Separate Cis-acting DNA Elements Control Cell Type- and Tissue-specific Expression of Collagen Binding Molecular Chaperone HSP47*

HSP47 is a collagen-binding heat shock protein and is assumed to act as a molecular chaperone in the biosynthesis and secretion of procollagen. As the synthesis of HSP47 is closely correlated with that of collagen in various cell lines and tissues, we performed a promoter/reporter assay using HSP47-producing and nonproducing cells. 280 base pairs (bp(s)) of upstream promoter were shown to be necessary for the basal expression but not to be enough for the cell type-specific expression. When the first and the second introns were introduced downstream of this 280-bp region, marked up-regulation of the reporter activity was observed in HSP47-producing cells but not in nonproducing cells. This was confirmed in transgenic mice by staining the lacZ gene product under the control of the 280-bp upstream promoter and the introns. Staining was observed in skin, chondrocytes, precursor of bone, and other HSP47/collagen-producing tissues. A putative Sp1-binding site at −210 bp in the promoter, to which Sp3 and an unidentified protein bind, was shown to be responsible for this up-regulation when combined with the introns. However no difference in the binding to this probe was observed between HSP47-producing and nonproducing cells. The responsible region for cell type-specific up-regulation was found to be located in a 500-bp segment in the first intron. On electrophoresis mobility shift assay using this 500-bp probe, specific DNA-protein complexes were only observed in HSP47-producing cell extracts. These results suggest that two separate elements are necessary for the cell type-specific expression of the hsp47 gene; one is a putative Sp1-binding site at −210 bp necessary for basal expression, and the other is a 500-bp region within the first intron, required for cell type-specific expression.

Heat shock proteins (HSPs) 1 are a highly conserved set of proteins that can be induced by heat shock and other environmental stresses. Generally, most stress proteins are expressed not only under stress conditions but also under normal growth conditions and known to play important roles in protein folding and assembly. They are now generally known as molecular chaperones. The mechanism for heat induction of HSPs is well studied and now known to be transcriptionally regulated by the interaction of heat shock factors with heat shock elements (HSEs), which locate in the promoter region of HSPs (1).
The expression of HSP47 is markedly increased by various stresses including heat shock. This heat induction is due to the presence of a well conserved HSE at Ϫ80ϳϪ62 bp in the promoter (16,17). On the other hand, the constitutive expression of HSP47 appears to be co-regulated with that of several types of collagens (15). For example, the synthesis of HSP47 and type IV collagen are coordinately increased during the differentiation of murine F9 teratocarcinoma cells (16,18), and the synthesis of both HSP47 and type I collagen are decreased following viral transformation of the fibroblasts. Collagen-nonproducing cells such as mouse myeloid leukemia M1cells and rat pheochromocytoma PC12 cells do not synthesize HSP47 (19,20). Immunocytochemical analysis has revealed that during murine tooth development, HSP47 synthesis coincides with type I collagen production (21,22) and that HSP47 is localized within developing tissues or cells that secrete various types of collagens (23,24). Masuda et al. (25) reported that the synthesis of HSP47 as well as of type I and III collagen increases during the progression of carbon tetrachloride-induced liver fibrosis. Marked induction of HSP47 was also reported in the fibrosis of kidney (26,27). These results suggest that the expression of HSP47 under nonstressed conditions is co-regulated with that of collagens in various cell lines and tissues as well as in pathophysiological conditions.
In this paper, we analyzed the promoter activity of the hsp47 gene in terms of co-expression with collagen and show that the basal promoter activity of HSP47 exists within a 280-bp promoter region, which was identified as a putative Sp1-binding site, and the introduction of a downstream intron region caused marked up-regulation of its activity only in HSP47/collagenproducing cells but not in nonproducing cells.
* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

MATERIALS AND METHODS
Screening of the Mouse Genomic DNA Library and Long PCR-The mouse genomic DNA library was purchased from CLONTECH.The recombinant phages carrying mouse DNA fragments were generated by partial digestion with Sau3AI followed by ligation to the EMBL-3 arms. This library was screened using a 1.
Cell Culture-Mouse BALB/c3T3 fibroblasts, human embryonal kidney 293 cells, human epitheloid carcinoma HeLa cells, human osteosarcoma HOS cells, and mouse embryonal carcinoma F9 cells were cultured in Dulbecco's modified Eagle's medium (with low glucose) supplemented with 10% fetal bovine serum. Mouse osteoblastic MC3T3-E1 cells were cultured in minimum essential ␣ medium supplemented with 10% fetal bovine serum. Human acute T cell leukemia Jurkat cells were cultured in RPMI 1640 medium supplemented with 10% fetal bovine serum.
