Intracellular Interaction of Hsp47 and Type I Collagenin Corneal Endothelial Cells

Xin Gu, MinHee K. Ko, and EunDuck P. Kay

PURPOSE. Previous studies by the current investigators showed that type I collagen was posttrans-lationally regulated in corneal endothelial cells (CECs). These cells synthesize type I procollagenand degrade it intracellularly; however, when CECs are modulated with fibroblast growth factor-2and/or corneal endothelium modulation factor, they synthesize and secrete type I collagen. Heatshock protein 47 (Hsp47), an endoplasmic reticulum resident protein, is known to function as amolecular chaperon in regulating procollagen folding and/or assembly. The interaction of Hsp47with type I procollagen synthesis in CECs was also studied.

METHODS. Expression of proteins was analyzed by radioactive labeling or immunoblot analysis. Thesteady state level of type I collagen and Hsp47 mRNAs was determined by northern blot analysis.Coprecipitation using immunoprecipitation followed by immunoblotting was performed to deter-mine the association profile between Hsp47 and type I procollagen. Subcellular localization ofHsp47 and type I procollagen was determined by immunofluorescent staining.

RESULTS. Normal and modulated cells expressed Hsp47 and Hsp70. The relative amount of Hsp47produced by modulated cells was much higher than that of control cells, but the expression levelof Hsp70 was the same in control and modulated cells. The steady state levels of type I collagentranscripts were higher in normal cells than in modulated cells, whereas modulated cells containeda much higher steady state level of Hsp47 mRNA. Type 1 procollagen was found to be associatedwith Hsp47 when analyzed by coprecipitation or cross-linking. The cytoplasmic localization profileof Hsp47 and type I collagen was different in normal cells, although a colocalization profile wasobserved to some degree. These two proteins were predominantly colocalized in the Golgi area inmodulated CECs.

CONCLUSIONS. Hsp47 may be involved in the synthesis and/or intracellular transport of type 1collagen in CECs. Modulated cells that secrete type I collagen into the extracellular matrix expressa higher level of Hsp47 than do control cells. {Invest Ophthalmol Vis Sci, 1999;40:289-295)

Corneal endothelium in vivo responds to diverse types ofpathologic conditions by converting to fibroblastlikecells.1'2 These morphologically modulated cells in turn

produce fibrillar collagens, the predominant species of whichis type I collagen, and deposit an abnormal fibrillar extracellu-lar matrix (retrocorneal fibrous membrane) between Descem-et's membrane and the corneal endothelium.3 The character-istic collagen phenotype in retrocorneal fibrous membranes istype I collagen, whereas types IV and VIII are the predominantcollagens in normal corneal endothelium in vivo and invitro.3'5 Thus, corneal endothelial modulation is involved inphenotypic switches in collagen gene expression. To elucidatethe mechanism of corneal endothelial modulation, we estab-lished an in vitro model in which a protein factor released bypolymorphonuclear leukocytes and/or fibroblast growth fac-

From the Doheny Eye Institute and the University of SouthernCalifornia School of Medicine, Los Angeles.

Supported in part by Grants EY06431 and EYO3O4O from theNational Institutes of Health, Bethesda, Maryland; and by an unre-stricted grant from Research to Prevent Blindness, New York, NewYork.

Submitted for publication April 29, 1998; revised August 10 andOctober 15, 1998; accepted October 27, 1998.

Proprietary interest category: N.Reprint requests: EunDuck Kay, Doheny Eye Institute, 1450 San

Pablo Street, DVRC 203, Los Angeles, CA 90033.

tor-2 modulated type IV collagen-synthesizing corneal endo-thelial cells (CECs) to type I collagen-synthesizing fibroblastlikecells.'3"8 Our previous study has shown that the steady statelevel of a2(l) collagen RNA in normal CECs is much higherthan that in the modulated cells that produce and secrete typeI collagen.910 Further analysis of the 5'- and 3'- untranslatedregions of a2(J) collagen RNA of normal CECs shows that thesequences in these regions are identical with those of modu-lated fibroblastlike CECs.'' Therefore, the possibility of trans-lational regulation based on mRNA sequences and the possibil-ity of transcriptional regulation of type I collagen expression inCECs has been eliminated.

Posttranslational regulation is an obvious mechanism toexplain this unique feature of type I collagen expression inCECs.11 Immunofluorescent staining of corneal tissue in vivohas shown that the corneal endothelium stains with anti-type Icollagen antibodies. There is no staining in the underlyingDescemet's membrane, however, which suggests that type Icollagen is synthesized in CECs in vivo, but there is no evi-dence of its secretion into the extracellular matrix. Pulse-chaseexperiments have further confirmed the synthesis and degra-dation of type I procollagen in normal CECs.'' These findingssuggest that the newly synthesized type I procollagen mustundergo intracellular degradation under physiologic condi-tions. The question to be examined is how type I procollagen

Investigative Ophthalmology & Visual Science, February 1999, Vol. 40, No. 2Copyright © Association for Research in Vision and Ophthalmology 289

290 Gu et al. IOVS, February 1999, Vol. 40, No. 2

is specifically degraded among the other collagens (types IVand VUI) present.

