Collagen Type I and I11 Occur Together in Hybrid Fibrils in the Space of Disse of Normal Rat Liver ALBERT

GEERTS,'DETLEFSCHUPPAN,3 SYLVIA LAZEROMS,' RONALDDE ZANGER2 AND EDDIEWISSE'

'Laboratory for Cell Biology and Histology, and "Laboratoryfor Physiology and Physiopathology, VrlJe Universiteit Brussel N.U.B.), B-1090 Brussels, Belgium; and "KlinikumSteglitz, Department of Gastroenterology, Freie Uniuersitat Berlin, Federal Republic of Germany

mostly found together with type I collagen in soft tissues such as the dermis, blood vessels, liver, lung, spleen, kidney and muscle (3). In rat liver and human liver, the collagens type I and I11 each represent at least one third of the total hepatic collagen. The collagens type IV,V and VI occur in lesser quantities (4-6). The collagens type I and I11 are interstitial collagens. They form fibrils with a typical striation pattern perpendicular to their longitudinal axis (7). The periodicity of the striation pattern equals 67 nm in wet samples and 64 nm in dry samples (8). The in vitro assembly of type I and type I11 collagen into interstitial fibrils has been studied extensively by x-ray crystallography and by electron microscopy (7, 8). From these studies, the model of a quarter-staggered array of the collagen molecules within the fibrils has evolved. The applied methods, however, do not allow us to conclude whether type I and type I11 molecules, when present together in a solution or in a tissue, may become integrated in the same or in separate fibrils because x-ray crystallography and cross-striation patterns of pure type I and pure type I11 collagen fibrils are almost identical. Several studies (9-14) dealing with the localization in the liver of collagen type I, collagen type I11 or both present indirect evidence that the liver may contain hybrid collagen fibrils consisting of both collagen type I and collagen type 111. In other studies (15-18),however, no evidence in support of the hybrid collagen fibril concept is found. In human skin, indirect evidence indicates that collagen type I and collagen type I11 may be localized in At least 13 genetically distinct collagen types of higher different fibrils (19-22). Staining for type I11 collagen is organisms have been characterized biochemically (1, 2). found preferentially on thin fibrils (diameter 20 to 60 The most prevalent collagen is type I collagen, occurring nm), whereas labeling for type I collagen is present on in bone and in most soft tissues (3).Type I11 collagen is approximately 80%of the fibrils with a diameter ranging up to 80 nm (21). However, in a recent study, Keene et al. (23) describe that collagen type I11 is present on all cross-striated fibrils in the human skin, tendon and Received May 22, 1989; accepted March 7, 1990. This work waa supported by "Fonds voor Geneeskundig Wetenschappelyk amnion, regardless of the fibril diameter. They, Onclerzoek" grants no. 30.0040.80 and no. 30.0028.86. therefore, suggest that collagen type I and collagen type Address reprint requests to: Dr. Albert Geerts, Laboratory for Cell Biology I11 form hybrid collagen fibrils. and Histologv, Vriie Universiteit Brussel (V.U.B.). Laarbeeklaan, 103, 8-1090. In tissues other than liver or skin, the data are equally Brussekdette, Belgium. 31/1/21610 confusing. For lymph nodes (24) and human leiomyoma

Collagen type I and procollagen type III were localized at the ultrastructural level on ultrathin frozen sections of rat liver by the protein A-gold technique using afhity-purified primary antibodies. Both collagen type I and procollagen type 111were localized on nearly all solitary and bundled fibrils in the space of Disse. Simultaneouslocalization of collagen type I and procollagen type 111 by a double-labeling procedure using protein A-gold probes of different sizes unequivocally demonstrated the presence of both collagens in the same fibrils. Measurement of the diameter of large numbers of collagen fibrils in the space of Disse of the rat liver showed a unimodal distribution of the fibril diameters around an average value of 62.4 nm (S.D. = 12.8 nm), and 91% of the collagen bundles contained less than 30 fibrils. Additional measurements on epoxy resinembedded material of five biopsy specimens of normal human liver showed a comparable unimodal distribution of the fibril diameters around an average value of 57.2 nm (S.D. = 9.6 nm), and 74% of the bundles contained less than 80 fibrils. The latter observation demonstratesthat human liver contains broader interstitial collagen bundles than rat liver. From these results, we conclude that the space of Disse of normal rat and human liver contains a uniform population of striated interstitial collagen fibrils. In the rat liver, these fibrils contain both collagen type I and procollagen type III. Therefore the concept that procollagen type IIIis predominantly localized in small diameter fibrils or bundles, whereas collagen type I is preferentially localized in thick ones, does not hold. (HEPATOLOGY 1990;12:233-241.)

