The chick type III collagen gene contains two ... - BioMedSearch

Download PDF
0 downloads 0 Views 168KB Size Report
Comment
ABSTRACT. Type III collagen is present in prechondrogenic mesen- chyme, but not in cartilages formed during endochon- dral ossification. However, cultured ...
2470–2477 Nucleic Acids Research, 1997, Vol. 25, No. 12

 1997 Oxford University Press

The chick type III collagen gene contains two promoters that are preferentially expressed in different cell types and are separated by over 20 kb of DNA containing 23 exons Yejia Zhang1, Zeling Niu, Arthur J. Cohen, Hyun-Duck Nah and Sherrill L. Adams* Department of Biochemistry, School of Dental Medicine and 1Graduate Program in Cell and Molecular Biology, University of Pennsylvania, Philadelphia, PA 19104-6003, USA Received January 28, 1997; Revised and Accepted May 5, 1997

ABSTRACT Type III collagen is present in prechondrogenic mesenchyme, but not in cartilages formed during endochondral ossification. However, cultured chick chondrocytes contain an unusual transcript of the type III collagen gene in which exons 1–23 are replaced with a previously undescribed exon, 23A; this alternative transcript does not encode type III collagen. This observation suggested that, although production of type III collagen mRNA is repressed in chondrocytes, transcription of the type III collagen gene may continue from an alternative promoter. To test this prediction, we isolated and characterized both the upstream and internal promoters of this gene and tested their ability to direct transcription in chondrocytes and skin fibroblasts. The upstream promoter is active in fibroblasts, but inactive in chondrocytes, indicating that repression of type III collagen synthesis during chondrogenesis is transcriptionally mediated. Additionally, sequences in intron 23, preceding exon 23A, function as a highly active promoter in chondrocytes; transcription from this promoter is repressed in fibroblasts. Thus transcriptional control of the type III collagen gene is highly complex, with two promoters separated by at least 20 kb of DNA that are preferentially expressed in different cell types and give rise to RNAs with different structures and functions.

INTRODUCTION Type III collagen, a fibril-forming collagen, is widely distributed in connective tissues in heterotypic fibrils with type I collagen (1–5). Type III collagen may play a role in chondrogenesis, since it has been observed in condensing prechondrogenic mesenchyme of the developing mouse limb (6) and in mesenchymal condensations formed during endochondral fracture repair (7–10). For example, it has been suggested that the type III

* To

DDBJ/EMBL/GenBank accession nos U72880 and AF00737

collagen-rich matrix of the condensing mesenchyme may be used as a scaffold for deposition of the cartilage matrix (7). However, type III collagen has not been observed in growth cartilages formed during endochondral ossification (7,11), indicating that, as mesenchymal cells differentiate and initiate production of the cartilage extracellular matrix, synthesis of this collagen is likely to cease. These results suggest that precisely regulated expression of the type III collagen gene may be important for cartilage development during endochondral bone formation. We recently reported that cultured chick chondrocytes from several embryonic growth cartilages, as well as limb mesenchyme from four day old embryos, contain a 4 kb alternative transcript of the type III collagen gene, much smaller than the 5.7 kb type III collagen mRNA (12) (Fig. 1A). The alternative transcript contains a previously undescribed exon, 23A, followed by most or all of exons 24–52, but lacks exons 1–23 (Fig. 1B). It contains at least three potential open reading frames, none of which encodes a normal type III collagen protein. These results suggested that the alternative transcript may arise through transcription initiation at an alternative promoter that is more active in cultured chondrocytes and limb mesenchyme than in other cells. Prior to the studies described below, virtually nothing was known about the molecular mechanisms that regulate type III collagen gene expression during chondrogenesis and prevent production of type III collagen in embryonic growth cartilage. Furthermore, no information was available regarding the location of exon 23A or the alternative promoter presumed to be responsible for production of the alternative transcript. Thus we isolated the upstream promoter of the chick type III collagen gene, which is responsible for production of type III collagen mRNA, and showed that it is active in skin fibroblasts, but inactive in chondrocytes, consistent with the repression of type III collagen synthesis during chondrogenesis and with the lack of type III collagen mRNA in chondrocytes. In addition, we isolated an internal promoter of the type III collagen gene that appears to be responsible for production of the 4 kb alternative transcript and demonstrated that this promoter is active in chondrocytes, but is repressed in fibroblasts.

whom correspondence should be addressed. Tel: +1 215 898 6569; Fax: +1 215 898 3695; Email: [email protected]

2471 Nucleic Acids Acids Research, Research,1994, 1997,Vol. Vol.22, 25,No. No.112 Nucleic

2471

U72880). This clone contained 1286 bp of intron 23, exon 24 and part of intron 24. Sequences in the GenBank/EMBL database homologous to intron 23 were identified and aligned using the FastA and BestFit programs of the Genetics Computer Group. Hybridization probes for library screening and Southern blotting were prepared by primer extension of template DNAs with the Klenow fragment of DNA polymerase I (15). Preparation of the exons 1–2 and 21–25 probes was described previously (12). The exon 1 template was prepared by PCR amplification of exon 1 using the 5′ primer 5′-CCGGTGCTGGAAGGGCAGGGA and the 3′ primer 5′-CGGGATCCCTGCCCTTCCAGCACCGG; the exon 23A template was prepared by amplification of exon 23A using the 5′ primer 5′-ATTTAAGGGTTGGCAGTGG and the 3′ primer 5′-GAATCCCAGGCAAACCCCTTAC. The probes were prepared by extension of the appropriate 3′ primers in the presence of [α-32P]dCTP. Cell culture and RNA isolation and analysis