Hsp47, a stress protein in the endoplasmic reticulum, isassumed to be a molecular chaperon specific to collagen.12"15

It has been shown that Hsp47 interacts with polysome-associ-ated a 1(1) procollagen chains, suggesting that Hsp47 has aspecific function relative to chain selection or completion ofstable folding in type I procollagen. A recent report showedthe coexpression of the Hsp47 gene and a 1(1) and al(JII)collagen genes in carbon tetrachloride-induced rat liver fibro-sis. The expression levels of Hsp47 mRNA were proportionalto those of collagen mRNAs during the progression of thefibrosis.l6 Therefore, we hypothesized that Hsp47 may bedifferentially involved in the synthesis of type I procollagenand in the intracellular degradation of the molecule in CECs.Furthermore, we hypothesized that the modulated cells thatsecrete type I collagen may use a common pathway withHsp47 for the intracellular transport of the molecule beforesecretion.

In die present study, we found that Hsp47 is associatedwith type I procollagen in normal and modulated endothelialcells. The amount of type I procollagen associated with Hsp47is much higher in modulated cells than in normal cells. Fur-thermore, there is differential subcellular localization of Hsp47and type I procollagen in normal CECs than in modulated cells,which suggests that there may be differential pathways for typeI procollagen fated for intracellular degradation.

MATERIALS AND METHODS

Cell Cultures and Metabolic LabelingIsolation and establishment of the rabbit corneal endothelialcell culture were performed as previously described.4 Briefly,the Descemet's membrane-corneal endothelium complex wastreated with 0.2% collagenase and 0.05% hyaluronidase(Worthington Biochemical, Lakewood, NJ) for 90 minutes at37°C. Cultured cells were maintained in Dulbecco's modifiedEagle's medium (DMEM; Gibco, Grand island, NY) supple-mented with 10% fetal bovine serum (Irvine Scientific, Irvine,CA) and 50 jug/ml gentamicin (DMEM-10) in a 5% CO2 incuba-tor. This method has been shown to promote cell proliferationduring the early phase of culture and to maintain the culture asa contact-inhibited monolayer when the cells reach conflu-ence. First-passage corneal endothelial cells were used for allexperiments. Modulated corneal endothelial cells were pre-pared by treating the cells with fibroblast growth factor-2(Intergen, Purchase, NY) and corneal endothelium modulationfactor when the cells were plated and then feeding the cellswith DMEM-10 supplemented with these two factors everyother day. All animal experiments were performed in compli-ance with the ARVO Statement for the Use of Animals inOphthalmic and Vision Research. The cells were labeled with200 /LtCi [35S]-methionine (1218 Ci/mmol; Amersham, Arling-ton Heights, IL) or 400 jiiCi [3H]-proline (27 Ci/mM; Amer-sham) in DMEM supplemented with 1% fetal bovine serum, 50/Ltg/ml ascorbic acid and 50 ptg/ml /3-aminopropionitrile for 16hours.

Preparation of Corneal Endothelium ModulationFactorPolymorphonuclear leukocytes were obtained from rabbits asdescribed previously.6'7 The purified polymorphonuclear leu-

kocytes were incubated in protein-free medium (UltraDOMA-PF; Biowhittaker, Walkersville, MD) supplemented with 10jag/ml bovine serum albumin and maintained in a 5% CO2

incubator for 16 hours. Partial purification of the polymorpho-nuclear leukocyte- conditioned medium by ammonium sulfateprecipitation to a final concentration of 80% was performed bychromatography on a heparin-Sepharose column. The fractioneluted with 0.6 M NaCl in phosphate-buffered saline (PBS) wasbrought to a protein concentration of 250 /xg/ml and used tomodulate the CECs at a concentration of 2.5 /xg/ml. Partialcharacterization of corneal endothelium modulation factor waspreviously described.7

Sodium Dodecyl Sulfate—Polyacrylamide GelElectrophoresis

The conditions of electrophoresis were as described by Lae-mmli, using the discontinuous Tris-glycine buffer system.17

The proteins were separated on gels: 9% for Hsp47, 4.5% forcollagen, or 4% to 12% gradient gel (Novex, San Diego, CA) forsimultaneous analysis of Hsp47 and collagen. Gels containingradioactive samples were exposed for autoradiography.18 Afterthe gels were exposed on radiographic films (BioMax RM film;Kodak, New Haven, CT), the relative density of the bands wasestimated using a one-dimensional image analyzer (LKB Ultra-scan XL; Pharmacia LKB Biotechnology, Pleasant Hill, CA).

Immunoblot Analysis

The same amount of proteins (30 pig) separated by SDS-PAGEwas transferred to a polyvinylidene difluoride membrane at0.22 A for 10 hours in a semidry transfer system (transferbuffer: 25 mM Tris [pH 8.3], 190 mM glycine, 20% MeOH).Immunoblot analysis was performed as described previously19

using a commercial ABC kit (Vectastain; Vector Laboratories,Burlingame, CA). All washes and incubations were performedat room temperature in TTBS (0.9% NaCl, 100 mM Tris-HCl [pH7.5], 0.1% Tween 20). Briefly, the polyvinylidene difluoridemembrane was immediately placed in the blocking buffer(0.5% nonfat milk in TTBS) for 1 hour. Membranes were incu-bated with primary antibody (1:5000 dilution) for 2 hours, withbiotinylated secondary antibody (1:2500 dilution) for 1 hour,and with ABC reagent for 1 hour. The enhanced chemilumi-nescence (ECL) reagent (Amersham, Buckinghamshire, En-gland) was used, and the ECL-treated membrane was exposedfurther to ECL film. Monoclonal anti-Hsp47 antibody was pur-chased from Stressgen Biotechnologies (Victoria, British Co-lumbia, Canada), and anti-type I collagen antibody was pur-chased from Chemicon (Temecula, CA).