233

234

GEERTS ET AL.

(23, evidence is available in favor of the existence of hybrid fibrils containing collagen type I and collagen type 111. On the contrary, in human term placenta and in rat lung two separate populations of collagen fibrils consisting of either collagen type I or collagen type I11 are described. In the placenta, collagen type I is present in 30 to 35 nm cross-banded fibrils, whereas collagen type I11 is organized in 10 to 15 nm beaded fibrils forming a meshwork encasing the collagen type I fibrils (26). In the rat lung, collagen type I11 is present in 15 to 20 nm beaded fibrils, whereas the cross-banded fibrils are not stained (27). The purpose of this study is to localize the collagens type I and I11 in the rat liver by the immunogold technique on ultrathin frozen sections. By using a well-established procedure based on labeling the collagens type I and I11 with protein A-gold probes of different sizes, we have examined the collagen fibrils in the space of Disse for the simultaneous occurrence of both collagen type I and 111.Additional morphometrical data on the collagen fibril diameters are collected to investigate the existence of populations of fibrils with different diameters.

MATERIALS AND lMETHODS Immunocytochemistry Antigens and primary antibodies. Purification of monkey and rat procollagen type I11 (pN-111, consisting of the aminoterminal propeptide linked to the collagen helix) from skins and immunization of rabbits were performed as described in previous reports (28-30). Monospecific antibodies against the rat aminoterminal propeptide of type I11 collagen were obtained by passing the antiserum over columns of CNBractivated Sepharose CL-4B (Pharmacia AB, Uppsala, Sweden) to which collagen type I or collagen type I11 was coupled. Specific antibodies were then eluted from a column to which pN-I11 was bound. As published previously (29, 301, sensitive radiobinding assays demonstrated a lack of cross-reactivity of the antibodies with collagen or procollagen type I, the helical part of collagen type 111, procollagen type IV, fibronectin or laminin. Monkey and rat collagen type I was extracted from skins with neutral buffers, and antisera were produced in rabbits as described previously (28-30). Monospecific antibodies against monkey collagen type I were obtained after passage of the antisera over CNBr-Sepharose affinity columns of human laminin P1, fibronectin, collagens type IV, V and VI and monkey procollagen type I11 (pN-111, lacking the carboxytermind propeptide). Finally, the antibodies were eluted from a column of purified neutral salt-soluble collagen type I. In sensitive radiobinding assays (29, 301, the eluted antibodies showed no reaction with the above-mentioned radiolabeled proteins. Immunoblotting was performed according to Dziadek et al. (31)and Schuppan et al. (32) usingneutral salt-solublerat and monkey collagen type I and procollagen type I11 containing pN and triple helical collagen type I11 as antigens. The affinitypurified antibodies to monkey collagen type I and rat procollagen type I11 peptide were applied at 1to 2 pg/ml.

Immunofluorescence Preparation of sections. The right lateral liver lobe of adult, male Wistar rats weighing 250 to 300 gm was removed with the