Figure 1. Skin fibroblasts contain only the authentic type III collagen mRNA, while chondrocytes contain exclusively the alternative transcript. (A) A Northern blot containing RNAs from cultured skin fibroblasts (Fib.) and attached lower sternal chondrocytes (Chond.) was hybridized with a genomic type III collagen clone containing exons 47–49 (13,14). Only the 5.7 kb type III collagen mRNA was detected in fibroblasts, while chondrocytes contained only the 4.0 kb alternative transcript. (B) Diagrams demonstrating the structure of type III collagen mRNA, containing exons 1–52, and the alternative transcript, containing exons 23A and 24–52 (12).

MATERIALS AND METHODS Isolation and characterization of genomic clones containing the upstream and internal promoters of the chick type III collagen gene Clones containing the upstream promoter were isolated by screening a chick genomic DNA library (Clontech) using a cDNA probe containing exons 1 and 2; probe preparation is described below. The inserted DNA from positive plaques was digested with EcoRI and cloned into the EcoRI site of pBluescriptII/SK– (Stratagene). A 3.2 kb EcoRI fragment was identified on Southern blots by hybridization with an exon 1 probe and relevant portions were sequenced (GenBank accession no. AF00737). This clone contained 2264 bp of 5′ flanking sequence of the type III collagen gene, exon 1 and part of the first intron. Potential transcription factor binding sites were identified using the FindPatterns program and homology with the 5′ flanking regions of the mouse and human type III collagen genes was identified using the BestFit and Gap programs of the Wisconsin Package v. 8.1 of the Genetics Computer Group (Madison, WI). Clones containing exon 23A and the internal promoter were isolated by rescreening the DNA library with a cDNA probe containing exons 21–25. The inserted DNA from positive plaques was digested with HindIII and cloned into the HindIII site of pBluescriptII/SK–. A 2.0 kb HindIII fragment was identified on Southern blots by hybridization with an exon 23A probe and relevant regions were sequenced (GenBank accession no.

Chondrocytes isolated from lower sternal cartilage of 18 day old chick embryos were grown for 7 days in suspension (16,17), trypsinized and allowed to attach in the presence of 4 U/ml hyaluronidase (18). Attached chondrocytes displayed a significant amount of the alternative transcript (Fig. 1A), as described previously for suspended chondrocytes (12). Fibroblasts from skin of 12 day old embryos were isolated and cultured as described previously (19,20). RNA was isolated by the method of Chomczynski and Sacchi (21). Northern hybridization. Ten µg of total RNA from cultured attached chondrocytes and skin fibroblasts was denatured with formaldehyde/formamide and fractionated by electrophoresis in 1.5% agarose gels (22), transferred to a Nytran membrane (Schleicher and Schuell) and crosslinked to the membrane with a Stratalinker UV crosslinker (Stratagene). The blot was hybridized at 50C overnight and washed under standard conditions (23). The hybridization probe, a genomic clone containing exons 47–49 of the chick type III collagen gene (13,14), was labeled with [α-32P]dCTP by random priming (24). Methylene blue staining was used to confirm that the blot contained similar amounts of chondrocyte and fibroblast RNA. Primer extension. An oligonucleotide complementary to exon 24 (5′-AGGCTTCCCATCACTTCCAGGACT) was labeled with [γ-32P]ATP using T4 polynucleotide kinase and annealed with 15 µg of total RNA from sternal chondrocytes; primer extension was performed under standard conditions (23) at 50C for 1 h with 0.5 U/µl Superscript II reverse transcriptase (Gibco/BRL); the products were analyzed on denaturing 6% polyacrylamide gels. RNase protection. The RNase protection probe contained a portion of intron 23, including exon 23A; in addition, due to the small size of exon 23A (69 nt), sequences from exons 24 and 25 were added to increase the stability of the RNA–RNA hybrids. A three-step procedure was used to construct the clone for preparation of this probe. (i) A DNA fragment containing 61 bp of intron 23 and 61 bp of exon 23A was amplified by PCR using as a template the 2.0 kb HindIII fragment containing intron 23 described above. The 5′ primer contained sequences from intron 23 and included an EcoRI site (5′-CGGAATTCTCAGGTACATGCAATTCATC); the 3′ primer contained sequences from exon 23A (5′-AGCCTCAGGTACAAAGGGAT). (ii) A DNA fragment containing 20 bp of exon 23A, exon 24 and 86 bp of exon

