Functional linkage between the endoplasmic reticulum protein Hsp47 and procollagen expression in human vascular smooth muscle cells.

Hsp47 is a heat stress protein that interacts with procollagen in the lumen of the endoplasmic reticulum, which is vital for collagen elaboration and embryonic viability. The precise actions of Hsp47 remain unclear, however. To evaluate the effects of Hsp47 on collagen production we infected human vascular smooth muscle cells (SMCs) with a retrovirus containing Hsp47 cDNA. SMCs overexpressing Hsp47 secreted type I procollagen faster than SMCs transduced with empty vector, yielding a greater accumulation of pro alpha1(I) collagen in the extracellular milieu. Interestingly, the amount of intracellular pro alpha1(I) collagen was also increased. This was associated with an unexpected increase in the rate of pro alpha1(I) collagen chain synthesis and 2.5-fold increase in pro alpha1(I) collagen mRNA expression, without a change in fibronectin expression. This amplification of procollagen expression, synthesis, and secretion by Hsp47 imparted SMCs with an enhanced capacity to elaborate a fibrillar collagen network. The effects of Hsp47 were qualitatively distinct from, and independent of, those of ascorbate and the combination of both factors yielded an even more intricate fibril network. Given the in vitro impact of altered Hsp47 expression on procollagen production, we sought evidence for interindividual variability in Hsp47 expression and identified a common, single nucleotide polymorphism in the Hsp47 gene promoter among African Americans that significantly reduced promoter activity. Together, these findings indicate a novel means by which type I collagen production is regulated by the endoplasmic reticulum constituent, Hsp47, and suggest a potential basis for inherent differences in collagen production within the population.

The production of type I procollagen is a multistep process that requires the participation of several enzymes and chaperones to ensure the faithful secretion of this trimeric protein.
Within the endoplasmic reticulum (ER), 1 two pro␣1(I) collagen chains and one pro␣2(I) collagen chain first associate at their globular carboxyl termini (1). This interaction is facilitated by protein-disulfide isomerase and stabilized by interchain disulfide bonds (2,3). Winding of the long triple helical domain then proceeds in the carboxyl to amino direction. The fidelity of this helix-folding reaction is dependent on hydroxylation of proline residues by prolyl-4-hydroxlase (4). If the activity of prolyl-4hydroxlase is impaired, as in ascorbate deficiency, the improperly folded procollagen chains are retained within the ER and secreted at a slower rate or targeted for degradation (5,6). Once the triple helix is formed, the protein is transported to the Golgi apparatus and brought to the cell surface after a period of Golgi cisternal maturation (7).
Hsp47 is a heat shock protein, present exclusively in the ER, that plays a vital role in procollagen processing. Hsp47 is expressed only by collagen-producing cells (8,9) and in vitro binds to collagens type I to V as well as gelatin (10). Within the ER, Hsp47 has been found to interact with nascent type I procollagen chains (11,12), with fully translated pro␣ collagen chains (13), with non-helical and poorly hydroxylated procollagen trimers (14) and with well-hydroxylated, triple helical procollagen (15,16). Once the procollagen-Hsp47 complex reaches the cis-Golgi, Hsp47 dissociates and is recycled back to the ER (13,17).
The contexts in which Hsp47 expression have been identified in humans are noteworthy. Increased levels of Hsp47 have been observed in atherosclerotic plaque (18), keloid lesions (19), fibrotic lungs (20), and diseased kidneys (21,22). Furthermore, the fundamental importance of Hsp47 in collagen biosynthesis is unequivocal: Hsp47-null mice die before birth and the embryos display ruptured blood vessels and a marked reduction in the amount of mature type I collagen (23). The exact mechanism by which Hsp47 contributes to the production of procollagen is unclear, however. Several possible roles have been proposed including facilitating pro␣ collagen chain elongation, preventing improper association of unassembled and possibly underhydroxylated pro␣ collagen chains (11,13,14), winding of the triple helix (24), maintaining thermal stability of the triple helix once formed (16,25), and diverting assembled procollagen into the cisternal maturation transport pathway (15,26). Although there is experimental support for each of these potential roles, a cohesive model has yet to emerge and not all of the data are consistent. For example, it remains unclear whether the interaction between Hsp47 and procollagen slows or accelerates procollagen transport. The former might be expected if Hsp47 serves to maintain the folded conformation during stress, whereas the latter would be more likely if Hsp47 plays a direct role in winding of the nascent pro␣ collagen chains (24). * This work was supported by grants from the Canadian Institutes of Health Research (MT11715) and Heart and Stroke Foundation of Canada (T4458). 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  Evidence for both speeding (27) and slowing (28) of procollagen secretion by Hsp47 has been reported.