Gelatin-Sepharose Affinity Chromatography

After the cells were labeled with [35S]-methionine, they werelysed in lysate buffer (50 mM Tris [pH 8.0] 150 mM NaCl, 1%NP-40, 5 mM EDTA, 2 mM iV-ethylmaleimide, and 2 mM phe-nylmethylsulfonyl fluoride) for 30 minutes on ice. The celllysate was centrifuged at 15,000g for 20 minutes and thesupernatant applied to a gelatin-Sepharose column containing1 mg/ml to 2 mg/ml gelatin. After the column was washed withthree column volumes of buffer, the proteins were eluted withSDS-PAGE sample buffer and boiled for 5 minutes.

Cross-Linking and Immunoprecipitation

Procedures for cross-linking have been reported.15 Dithiobissuccinimidyl propionate (DSP), a substance that has been

IOVS, February 1999, Vol. 40, No. 2

shown to cross cell membranes, was selected for this study.The [35S]-methionine-labeled cells were sequentially treatedwith 0.2% trypsin containing 5 mM EDTA and 0.1% collagenase(Worthington Biochemical) to ensure removal of extracellularcollagen. Cells were then combined with 2 /J,1 DSP (stored at aconcentration of 0.1 M in dimethyl sulfoxide), vortexed, andplaced on ice for 30 minutes. After the reaction, cells wererinsed with 2 mM glycine in PBS, to block the DSP activity, andthen rinsed with PBS. The cells were lysed with lysis buffer (50mM Tris-HCl [pH 8.0], 150 mM NaCI, 1% NP-40, 5 mM EDTA,1 mM PMSF, 1 mM iV-ethylmaleimide, 1 jwg/ml leupeptin, and1 jxg/ml pepstatin) on ice for 15 minutes. Five micrograms ofundiluted anti-Hsp47 antibodies was added to the cell lysatesand incubated for 2 hours at 4°C, then precipitated with 50 JLLI

protein G-Sepharose beads (Sigma, St. Louis, MO) for 1 hour at4°C. The precipitated immunocomplexes were washed threetimes with lysis buffer. The proteins bound to the Sepharosebeads were eluted with Laemmli's sample buffer containingdithiothreitol, boiled for 5 minutes, and applied to an SDS-polyacrylamide gel for electrophoresis.

Immunofluorescent Staining

Normal and modulated comeal endothelial cells (3 X 10^/chamber) were seeded on the four-well chamber slide andmaintained in culture until they reached 80% confluence. Thecells were washed with PBS and fixed with 3% paraformalde-hyde in PBS for 5 minutes at room temperature. Cells werethen permeabilized with 0.5% Triton X-100 in PBS for 5 min-utes and blocked with 2% bovine serum albumin for 1 hour.The slides were simultaneously incubated with monoclonalantibody against Hsp47 (1:50 dilution) and polyclonal goatanti-type I collagen (1:50 dilution) for 2 hours at room temper-ature, then washed with PBS. Cells were then incubated withbiotinylated anti-goat immunoglobulins (1:50 dilution) for 1hour, followed by incubation with rhodamine conjugated withavidin (1:50 dilution) and fluorescein conjugated with anti-mouse imiminoglobulins (1:50 dilution) for 30 minutes at roomtemperature. After extensive washing, the slides were exam-ined under a confocal microscope (Carl Zeiss, Qberkochen,Germany).

Collagenase Treatment

Immunocomplexes precipitated with protein G-Sepharose (25yl bead volume) were suspended in 30 JLLI collagenase buffer(100 mM Tris [pH 7.5], 5 mM CaCl2, and 10 mM PMSF)containing three units of bacterial collagenase form III (Ad-vance Biofactures, Lynbrook, NY), and the samples were incu-bated at 37°C for 30 minutes. The samples were then boiled for5 minutes in Laemmli's sample buffer containing dithiothreitol.