Construction of Luciferase Expression Plasmids-DNA from the hsp47 gene containing 5.5 kb of upstream promoter and 38 bp of exon I was cloned into a luciferase expression vector, pGL2-basic (Promega), to create pLUC5.5. pLUC2.2, pLUC280, and pLUC13 were generated from pLUC5.5 using the NcoI restriction site at Ϫ2.2, SacII site at Ϫ280, and SmaI site at Ϫ13, respectively. pLUC50 was generated by PCR cloning using the pLUC50 primer (5Ј-AAGTCGACGGGGGTGGG-GCCAGCC-3Ј). pLUC890 and pLUC210 were generated with a doublestranded nested deletion kit (Amersham Pharmacia Biotech). Introncontaining constructs were generated by introducing a SalI restriction site just before the translational start site in the third exon of the hsp47 gene to create pLUC5.5(III) and pLUC280(III). pLUC280(III)Sp1m was generated by site-directed mutagenesis (28). Constructs lacking various regions in the introns were generated by using restriction sites as shown in Fig. 9. All plasmids were purified by ultracentrifugation twice and/or a Qiagen Maxi Kit, and it was confirmed that the transfection efficiencies were equal.
Transient Transfection and Luciferase Assay-Cells were plated at a density of 7 ϫ 10 4 /35-mm culture dish 16 h before transfection and transfected by the calcium phosphate method with 5 g of the reporter plasmid and 1 g of pact-␤-galactosidase plasmid bearing the chicken ␤-actin promoter upstream of the lacZ gene (29) to normalize the differences in transfectional efficiencies. The precipitate was removed after 4 h, and fresh medium was added. At 48 h after the transfection, cells were harvested to perform the luciferase assay. Quantitation of luciferase activities was carried out by Lumat LB9501 (Berthold). ␤-Galactosidase assay was performed according to Maniatis et al. (28). Transfection into Jurkat cells was performed with LipofectAMINE (Life Technologies, Inc.).
Generation of Transgenic Mice-Transgenic mice were produced by microinjecting each of the linearized DNA (5.5(III)Z␤A, 280(III)Z␤A, 280Z␤A) into the pronuclei of fertilized eggs from F1 hybrid mice (C57BL/6 ϫC3H) as described (30). Founder mice were identified by PCR assays of genomic DNA extracted from the tail. The DNA was subjected to lacZ-specific PCR with a set of primers to amplify an 822-bp product (31). Positive founder mice were mated with wild-type mice to generate hemizygous embryos, and positive embryos were identified by PCR using placenta DNA.
Electrophoresis Mobility Shift Assay-Cell extract (3 g) in 20 ml of gel shift buffer (20 mM HEPES, pH 7.9, 1 mM MgCl 2 , 60 mM KCl, 12% glycerol) was incubated on ice for 15 min in the presence of 1 g of poly(dI-dC), and then 32 P-end-labeled probes were added, and the mixture was incubated at room temperature for an additional 20 min with or without nonradiolabeled competitors. Protein-DNA complexes were resolved on 4% polyacrylamide gels in electrophoresis buffer (22.5 mM Tris, 22.5 mM boric acid, 0.5 mM EDTA) and electrophoresed for 2 h. The sequences of the oligonucleotide probe and competitors are given below: HSP47Sp1 spanning nucleotides Ϫ228ϳϪ194, CCCCCTCTACCTAGT-AGGGCGAGGGGGCAGTAGGAC; HSP47Sp1m, CCCCTCTACCTAG-TAGAATGAGGGGGCAGTAGGAC; consensus Sp1, TGGATTCGATC-GGGGCGGGGCGAG. Antibodies against Sp1, Sp2, Sp3, and Sp4 were purchased from Santa Cruz Biotechnology. In the case of electrophoresis mobility shift assay (EMSA) using the 500-bp intron, protein-DNA complexes were resolved on 2.5% polyacrylamide gel.