To examine the effects of Hsp47 on type I collagen production, we have introduced the human Hsp47 gene into a collagen-producing human cell. For this, we have used human vascular smooth muscle cells (SMCs), which are critical to the elaboration of collagen in the vessel wall (29,30). This allowed us to evaluate the response to elevated levels of Hsp47 in the context of a fully functional pathway for procollagen biosynthesis. The results indicate that Hsp47 has the capacity to drive collagen production by a mechanism that involves up-regulation of pro␣1(I) collagen gene expression and stimulation of procollagen chain synthesis, in concert with enhanced procollagen secretion. We also provide evidence for interindividual genomic variants associated with Hsp47, which, together with the identified effects on procollagen production, implicate Hsp47 as a determinant of collagen elaboration in the population.

EXPERIMENTAL PROCEDURES
Cell Culture-A human vascular SMC clonal line, designated HITB5, was generated from the human internal thoracic artery, as described previously (31). This line bears remarkable similarity to SMCs in the vessel wall, including the expression of smooth muscle ␣-actin, smooth muscle-myosin heavy chain isoforms SM1 and SM2, calponin, heavy caldesmon, and metavinculin, as well as the capacity to contract. HITB5 SMCs were maintained in M199 (Invitrogen) supplemented with 10% fetal bovine serum. Mouse embryonic dermal fibroblasts were kindly provided by Dr. L. Dagnino (University of Western Ontario, London, ON) and were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum.
Overexpression of Hsp47 in Human SMCs-A retroviral gene delivery system was utilized to generate human SMCs expressing Hsp47 under the control of the P CMV promoter. The full-length cDNA encoding human Hsp47, together with 87 bp of the 5Ј-untranslated region and 707 bp of the 3Ј-untranslated region, was excised with HindIII and SmaI from the plasmid cln9-33 (kindly provided by Dr. K. Nagata, Kyoto University, Kyoto, Japan) (32). This fragment was inserted into HindIII and StuI sites of the retroviral expression vector, pLNCX2 (CLONTECH, Palo Alto, CA). Orientation and sequence accuracy of the pLNCX2.Hsp47 construct was verified by cDNA sequencing using a ABI 377-XL stretch DNA sequencer and ABI sequence navigator software (PE Applied Biosystems, Foster City, CA).
Retrovirus containing pLNCX2.Hsp47 or pLNCX2 was generated by calcium phosphate-mediated transfection of the Phoenix-amphotrophic retrovirus packaging cell line (kindly provided by Dr. G. P. Nolan, Stanford University Medical School, Stanford, CA, distributed by ATCC, Manassas, VA) (33). The virus-containing supernatant was harvested 48 -72 h later and, after centrifugation and filtration (0.45-m pore size), was incubated with proliferating HITB5 SMCs for 48 h in the presence of 20% fetal bovine serum and 5 g/ml Sequa-brene™ (Sigma-Aldrich). Infection efficiency was estimated at 40% based on infection of parallel cultures with pLNCX2.EGFP. Cells stably expressing Hsp47 and control cells infected with pLNCX2 alone were selected with 500 g/ml G418 for 2 weeks. Overexpression of Hsp47 was confirmed before every experiment by Western blot analysis.
Western Blot Analysis of Hsp47 and Type I Collagen Expression-Analysis of type I collagen and Hsp47 expression was determined by Western blot analysis, as described (29). Briefly, cells and conditioned media were harvested in the presence of 0.1 mM phenylmethylsulfonyl fluoride and 10 g/ml leupeptin. Equal amounts of protein were resolved on 6 and 12% polyacrylamide gels for type I collagen and Hsp47, respectively. Type I collagen was detected using a polyclonal rabbit antiserum to the C-telopeptide region of the ␣1(I) chain of human type I collagen (LF67, 1:8000 dilution, gift of Dr. L. W. Fisher, National Institute of Dental Research, Bethesda, MD) (34,35). Hsp47 was detected using a mouse monoclonal antibody to rat Hsp47 (1:4000 dilution, gift of Dr. B. D. Sanwal, University of Western Ontario, London, ON). Membranes were incubated with antibodies overnight at 4°C. An antirabbit peroxidase-conjugated IgG or antimouse peroxidase-conjugated Fab fragment was used to detect bound antibody by chemiluminescence (Promega Corp., Madison, WI). Washed membranes were exposed to x-ray film (Kodak XAR-5, Kodak, Toronto, ON).