Northern Blot Analysis

PolyA+ RNA was isolated using a commercial kit (Mini RiboSepUltra mRNA Isolation Kit; Becton Dickinson Labware, Bedford,MA), as described in the protocol provided with the kit. Briefly,cells were lysed with proteinase-containing lysis buffer, andthe lysate was incubated at 45°C for 2 hours. The NaCI con-centration of the lysate was adjusted to 0.5 M and applied to anoligo(dT) cellulose column. PolyA+ RNA was eluted and sub-sequently precipitated with two volumes of 100% ethanol inthe presence of sodium acetate. The sample was centrifuged at12,000g at 4°C for 15 minutes. The resultant RNA precipitate

Association of Hsp47 and Type I Procollagen 291

kDa Hsp 47 Hsp 70

83

45

M

* • «

1 2 1 2FIGUKE 1. lmmunoblot analysis of Hsp47 and Hsp70 in comeal endo-thelial cells (CECs). Cell lysates prepared from normal and modulatedCECs were separated on a 9% SDS-polyacrylamide gel and transferredto polyvinylidene difluoride membrane followed by inimunoblot anal-ysis with anti-Hsp47 or anti-Hsp70 antibody. Lane 7, normal CECs;lane 2, modulated CECs. M, protein size marker. Data are representa-tive of five experiments.

was dissolved in sterile water, and the RNA concentration wasmeasured by absorbance at 260 nm to 280 nm. One microgramof PolyA+ RNA was denatured by treatment with formalcle-hyde-formamide, separated by SDS-PAGE on an 0.8% agarosegel, and blotted to a nitrocellulose filter. The RNA was prehy-bridized, hybridized, and washed as described previously.9 Rata 1(1) and a2(l) cDNA clones were gifts from Dr. David Rowe(University of Connecticut, Farmington, CT) and were previ-ously used in CECs.9 The Hsp47 cDNA clone was a gift fromDr. Kazuhiro Nagata (Kyoto University, Kyoto, Japan).20

RESULTS

Expression of Hsp47 in Comeal Endothelial Cells

Hsp47 is known to be involved in the translation-txanslocationmachinery for the production of type I procollagen, and othercollagen types.12"16'21 Beginning with the assumption thatCECs that must intracelluiarly degrade type I procollagen mayhave a diminished ability to produce Hsp47, we compared theexpression level of Hsp47 in CECs with that of the modulatedcells that secrete type I collagen into the extracellular matrix.The inimunoblot analysis showed that normal and modulatedCECs produced Hsp47 under tissue culture conditions (Fig. 1.To compare the relative expression of Hsp47 in normal andmodulated CECs, the same amount of protein was analyzed. Agreater amount of Hsp47 was observed in the modulated cellsthan in the normal cells, suggesting that type I collagen-secret-ing cells contain a higher level of Hsp47 than cells that do notsecrete type I collagen. When the relative density of the bandscorresponding to Hsp47 was estimated, there was approxi-mately a twofold increase in the level of Hsp47 in the modu-lated cells (Table 1. Of interest, the molecular size of Hsp47produced by both cells was approximately 50 kDa to 51 kDa,migrating slower than the Hsp47 marker provided by Stressgen(47 kDa, not shown). In comparison, no difference was foundin the amount of Hsp70 produced by normal or modulatedCECs (Fig. 1, Table 1).

292 Gu et al.

TABLE 1. Densitometric Quantification of theAutoradiographs of Hsp47, Hsp70, and Hsp47Transcript

Hsp47 Hsp70 Hsp47 Transcript 28 S —

Normal CECModulated CEC

1.713-25

3-633-68

0.370.93

After the gels were exposed on radiographic film, the relativedensity of the bands of Hsp47 and Hsp70 in Figure 1 and Hsp47transcript in Figure 3 was estimated using a one-dimensional imageanalyzer.

When 35S-methionine-Iabeled cell lysates (2 X 106 cpm/sample) obtained from normal and modulated CECs were ap-plied to a gelatin-Sepharose affinity column and the boundproteins were analyzed by SDS-PAGE and autoradiography, themajor gelatin-binding protein was the 50-kDa peptide band inthe normal and the modulated CECs (Fig 2. ). Normal CECscontained less of the peptide band than the modulated cells,whereas corneal stromal fibroblasts that predominancy pro-duce type I collagen contained high levels of the 50-kDa band.Densitometric quantification of these bands showed that therewas approximately a twofold increase in the intensity of theband in modulated CECs (data not shown). Analysis of the celllysate in the absence of gelatin-Sepharose affinity chromatog-raphy showed that the 50-kDa band was missing in all three cellsystems. There were other gelatin-binding proteins in CECs,the identities of which were unknown,

Expression of Hsp47 mRNA and type I collagen RNAs wasexamined by northern blot analysis of polyA"*" RNA extractedfrom normal and modulated CECs. The steady state level ofa 1(1) collagen RNA was similar in both cells, whereas therelative amount of a2(T) collagen RNA present as characteristicdoublets differed greatly, with the a2<J) collagen being mark-edly decreased in modulated cells in which the lower tran-script was more abundant (Fig. 3). This unusual expression

kDa

83 -

45 -

M

- Hsp47

1 2 3 4 5 6FIGURE 2. Gelatin-Sepharose affinity chromatography of gelatin-bind-ing components of the [35S]-methionine-labeled CECs. Cell lysates (2 X106 cpm) were applied to a gelatin-Sepharose column. Proteins wereeluted with electrophoresis sample buffer and applied to a 9% SDS-polyacrylamide gel. Lanes 1 through 3 are eluted samples from gelat-in-Sepharose affinity; lanes 4 through 6 are cell lysates before process-ing through the gelatin-Sepharose affinity column. Lane J, normalCECs; lane 2, modulated CECs; lane 3, corneal stromal nbrobtasts;lane 4, normal CECs; lane 5, modulated CECs; lane 6, corneal stromalfibroblasts. Data are representative of three experiments.