2472 Nucleic Acids Research, 1997, Vol. 25, No. 12 25 was amplified using a previously described cDNA clone containing exons 23A–25 (12) as template. The 5′ primer (5′-GTACCTGAGGCTTTGTAAGG) was partially complementary to the exon 23A 3′ primer used in step 1; the 3′ primer contained sequences from exon 25 and included a SacI site (5′-GCGAGCTCGGACCAGGAAAACCCATTAC). (iii) The products of steps 1 and 2 were denatured and renatured, permitting annealing of the complementary sequences in exon 23A; the annealed products were reamplified using the intron 23 5′ primer and exon 25 3′ primer described above. The resulting 270 bp DNA fragment was cloned into the SmaI site of pBluescript II/SK– and sequenced. The plasmid was linearized with EcoRI and transcribed with T3 RNA polymerase in the presence of [α-32P]UTP to generate a 290 nt antisense RNA. The probe was annealed to 10 µg of total RNA from chondrocytes and skin fibroblasts and single-stranded RNA was removed by digestion with 40 µg/ml RNase A and 600 U/ml RNase T1 (25); protected RNAs were fractionated on a 6% denaturing polyacrylamide gel. Expression vectors containing the upstream and internal promoters of the chick type III collagen gene DNA fragments for analysis of the upstream promoter were derived from the exon 1-containing 3.2 kb genomic EcoRI DNA fragment described above. A 2308 bp DNA fragment extending from –2264 to +44 in exon 1 was prepared by PCR amplification using Vent DNA polymerase (New England Biolabs) with a 5′ primer from the T3 promoter region of pBluescriptII/SK– (Stratagene) and a 3′ primer from exon 1 (5′-CGGGGATCCCTGCCCTTCCAGCACCGG). This fragment was digested with BamHI to remove vector sequences, blunt-ended using the Klenow fragment of DNA polymerase I and cloned in both the forward and reverse orientations into the SmaI site of the expression vector pCATBasic1 to create pF–2264/+44 and pF+44/–2264, respectively. The F designation reflects the utilization of this promoter in fibroblasts. The pCATBasic1 vector was modified from pCATBasic (Promega) by replacement of the PstI–XbaI fragment of the multiple cloning site with the PstI–XbaI fragment of pBluescript II/SK–, which contains a SmaI site. Several 5′ end deletion mutants were prepared as follows. pF–999/+44 was created by digestion of pF–2264/+44 with PstI, followed by religation of the relevant fragment. pF–798/+44 and pF–239/+44 were constructed by PCR amplification using pF–2264/+44 as a template; the 5′ primers were 5′-GGGGTACCATGAACCTTCTTATTCTCAGTG and 5′-GGGGTACCTCACATCCATACATCTCG, respectively; the 3′ primer for both constructs was 5′-CTGCGATCTAAGTAAGCTT. The PCR products were digested with BglII to remove excess vector sequences, blunt-ended and cloned into the SmaI site of pCATBasic1. pF–580/+44 and pF–480/+44 were constructed by digestion of pF–798/+44 with PstI and partial digestion with AvrII. There are two AvrII sites, located at –581 and –481; thus religation of the relevant fragments produced two constructs containing 580 and 480 bp of 5′ flanking sequence. DNA fragments for analysis of the internal promoter were derived from the 2.0 kb HindIII fragment containing intron 23 described above. A 559 bp DNA fragment extending from –518 (relative to the previously identified 5′ end of exon 23A) to +41 was isolated by digestion with HindIII and EcoRV (which cuts within exon 23A), blunt-ended and cloned in both the forward and reverse orientations into the SmaI site of the expression vector

p0CATntLPA3Zf(+) (p0CAT) (18). These constructs are referred to hereafter as pC–518/+41 and pC+41/–518; the C designation reflects the utilization of this promoter in chondrocytes. pC–430/+41, pC–259/+41 and pC–64/+41 were made by digestion of pC–518/+41 with NdeI, SmaI and SacI, respectively, followed by religation of the relevant fragments. pC–145/+41 was made by PCR amplification using the 2.0 kb HindIII fragment as template with an intron 23 5′ primer containing a KpnI site (5′-AGTGGTACCAATGACATTTGTAGCAGA) and a 3′ primer from exon 23A (5′-AGCCTCAGGTACAAAGGGAT). The PCR product was digested with KpnI and EcoRV and cloned into the KpnI and SmaI sites of p0CAT. Functional analysis of promoter constructs Skin fibroblasts and chondrocytes attached as described above were transfected with plasmid DNAs in the presence of Lipofectamine (GIBCO BRL), as described previously (18). For analysis of the upstream promoter, 5 µg of pF–999/+44 or equimolar amounts of the other constructs were used; for analysis of the internal promoter, 13 µg of pC–518/+41 or equimolar amounts of the other constructs were used. Cells were cotransfected with 0.5 µg of pCh110 (26), which expresses β-galactosidase under control of the SV40 early promoter. CAT activity of cell extracts containing equal amounts of β-galactosidase activity was determined using thin layer chromatography (27) or xylene extraction (28). The activities of various constructs were compared pair-wise using an unpaired t-test; values of P ≤ 0.05 were considered statistically significant. RESULTS AND DISCUSSION The upstream promoter of the chick type III collagen gene is active in fibroblasts, but inactive in chondrocytes Since type III collagen is not found in growth cartilages formed during endochondral ossification (7,11), we predicted that the upstream promoter of the type III collagen gene, which gives rise to the type III collagen mRNA, would be inactive in chondrocytes isolated from embryonic sternal cartilage, a growth cartilage. To test this prediction, a DNA fragment containing 2264 bp of 5′ flanking sequence and 44 bp of exon 1 was cloned into the reporter plasmid pCATBasic1 in both orientations, producing pF–2264/+44 and pF+44/–2264 (Fig. 2A); these constructs were analyzed by transfection into embryonic chick skin fibroblasts, which contain type III collagen mRNA, and attached sternal chondrocytes, which do not (Fig. 1A). This DNA fragment directed a significant level of CAT activity in skin fibroblasts in the forward, but not in the reverse, orientation (Fig. 2B). The forward construct, pF–2264/+44, was 30% as active as pCATControl (Promega), in which CAT expression is driven by the SV40 early promoter and enhancer (not shown). The activity of pF–2264/+44 in chondrocytes was only 15% of that in fibroblasts (Fig. 2B), a significant difference (P ≤ 0.00005); its activity in chondrocytes was similar to the activity of pCATBasic1 without a promoter. The low activity of pF–2264/+44 in chondrocytes was not due to poor transfection efficiency in these cells, since pCATControl (Promega) was equally active in chondrocytes and fibroblasts (not shown). These results indicate that the DNA fragment in pF–2264/+44 constitutes