Cloning of the Promoter Region of the Mouse hsp47
Gene-We used a 1.2-kb promoter region of the hsp47 gene (17) to screen a mouse genomic library generated by partial digestion with Sau3AI (CLONTECH). Four independent positive clones were isolated from 2 ϫ10 6 phage, and the restriction map was determined for the clone (Fig. 1). This clone contained a 5.5-kb promoter region and the entire hsp47 gene. Long distance PCR of mouse BALB/c3T3 genomic DNA confirmed that this clone does not contain chimeric inserts (data not shown). We used this clone for subcloning and further analysis.
Determination of a Minimum Promoter in the Upstream Re- gion-Mouse BALB/c3T3 fibroblasts produce high levels of both type I collagen and HSP47 (6). On the other hand, human embryonal kidney 293 cells do not produce any type of collagen nor HSP47 (33,34). We confirmed this by immunoblot analysis as shown in Fig. 2A using anti-HSP47 antibody and anti-type I collagen antibody, which can react with both mouse and human HSP47 and type I collagen, respectively (also see Fig. 4A). DNA sequences from Ϫ5.5 kb to ϩ38 bp were fused to the luciferase gene to create a fusion plasmid (pLUC5.5) to determine whether DNA sequences located in the 5Ј-flanking region of the hsp47 gene are involved in its cell type-specific expression. Several 5Ј-deletion constructs (pLUC2.2, pLUC890, pLUC280, pLUC50, and pLUC13) were also created as shown in Fig. 2B. These plasmids were transiently transfected into HSP47/collagen-producing BALB/c3T3 cells or nonproducing 293 cells by the calcium phosphate method, and luciferase assays were performed after 48 h. Luciferase activity in both cells was normalized by the luciferase activity of the pSV40-LUC carrying both the promoter and enhancer of SV40. In both BALB/ 3T3 and 293 cells, the luciferase activity of pLUC280 containing HSE and putative Sp1-binding sites was higher than that of pLUC50, which contained only a TATA box. The luciferase activity was hardly detected in pLUC13, which did not a contain TATA box. In both BALB/c3T3 and 293 cells, pLUC5.5 did not exhibit any more activity than pLUC280. The luciferase activity of other 5Ј-deletion constructs (pLUC2.2 and pLUC890) was comparable with that of pLUC5.5 or pLUC280. Thus, the 280-bp promoter was postulated as a minimum promoter. However, no differences between BALB/c3T3 cells and 293 cells were observed in the luciferase activity of pLUC280 or pLUC5.5. These results suggest that 5.5 kb or 280 bp of the upstream region, which confers the basic promoter activity, is not sufficient for the cell type-specific transcriptional activation.
Introduction of the Intron Region Up-regulates the Promoter Activity-The hsp47 gene consists of 6 exons separated by 5 introns, and the translational start site is in the third exon (17,35). We introduced the first and second introns (3.7 kb containing the second exon) between the 280 bp or 5.5 kb upstream promoter region and the luciferase gene (Fig. 3). Transient transfection and luciferase assays were performed in BALB/ c3T3 cells and 293 cells. The luciferase activity of intron-containing constructs (pLUC280(III) and pLUC5.5(III)) was 5ϳ7fold higher than that of the constructs without introns (pLUC280 and pLUC5.5, respectively) in BALB/c3T3 cells. This up-regulation in the presence of the introns was not observed in 293 cells. pLUC280(␤-glo) carrying the 280-bp promoter of hsp47 and a rabbit ␤-globin intron did not show the level of activity seen for pLUC280(III), suggesting that the splicing per se is not important for this up-regulation. These results suggest that there is a positive regulatory element(s) in the introns. Up-regulation by the Introns Is Observed in HSP47/Collagen-producing Cells-To confirm whether the difference in the up-regulation of the luciferase activity between BALB/c3T3 cells and 293 cells is also generally observed between HSP47producing cells and nonproducing cells, we performed the luciferase assay using pLUC280 and pLUC280(III) in other cell lines. As shown in Fig. 4A, mouse BALB/c3T3 fibroblasts, mouse osteoblastic MC3T3-E1 cells, human epitheloid carcinoma HeLa cells, and human osteosarcoma HOS cells synthesize both HSP47 and type I collagen at high levels. On the other hand, neither HSP47 nor collagen was detected in mouse embryonal carcinoma F9 cells, human embryonal kidney 293 cells, and human acute T cell leukemia Jurkat cells. Fig. 4B shows the luciferase activity in these 7 cell lines. In BALB/c3T3 cells, MC3T3-E1 cells, HeLa cells, and HOS cells, the luciferase activity of pLUC280(III) was 5ϳ7-fold pLUC280. On the other hand, the luciferase activity of pLUC280(III) was comparable with that of pLUC280 in F9 cells, 293 cells, and Jurkat cells. This result indicates that pLUC280(III) exhibits a high level of luciferase activity in HSP47/collagen-producing cells but not in nonproducing cells.