HEPATOLOGY

rats under ether anesthesia. As quickly as possible, 3 x 3 x 2 mm3 tissue blocks were cut, mounted on holders with Tissue-Tek (Reichert-Jung, Buffalo, NY) and submerged in liquid nitrogen. Sections 5 km thick were cut using a cryostat (American Optical Corp., Southbridge, MA). The sections were briefly dried, fixed in acetone at - 20" C for 10 min and dried again. The sections were then rinsed in PBS, covered with 2% BSA fraction V (Sigma Chemical Co., St. Louis, MO) for 15 min, incubated with primary antibodies (- 25 p,g/ml) for 1hr, washed three times for 5 min in PBS, incubated with FITC-labeled affinity-purified goat antirabbit antibodies (Miles Scientific, Elkhart, IN), washed three times for 5 min in PBS and mounted in glycerol and water 8 :2 (voVvo1). Control incubations included sections incubated in nonimmune rabbit IgG (Sigma Chemical Co.), but otherwise treated as described above, and sections incubated with FITC-conjugated antibodies only. The sections were viewed using a Leitz Orthoplan microscope equipped for epifluorescence (Leitz KG, Hamburg, F.R.G.). Fluorescent images were recorded with a Bosch TYK 9A1 Newvicon video camera (Robert Bosch GMBH Photo Abt., Stuttgart, F.R.G.). The digitized images were stored in a Masscomp 55208 (Concurrent Computer Corp., Westford, MA) computer. Photographic negatives of the images were made by transmitting the digitized and rescaled images to the high-resolution photo-monitor of a Philips 505 scanning electron microscope (Philips Electron Optics, Eindhoven, The Netherlands). Fixation of rat liver. Male Wistar rats, weighing 200 to 250 gm, had free access to water but were fasted overnight before they were killed. Low-pressure perfusion fixation of the liver through the portal vein was performed with the animals under ether anesthesia according to the method of De Zanger and Wisse (33).The liver was preperfused with isotonic buffer for 30 sec and subsequently perfusion-fixed for 10 min with 2% or 4%p-formaldehydedissolved in 0.1 m o m Na and K phosphate buffer, pH 7.4.After the perfusion small tissue blocks were kept in fixative at 4" C for up to 3 hr. Cryoultramicrotomy. Tissue blocks, 0.5 x 0.5 x 0.5 mm3, were immersed in 2.3 molL sucrose for 1 hr before snapfreezing in liquid nitrogen. Cryosectioning was performed on a LKB 5 ultracryotome (LKB Produkter AB, Bromma, Sweden) according to the method of Tokuyasu (34). Immunoelectron microscopy. Sections 100 to 200 nm thick were cut at approximately -90" C and mounted on formvar (Fluka AG, Buchs, Switzerland)-coated hexagonal copper grids. To quench free aldehyde groups, the sections were preincubated with 0.1 m o m glycine dissolved in PBS and with 2% BSA to block unspecific binding sites. The sections were then incubated for 1hr with primary antibodies ( 25 kg/ml), washed 3 times for 5 min in PBS and labeled for 1 h r with 5 nm, 8 nm or 10 nm protein A-gold complexes (PA-Au5, PA-Au8 and PA-AulO, respectively). The PA-Au5,PA-Au8 and PA-AulO complexeswere prepared as described previously (35, 36). These reagents were applied after dilution in PBS containing 1%BSA so that the optical density read at A = 520 nm over 1cm distance was approximately 0.1 ( A u ~ )0.15 , (Au8) and 0.05 (AulO),respectively. Adding the BSA to the diluted colloidal gold suspension minimized the tendency of the gold particles to adhere unspecifically to the tissue. Double labeling of the sections was performed according to the method of Geuze and Slot (37, 38). Briefly, after labeling the sections with anti-collagentype I antibodies followedby the PA-Au8 or PA-AulO reagent, the sections were incubated for 15 min with free protein A (50 pg/ml) to bind remaining free Fc fragments of IgG molecules. After they were thoroughly washed with PBS to remove unbound protein A, the sections

-

COLLAGEN TYPE I AND 111 IN HYBRID FIBRILS

Vol. 12, No. 2, 1990

235

were incubated with antiprocollagen type 111 antibodies and then incubated with the PA-Au5 reagent. Immunostaining was followed by washng the sections five times for 5 min in PBS and one time for 5 min in distilled water, staining for 3 min with 2% uranyl oxalate, briefly washing in distilled water, staining for 3 min with 2% to 4% uranyl acetate and embedding in 1.3%methylcellulose(Tyllose MH 300, Fluka AG). Control incubations included sections incubated in nonimmune rabbit IgG (Sigma Chemical Co.;25 p,g/ml), but otherwise treated as described, and sections incubated with PA-Au complex only. The sections were viewed in a Phdips 400 transmission electron microscope at 80 or 100 kV.

Morphometry Rat livers. Five normal rat livers were perfusion-fixed through the portal vein for at least 5 min by a routine low-pressure perfusion procedure (33). The fixative consisted of 1.5% glutaraldehyde, 0.01% CaC1,. 2H,O and 2% sucrose dissolved in 0.1 m o m cacodylate buffer, pH 7.4. Small pieces of tissue were further immersion fixed in 1%Millonig's OsO, for 1 hr, dehydrated in alcohol and propylene oxide, and embedded in epoxy resin. Sections 60 nm thick were cut with a diamond knife, stained for 20 min with uranyl acetate, stained for 10 min with lead citrate and viewed at 80 kV in a Philips 400 transmission electron microscope (Philips Electron Optics). Human livers. Liver needle biopsy specimens, taken under laparoscopy in the Gastroenterology Unit of the University Hospital of the Free University Brussels, were collected over a period of 2 yr (1982 to 1984). The experimental procedures on human tissue were performed with the consent of the Ethical Committee of the Faculty of Medicine and Pharmacy of the Free University of Brussels. Part of each biopsy specimen was puncture perfusion-fixed according to the procedure described previously (39)within 30 sec after removal from the organ. The specimens were fixed with the same fixative as described for rat livers. Further processing of the tissue was identical to the procedure mentioned in the previous paragraph. From a series of 35 specimens from 35 patients, five specimens were selected in which no obvious histological abnormalities were detected. Morphometry of collagen fibrils and bundles. From each human or rat liver, sections were cut from at least three blocks. In each section, cross-sectioned sinusoids were selected at random. All cross-sectionedfibrils present in the space of Disse of these sinusoids were photographed and printed at a final magnification of 97,200. The minimal diameter of each fibril and the number of fibrils per bundle were measured semiautomatically and statistically evaluated on an Interactives Bild Analyse System 1 (IBAS-1) computer system (Kontron, Analytik GmbH, Munich, FRG). At least 1,000 fibrils present in about 50 bundles were analyzed per liver.