IOVS, February 1999, Vol. 40, No. 2

Type I collagen Hsp47

I-118s-

1 2 1 2 1 2FIGURE 3- Northern blot analysis of polyA+RNA (1 jxg/lane) isolatedfrom normal (lane 1) and modulated (lane 2) corneal endothelial cells.The RNA was electrophoresed on an 0.8% agarose gel in formaldehyde-formamide, transferred to a nitrocellulose filter, and hybridized to therespective cDNA probes. For collagen DNA probes, pro al(I) mousecDNA clone (A) and pro a2(I) mouse cDNA clone (B) were used. In aparallel slot, rRNAs were electrophoresed and visualized by ethidiumbromide staining. Data are representative of two experiments.

profile of a2(I) collagen RNA was consistent with results in ourprevious report.10 In contrast to type I collagen RNAs, thesteady state level of the Hsp47 transcript was much higher inthe modulated cells than in the normal CECs. The Hsp47mRNA was detected as a single transcript of approximately 2.5kb, similar to the previous rinding.22 Quantitative examinationmore clearly showed the increased level of Hsp47 transcript inthe modulated cells, when compared with that of normal cells(Table 1).

Association of Type I Collagen and Hsp47

To investigate the association profile of Hsp47 and type Icollagen in CECs further, the cell lysates were analyzed withimmunoprecipitation followed by immunoblot analysis. Celllysates were precipitated with anti-Hsp47 antibody in the pres-ence of protein-G Sepharose, and the complexes were sepa-rated on an SDS-polyacrylamide gel followed by immunoblotanalysis with anti-type I collagen antibody. Another fraction ofthe cell lysates was first precipitated with anti-type I collagenantibody followed by immunoblot analysis with anti-Hsp47antibody. The immune complexes precipitated with anti-Hsp47 antibody contained type I procollagen (Fig- 4), whereasthe complexes immunoprecipitated with anti-type I collagenantibody contained the 50-kDa form of Hsp47. Because thesame amount of protein was used for both analyses, it isassumed that the modulated cells contained more protein com-plexes of Hsp47 and type I procollagen than the normal cells.Of interest, the'immune complexes precipitated with anti-typeI collagen antibody showed a less variable association withHsp47 between normal and modulated cells. Corneal stromalfibroblasts, positive control cells, showed almost the samelevel of association of these two proteins as the modulatedcorneal endothelial cells. Quantitative examination of the proa 1(1) collagen band (Fig. 4) immunoprecipitated with anti-Hsp47 antibody showed that there was a significant increase ofpro ctl(I) collagen associated with Hsp47 in modulated cells.An approximate sevenfold increase was observed in modulatedcells, In contrast, there was only a twofold difference in Hsp47immunoprecipitated with anti-type I collagen antibody be-tween normal and modulated cells (Table 2. This finding is inagreement with the previous report21 in which a differentialpotential in the association of the two proteins was observed.

IOVS, February 1999, Vol. 40, No. 2

kDa

205 - *

M 1 2 3

- r

1 2 3

kDa- 8 3

- 45

M

Ip

WB

Hsp47

Type I

Type I

Hsp47FIGURE 4. Coprecipitation of type I collagen with Hsp47. Cell lysatesfrom normal or modulated CECs were immunoprecipitated with anti-body against Hsp47, followed by immunoblot analysis with anti-type Icollagen antibody, or cell lysates were precipitated with type I collagenantibody followed by immunoblot analysis with anti-Hsp47 antibody.The immunoprecipitates were electrophoresed on a 4% to 12% gradi-ent gel, and the proteins were transferred to polyvinylidene difluoridemembrane followed by immunoblot analysis. Lane 1, normal CECs;lane 2, modulated CECs; lane 3, corneal stromal nbroblasts. Arrowsindicate pro al(I) and pro a2<J) bands. IP, immunoprecipitation; WB,western blot analysis; M, protein size maker.

To confirm that the binding of type I procollagen to Hsp47was not the result of cell lysis, cross-linking with DSP, whichhas been shown to cross cell membranes, was used to demon-strate their relationship in living cells. [3H]-proline-labeled celllysates were prepared and immunoprecipitated with anti-Hsp47 antibody. A portion of the immunoprecipitates wassubjected to digestion with bacterial collagenase. Normal andmodulated CECs contained bacterial collagenase-susceptiblepeptide bands in the region of pro al(T) and pro a2(T) (Fig. 5.There were several proteins associated with Hsp47 in both cellsystems. These were not collagenous proteins, as determinedby their insensitivity to bacterial collagenase.

Subcellular Localization of Hsp47 and Type ICollagenIt has been reported that Hsp47 is colocalized with procolla-gen in what appeared to be pre-Golgi vesicular structuresduring tooth development.23 With an interesting phenotype oftype I collagen expression in corneal endothelial cells, wefurther investigated whether there was differential localization

TABLE 2. Densitometric Quantification of Pro a(I)Collagen Associated with Hsp47 and Hsp47 Associatedwith Type I Collagen

Pro a(I) Collagen Hsp47

Normal CECModulated CECCorneal stromal fibroblast

0.483.253.67

0.360.791.71

The relative density of the bands of pro a(I) collagen and Hsp47in Figure 4 was estimated using a one-dimensional image analyzer.Corneal stromal nbroblasts served as positive control cells that synthe-size and secrete type I collagen into the medium.