2473 Nucleic Acids Acids Research, Research,1994, 1997,Vol. Vol.22, 25,No. No.112 Nucleic

Figure 2. The upstream promoter of the type III collagen gene is active in skin fibroblasts, but inactive in chondrocytes. (A) A DNA fragment containing 2264 bp of DNA preceding exon 1 and sequential 5′ end deletions of that fragment were cloned into the expression vector pCATBasic1 (pCATB1). Constructs are identified by the nucleotide numbers encompassed in the promoter fragments. (B) Promoter constructs were analyzed by transfection into cultured skin fibroblasts and sternal chondrocytes (18). CAT activity is expressed as percent acetylation of chloramphenicol per unit of β-galactosidase activity; each value represents the average ± standard error of several experiments (indicated by the number n above each bar) performed with independent batches of primary cells in which each construct was analyzed in duplicate. n.d., not determined.

a functional promoter in skin fibroblasts and support our prediction that this promoter is inactive in chondrocytes. We began to identify the sequences responsible for preferential expression of this promoter in fibroblasts by constructing a series of 5′ end deletion mutants (Fig. 2A) which were also analyzed by transfection into fibroblasts and chondrocytes. Deletion of nt –2264 to –1000 in pF–999/+44 resulted in a 41% decrease in activity in skin fibroblasts (P ≤ 0.005), suggesting the existence of a positive element in this region that is partially responsible for promoter activity in these cells. Further deletion to –798 did not have an additional effect on promoter activity; construct pF–798/+44 retained a significant amount of activity in fibroblasts (53% of the full-length promoter), while activity in chondrocytes remained low (15% of the activity in fibroblasts, P ≤ 0.005). Therefore, a promoter containing 798 bp of 5′ flanking sequence is functional and appears to contain many of the sequences that are necessary for preferential transcription in fibroblasts.

2473

Figure 3. Sequence of the upstream promoter of the type III collagen gene. The sequence of the chick (C) gene from –799 to +44 is shown in its entirety; the sequences of the mouse (M) and human (H) genes from –239 to +44 are shown where they differ from the chick sequence. The 5′ flanking and exon sequences are shown in lower and upper case letters, respectively; dots indicate deletions. The arrowheads followed by nucleotide numbers indicate the 5′ ends of promoter constructs; a star indicates the 5′ boundary of the evolutionarily conserved region of the promoter. Potential transcription factor binding sites are overlined and labeled above the line.

Deletion of the sequences between –798 and –580 in pF–580/+44 resulted in a further 65% decrease in promoter function in fibroblasts (P ≤ 0.005), indicating that this region contains additional elements that contribute to the activity of the upstream promoter. This domain contains potential binding sites for the transcription factors Sp1, GATA-1, NF-1 and C/EBPβ (Fig. 3). The activity of pF–580/+44, pF–480/+44 and pF–239/+44 in skin fibroblasts was very low, comparable to the full-length promoter in the reverse orientation, pF+44/–2364. These results indicate that, while some sequences between –580 and +44 may be necessary for promoter function, they clearly are not sufficient in the absence of upstream elements. The low activity of all constructs in chondrocytes suggests that these cells may be missing one or more transcription factors that are essential for activity of the upstream promoter.

2474 Nucleic Acids Research, 1997, Vol. 25, No. 12 The proximal region of the upstream promoter of the type III collagen gene is highly conserved among chick, mouse and human The upstream promoter of the type III collagen gene from –187 (star in Fig. 3) to –1 is highly conserved across species. This region of the chick promoter displays 70% homology with the promoters of the mouse (29) and human (30) type III collagen genes; in contrast, the region of the chick promoter from –798 to –187 displays only 32 and 34% homology, respectively, with the promoters of the mouse and human genes. Two positive elements have been identified in the proximal promoter of the mouse type III collagen gene, one bound by an AP-1 like factor and one bound by a newly characterized factor, BBF (31,32). These elements are highly conserved in the upstream promoter of the chick type III collagen gene, suggesting that they may be important for its function as well. Additional highly conserved sequences in this region include a binding site for a member of the basic helix–loop–helix (bHLH) class of transcription factors (CAGATG) embedded within a glucocorticoid response element (GRE), an AP2 site and a domain including a modifed TATA box (TATTTAA). The high degree of evolutionary conservation of these sequences suggests that they are likely to be important in transcriptional regulation of the upstream promoter of the chick type III collagen gene. However, these elements are obviously not sufficient to activate this promoter, since pF–239/+44, which contains all of the evolutionarily conserved elements, is essentially inactive in skin fibroblasts. This region of the chick promoter also contains a potential CCAATT box at –40 to –35 nt, much closer to the transcription start site than this element is normally found (33). The human (30) and mouse (29,34) type III collagen genes do not contain a CCAATT box at this site due to a 2 nt substitution, resulting in the sequence CCAAAC. Thus the functionality of the CCAATT box in the upstream promoter of the chick type III collagen gene remains to be determined. The only species in which the function of the upstream promoter of the type III collagen gene has been analyzed previously is the mouse. A fragment of the mouse type III collagen gene promoter extending from –250 to +16 was active in fibroblastic NIH 3T3 cells (35); no other cell types were analyzed. Despite the similarities in the proximal promoters of the chick and mouse genes (Fig. 3), the chick promoter construct pF–239/+44 is inactive in skin fibroblasts and appears to require sequences between –798 and –580 for activity in these cells (Fig. 2B). The reason for this difference is not known; it may be due to the divergent nature of the sequences upstream of –187 or to differences between primary chick skin fibroblasts and mouse 3T3 fibroblasts. The alternative transcript of the type III collagen gene in chondrocytes initiates within intron 23 Although embryonic growth cartilages do not contain type III collagen, the gene appears to be actively transcribed in chondrocytes to produce an alternative transcript in which exons 1–23 have been replaced by exon 23A (Fig. 1). We predicted that the appearance of the alternative transcript results from transcription initiation at an alternative promoter. To test this prediction, intron 23 was isolated and characterized; a region essentially identical to the previously determined exon 23A cDNA sequence was found within the intron (Fig. 4), 699 bp upstream from exon 24.