HSP47 lacZ Transgenes Direct Tissue-specific Expression in Vivo-To assess the pattern of expression of transgenes in vivo, we stained transgenic embryos with X-gal. Fig. 5 shows lacZ constructs containing various cis-regions of the hsp47 gene. All constructs carry 0.7 kb of the rabbit ␤-globin intron and 3.3 kb of the hsp47 3Ј-region containing the poly(A) additional signal. The number of copies of the transgenes in each line was determined by Southern blot analysis (Table I). Three out of four lines carrying the construct containing the 5.5-kb promoter and introns exhibited ␤-galactosidase activity. Three out of five lines carrying the construct containing the 280-bp promoter and introns exhibited marked ␤-galactosidase activity. However, no significant staining was detectable in the lines with the 280-bp promoter (280Z␤A). 5.5(III)Z␤A, containing the 5.5-kb promoter and introns, directed lacZ expression exclusively in the skin (Fig. 6a). Deletion of 5.2 kb of the upstream sequence of the promoter region (280(III)Z␤A) did not affect the pattern of lacZ expression in the skin (Fig. 6b). On the other hand, further deletion of the intron sequence (280Z␤A) led to the elimination of the lacZ expression completely (Fig. 6c). By E13.5, chondrogenesis had commenced in the limbs, axial skeleton, and thorax. By removing the skin, staining of the axial skeleton and thorax are detectable in mice harboring 5.5(III)Z␤A and 280(III)Z␤A constructs but not in mice harboring 280Z␤A (Fig. 6, d and e compared with f). The radius, ulna, humerus, and scapula of the forelimb (Fig. 6, g and h compared with i) and the fibula, tibia, femur, and ilium of the hind limb (j and k compared with l) are also stained in mice harboring 5.5(III)Z␤A and 280(III)Z␤A constructs at this stage after skin removal. Neither carpal bones nor tarsal bones were stained in both embryos for unknown reasons. Other areas of transgene expression included the external portion of the developing ear and connective tissue surrounding developing hair follicles in the nose. This staining pattern is very similar to that reported for embryos expressing lacZ under the promoter of pro␣2(I), pro␣1(II), and pro␣2(XI) collagens (36 -38). These results indicate that the 280-bp upstream promoter region combined with the first and the second introns contains the cis-regulatory elements necessary for the tissue-specific expression of the hsp47 gene.
A Putative Sp1-binding Site at Ϫ210 Is Necessary for the High Level Promoter Activity in BALB/c3T3 Cells-To narrow down the essential region of the upstream promoter, we prepared several 5Ј-flanking constructs. Transient transfection and luciferase assays were performed in BALB/c3T3 cells (Fig.  7A).The luciferase activity of pLUC210(III) was obviously lower than that of pLUC280(III) and comparable with that of pLUC210 or pLUC280. A 70-bp region (Ϫ280ϳϪ210) is necessary for the high level expression of luciferase activity. In this 70-bp region a putative Sp1-binding site was located at Ϫ210. We created a putative Sp1 site-mutated construct (pLUC280(III)Sp1m) from pLUC280(III) by mutagenesis (28) and performed the luciferase assay in BALB/c3T3 cells (Fig.  7B). The luciferase activity of pLUC280(III)Sp1m was obviously lower than that of pLUC280(III) and comparable with that of pLUC280. This indicates that a putative Sp1-binding site at Ϫ210 is necessary for strong transcriptional activity specific for HSP47/collagen-producing cells.