Association of Hsp47 and Type I Procollagen 293

b.col. 1 h

Hsp47-

— 0,2(1)

1 1 2 2FIGURE 5. Coprecipitation of cross-linked Hsp47 and type 1 collagen.CECs were labeled with [3H]-proline for 16 hours, collected afterdigestion with trypsin and collagenase and cross-linked with DSP. Celllysates were immunoprecipitated with anti-Hsp47 antibody and sub-jected to digestion with bacterial collagenase; the resultant precipitateswere applied to an SDS 4% to 12% gradient polyacrylamide gel. lane I,normal CECs; lane 2, modulated CECs. b. col., bacterial collagenasedigestion.

of type I procollagen fated for intracellular degradation in CECsfrom that of the molecule bound for secretion in modulatedCECs. Corneal endothelial cells showed that Hsp47 and type Iprocollagen are either colocalized or their localization is notcoincidental, even at the single-cell level (Fig, 6). In contrast,the modulated cells showed a marked colocalizatton profile ofHsp47 and type I procollagen. The subcellular localizationseemed to be in the Golgi area. Our recent study using doublelabeling with anti-Golgi 58 K protein antibody and anti-type 1collagen (or anti-Hsp47) antibody confirmed this observation(data not shown). However, close scrutiny is required to iden-tify the cellular organelles involved in these distinctive path-ways in both cells. Such a study is currently under way withorganelie-specific enzymes used as markers.

DISCUSSION

In previous studies, we showed that there are unique featuresto the expression of type 1 collagen in corneal endothelial cells.Type 1 procollagen with a correct composition of two proal(T) chains and one pro a2(I) chain is synthesized by CECs.The intracellular presence of the molecule in vivo was shownby immunocytochemical analysis. Pulse-chase experimentshave further shown that type I procollagen is intracellularlydegraded; thus, secretion and subsequent deposition of thebiologically undesirable type I collagen into the basementmembrane environment is blocked. One question asked re-garding this unique posttranslational regulation is how type Icollagen is selectively degraded in the presence of other col-lagens (types IV and VHI). In the present study, an attempt wasmade to investigate the fate of the intracellular type I procol-lagen. It has been proposed that recently discovered chaperon

294 Gu et al. IOVS, February 1999, Vol. 40, No. 2

FIGURE 6. Subcellular localization of Hsp47 and type I collagen. Normal or modulated corneal endothelialcells were treated with Triton X-100, bovine serum albumin, and primary antibodies, as described in theMethods section. Fluorescein signals (green) are Hsp47 positive; rhodamine signals (red) are type Icollagen positive; yellow signals show the colocalization of Hsp47 and type I collagen. (A, B, C) normalCECs; (D, E, F) modulated CECs. Bar, 18 joun.

molecules perform numerous functions, including stabilizationof the newly synthesized or unfolding polypeptides, facilitatingtranslocation of nascent chains across the membrane, mediat-ing assembly or disassembly of multimeric protein complexes,and targeting proteins for degradation within lysosomes.24"26

Among molecular chaperons, it is known that Hsp47, an en-doplasmic reticulum resident protein that transiently associateswith procollagen, is involved in collagen processing and/orsecretion under normal conditions. Under conditions of stress,Hsp47 is part of the quality control system for procollagen,including the prevention of the secretion of procollagen withabnormal conformation.12 Furthermore, Hsp47 is reportedlyinvolved in the translation-translocation machinery in the pro-duction of pro al(T) chains of type I collagen.14-27"29

Assuming that CECs, which must intracellulary degradetype I procollagen, may have a diminished ability to produceHsp47, we compared the expression level of Hsp47 in CECswith that of the modulated cells that secrete type I collageninto the matrix. Increased expression of Hsp47 in the modu-lated cells has been shown by immunoblot analysis of theprotein, and the steady state level of Hsp47 mRNA is higher inthe modulated cells than it is in the normal cells. The molecularsize of the transcript (2.5 kb) is identical with the previouslyreported size.22 However, the molecular weight of Hsp47 pro-duced by normal and modulated CECs is 50 kDa to 51 kDa,which is slightly larger than the reported size, probably be-cause of the differential degree of glycosylation of the moleculein CECs (personal communication with Kazuhiro Nagata), be-cause Hsp47 has two sites that can be glycosylated with high-mannose-type sugar moieties.30 It may be because of thedifference in other conditions (species, culture conditions,etc.), because corneal stromal fibroblasts that were used forpositive control have a size identical with that of Hsp47, as dothe corneal endothelial cells. Yet, close scrutiny was requiredto identify the discrepancy in the molecular weight of Hsp47synthesized by CECs. In contrast, there was no difference

between the expression levels of Hsp70 in normal and modu-lated CECs. Hsp47 synthesized by CECs, regardless of thesource (normal CECs versus modulated CECs), showed char-acteristic features of the molecule: binding affinity to gelatin,transient binding of Hsp47 with type I procollagen in theabsence of cross-linking agent, and protein complex formationwith type I procollagen in living cells. Immunoprecipitation ofcell extract with anti-Hsp47 or anti-type I collagen antibodyshowed that Hsp47 and type I procollagen coprecipitate witheither antibody, implying an association between Hsp47 andtype I procollagen within the cells. Our data further show thatthere seemed to be more association of type I procollagen andHsp47 in the modulated cells than in the normal CECs. To-gether, these findings suggest that a sufficient amount of Hsp47may be critical for trafficking type I procollagen until it issecreted.