Figure 4. Exon 23A is located within intron 23 of the type III collagen gene. The sequence of a portion of intron 23 is shown; exon and intron sequences are shown in upper and lower case letters, respectively. Nucleotides are numbered relative to the 5′ end of exon 23A previously determined by cDNA cloning (12). Potential transcription factor binding sites are underlined and labeled above the line; potential initiator elements are indicated by double underlines; shaded sequences (–250 to –147) are identical to a region of intron 1 of the chick α2(VI) collagen (39) and ovalbumin X (40) genes. The large and small stars indicate major and minor 5′ ends of the alternative transcript mapped by primer extension; the large and small open stars within circles indicate major and minor 5′ ends mapped by RNase protection. The arrowheads followed by nucleotide numbers indicate the 5′ ends of promoter constructs.

The only difference between the previously described cDNA sequence of exon 23A (12) and the intron 23 sequence described here is that the cDNA sequence initiated with the dinucleotide AA, while there is only one A in the intron sequence. It seems likely that the 5′ most A in the cDNA clones was added during PCR amplification (36). The sequences in intron 23 preceding exon 23A do not resemble a conventional promoter in that there is no CCAAT box or TATA box, nor is there a GC-rich region such as that often found in promoters without a TATA box (37); in fact the overall GC-content of this region of intron 23 is rather low (38%). However, there are several sequences that resemble transcriptional initiator elements with the consensus sequence PyPyA+1NT/APyPy (38) (double underline in Fig. 4). Potential initiator elements are found on either side of the previously predicted 5′ end of exon 23A and in the middle of the exon. Interestingly, there is a 104 bp region that displays 80% homology to a portion of intron 1 of the chick α2(VI) collagen gene (39) and intron 1 of the chick ovalbumin X gene (40) (shaded sequences in Fig. 4). This homology seems unusually high for intron sequences of distantly related or unrelated genes. Since the sequences preceding exon 23A do not resemble conventional promoter elements, we performed primer extension and RNase protection assays to determine whether exon 23A indeed represents the 5′ end of the alternative transcript or whether additional sequences could be present in the RNA that were not represented in our previously isolated cDNA clones. A

2475 Nucleic Acids Acids Research, Research,1994, 1997,Vol. Vol.22, 25,No. No.112 Nucleic

Figure 5. 5′ end mapping of the alternative transcript by primer extension and RNase protection. (A) An exon 24 primer was used for primer extension analysis. A 114 nt primer extension product was predicted based on the sequence of the previously isolated cDNA clones (12). (B) A cloned PCR product containing part of intron 23, exons 23A and 24 and part of intron 25 was used to prepare a probe for RNase protection analysis. A 209 nt protected fragment was predicted for the alternative transcript, based on the previously determined cDNA sequence, while a 140 nt fragment was predicted for the type III collagen mRNA. The faint band of 290 nt represents undigested full-length probe. (C) Structure of the clone used to make the probe for RNase protection and predicted fragments protected by the alternative transcript and type III collagen mRNA. Chond., chondrocyte; Fib., fibroblast.

primer complementary to exon 24 was predicted to direct synthesis of a 114 nt cDNA with chondrocyte RNA as the template if the alternative transcript initiates at the previously identified site at the 5′ end of exon 23A. Several primer extension products were observed, a major band of 84 nt and minor bands of 113 and 123 nt (Fig. 5A; Chond.); the transcription start sites predicted by the major and minor primer extension products are indicated by large and small stars, respectively, in Figure 4. The 113 nt product is consistent with the predicted size of 114 nt based on the previously determined cDNA sequence (12). The synthesis of multiple primer extension products suggests either the existence of multiple start sites for the alternative transcript or premature termination of cDNA synthesis, perhaps due to secondary structure of the template RNA. Nonetheless, this result indicated that few transcripts extend beyond the previously defined 5′ end of exon 23A.