Analysis of Procollagen Synthesis and Secretion Rates by Metabolic Labeling and Immunoprecipitation-Duplicate cultures of SMCs incu-bated for 1 h in methionine-free Dulbecco's Modified Eagle's Medium (Invitrogen) were then pulsed for 1 h in methionine-free Dulbecco's modified Eagle's medium supplemented with 100 Ci/ml [ 35 S]methionine (ICN). Following three washes with Dulbecco's phosphate-buffered saline, the cultures were chased with Dulbecco's modified Eagle's medium containing 4 mM methionine (Invitrogen). At designated times, the medium was harvested with 0.1 mM phenylmethylsulfonyl fluoride and 10 g/ml leupeptin, and washed cells were solubilized in radioimmune precipitation assay buffer (1% Igepal CA-630 (Sigma-Aldrich), 0.5% sodium deoxycholate, 0.1% SDS in phosphate-buffered saline) containing protease inhibitors. The media and cell lysates were clarified by centrifugation and precleared with 5 l of protein A-agarose (Santa Cruz Biotechnology, Inc., Santa Cruz, CA) for 30 min. Equal amounts of protein were immunoprecipitated using LF67 (1:500) and 15 l of protein A-agarose overnight at 4°C. Immunoprecipitates were washed extensively with radioimmune precipitation assay buffer and resolved on a 6% polyacrylamide gel. Quantification of radioactive bands in dried gels was performed using a phosphorimager screen and Image-Quant software (Amersham Biosciences).
Northern Blot Analysis-HITB5 SMCs transduced with pLNCX2 or pLNCX2.Hsp47 were harvested using Trizol™ reagent (Invitrogen). Total RNA was separated on a 1.2% agarose-formaldehyde gel and transferred to Zetaprobe GT membrane (Bio-Rad Laboratories). Membranes were incubated overnight in hybridization solution containing 100 ng of cDNA probe labeled by random-hexamer priming with [␣-32 P]dCTP. Hsp47 mRNA was detected using a plasmid containing a cDNA clone for rat Hsp47 (pIP1) (36). Pro␣1(I) collagen mRNA was detected using a human pro␣1(I) collagen cDNA derived from Hf677 (ATCC). Fibronectin mRNA was detected using mouse fibronectin cDNA (pC9912, gift of Dr. G. Kidder, University of Western Ontario, London, ON) and a cDNA probe for human glyceraldehyde-3-phosphate dehydrogenase (pHcGAP, ATCC) was used as a control for RNA loading. Blots were washed, and mRNA bands were quantified after phosphorimaging.
Immunohistochemistry-Transduced SMCs were grown on glass coverslips until confluent and fixed for 20 min in ice-cold methanol. Cells were rehydrated for 10 min in phosphate-buffered saline, blocked for 30 min in 10% normal goat serum, and incubated with LF67 (1:200) for 1 h. After washing, bound primary antibody was detected with an fluorescein isothiocyanate-labeled goat-antirabbit IgG (1:500) that was applied for 1 h. Nuclei were stained with 2.5 g/ml Hoechst 33258 in phosphatebuffered saline for 5 min. Following three 10-min washes in phosphatebuffered saline, coverslips were mounted on glass slides with Perma-Flour (Immunon™, Pittsburgh, PA). Collagen fibrils were visualized by confocal microscopy using a Zeiss LSM 410 microscope and a krypton/ argon laser emitting at 488 nm. An argon ion ultraviolet laser emitting at 351 nm for the detection of Hoechst 33258 was used to image the nuclei.
