis of the internal promoter identified an activation element containing a potential AP1 site within nt -130 to -115 (Fig. 1). Several experiments suggest that AP1 is essential for activation of this promoter: (i) replacement of the AP1 site in the mutant smAP1 decreased promoter activity in chondrocytes to 14% of the wild-type promoter pC-259 (Fig. 1); (ii) a single nucleotide change that abrogates AP1 binding to the osteocalcin promoter (24) abolished protein binding to the activation element (Fig. 2A, lanes 2 and 4); (iii) protein-DNA complexes formed on an oligonucleotide containing the activation element were supershifted by antibodies against various AP1 components (Fig. 2B, lanes 2-4).
This is the first identified chondrocyte-preferential promoter whose activity requires AP1. Interestingly, numerous experiments have suggested that the c-Fos component of AP1 plays a critical role in cartilage development. During mouse embryogenesis, c-Fos is detected primarily in the growth regions of developing cartilage (40) and mice lacking c-Fos display growth retardation and disorganized growth cartilage (41). Additionally, retrovirus-mediated expression of c-Fos prevents maturation and hypertrophy of cultured chondrocytes (42) and causes skeletal dysplasia in vivo (43). Despite such experimental evidence suggesting a role for c-Fos in cartilage differentiation, the target genes for c-Fos in cartilage have not been identified, and chondrocyte-characteristic genes whose expression requires c-Fos specifically or AP1 in general have not previously been identified.
Although AP1 is required for activation of the internal promoter of the type III collagen gene, lack of AP1 is not responsible for the low level of activity in skin fibroblasts. Ablation of the AP1 site decreased promoter activity in fibroblasts, both in the presence and absence of the repressor element (smAP1 and smAP/LBP in Fig. 1). Furthermore, gel shift assays demonstrated that fibroblasts contain functional AP1 (Fig. 2), consistent with previous reports (29,32-35). Thus the sequences required for repression of transcriptional activity in fibroblasts must reside elsewhere in the promoter.
Repression of the internal promoter in fibroblasts requires an element containing an LBP1 site
Replacement of an 18 bp region extending from -73 to -56 relieved the repression of this promoter in fibroblasts. This repressor element contains two 5 bp inverted repeats separated by 5 bp (Fig. 3) that on initial examination appeared to form half-sites for an LBP1-like protein (10,23). LBP1 encompasses a family of transcription factors that regulate the activity of the HIV LTR (10,23,44), the simian virus 40 (SV40) late promoter (45-47), and the promoters of the [alpha]- (48-50) and [beta]-globin (51), MHC class II Ea and Dra (52) and P450 (53) genes. It has been referred to variously as LBP1 (10), untranslated region binding protein 1 (44), late stimulatory factor 1 (45), CP2 (48) and NF-E4 (51). LBP1 family members are related to a family of Drosophila transcription factors known as Elf-1 (54), NTF-1 (55) or grainyhead (56). There are multiple splice variants of these factors (23,47,54), including at least one form that does not bind DNA (23,47). The vertebrate family members most often recognize direct repeats with the consensus sequence (A/T)CTG(G/C) separated by 5 bp; the 10 bp center-to-center spacing appears to be important (23,46), suggesting binding on a single face of the DNA helix. However, the sequence requirements appear to be flexible, since the LSF1 binding sites in the SV40 late promoter bear little resemblance to the consensus sequence (46). Interestingly, although the characterized mammalian factors require two half-sites (23,46,56,58), the Drosophila counterparts bind to single half-sites (56,58).
The protein binding to the fibroblast-specific repressor of the internal promoter appears related to LBP1, since it is recognized by an antiserum prepared against a highly conserved domain of recombinant human LBP1a (23) (Figure 3A, lane 5, and B, lanes 3 and 4). However, it differs from previously described LBP1 family members in that it appears to require only a single half-site for binding. Oligonucleotides in which a single half-site was mutagenized did not compete with the wild-type oligonucleotide for protein binding (Fig. 3A, lanes 3 and 4) and mutagenesis of only the left half-site did not result in the expected derepression of transcriptional activity in fibroblasts (smLBP-L in Fig. 1). Thus this protein appears to more closely resemble the Drosophila counterpart NTF1 (56,58) than the previously characterized vertebrate LBP1 family members. Since avian LBP1 family members have not yet been cloned, we do not know whether this indicates a difference between the mammalian and avian factors or whether the avian protein partially characterized here represents a previously unknown LBP1 family member. Nonetheless, it seems likely that the two inverted half-sites may actually represent two individual protein binding sites, and that transcriptional repression requires one or the other site, but not both.