RESULTS Characterization of the antibodies

The specificity of the affinity-purified antibodies used in this study was demonstrated by immunoblotting (Fig. 1). The antibodies to monkey collagen type I reacted with the cxl(1) and a2(I) chains of both rat and monkey collagen type I (Fig. 1, lanes 1and 21, but not with the aminoterminal propeptide or the helical portion of rat or monkey procollagen type I11 (Fig. 1, lanes 3 and 4) because pN-collagen type 111, the main

1 2 3 4 5 6 7 FIG.1. Immunoblot demonstrating the specificity of the antibodies to collagen type I and to the aminoterminal propeptide of procollagen type 111. Lanes 1 and 5, rat collagen type I; lanes 2 and 6, monkey collagen type I; lanes 3 and 7, rat procollagen type 111; lanes 4 and 8, monkey procollagen type 111. The nitrocellulose blots were incubated with anity-purified antibodies against monkey collagen type I flanes 1, 2, 3 and 4) and with antibodies against the aminoterminal propeptide of procollagen type I11 (lanes 5 to 8). Note the lack of cross-reactivity of the antibodies. Antibodies to collagen type I do not recognize epitopes on the aminoterminal propeptide or on the helical portion of procollagen type 111 because pN-collagen type 111, the major protein in lanes 3 and 4, is not recognized. The slight reaction seen in these lanes is caused by minute contamination of the procollagen type I11 preparations by collagen type I. Antibodies to the aminoterminal propeptide of procollagen type I11 react with pN-collagen type 111 extracted from rat and monkey skin, but not with collagen type I. Note also the interspecies reactivity of the antibodies. Antibodies prepared by immunization of rabbits with rat collagen also react with the homologous molecules from monkey and vice versa.

protein present in lanes 3 and 4, was not recognized. The faint bands seen in lanes 3 and 4 were the cxl(1) chains of collagen type I. The latter collagen type was present as a minor contaminant of the procollagen type I11 preparations. The antibodies directed against the aminoterminal propeptide of procollagen type I11 reacted with rat and monkey pNal(II1) (Fig. 1, lanes 7 and 8) but not al(1) or a2(I) chains (Fig. 1, lanes 5 and 6). Immunocytochemistry

By immunofluorescence, collagen type I was found on bundles of varying diameter in the space of Disse and between adjacent parenchymal cells (Fig. 2). Procollagen type I11 was also present in the space of Disse on bundles with different diameters (Fig. 3). The intensity of the reaction obtained by staining the sections with antibodies directed against procollagen type 111was stronger than the intensity obtained with antibodies to collagen type I. The fluorescent images obtained with antibodies

236

GEERTS ET AL.

HEPATOLOGY

FIG.2. A 5 &msection of rat liver stained for collagen type I by the indirect immunofluorescence technique. Collagen type I is clearly present along the sinusoids and between adjacent parenchymal cells. The immunocytochemical reaction is clearly not restricted to one class of collagen bundles. By computerized rescaling of the digitized fluorescence image, we were able to visualize faintly stained thin bundles as well as thick bundles. (Magnification x 812.) FIG.3. A 5 km section of rat liver stained for the N-terminal propeptide of procollagen type I11 by indirect immunofluorescence. Procollagen type I11 is present along the sinusoids and between adjacent parenchymal cells. Thick and thin collagen bundles are labeled. (Magmiication x 812.)

to collagen type I were rescaled to visualize the faintly stained thin collagen bundles. After immunogold labeling for collagen type I, we found consistent labeling of the majority of the striated interstitial collagen fibrils in the space of Disse (Fig. 4). In general, the fibrils showed no periodic labeling. No obvious difference in staining intensity between solitary fibrils and fibrils organized in bundles was observed. In control sections that were incubated with nonimmune rabbit IgG solution (25 p,g/ml) instead of affinitypurified antibody solution, gold labeling of collagen fibrils or bundles was negligible (Fig. 5). With antibodies directed against procollagen type 111, we found labeling of all striated interstitial collagen fibrils present in the space of Disse or in the spaces

between adjacent parenchymal cells (Fig. 6). The intensity of the staining was stronger than observed with antibodies against collagen type I. Some of the solitary fibrils and some peripheral fibrils in thicker bundles showed periodic labeling (Fig. 6). The average distance between the gold particles on periodically labeled fibrils was 65 nm. When collagen fibrils or bundles were sectioned perpendicular to their longitudinal axis, gold particles surrounding the fibrils and gold particles on top of the cross-sectioned fibrils were observed (Fig. 7). By double labeling, we found that most of the fibrils contained both antigens (Fig. 8).Labeling of procollagen type I11 was again more intense than labeling of collagen type I.