Identification of CBP2 Promoter Single Nucleotide Polymorphisms in Humans-Genomic DNA was prepared from leukocytes isolated from peripheral blood samples of 26 control subjects of various ethnic groups. Two sets of sequencing primers (P1, 5Ј-CCACTGTCGCCCAGAT-TATTTA-3Ј and 3Ј-CAGTGCCCTTCTCCATACTTGT-5Ј; P2, 5Ј-CAGG-TACCGGGTCTGGTCT-3Ј and 3Ј-GTCTCCCGCCCCTCACCT-5Ј) were designed to cover a 1-kb region upstream from the CBP2 gene, as reported by Ikegawa et al. (37). Amplification conditions were as follows: 94°C for 5 min, 30 cycles of 30 s incubations at 94°C, 58°C and 72°C, and a final 10 min extension step at 72°C. The expected fragment sizes were 624 and 640 bp for the P1 and P2 reaction, respectively. DyeNamic ET Terminator sequencing kit was used according to the manufacturer's instructions and samples loaded onto ABI 377-XL stretch DNA sequencer. ABI Sequence Navigator software (PE Applied Biosystems) was used to align and compare DNA fragments for sequence differences.
The CBP2 promoter [Ϫ656]CϾT SNP was genotyped in an additional 200 people by amplifying genomic DNA using the P1 primers and the above amplification protocol. The 624-bp product was digested with endonuclease ApaL1, which yielded one fragment for the [Ϫ656]CϾT SNP and two fragments with sizes of 446 and 178 bp for the wild-type allele. The fragments were resolved in 2% agarose gels. SAS version 6.12 (SAS Institute, Cary, NC) was used for statistical analyses. Allele frequencies were determined from electrophoretogram tracings of genomic DNA sequence, except for the [Ϫ656]CϾT SNP, which was assayed using restriction digestion. Chi-square analysis tested the deviation of genotype frequencies from Hardy-Weinberg predictions, with the nominal p Ͻ 0.05.
In Vitro Analysis of Promoter Function-The effect of the [Ϫ656]CϾT SNP on promoter function was evaluated by transfecting embryonic dermal fibroblasts with luciferase reporter constructs. A genomic DNA fragment containing a 1 kb region upstream of the first CBP2 exon with the [Ϫ656]CϾT SNP was generated using a two-primer pair method. The mutagenesis primer pairs were: 5Ј-GGAGGAGTA-CACAGGAAGGAAAACCTG-3Ј and 3Ј-GCCAGGTTTTCCTTCCTGT-GTACTCCT-5Ј (underlined nucleotide indicates the mutated position). Sequencing primer sets P1 and P2 were used to amplify the remaining regions. The DNA fragment containing the mutation, a wild-type fragment, and an antisense fragment were subcloned into the PGL3 vector (Promega Corp.) and appropriate sequences verified by DNA sequence analysis.
Embryonic dermal fibroblasts were transfected with reporter constructs using calcium phosphate precipitation. The fibroblasts were serum-deprived for 48 h and incubated with or without all-trans retinoic acid for 20 h. Luciferase activity was measured according to the manufacturer's instructions (Promega Corp.) on a lumat LB9507 luminometer (Berthold Technologies, Oak Ridge, TN) and expressed relative to total protein content (BCA protein assay kit, Pierce).

SMCs Overexpressing Hsp47 Generate Increased Levels of Intracellular and Extracellular pro␣1(I) Collagen
-To evaluate the consequences of augmented Hsp47 expression, we introduced Hsp47 cDNA into a vascular SMC line, HITB5. This line was chosen for study because of its human origin, the importance of collagen production by vascular SMCs, and because HITB5 SMCs express copious type I collagen (38). We thus could evaluate the effect of Hsp47 overexpression in the context of a functionally intact pathway for collagen biosynthesis, as well as determine its effect when procollagen hydroxylation is suboptimal, as with ascorbate depletion.
Infection of HITB5 SMCs with retrovirus containing Hsp47 cDNA yielded SMCs that expressed, on average, 10 -12-fold greater levels of Hsp47 protein than SMCs infected with vector alone (Fig. 1). To determine the consequences of this for type I procollagen abundance, cells and conditioned media were immunoblotted using LF67, a polyclonal antibody that detects pro␣1(I) collagen, partially processed pro␣1(I) collagen, and mature ␣1(I) collagen. As shown in Fig. 1, cells transduced with vector alone and cultured in the absence of ascorbate produced pro␣1(I) collagen that was detectable within the cell and also in media that had been conditioned for 48 h. Most of the collagen in the media was in the form of procollagen or partially cleaved procollagen with much less fully processed collagen, consistent with previous in vitro studies (29,39,40). Compared with vector-transduced cells (HITB5-vector), SMCs overexpressing Hsp47 (HITB5-Hsp47) contained a substantially greater amount of intracellular pro␣1(I) collagen. However, this elevated level of procollagen could not necessarily be interpreted as a result of intracellular retention of target protein by Hsp47, as observed after overexpression other ER-resident chaperones (41), because the amount of pro␣1(I) collagen that accumulated on the outside of the cell was also substantially increased (Fig.  1, bottom panel).