The fact that chondrocytes also contain an LBP1-related protein that binds to the repressor element is puzzling, since promoter activity is not repressed in these cells. It is possible that the LBP1-related proteins in chondrocytes and fibroblasts are not identical; for example, they may represent different splice variants or the products of different genes. Additionally, it is possible that the LBP1-related protein binding to the repressor element may display cell type-specific interactions with other transcription factors that confer its differential activity in fibroblasts and chondrocytes. Several observations indicate that LBP1 function can be modulated by interactions with other transcription factors, e.g. LBP1 interacts with YY1 to repress transcription from the HIV1 LTR (59) and activation of the [gamma]-globin gene promoter requires interaction of LBP1 with an unidentified developmentally regulated factor (60). In the absence of reagents specific for avian LBP1 family members, it is not possible at present to distinguish among these possibilities.
The internal promoter of the type III collagen gene differs from other chondrocyte-specific promoters
The sequences defined above as important for regulation of the internal promoter bear little obvious resemblance to elements that are important for chondrocyte-specific expression of other genes (13,61-66). The best-characterized such gene is that encoding type II collagen, the major fibril-forming collagen of cartilage. The chondrocyte specificity of type II collagen gene transcription is dictated in part by an enhancer in the first intron (67-72) and in part by silencer or repressor elements in the 5[prime] flanking region that repress activity in non-chondrocytic cells (73).
The internal promoter of the type III collagen gene resembles the regulatory sequences of the type II collagen gene in that it contains an enhancer or activation element that is required for the high level of activity in chondrocytes and a repressor that suppresses transcription in fibroblastic cells. However, the enhancer in the mouse gene is cooperatively activated by L-Sox5, Sox6 and Sox9 (74,75). Although a motif for binding of Sox proteins (A/T)(A/T)CAA(A/T)G is found in the internal promoter of the type III collagen gene at nt -425 to -419 (8), it seems unlikely that Sox proteins play a role in activation of the internal promoter, since pC-259, which is highly active in chondrocytes (Fig. 1), does not contain those sequences (8).
The silencer regions of the rat type II collagen promoter extend from -700 to -620 and from -460 to -360 nt upstream from exon 1 (73). It has been reported that HeLa cell nuclear extracts contain proteins that bind to these elements, while chondrocytes do not (73); however, the identity of these proteins has not been determined. Our analysis of the inhibitory sequences in the rat type II collagen gene identified the presence of several potential LBP1 half-sites. Thus it is possible that an LBP1-like protein may also play a role in chondrocyte-specific transcription of the type II collagen gene by repressing promoter function in non-chondrocytic cells.
ACKNOWLEDGEMENTS
We dedicate this manuscript to Vickie Bennett, a wonderful colleague and friend who initiated this project as a post-doctoral fellow in our laboratory. Vickie died January 5, 1998, after an 8 year battle with breast cancer, and is greatly missed by all of us. We gratefully acknowledge Jim Alwine and the other members of Yejia's thesis committee (Carolyn Gibson, Mon-Li Chu, Joel Rosenbloom and Maurizio Pacifici) for many helpful suggestions and Robert Roeder for generously providing the LBP1 antibody. This work was supported by NIH grants AR20553 and AR43162.
REFERENCES
*To whom correspondence should be addressed. Tel: +1 215 898 6569; Fax: +1 215 898 3695; Email: sherri{at}biochem.dental.upenn.edu
This page is run by Oxford University Press, Great Clarendon Street, Oxford OX2 6DP, as part of the OUP Journals
Comments and feedback: jnl.info{at}oup.co.uk
Last modification:
Copyright© Oxford University Press, 1999.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  H. Acloque, A. Mey, A. M. Birot, H. Gruffat, B. Pain, and J. Samarut Transcription factor cCP2 controls gene expression in chicken embryonic stem cells Nucleic Acids Res., April 23, 2004; 32(7): 2259 - 2271. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. M. Dorman and S. E. Johnson Activated Raf Inhibits Myogenesis through a Mechanism Independent of Activator Protein 1-mediated Myoblast Transformation J. Biol. Chem., August 25, 2000; 275(35): 27481 - 27487. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||