COLLAGEN TYPE I AND 111 IN HYBRID FIBRILS

Vol. 12, No. 2, 1990

237

t \\\'

J

b b

FIG.4. A 100 to 200 nm section of rat liver stained for collagen type I by the immunogold procedure using 5 nm gold particles. All individual fibrils of the collagen bundle are decorated. EC = endothelial cell process. (Magnification x 64,090). FIG.5. A 100 to 200 nm section of rat liver incubated with nonimmune rabbit IgG solution (25 p,g/ml) instead of affinity-purified antibody and otherwise treated according to the same protocol. Gold labeling of the collagen fibrils is negligible. (Magnification x 58,905). FIG.6. A 100 to 200 nm section of rat liver stained for the N-terminal propeptide of procollagen type I11 using 5 nm gold particles. All individual fibrils of this collagen bundle, located between two adjacent parenchymal cells, are decorated with gold particles (arrowheads).The labeling of some fibrils shows periodicity (arrowheads).The average distance between the gold particles in this periodic pattern is 65 nm. This distance comes very close to the distance (64 nm) over which adjacent collagen molecules are shifted in the D-staggered array in dry specimen. P = parenchymal cell. (Magmfication x 64,090). FIG.7. A 100 to 200 nm section of rat liver stained for the N-terminal propeptide of procollagen type I11 using 5 nm gold particles. The collagen fibrils composing this bundle are sectioned perpendicular to their longitudinal axis. Although most of the gold particles surround the fibrils, some gold particles are on top of a cross-sectional fibril (amwheadsj. This observation demonstrates that pN-collagen type I11 is not only present in the outer layers of the fibrils but is also incorporated into it. mv = microvillus of a parenchymal cell. (Magnification x 53,550).

Morphometry

In human and rat liver, the distribution of the minimal diameters of randomly selected, crosssectioned, interstitial collagen fibrils in the space of Disse was unimodal (Figs.9 and 10).In human liver, the average fibril diameter equaled 57.2 nm (S.D. = 9.6 nm; n = 5,852) and in the rat 62.4 nm (S.D. = 12.8 nm; n = 6,200). The obtained distributions differed slightly from a gaussian distribution with the same mean value

and S.D. In the Kolmogoroff-Smirnov test, the difference was significant (p < 0.01). The frequency distribution of the number of fibrils per bundle present in the space of Disse was highly asymmetrical. In rat liver, small collagen bundles containing less than 30 fibrils accounted for 91% of the total number of bundles (Table 1, row 2). The minimal diameter of these bundles, measured in cross-section, did not exceed 600 nm. These small bundles contained

238

GEERTS ET AL.

HEPATOLOGY

. . *

.

.

*

d

.
150

37

37

14

8

2

2

14

32

22

15

4

6

a Frequency distribution of the interstitial collagen bundles in human liver according to the number of fibrils they contain and percentage of fibrils organized in bundles of different sizes.

that the pN-collagen type I11 molecules were either in the process of being assembled in the fibrils or were already incorporated. If free N-terminal propeptide of procollagen type I11 had been present or if the pN-

collagen type I11 molecules had been near the fibds but not really incorporated into them, the gold label should have been distributed at random. Also, the incorporation of pN-collagen type I11 inside fibrils was further docu-

240

HEPATOLOGY

GEERTS ET AL.