It was also apparent that the effect of Hsp47 overexpression was distinct from that of ascorbate. Incubation of vector-transduced SMCs with 100 M ascorbate resulted in a decline in the level of intracellular pro␣1(I) collagen and an increase in the level of the extracellular procollagen. This profile is consistent with more efficient secretion of procollagen out of the cell but contrasted with the parallel increases in intra-and extracellular pro␣1(I) collagen observed when Hsp47 was overexpressed. As well, the procollagen species in ascorbate-supplemented cultures migrated through SDS-PAGE faster and with clearer delineation of the procollagen cleavage intermediates (Npro␣1(I) collagen, Cpro␣1(I) collagen), consistent with structural optimization imparted by greater hydroxylation of procollagen. The combination of Hsp47 overexpression and ascorbate treatment was additive, producing a substantially increased amount of extracellular pro␣1(I) collagen, and a level of intracellular pro␣1(I) collagen below that of HITB5-Hsp47 SMCs in ascorbate-deficient media but above that of HITB5-vector SMCs incubated with ascorbate.
Taken together therefore, overexpression of Hsp47 in SMCs has a profound effect on the elaboration of type I procollagen, but this effect cannot be categorized as one that acts solely by enhancing, or delaying, procollagen transport.
Overexpression of Hsp47 in SMCs Increases the Rate of Procollagen Secretion-To clarify the basis of the observed effects of Hsp47 overexpression, type I procollagen secretion kinetics were assessed by pulse-chase experiments. Vector-infected and Hsp47-overexpressing HITB5 SMCs were incubated for 4 days with or without 100 M ascorbate and labeled for 1 h with [ 35 S]methionine, and secretion of newly synthesized type I procollagen was tracked by immunoprecipitation of procollagen and collagen in the culture media using LF67, followed by phosphorimaging. As shown in Fig. 2, type I procollagen was secreted from cells overexpressing Hsp47 at a substantially greater rate than from vector-transduced SMCs. As expected, the addition of ascorbate to control vector-transduced SMCs also increased the procollagen secretion rate from these SMCs (Fig. 2B). Addition of ascorbate to HITB5-Hsp47 cells resulted in an even greater rate of secretion, i.e. greater than from HITB5-Hsp47 cells not incubated with ascorbate and also greater than from HITB5-vector SMCs incubated with ascorbate.
Overexpression of Hsp47 in SMCs Increases the Intracellular Procollagen Transport Kinetics but Also Enlarges the Pool of Newly Synthesized Procollagen Chains-The intracellular transport kinetics were also determined by pulse-chase analysis. As illustrated in Fig. 3A, the clearance rate of type I procollagen chains was increased in Hsp47-overexpressing cells. The profiles followed monoexponential decays, which  (HITB5-Hsp47). Cell lysates and medium conditioned for 4 days were analyzed by Western blot analysis using a monoclonal antibody to Hsp47 and a polyclonal antibody to type I collagen, LF67. The consequences of Hsp47 overexpression on ascorbate-deficient and ascorbate-supplemented SMC cultures were assessed by addition of 100 M ascorbate 4 days prior to harvesting.
when fitted with first order rate constants yielded a t1 ⁄2 of 56 min for HITB5-vector SMCs, 28 min for HITB5-Hsp47 cells, and 35 min for HITB5-vector SMCs incubated with ascorbate (Fig. 3B). Addition of ascorbate to Hsp47-overexpressing SMCs further increased the type I procollagen clearance rate, with kinetics best described by two exponential components corresponding to t1 ⁄2 values of 11 and 106 min for the fast and slow components, respectively. Addition of the lysosomal inhibitor, chloroquine (25 M) or the proteosomal inhibitor, ALLN (100 M), had minimal effect on the clearance rates of type I procollagen under basal circumstances or in SMCs overexpressing Hsp47 (data not shown), indicating that intracellular degradation is a relatively minor pathway for procollagen in these cells. The increased rate of intracellular disappearance of procolla- Precipitates were resolved on a 6% polyacrylamide gel, and intensity of the bands for the pro␣1 (1) collagen chain and the pro␣2(I) collagen chain, with which it immunoprecipitates, were quantified. Curves were fitted using least squares regression analysis (r Ͼ 0.98 for all plots), and data are the mean of duplicate culture wells. Two other experiments gave similar results.