Sp3 and an Additional Unidentified Factor Bind to the Putative Sp1-binding Site at Ϫ210 -We performed an EMSA with a probe (hsp47 promoter) covering bases Ϫ225 to Ϫ196, which contained a single putative Sp1-binding site. Three major DNA-protein complexes were observed with HSP47-producing BALB/c3T3 (Fig. 8A, lane 4) and HeLa cell extracts (lane 6). The same band-shift was observed with the cell extracts from HSP47-nonproducing cells such as F9 and 293 cells (Fig. 8,  lanes 8 and 10). A DNA-protein complex with recombinant human Sp1 protein (Fig. 8A, lane 2) was supershifted by adding anti-Sp1 antibody (Fig. 8A, lane 3), whereas no supershifted bands were observed with BALB/c3T3, F9, HeLa, and 293 cell extracts (Fig. 8A, lanes 5, 7, 9, and 11). The labeled probe and BALB/c3T3 cell extract were incubated with an excess of unlabeled probe or antibodies. The shifted bands (Fig. 8B, lane 2) were competed by addition of excess unlabeled wild-type promoter probe (lane 3) and unlabeled consensus Sp1 binding probe (lane 5) but not by unlabeled mutated promoter probe (lane 4). The Sp1 gene family includes the Sp1 gene and three other genes encoding the Sp2, Sp3, and Sp4 proteins. The consensus binding sequences of these proteins are similar (39).
Monoclonal antibodies against Sp1 and polyclonal antibodies against Sp2, Sp3, and Sp4 were used for the supershift experiment in EMSA. The anti-Sp3 antibody supershifted two faster migrating bands (Fig. 8B, lane 9), whereas no supershifted bands were observed by adding antibodies against Sp1 (lane 7),   10). On the addition of all four antibodies, one band remained (lane 11). The same results were obtained with 293 cell extracts (data not shown). These results indicate that Sp3 and an unidentified factor bind to this probe, but that there was no difference in band shifts between HSP47/ collagen-producing and -nonproducing cell extracts. 500 bp in the First Intron Are Necessary for High Level Luciferase Activity in BALB/c3T3 Cells-To narrow down the essential region in the downstream introns, first the 3.7-kb introns were separated into three overlapping regions as shown in Fig. 9, and the luciferase assay was performed in BALB/c3T3 cells after the transient transfection of these constructs combined with the 280-bp upstream element. The deletion of the PstIϳPmlI region failed to sustain an enhanced luciferase activity, showing that this region is essential for the up-regulation of the hsp47 gene. Then, the PstIϳPmlI 1.5-kb region was further divided in four. Luciferase assay clearly showed that the BstXIϳSfiI 500-bp region was responsible for the up-regulation of the hsp47 gene in BALB/c3T3 cells when combined with the 280-bp upstream element (Fig. 9).
EMSA Using the 500-bp Intron Fragment-We performed EMSA by using the 500-bp intron DNA (BstXIϳSfiI) as a probe (Fig. 10). Three DNA-protein complexes could be detected with the extracts from four HSP47-producing cell lines (Fig. 10,  lanes 2, 3, 5, and 6) but not with those from HSP47-nonproducing cell lines (lanes 4, 7, and 8). These bands were competed out  11). Recombinant human Sp1 protein (0.2 g) was similarly analyzed (lanes 2 and 3). Anti-Sp1 monoclonal antibody was added in lanes 3, 5, 7, 9, and 11 by the presence of excess unlabeled 500-bp probe in the reaction mixture (Fig. 10, lanes 9 and 10). These results suggest that only HSP47/collagen-producing cells express specific DNA binding protein(s) for the 500-bp intron region, which is assumed to be the positive regulatory element. DISCUSSION We showed here two separated elements are necessary for tissue-and cell type-specific transcriptional activation of the hsp47 gene; one is a putative Sp1-binding site at Ϫ210 bp in the promoter region, and the other is a 500-bp region located in the first intron. Transgenic mice harboring these two elements express the reporter activities only in collagen-producing tissues. The assay for transcriptional activity using the luciferase reporter gene as well as gel mobility shift analysis revealed that the upstream region is necessary for the basal activity and the 500-bp intron region for cell type-specific expression.
HSP47 is a unique molecular chaperone in the following respects. First, it binds specifically to various collagens including types I to V in vitro as well as in vivo (3,4,6,40). Recently, a synthetic peptide approach using collagen model peptides revealed that HSP47 recognizes (Pro-Pro-Gly) n repeat when n is no less than 7 (56). Interestingly, when the proline residue at the second position is hydroxylated, HSP47 cannot bind to these peptides any more. That is, hydroxylation may be a regulatory mechanism for association and dissociation of HSP47 with collagen. Second, HSP47 is the only heat-inducible protein residing in the ER (3). It is induced by various stresses including heat shock through heat shock factor-HSE interaction. However it cannot be induced by ER stress, including the treatment with tunicamycin. hsp47 has an HSE in the promoter region (17,35) but no unfolded protein response element. Other ER stress proteins so far identified are induced only through an unfolded response mechanism. Third, the constitutive expression of hsp47 is always closely correlated with those of collagens under nonstressful conditions (15). In addition, the correlation in the expression of HSP47 and collagen is observed during the development of mouse embryos (41) as well as under pathophysiological conditions such as liver cirrhosis or renal fibrosis (25,42,43,27).