2475

This finding was confirmed by the RNase protection assay shown in Figure 5B, utilizing a uniformly labeled riboprobe containing 61 nt of intron 23, exons 23A and 24, and 86 nt of exon 25 (Fig. 5C). The predicted size of the RNA fragment protected by the alternative transcript is 209 nt, based on the previously determined cDNA sequence (12). Three protected fragments of 211, 202 and 178 nt were observed (Fig. 5B; Chond.); the transcription start sites predicted by the major and minor RNase protection products are indicated by large and small open stars within circles in Figure 4. The longest protected fragment of 211 nt is consistent with the predicted size of 209 nt based on the previously identified cDNA clones (12). The site identified by the major protected fragment of 178 nt is identical to the major site identified by the primer extension assay (Fig. 5A). Type III collagen mRNA, which is abundant in skin fibroblasts (Fib.) and may be present in minute amounts in chondrocytes, protects a 140 nt fragment, as expected, because it does not contain exon 23A. The 5′ ends of the alternative transcript predicted by primer extension and RNase protection assays are in good agreement; furthermore, the potential 5′ ends identified by both methods coincide well with the sites of consensus initiator elements. These results suggest that there are multiple start sites for the alternative transcript; the major start site is within exon 23A, at +31 relative to the previously identified 5′ end of the exon. The existence of multiple start sites is not unusual for a promoter without a TATA box (reviewed in 37 and 41). The fact that the major transcription start site appears to lie within exon 23A, rather than at its 5′ end, is not inconsistent with the previously published cDNA sequence (12), since in that study only the longest clones were selected for sequence analysis. These results indicate that the region of intron 23 preceding exon 23A is likely to represent a functional promoter that gives rise to the alternative transcript in chondrocytes. The presumptive internal promoter is functional and is more active in chondrocytes than in fibroblasts To determine whether the sequences preceding exon 23A could function as a promoter, we inserted a 559 bp DNA fragment (containing 518 bp of 5′ flanking sequence and 41 bp of exon 23A) in both the forward and reverse orientations into the expression vector p0CAT, creating pC–518/+41 and pC+41/–518 (Fig. 6A). The activity of these constructs was examined in cultured chondrocytes, which contain the alternative transcript, and in skin fibroblasts, which do not. This DNA fragment was highly active in chondrocytes in the forward, but not the reverse, orientation (Fig. 6B); the activity of pC–518/+41 in chondrocytes is ∼10 times the activity of the upstream promoter, pF–2264/+44, in skin fibroblasts. Furthermore, the activity of pC–518/+41 was quite low in skin fibroblasts, 22% of the activity in chondrocytes (P ≤ 0.001). These results, together with the primer extension and RNase protection experiments described above, indicate that the sequences of intron 23 preceding exon 23A represent a functional promoter whose activity is repressed in skin fibroblasts. This internal promoter is likely to be at least 20 kb away from the upstream promoter of the type III collagen gene, and possibly as much as 30 kb, since the first intron of the gene appears to be quite large and has never been isolated in its entirety (42). We next constructed sequential 5′ end deletions of this promoter (Fig. 6A), to begin to identify the sequences responsible

2476 Nucleic Acids Research, 1997, Vol. 25, No. 12 for the observed pattern of expression. Functional analysis of these deletion mutants by transfection into chondrocytes and fibroblasts identified several regions of this promoter that are likely to play a role in its regulation. For example, deletion of the sequences from –518 to –430 and from –259 to –145 resulted in decreased activity in chondrocytes (P ≤ 0.05), suggesting the existence of positive elements in these regions. Interestingly, the region from –259 to –145 consists entirely of the sequences that are homologous to a portion of intron 1 of the chick α2(VI) collagen gene (39) and the ovalbumin X gene (40) (Fig. 4). While

the function of these sequences is not known, our results indicate that they may play a role in transcriptional control. In addition, deletion of the sequences from –430 to –259 resulted in a significant increase in activity in chondrocytes (P ≤ 0.05), suggesting the existence of one or more negative elements in this region. The construct pC–145/+41 retained some chondrocytepreferential activity, since it was about twice as active in chondrocytes as in fibroblasts (P ≤ 0.05). However, pC–64/+41 displayed a similar level of activity in both cell types. These results suggest that the activity of this promoter is repressed in fibroblasts and this repression involves sequences between –145 and –64. This domain contains numerous potential transcription factor binding sites, including a binding site for a bHLH factor and recognition sites for AP1 and AP2 (Fig. 4). The minimal promoter pC–64/+41, which was highly active in both cell types, contains binding sites for bHLH, AP1 and NF-1 transcription factors, as well as the initiator elements, some or all of which may be important for transcriptional activity. These results confirm that the alternative transcript arises through use of an internal promoter located within intron 23 which is more active in chondrocytes than in skin fibroblasts. This promoter is quite different from other promoters that are preferentially expressed in chondrocytes. The type II collagen gene requires enhancer sequences in the first intron to confer activity (43–46); the link protein gene is regulated in part by a glucocorticoid-response element and an AT-rich element (47,48). The chick α2(I) collagen gene contains an alternative promoter located in intron 2 which is expressed preferentially in chondrocytes (18,49). The sequences identified as important for transcriptional activity of these three promoters and enhancers bear no homology to the sequences in the internal promoter of the type III collagen gene identified in the work described here. Thus a variety of combinations of transcription factors and their cognate DNA sequences appear to confer preferential transcriptional activity in chondrocytes. Summary

Figure 6. The internal promoter of the type III collagen gene is highly active in chondrocytes, but much less active in skin fibroblasts. (A) A 516 bp DNA fragment preceding exon 23A and sequential 5′ end deletions of that fragment were cloned into the expression vector p0CAT and analyzed by transfection into cultured skin fibroblasts and sternal chondrocytes. (B) CAT activity is expressed as percent acetylation per 0.1 U of β-galactosidase activity. Each value represents the average ± standard error of several experiments (indicated by the number n above each bar) performed with independent batches of primary cells in which each construct was analyzed in duplicate.