Another assumption was that the type I procollagen that isdestined for intracellular degradation may have a differentintracellular transport pathway than the molecules secretedinto the matrix. This led to a most interesting observation onthe subcellular localization of Hsp47 and type I procollagen. Innormal CECs, Hsp47 and type I procollagen were colocalizedin some, but not most, cytoplasmic areas. These findings sug-gest that the two molecules may use a common transportpathway to a certain degree, beyond which they use differenttransport pathways. Perhaps Hsp47 is recruited from the Golgiapparatus for recycling, whereas type I procollagen uses an-other pathway to reach the degradation site. Conversely, mod-ulated cells that secrete type I collagen into the matrix showthat Hsp47 and type I procollagen are colocalized to the Golgiarea. The subcellular localization of these proteins was identi-cal, suggesting that type I collagen destined for secretionshares a common intracellular transport pathway with Hsp47.At present, a further study is under way to identify subcellularorganelles using specific enzyme markers while specific inhib-itors are used to block the intracellular trafficking process. It

IOVS, February 1999, Vol. 40, No. 2 Association of Hsp47 and Type I Procollagen 295

has recently been reported that degradation of the newlysynthesized procollagen takes places in the proximal region ofthe secretory pathway, such as the endoplasmic reticulum,czs-Golgi network, or czs-Golgi and in a distal region of thesecretory pathway, such as the trans-Go\g\ or trans-Go\g\ net-work, in addition to the endosome-lysosome system.31 It islikely that type I procollagen in CECs is not only specificallydegraded, perhaps through the Hsp47-mediated pathway, butis degraded through nonspecific pathways in endoplasmic re-ticulum or Golgi. Nevertheless, it is unknown what directs thespecific degradation of type I procollagen in the presence ofthe physiologic type IV and type VIII collagen. There are a fewproblems in assigning Hsp47 to the specificity on type I colla-gen: The association constant of Hsp47 with types I-V collag-ens are in the same range,12 and Hsp47 is also involved in thetranslation-translocation of type IV collagen.32 It is possiblethat other endoplasmic reticulum resident chaperon moleculesare also involved in determining the specificity, especiallybecause Grp78 and Grp94 are involved in type IV collagensynthesis along with Hsp47.32 Despite this uncertainty, how-ever, the present study provides further insight into the role ofHsp47 in the trafficking of type I procollagen. The amount ofHsp47 may be critical, and the close association of these twoproteins in certain organelles is important.

Acknowledgments

The authors thank Kazuhiro Nagata, Kyoto University, Japan; ArtherVeis, Northwestern University, Chicago, Illinois; John Sauk, Universityof Maryland, Baltimore; and Robert S. Bienkowski, Albert EinsteinCollege of Medicine, New Hyde Park, New York, for their advice.

References

1. Michels RG, Kenyon KR, Maumenee AE. Retrocorneal fibrousmembrane. Invest Opbthalmol. 1972;11:822-831.

2. Brown SI, Kitano S. Pathogenesis of the retrocorneal membrane.Arch Ophtbalmol. 1966;75:518-525.

3. Kay EP, Cheung C-C, Jester JV, Nimni ME, Smith RE. Type Icollagen and fibronectin synthesis by retrocorneal fibrous mem-brane. Invest Opbthalmol Vis Sci. 1982;22:200-212.

4. Kay EP, Smith RE, Nimni ME. Basement membrane collagen syn-thesis by rabbit corneal endothelial cells in culture. Evidence for analpha chain derived from a larger biosynthetic precursor. / BiolCbem. 1982;257:7116-7121.

5. Sawada H, Konomi H, Hirosawa K. Characterization of the collagenin the hexagonal lattice of Descemet's membrane: its relation totype VIII collagen./ Cell Biol. 1990;110:219-227.

6. Kay EP, Gu X, Ninomiya Y, Smith RE. Corneal endothelialmodulation: a factor released by leukocytes induces basic fibro-blast growth factor that modulates cell shape and collagen. InvestOphthalmol Vis Sci. 1993;34:663-672.

7. Kay EP, Gu X, Smith RE. Corneal endothelial modulation: bFGF asdirect mediator and corneal endothelium modulation factor asinducer. Invest Ophthalmol Vis Sci. 1994;35:2427-2435.

8. Kay EP, Rivela L, He YG. Corneal endothelial modulation factorreleased by polymorphonuclear leukocytes. Partial purificationand initial characterization. Invest Ophthalmol Vis Sci. 199O;31:313-322.

9. Kay EP. Expression of types I and IV collagen genes in normal andin modulated corneal endothelial cells. Invest Ophthalmol Vis Sci.I989;30:260-268.