mented by the observation that labeling was also present on top of cross-sectionedfibrils. This observation, taken together with the fact that collagen fibrils were labeled periodically, argued against the possibility that the antibodies reacted either with free aminoterminal propeptide associated with the collagen fibrils without being part of the fibril itself, or with free aminoterminal propeptide passively trapped inside nascent fibrils. Sato, Leo and Lieber (13) reached the same conclusion by studying the ultrastructural localization of the aminoterminal propeptide of procollagen type I11 in baboon liver. Also, when rats were injected with '2SI-labeled PIIIP, no evidence for trapping of PIIIP by nascent collagen fibrils was found (45). In contrast, 82.5% of the radiolabeled PIIIP retained by the liver was rapidly taken up by sinusoidal endothelial cells and degraded in the lysosomal compartment. The remainder was taken up by parenchymal and Kupffer cells. Our morphometrical data confirmed the conclusion that the liver perisinusoidal space contained a fairly homogeneous population of interstitial collagen fibrils. When our data were compared to those obtained for human skin (211, differences became apparent. The diameter of the fibrils in the skin ranged from 20 to 120 nm. In the dermis of human skin, anticollagen and antiprocollagen type I11 antibodies reacted preferentially with thin fibrils, that is, fibrils with a diameter of 20 to 60 nm. Antibodies against type I collagen decorated about 80%of the fibrils, their diameter ranging from 20 to 80 nm. Originally, the authors concluded that skin contained fibrils consisting of either type I or type I11 collagen. Recently, however, Keene et al. (23) presented indirect ultrastructural evidence for the presence of mixed collagen fibrils in the normal human skin, tendon and amnion. In cell culture and in several tissues other than liver or skin, the presence of hybrid fibrils was described. Furcht et al. (46) reported that human fibroblasts in culture formed fibrils composed of procollagen types I and I11 and of fibronectin. Collagen fibrils of mixed composition occurred in leiomyoma (251, because intermolecular covalent cross-links between type I and type I11 collagen molecules have been demonstrated in this tissue. Konomi, Sano and Nagai (24) have demonstrated by double staining immunofluorescence that type I and type I11 collagen were present on the same interstitial fibers (i.e., bundles of fibrils) in the lymph node. However, no immunoelectron microscopic observations were made. Therefore, the possibility of fibrils, consisting of pure type I and pure type I11 collagen, organized in the same bundles, could not be excluded. In the developing avian cornea, hybrid fibrils consisting of type I and type I1 collagen and of type I and type V collagen were demonstrated by using monoclonal antibodies against these collagens (47,48). In chick embryo sternal cartilage, the presence of mixed fibrils consisting of the collagen types 11, IX and XI was demonstrated (49). The mechanism by which fibrils are assembled into bundles is not known. Lapike, Nusgens and Pierard

( 5 0 ) have studied in uztro the self-assembly of collagen type I and I11 in bundles by thermal induction. They

have demonstrated that pure preparations of type I collagen tend to form large bundles, whereas type I11 collagen assembles into thinner bundles. Mixtures of both collagens form bundles of intermediate size. This observation, combined with our finding that many collagen fibrils in the space of Disse contain both collagen type I and procollagen type 111, may provide an explanation why collagen bundles in the liver lobule have a smaller diameter than those of skin or tendon. In conclusion, we have demonstrated that the striated interstitial collagen fibrils in the space of Disse are composed of at least two collagen types, that is, collagen type I and procollagen type 111. The biological significance of the presence of these hybrid fibrils in the space of Disse of the liver is not understood. Mixing the two collagens may accelerate or retard the growth of the fibrils, but it may also modulate the structural properties, functional properties or both of the fibrils.

Acknowledgments: We thank Dr. J.-W. Slot, Dr. H. Geuze and R. Willemsen (State University Utrecht, The Netherlands) for teaching A.G. how to use the L.K.B. cryotome and for supplying us with the PA-gold reagents. We also thank Mrs. C. Derom for her excellent photographic work. REFERENCES 1. Schuppan D, Hahn E. Components of the extracellular matrix (collagens,elastin, glycoproteins and proteoglycans). In: Wolff J, Sievers J, Berry M, eds. NATO AS1 Series, Vol. H5. Heidelberg: Springer-Verlag 1987:3-29. 2. Burgeson R. New collagens, new concepts. Annu Rev Cell Biol 1988;4:551-577. 3. Miller M, Rhodes R. Preparation and characterization of the different types of collagen. Methods Enzymol 1982;82,PartA3364. 4. Rojkind M, Ponce-NoyolaP. The extracellular matrix of the liver. Coll Re1 Res 1982;2:151-175. 5. Schuppan D, Hahn E. Connective tissue proteins in the liver: collagens and glycoproteins.In: Waldschmidt J, ed. Cholestasis in neonates. Munich: Zuchschwerdt 1988:97-115. 6. Schuppan D. Structure of the extracellular matrix in normal and fibrotic liver: collagens and glycoproteins. Semin Liver Dis 199O;lO:1-10, 7. Brodsky B, Eikenberry E. Characterization of fibrous forms of collagen. In: Cunningham L, Frederiksen D, eds. Methods in enzymology, Vol. 82, Part A. New York Academic Press 1982: 127-174. 8. Piez K. Molecular and aggregate structures of the collagens. In: Piez K, Reddy A, eds. Extracellular matrix biochemistry. New York Elsevier Science Publishing Co.: 1984, 1-39. 9. Konomi H, Sano J, Nagai Y. Immunohistochemicallocalization of type I, I11 and IV (basement membrane) collagens in the liver. Acta Pathol Jpn 1981;31:973-978. 10. Sano J, Sat0 S, Ishizaki M, Yajima G, Konomi H, Fujiwara S, Nagai Y. Types I, I11 and IV (basement membrane)collagens in the bovine liver parenchyma: electron microscopical localization by the peroxidase, labeled antibody method. Biomed Res 1981;2:546551. 11. Geerts A, Geuze H, Slot J, Voss B, Schuppan D, Schellinck P, Wisse E. Immunogold localization of procollagen type 111, fibronedin and heparan sulfate proteoglycan on ultrathin frozen sections of the normal rat liver. Histochemistry 1986;84:355-362. 12. Geerts A, Schellinck P, De Zanger R, Schuppan D, Wisse E. Fine