The present study provides a molecular basis for the transcriptional regulation of hsp47 in a cell-and tissue-specific manner. Gel mobility shift analysis and supershift experiments using specific antibodies suggested that Sp3 and other unidentified protein(s) bind to the putative Sp1-binding site in the upstream region at Ϫ210 bp, which was shown to be important for the basal activity of hsp47 expression. It has been shown that Sp1 forms homodimeric and heteromeric complexes and potentially forms DNA loops when binding to GC boxes of similar sequences (44 -47). Such binding sites are common in the promoter region of many genes and are usually repeated at different locations. Sp1 combined with sterol regulatory element-binding protein 1 of the low density lipoprotein promoter or with bovine papillomavirus enhancer E2 protein mediates the synergistic activation of transcription (48,49).
For cell type-specific expression of hsp47 in the cultured cells and tissue-specific expression in transgenic mice, the intron region was required in addition to the Sp1-binding site, and the region responsible for cell type-specific expression was narrowed down to a 500-bp region in the first intron. A similar dual regulatory mechanism has been reported in some collagen genes. In chondrocytes, both the Sp1 binding element in the promoter and the 100-bp enhancer in the first intron of the Col2a1 gene are necessary for high level, cell type-specific expression of Col2a1 (50,51). Sox9 protein has recently been shown to bind to sequences in the first intron of the Col2a1 gene and regulate the tissue-specific expression of this gene in transgenic mice (52). In the case of type XI collagen, the upstream 742 bp and 2.3 kb of the first intron segment of the mouse Col11a2 gene contain the information necessary to confer high level, tissue-specific expression in primordial cartilage in mouse embryos (38). Two type IV collagen chains (a1(IV) and a2(IV)) chains) use a common bidirectional promoter with a length of 127 bp containing Sp1-binding sites in each direction (53). The expression of the a2(IV) chain was activated by the 500-bp enhancer in the first intron of the Col4a2 gene. This enhancing effect is strictly dependent on the intact genomic structure of this gene; alteration of the orientation or distance to the promoter abolishes the activity completely (54). This 500-bp region shares no homology with promoters or introns of various collagen and other reported genes.
The finding that the expression of hsp47 is regulated by two separated elements similar to that of several collagens is of interest in that HSP47 is a collagen-specific molecular chaperone, and its expression is closely correlated with that of collagens. As reported for the Col4a2 gene (54), the 500-bp segment in the first intron of the hsp47 gene did not exert the enhancer activity when it was placed upstream of the 280-bp promoter or downstream of the luciferase reporter gene (data not shown). This suggests that the position or distance from the putative Sp1-binding site is important for this 500-bp segment to function as the enhancer.
We confirmed that these two separate elements regulate the expression of the reporter lacZ gene in the mouse by transgenic analysis. When the reporter gene is introduced with the 280-bp promoter and the intron into mice, reporter lacZ activities are observed in skin, axial skeleton, thorax, humerus, and other collagen-producing tissues in E13.5 embryos. HSP47 is reported to be expressed in these tissues during the development of mouse and chick embryos (23,41). The X-gal staining pattern is very similar to that in lacZ-harboring embryos under the control of ␣2(I), ␣1(II), and ␣2(XI) collagen promoters (36 -38, 55).
Recently we succeeded in disrupting the hsp47 gene in mice. The disruption resulted in embryonic lethality in mice by 11.5 days post-coitus and caused a molecular abnormality in procollagens. Type I procollagen chains containing propeptides accumulated in the tissues, but the mature collagen chains normally processed were scarcely observed. 2 Collagen fibrils and basement membranes were hardly detected in the tissues, and apoptosis was clearly observed after 10.5 days post-coitus in the mesenchymal tissues in hsp47-disrupted mice. Thus these results suggest that HSP47 is essential as a collagen-specific molecular chaperone for the proper processing of procollagen molecules, and the hsp47 gene is needed for the normal development of mouse embryo.