The chick type III collagen gene displays a complex structure with a correspondingly complex pattern of transcriptional control (Fig. 7). It contains two promoters separated by at least 20 kb of DNA containing 23 exons. The upstream promoter, which gives rise to the authentic type III collagen mRNA, is active in skin fibroblasts, which synthesize type III collagen, but is inactive in sternal chondrocytes, which do not. Since prechondrogenic mesenchymal cells synthesize type III collagen (6), while differentiated growth plate chondrocytes do not (7,11), our results suggest that the transcriptional inactivation of the upstream promoter of the type III collagen gene in chondrocytes may be

Figure 7. The chick type III collagen gene contains two functional, cell-type preferential promoters. The upstream transcription start site at the beginning of exon 1, which gives rise to type III collagen mRNA, and the internal start site at the beginning of exon 23A, which gives rise to the alternative transcript, are indicated; they are separated by at least 20 kb of DNA containing 23 exons (42). Note that exon 23A is spliced out of the type III collagen mRNA.

2477 Nucleic Acids Acids Research, Research,1994, 1997,Vol. Vol.22, 25,No. No.112 Nucleic due to loss of a transcription factor that is required for activation of this promoter. The internal promoter of the chick type III collagen gene, which gives rise to the alternative transcript, is highly active in chondrocytes, but displays lower activity in skin fibroblasts. The function of the alternative transcript is not known. It displays a complex structure in terms of protein coding potential, with at least three open reading frames, two of which are out of frame with the collagen coding sequence; the third potentially encodes a truncated collagen that would be missing the first 576 amino acids (12). It is not possible at present to predict whether any of the open reading frames in the alternative transcript are translated to produce the predicted noncollagenous proteins or the truncated type III collagen; however, the collagen reading frame appears to be the least likely to be translated efficiently, since the initiation codon overlaps with the termination codon of the preceding open reading frame (50). The type III collagen gene is one of several that appear to have functions in addition to encoding collagen proteins. We demonstrated previously that the chick α2(I) collagen gene gives rise to an alternative transcript in cartilage that does not encode α2(I) collagen (49). In addition, it has recently been discovered that the human α1(XVI) collagen gene gives rise to an abundant testis-specific alternative transcript initiating in intron 11 that does not encode type XVI collagen (M.-L.Chu, personal communication). While the existence of alternative transcripts as products of these three genes suggests some common features, it is important to note that the structure and developmental appearance of the various alternative transcripts, as well as the regulation of the alternative promoters, are quite different. These results suggest that several members of the collagen gene family have independently evolved mechanisms utilizing alternative promoters and production of alternative transcripts to provide possible additional regulatory or structural functions. ACKNOWLEDGEMENTS We gratefully acknowledge the assistance of Kim Pallante for cell culture and development of transfection protocols, Christina Kelly for library screening, Jim Alwine for constructive suggestions, and Mon-Li Chu for sharing results prior to publication. Automated DNA sequence analysis was performed by the Polymer Analysis Laboratory in the School of Dental Medicine. This work was supported by a subproject of Program Project AR20553 to S.L.A.; A.J.C. and H.-D.N. were supported in part by training grant AR70490.

REFERENCES 1 Henkel,W. and Glanville,R.W. (1982) Eur. J. Biochem. 122, 205–231. 2 Keene,D.R., Sakai,L.Y., Bachinger,H.P. and Burgeson,R.E.. (1987) J. Cell. Biol. 105, 2392–2402. 3 Fleischmajer,R., Perlish,J.S., Burgeson,R.E., Shaikh-Bahai,F. and Timpl,R. (1990) Ann. N. Y. Acad. Sci. 580, 161–165. 4 Keene,D.R., Sakai,L.Y. and Burgeson,R.E. (1991) J. Histochem. Cytochem. 38, 59–69. 5 Birk,D.E. and Mayne,R.A. (1997) Eur. J. Cell Biol. 72, 352–361. 6 Silver,M.H., Foidart,J.-M. and Pratt,R.M. (1982) Differentiation 18, 141–149. 7 Lane,J.M., Suda,M., von der Mark,K. and Timpl,R. (1986) J. Orthop. Res. 4, 318–329. 8 Page,M., Hogg,J. and Ashhurst,D.E. (1986) Histochem. J. 18, 251–265.