Vol. 12, No. 2, 1990

COLLAGEN TYPE I AND 111 IN HYBRID FIBRILS

structural distribution of procollagen type 111 in the normal rat liver: critical reinvestigation of the reticular fiber concept. In: Kirn A, Knook D, Wisse E, Eds. Cells of the hepatic sinusoid, Vol. 1. Rijswijk, The Netherlands: Kupffer Cell Foundation, 1986:221226. 13. Sat0 S, Leo M, Lieber C. Ultrastructural localization of type I11 procollagen in baboon liver. Am J Pathol 1986;122:212-217. 14. Peyrol S, Grimaud J. Perisinusoidal connective tissue: immunohistochemical mapping of the major matricial components. In: Bioulac-Sage P, Balabaud C, eds. Sinusoids in human liver: health and disease. Rijswijk, The Netherlands: Kupffer Cell Foundation, 1988~323-340. 15. Grimaud J, Druguet M, Peyrol S. Chevalier 0, Herbage D, El Badrawy N. Collagen immunotyping in the human liver. J Histochem Cytochem 1980;28:1145-1156. 16. Martinez-Hernandez A. The hepatic extracellular matrix. I. Electron immunohistochemical studies in normal rat liver. Lab Invest 1984;51:57-74. 17. Clement B, Emonard H, Rissel M, Druguet M, Grimaud J, Herbage D, Bourel M, et al. Cellular origin of collagen and fibronectin in the liver. Cell Mol Biol 1984;30:489-496. 18. Clement B, Rissel M, Peyrol S , Mazurier Y, Grimaud J, Guillouzo A. A procedure for light and electron microscopic intracellular immunolocalization of collagen and fibronectin in rat liver. J Histochem Cytochem 1985;33:407-415. 19. Nowack H, Gay S , Wick G, Becker U, Timpl R. Preparation and use of immunohistology of antibodies specific for type I and type I11 collagen and procollagen. J Immunol Methods 1976;12:117-124. 20. Fleischmajer R, Gay S , Perlish S, Cesarini J. Immunoelectron microscopy of type I11 collagen in normal and scleroderma skin. J Invest Dermatol 1980;75:189-191. 21. Fleischmajer R, Timpl R, Tuderman L, Raisher L, Wiestner M, Perlish J, Graves P. Ultrastructural identification of extension aminopropeptides of type I and type I11 collagens in human skin. Proc Natl Acad Sci USA 1981;78:7360-7364. 22. Perlish J, Lemlich G, Fleischmajer R. Identification of collagen fibrils in scleroderma skin. J Invest Dermatol 1988;90:48-54. 23. Keene D, Sakai L, Bachinger H, Burgeson R. Type I11 collagen can be present on banded collagen fibrils regardless of fibril diameter. J Cell Biol 1987;105:2393-2402. 24. Konomi H, Sano J, Nagai Y. Immunocytochemical localization of types I, 111, IV (basement membrane) collagens in the lymph node: co-distribution of types I and I11 collagens in the reticular fibers. Biomed Res 1981;2:536-545. 25. Henkel W, Glanville R. Covalent crosslinking between molecules of type I and type I11 collagen: the involvement of the N-terminal, non-helical regions of al(I)and ul(II1) chains in the formation of intermolecular crosslinks. Eur J Biochem 1982;122:205-213. 26. Amenta P, Gay S, Vaheri A, Martinez-Hernandez A. The extracellular matrix is an integrated unit: ultrastructural localization of collagen types I, 111, IV, V, VI, fibronectin, and laminin in human term placenta. Coll Re1 Res 1986;6:125-152. 27. Amenta P, Gil J, Martinez-Hernandez A. Connective tissue of rat lung: ultrastructural localization of collagen types 111, IV and VI. J Histochem Cytochem 1988;36:1167-1173. 28. Sage H, Bornstein P. Preparation and characterization of procollagen and procollagen-collagen intermediates. In: Cunningham L, Frederiksen D, eds. Methods in enzymology. New York: Academic Press 1982:86-127. 29. Becker J, Schuppan D, Eenzian H, Bals T, Hahn E, Cantaluppi C. Immunohistochemical distribution of collagens type IV,V and VI and of procollagens type I and I11 in human alveolar bone and dentine. J Histochem Cytochem 1986;34:1417-1429. 30. Schuppan D, Dumont J, Kim K, Hennings G, Hahn E. Serum concentration of the aminoterminal procollagen type I11peptide in