FIG. 3. Hsp47 overexpression stimulates intracellular trafficking of type I procollagen and augments the pool of newly produced procollagen chains.
Control and Hsp47-overexpressing HITB5 SMCs were incubated for 4 days with or without 100 M ascorbate and then pulse-labeled with [ 35 S]methionine for 1 h. Samples were harvested at designated times after washing and reintroducing media with 4 mM unlabeled methionine and immunoprecipitated using LF67. A, phosphorimages of immunoprecipitated pro␣1(I) collagen and co-immunoprecipitated pro␣2(I) collagen resolved on a 6% polyacrylamide. B, band intensities of labeled procollagen chains quantified by phosphorimaging and normalized to the baseline signal immediately following the pulse. Curves were fitted using least squares regression analysis (r Ն0.93 for all plots). Inset, half-life of type I procollagen was calculated using the equation for the linear regression line following natural logarithm transformation. C, pool sizes of newly synthesized pro␣1(I) collagen and co-precipitating pro␣2(I) collagen chains immediately following the 1 h incubation with [ 35 S]methionine. Data are the mean of three separate experiments, each of which was performed in duplicate. *, p Ͻ 0.05 versus HITB5-vector SMCs in the presence or absence of ascorbate.
gen by Hsp47 was thus primarily accounted for, and in keeping with, the augmented secretion kinetics described above.
In addition to the effect on transport kinetics, inspection of the radio-immunoblots revealed that immediately following the pulse, the pool size of labeled pro␣1(I) collagen, and the pro␣2(I)collagen chain with which it co-immunoprecipitates, was greater in Hsp47-overexpressing SMCs than in vectortransduced cultures. This surprising observation was replicated three times using SMCs from different virus-infected cultures. On average, the pool of newly produced pro␣ collagen chains was 2.6-fold greater in Hsp47-overexpressing cells than control vector-transduced cells (Fig. 3C). This increase was observed in both ascorbate-deficient and ascorbate-supplemented cultures, and it raised the possibility that the production rate of individual pro␣ collagen chains was increased by Hsp47.
Hsp47 Overexpression Increases pro␣1(I) Collagen mRNA Expression in HITB5 SMCs-To further explore the basis for the increased pool size of newly produced type I procollagen chains in Hsp47-overexpressing SMCs, pro␣1(I) collagen transcript expression was evaluated by Northern blot analysis. As shown in Fig. 4, there was a 2.5-fold increased in pro␣1(I) collagen mRNA in Hsp47-overexpressing SMCs. This response appeared to be selective for procollagen, as expression of fibronectin mRNA was unchanged. Therefore, the enlarged pool of newly synthesized type I procollagen appears to be due to increased procollagen expression and chain synthesis. Moreover, these data reconcile the apparent paradox of a chaperone elevating both intracellular and extracellular levels of the target protein, because they establish that in Hsp47-overexpressing cells procollagen production is increased in addition to procollagen secretion.

Hsp47 Overexpression Augments Type I Collagen Fibril
Formation-To determine if the augmentation of Hsp47 expression influenced the formation of a type I collagen fibril matrix, confluent HITB5 SMCs were incubated for 9 days in the presence or absence of 100 M ascorbate, added freshly every 3 days, and cultures were then immunostained using LF67. As shown in Fig. 5A, vector-infected SMCs in ascorbate-deficient media produced collagen fibrils that accumulated in a heterogeneous pattern, which included relatively thick and amorphous aggregations. In SMCs overexpressing Hsp47, the amorphous appearance of the thicker fibril bundles could still be appreciated. However, there was a striking increase in the overall number of collagen fibrils produced. Vector-transduced SMCs incubated with ascorbate also yielded more fibrils than vector-transduced SMCs in ascorbate-deficient conditions. However these fibrils were also finer and organized in a more intricate network. These features likely reflect a fibril structure arising from more complete hydroxylation, consistent with the clearer delineation of the pro␣1(I) collagen forms on SDS-PAGE. As shown in Fig. 5D, SMCs that overexpressed Hsp47 and were incubated with ascorbate produced an even more extensive network than observed with either manipulation alone, with a dense mat of fibrils.