2477

9 Multimaki,P., Aro,H. and Vuorio,E. (1987) Biochem. Biophys. Res. Commun. 142, 536–541. 10 Sandberg,M., Aro,H., Multimaki,P., Aho,H. and Vuorio,E. (1989) J. Bone Joint Surg. 71A, 69–77. 11 von der Mark,K. and von der Mark,H. (1977) J. Bone Joint Surg. 59, 458–464. 12 Nah,H.-D., Niu,Z. and Adams,S.L. (1994) J. Biol. Chem. 269, 16443–16448. 13 Yamada,Y., Mudryj,M. and de Crombrugghe,B. (1983) J. Biol. Chem. 258, 14914–14919. 14 Yamada,Y., Mudryj,M., Sullivan,M. and de Crombrugghe,B. (1983) J. Biol. Chem. 258, 2758–2761. 15 Nah,H.-D. and Upholt,W.B. (1991) J. Biol. Chem 266, 23446–2352. 16 Adams,S.L., Boettiger,D., Focht,R.J., Holtzer,H. and Pacifici,M. (1982) Cell 30, 373–384. 17 Pacifici,M., Golden,E.B., Iwamoto,M. and Adams,S.L. (1991) Exp. Cell Res. 195, 38–46. 18 Pallante,K.M., Niu,Z., Zhao,Y., Cohen,A.J., Nah,H.-D. and Adams,S.L. (1996) J. Biol. Chem. 271, 25233–25239. 19 Focht,R.J. and Adams,S.L. (1984) Mol. Cell. Biol. 4, 1843–1852. 20 Bennett,V.D. and Adams,S.L. (1987) J. Biol. Chem. 262, 14806–14814. 21 Chomczynski,P. and Sacchi,N. (1987) Anal. Biochem. 162, 156–159. 22 Lehrach,H., Diamond,D., Wozney,J.M. and Boedtker,H. (1977) Biochemistry 16, 4743–4751. 23 Sambrook,J., Fritsch,E.F. and Maniatis,T. (1989) Molecular Cloning: A Laboratory Manual, Cold Spring Harbor Laboratory, Cold Spring Harbor, NY. 24 Feinberg,A.P. and Vogelstein,B. (1983) Anal. Biochem. 132, 6–13. 25 Gilman,M. (1987) in Ausubel,F.M., Brent,R., Kingston,R.E., More,D.D., Seidman,J.G., Smith,J.G. and Struhl,K. (eds) Current Protocols in Molecular Biology. John Wiley and Sons, New York, pp. 4.7.1–4.7.8. 26 Herbomel,P., Bourachot,B. and Yaniv,M. (1984) Cell 39, 653–662. 27 Gorman,C.M., Moffat,L.F. and Howard,B. (1983) Mol. Cell. Biol. 2, 1044–1051. 28 Neumann,J.R., Morency,C.A. and Russian,K.O. (1987) BioTechniques 5, 445–447. 29 Liau,G., Mudryj,M. and de Crombrugghe,B. (1985) J. Biol. Chem. 260, 3773–3777. 30 Benson-Chanda,V., Su,M-W., Weil,D., Chu,M-L. and Ramirez,F. (1989) Gene 78, 255–265. 31 Ruteshouser,C. and de Crombrugghe,B. (1989) J. Biol. Chem. 264, 13740–13744. 32 Ruteshouser,C. and de Crombrugghe,B. (1992) J. Biol. Chem. 267, 14398–14404. 33 Bucher,P. (1990) J. Mol. Biol. 212, 563–578. 34 Hatamochi,A., Paterson,B. and de Crombrugghe,B. (1986) J. Biol. Chem. 261, 11310–11314. 35 Mudryj,M. and de Crombrugghe,B. (1988) Nucleic Acids Res. 16, 7513–7526. 36 Clark,J.M. (1988) Nucleic Acids Res. 16, 9677–1986. 37 Basu,A., Park,K., Atchison,M.L., Carter,R. and Avadhani,N.G. (1993) J. Biol. Chem. 268, 4188–4196. 38 Javahery,R., Khachi,A., Lo,K., Zenzie-Gregory,B. and Smale,S.T. (1994) Mol. Cell.Biol. 14, 116–127. 39 Hayman,A.R., Koppel,J. and Trueb,B. (1991) Eur. J. Biochem. 197, 177–184. 40 Heilig,R., Muraskowsky,R. and Mandel,J.-L. (1982) J. Mol. Biol. 156, 1–19. 41 Gilinger,G. and Alwine,J.C. (1993) J. Virol. 67, 6682–6688. 42 Yamada,Y., Liau,G., Mudryj,M., Obici,S. and de Crombrugghe,B. (1984) Nature 310, 333–337. 43 Horton,W., Miyashita,T., Kohno,K., Hassell,J.R. and Yamada,Y. (1987) Proc. Natl. Acad. Sci. USA 84, 8864–8868. 44 Wang,L., Balakir,R. and Horton,W.E.,Jr (1991) J. Biol. Chem. 266, 19878–19881. 45 Mukhopadhyay,K., Lefebvre,V., Zhou,G., Garofalo,S., Kimura,J.H. and de Crombrugghe,B. (1995) J. Biol. Chem. 270, 27711–27719. 46 Krebsbach,P.H., Nakata,K., Bernier,S.M., Hatano,O., Miyashita,T., Rhodes,C.S. and Yamada,Y. (1996) J. Biol. Chem. 271, 4298–4303. 47 Rhodes,C. and Yamada,Y. (1995) Nucleic Acids Res. 23, 2305–2313. 48 Doege,K.J., Garrison,K., Coulter,S.N. and Yamada,Y. (1994) J. Biol. Chem. 269, 29232–29240. 49 Bennett,V.D. and Adams,S.L. (1990) J. Biol. Chem. 265, 2223–2230. 50 Kozak,M. (1987) Mol. Cell. Biol. 7, 3438–3445.