24 1

the rat reflects early formation of connective tissue in experimental liver fibrosis. J Hepatol 1986;3:27-37. 31. Dziadek M, Richter H, Schachtner M, Timpl R. Monoclonal antibodies used as probes for the structural organization of the central region of fibronectin. FEBS Lett 1983;155:321-325. 32. Schuppan D, Besser M, Schwarting R, Hahn E. Radioimmunoassay for the carboxyterminal crosslinking domain of type IV (basement membrane) procollagen in body fluids. J Clin Invest 1986;78:241-248. 33. De Zanger R, Wisse E. The filtration effect of rat liver fenestrated sinusoidal endothelium on the passage of (remnant) chylomicrons to the space of Disse. In: Knook D, Wisse E, eds. Sinusoidal liver cells. Amsterdam: Elsevier Biomedical Press 1982:69-76. 34. Tokuyasu K, Singer S. Improved procedures for immunoferritin labeling of ultrathin frozen sections. J Cell Biol 1976;71:894-906. 35. Slot J, Geuze H. Sizing of protein A-colloidal gold probes for immuno-electron microscopy. J Cell Biol 1981;90:533-536. 36. Slot J, Geuze H. A new method of preparing gold probes for multiple-labehg cytochemistry. Eur J Cell Biol 1985;38:87-93. 37. Geuze H, Slot J, Van der Ley P, Scheffer R. Use of colloidal gold particles in double-labeling immunoelectron microscopy of ultrathin frozen sections. J Cell Biol 1981;89:653-665. 38. Slot J, Geuze H, Weerkamp A. Localization of macromolecular components by application of the immunogold technique on cryosectioned bacteria. Methods Microbiol 1988;20:211-236. 39. De Wilde A, Van Der Spek P, Devis G, Wisse E. On the fixation of needle biopsies of rat liver tissue as a model to study the h e structure of sinusoidal cells. In: Knook D, Wisse E, eds.Sinusoidal liver cells. Amsterdam: Elsevier Biomedical Press, 1982:85-92. 40. Clement B, Grimaud J, Campion J, Deugniez Y, Guillouzo A. Cell types involved in collagen and fibronectin production in normal and fibrotic human liver. HEPATOL~CY 1986;6:225-234. 41. Timpl R. Immunology of the collagens. In: Piez K, Reddy A, eds. Extracellular matrix biochemistry. New York: Elsevier Science Publishing Co 1984:159-190. 42. Kent G, Gay S, Inouye T, Bahu R, Minick 0, Popper H. Vitamin A-containing lipocytes and formation of type I11 collagen in liver injury. Proc Natl Acad Sci USA 1976;73:3719-3722. 43. WickG, Brunner H, Penner E, Timpl R. The diagnosticapplication of specific antiprocollagen sera. 11. Analysis of liver biopsies. Int Arch Allergy Appl Immunol 1978;56:316-324. 44. Sakakibara K, Ooshima A, Igarashi S, Sakakibara J. Immunolocalization of type I11 collagen and procollagen in cirrhotic human liver using monoclonal antibodies. Virchows Arch [A] 1986;409: 37-46. 45. Smedsrod B. Aminoterminal propeptide of type I11 procollagen is cleared from the circulation by receptor-mediated endocytosis in liver endothelial cells. Collagen Re1 Res 1988;42:375-388. 46. Furcht L, Wendelschafer-Crabb G, Mosher D, Foidart J. An axial periodic fibrillar arrangement of antigenic determinants for fibronedin and procollagen on ascorbate treated human fibroblasts. J Supramol Struct 1980;13:15-33. 47. Linsenmayer T, Fitch J, Gross J, Mayne R. Are collagen fibrils in the developing avian cornea composed of two different collagen types? Ann NY Acad Sci 1985;386:232-245. 48. Birk D, Fitch J, Babiarz P, Linsenmayer T. Collagen type I and V are present in the same fibrils in avian corneal stroma. J Cell Biol 1988;106:999-1008. 49. Mender M, Eich-Bender S, Vaughan L, Winterhalter K, Bruckner P. Cartilage contains mixed fibrils of collagen types 11, M and XI. J Cell Biol 1989;108:191-197. 50. Lapiere C, Nusgens B, Pierard G. Interaction between collagen type 1 and type 111in conditioning bundles organization. Connect Tissue Res 1977;1977:21-29.