Human CBP2 Promoter Region Shows Genetic Polymorphism-The amplified production of type I collagen fibrils by increased Hsp47 represents a novel corollary to previous data showing that selectively decreasing the level of Hsp47 leads to reduced production of type I collagen (23). Although the mechanistic basis of these two responses may differ, the data together point to the actions of Hsp47 as a primary regulator of FIG. 4. Hsp47 overexpression stimulates pro␣1(I) collagen gene expression by human SMCs. Northern blots of total RNA from SMCs transduced with pLNCX2 or pLNCX2.Hsp47 were probed for Hsp47, pro␣1(I) collagen, fibronectin, and glyceraldehyde-3-phosphate dehydrogenase. Procollagen transcript abundance was increased in SMCs overexpressing Hsp47.

FIG. 5. Overexpression of Hsp47 in human SMCs induces the formation of a robust collagen fibril network. Control HITB5
SMCs and SMCs overexpressing Hsp47 were plated on glass coverslips and incubated with or without 100 M ascorbate for 9 days. Methanolfixed SMCs were then immunostained with a polyclonal antibody to type I collagen, LF67 and visualized using a fluorescein isothiocyanatelabeled secondary antibody. Nuclei were visualized with Hoechst 33258. Amorphous aggregations of fibrils (arrows) were apparent in cultures of HITB5-vector SMCs that were incubated in the absence of ascorbate (A). This morphology was also present in ascorbate-deficient HITB5-Hsp47 SMC cultures (B); however, collagen fibril abundance was substantially greater. Addition of 100 M ascorbate to HITB5-vector cultures yielded finer fibrils (C), and the combination of ascorbate and Hsp47 overexpression resulted in the elaboration of an even more extensive network of fine fibrils (D). collagen production. This raises the question of whether collagen production in the population might be impacted by interindividual variation in Hsp47 production. To assess the possibility that Hsp47 expression is variable in the population, we screened for the presence of SNPs in the 1-kb upstream promoter region of CBP2, the gene that encodes Hsp47 (37,42). Genomic DNA sequence analysis in 26 unrelated individuals demonstrated the presence of 5 common SNPs (Fig. 6A). The allele frequencies in subjects from various ethnic groups are summarized in Table I Fig. 6B, luciferase activity in embryonic dermal fibroblasts transfected with the [Ϫ656]CϾT SNP-containing construct was 0.23 that of cells transfected with the wild-type promoter (p Ͻ 0.01). Stimulation of cells transfected with the wild-type promoter construct with all-trans retinoic acid significantly increased luciferase activity. All-trans retinoic acid also increased promoter activity of the [Ϫ656]CϾT SNP construct, suggesting a role for additional promoter sites in retinoic acid responsiveness. However, total luciferase activity remained strikingly suppressed, suggesting that this SNP has global consequences for CBP2 promoter activity.

DISCUSSION
Hsp47 has a number of attributes that distinguish it from other ER-resident chaperones. These include its specificity for procollagens as well as an atypical binding profile whereby it interacts with both folded and unfolded conformations of its target protein (11,13,14,16). The results of the current study indicate that Hsp47 can act as a primary stimulator of type I collagen production. Selectively increasing the amount of Hsp47 within SMCs reconfigured the collagen production phenotype such that both intracellular and extracellular steadystate levels of type I procollagen were increased. This profile was associated with up-regulated pro␣1(I)collagen gene expression, an increased rate of type I procollagen chain synthesis, and faster procollagen secretion. The net effect was the elaboration of an extensive network of type I collagen fibrils. These findings point to a novel pathway for controlling type I collagen production and suggest a linkage between an ER-resident protein and the machinery for procollagen expression.