10. Kay EP, He YG. Post-transcriptional and transcriptional control ofcollagen gene expression in normal and modulated rabbit cornealendothelial cells. Invest Ophthalmol Vis Sci. 1991;32:1821-1827.

11. Kay EP, Gu X, Choi SH, Ninomiya Y. Posttranslational regulation oftype I collagen in corneal endothelial cells. Invest Ophthalmol VisSci. 1996;37:11-19.

12. Nagata K. Hsp47: a collagen-specific molecular chaperone. TrendsBiochem Sci. 1996;21:22-26.

13. Natsume T, Koide T, Yokota S, Hirayoshi K, Nagata K. Interactionsbetween collagen-binding stress protein HSP47 and collagen.J Biol Chem. 1994;269:31224-31228.

14. Sauk JJ, Smith T, Norris K, Ferreira L. Hsp47 and the translation-translocation machinery cooperate in the production of al( l)chains of type 1 procollagen. J Biol Chem. 1994;269:394l-3946.

15. Nakai A, Satoh M, Hirayoshi K, Nagata K. Involvement of theprotein HSP47 in the procollagen processing in the endoplasmicreticulum./ Cell Biol. 1992;117:903-9l4.

16. Masuda H, Fukumoto M, Hirayoshi K, Nagata K. Coexpression ofthe collagen-binding stress protein HSP47 gene and the a 1(1) anda 1(111) collagen genes in carbon tetrachloride-induced rat liverfibrosis. JClin Invest. 1994;94:2481-2488.

17. Laemmli UK. Cleavage of structural protein during the assembly ofthe head of bacteriophage T4. Nature. 1970;227:680-685-

18. Laskey RA, Mills AD. Quantitative film detection of 3H and U C inpolyacrylamide gels by fluorography. Eur J Biochem. 1975;56:335-341.

19. Gu X, Seong GJ, Lee YG, Kay EP. Fibroblast growth factor-2 usesdistinct signaling pathways for cell proliferation and cell shapechanges in corneal endothelial cells. Invest Ophthalmol Vis Sci.1996;37:2326-2334.

20. Takechi H, Hirayoshi K, Nakai A, Kudo H, Saga S, Nagata K.Molecular cloning of a mouse 47-kDa heat-shock protein (HSP47),a collagen-binding stress protein, and its expression during thedifferentiation of F9 teratocarcinoma cells. Eur J Biochem. 1992;206:323-329.

21. Satoh M, Hirayoshi K, Yokota S, Hosokawa N, Nagata K. lntracel-lular interaction of collagen-specific stress protein Hsp47 withnewly synthesized procollagen. / Cell Biol. 1996; 133:469-483.

22. Tanaka Y, Kobayashi K, Kita M, et al. Expression of 47 kDa heatshock protein (HSP47) during development of mouse cornea. ExpEye Res. 1996:63-383-393.

23. Smith T, Ferreira LR, Hebert C, Norris K, Sauk JJ. Hsp47 andcyclophilin B traverse the endoplasmic reticulum with procollageninto pre-Golgi intermediate vesicles. A role for Hsp47 and cyclo-philin B in the export of procollagen from the endoplasmic retic-ulum. J Biol Chem. 1995;270:18323-18328.

24. Becker J, Craig EA. Heat-shock proteins as molecular chaperones.Eur J Biochem. 1994;219:11-23-

25. Ellis RJ, van der Vies SM. Molecular chaperones. Anna Rev Bio-chem. 1991;60:321-347.

26. Ellis RJ. The general concept of molecular chaperones. PbilosTrans R Soc Lond B Biol Sci. 1993;339:257-26l.

27. Shroff B, Smith T, Norris K, Pileggi R, SaukJJ. Hsp47 is localized toregions of type I collagen production in developing murine femursand molars. Connect Tissue Res. 1993;29:273-286.

28. Clarke EP, Jain N, Brickenden A, Lorimer IA, Sanwal SD. Parallelregulation of procollagen I and colligin, a collagen-binding proteinand a member of the serine protease inhibitor family. / Cell Biol.1993;121:193-199.

29. Hu G, Gura T, Sabsay B, Sauk J, Dixit SN, Veis A. Endoplasmicreticulum protein Hsp47 binds specifically to the N-terminal glob-ular domain of the amino-propeptide of the procollagen 1 al(I>chain. / Cell Biochem. 1995;59:35O-367.

30. Hughes RC, Taylor A, Sage H, Hogan BLM. Distinct patterns ofglycosylation of colligin, a collagen-binding glycoprotein, andSPARC (osteonectin), a secreted Ca2+-binding glycoprotein. Evi-dence for the localisation of collagen in die endoplasmic reticu-lum. Eur J Biochem. 1987;l63:57-65.

31. Ripley CR, Bienkowski RS. Localization of procollagen I in thelysosome/endosome system of human fibroblasts. Exp Cell Res.1997;236:147-154.

32. Ferreira LR, Norris K, Smith T, Hebert SC, SaukJJ. Hsp47 and otherER-resident molecular chaperones form heterocomplexes witheach other and with collagen type IV chains. Connect Tissue Res.1996;33:265-273.