The pool of intracellular procollagen that formed in SMCs within 1 h of adding labeled methionine was substantially increased in Hsp47-overexpressing SMCs compared with control SMCs. Expansion of this particular pool of procollagen could be due either to an increase in the procollagen chain synthesis rate or a decrease in intracellular degradation of nascent procollagen chains. The contribution of the latter pathway is likely small, given that significant co-translational degradation of pro␣ collagen chains has not been identified, and the proportion of the entire intracellular procollagen pool that is degraded within the cell is reported to be only 10-20% (43)(44)(45). Furthermore, addition of lysosomal and proteosome inhibitors had little effect on the procollagen transport kinetics in SMCs. In addition, the 2.6-fold increase in the pool of newly produced procollagen was associated with a 2.5-fold increase in pro␣1(I) collagen transcript abundance. Taken together therefore, the expanded pool of newly synthesized procollagen in Hsp47-overexpressing SMCs reflected an increase in the expression and production rate of procollagen chains.
Increased Hsp47 also influenced the kinetics of procollagen secretion, with an increased intracellular clearance rate and an increased rate of secretion out of the cell. Because these

Hsp47 Stimulates Collagen Production
changes were in the context of increased procollagen synthesis, it is difficult to determine if they were a direct result of the procollagen-Hsp47 interaction in the ER or a more integrated response of several ER proteins ensuring that procollagen was efficiently transported out of the cell, in the face of increased procollagen synthesis. Importantly however, the effect of augmented expression of Hsp47 on transport kinetics did not depend on ascorbate concentration. Under conditions of ascorbate depletion and supplementation, procollagen secretion increased by a similar extent, and the enhanced secretion induced by ascorbate was additive to that induced by Hsp47 alone. In addition, unlike ascorbate, Hsp47 overexpression did not abrogate the production of procollagen forms that are presumably imperfectly structured, suggested by the banding pattern on Western blot analysis. Similarly, the morphology of the assembled collagen fibrils in cultures of Hsp47-overexpressing cells suggested that the collagen structure was not optimized, in contrast to that in ascorbate-supplemented cultures. Although Hsp47 has been found to bind poorly hydroxylated procollagen (13,46), the current observations suggest that Hsp47 does not have a corrective role in the context of suboptimal procollagen hydroxylation. Because SMCs overexpressing Hsp47 elaborated a greater extracellular network of type I collagen fibrils, it is possible that the enhanced procollagen expression and synthesis was a secondary consequence of the modified external milieu. However, extracellular collagen fibrils have been well established to down-regulate, not increase, expression of pro␣1(I) collagen (47)(48)(49) and this, therefore, is an unlikely explanation. There is, however, precedent for linking events within the ER to activation of genome. The unfolded protein response, for example, entails the stimulated expression of ER-resident proteins consequent to the release of the chaperone BiP from its ER receptor. This is followed by receptor clustering and autophosphorylation and activation of JNKs (50). Recently, overexpression of calreticulin in myoblasts was associated with up-regulation of protein phosphatase 2A, further highlighting the potential for ER events to be transduced outside that compartment (51).
The augmentation of collagen production by increased Hsp47 levels has interesting implications. It is well established that an absence or reduction of Hsp47 leads to reduced collagen production (12,23). Taken together with the current findings therefore, any perturbation that selectively alters the amount of Hsp47 in the ER seems to reset the amount of collagen that is produced. It is noteworthy therefore that within the human population there are common nucleotide polymorphisms in the promoter region of the CBP2 gene encoding Hsp47. The CBP2 [Ϫ656]CϾT SNP is a mutation that was private to African Americans and resulted in a substantial reduction in CBP2 promoter activity. The functional consequences of depressed CBP2 promoter function in this population warrant further study, but the finding suggests a genomic basis for interindividual differences in collagen production within the population and may, in part, underlie individual susceptibility to conditions characterized by aberrant healing. Phenotypes to be studied for possible association with their functional differences in the CBP2 promoter would include cutaneous wound healing, keloid formation, and vascular phenotypes.
The current findings may also have therapeutic implications. There may be circumstances in which augmented or accelerated production of collagen would be beneficial. An important example of this is the production of a structurally sound fibrous cap of on the surface of atherosclerotic plaques. Caps with insufficient fibrillar collagen are prone to rupture which can lead to thrombus formation and heart attacks (30). Surpris-ingly, few defined stimuli are known to drive collagen synthesis, and these are typically multifunctional proteins, such as TGF-␤, which influence multiple cell types diversely. The increase in collagen production by Hsp47 observed in these studies thus raises a potential strategy for accelerating collagen production by vascular cells, possibly in a specific manner.