Epitope-Defined Monoclonal Antibodies Against Multiplexin Collagens Demonstrate That Type XV and XVIII Collagens Are Expressed in Specialized Basement Membranes

Epitope-defined Monoclonal Antibodies against Multiplexin Collagens Demonstrate that Type XV and XVIII Collagens are Expressed in Specialized Basement Membranes

Yasuko Tomono1, Ichiro Naito2,5∗, Kaori Ando1, Tomoko Yonezawa5, Yoshikazu Sado3, Satoshi Hirakawa4,5, Jirô Arata4, Tohru Okigaki6, and Yoshifumi Ninomiya5 Divisions of 1Molecular and Cell Biology, 2Ultrastructural Biology, and 3Immunology, Shigei Medical Research Institute, 2117 Yamada, Okayama 701-0202, Japan; Departments of 4Dermatology, and 5Molecular Biology and Biochemistry, Okayama University Graduate School of Medicine and Dentistry, 2-5-1 Shikata-cho, Okayama 700-8558, Japan; and 6Division of Social Welfare, Faculty of Social Welfare, Kinki Welfare University, Fukusaki, Hyogo 679-2217, Japan

ABSTRACT. Type XV and type XVIII collagens are classified as part of multiplexin collagen superfamily and their C-terminal parts, endostatin and restin, respectively, have been shown to be anti-angiogenic in vivo and in vitro. The α1(XV) and α1(XVIII) collagen chains are reported to be localized mainly in the basement membrane zone, but their distributions in blood vessels and nonvascular tissues have yet to be thoroughly clarified. In the present study, we raised monoclonal antibodies against synthetic peptides of human α1(XV) and α1(XVIII) chains and used them for extensive investigation of the distribution of these chains. We came to the conclusion that nonvascular BMs contain mainly one of two types: subepithelial basement membranes that contained type XVIII in general, or skeletal and cardiac muscles that harbored mainly type XV. But basement membranes surrounding smooth muscle cells in vascular tissues contained one or both of them, depending on their locations. Interestingly, continuous capillaries contained both type XV and type XVIII collagens in their basement membranes; however, fenestrated or specialized capillaries such as glomeruli, liver sinusoids, lung alveoli, and splenic sinusoids expressed only type XVIII in their basement membranes, lacking type XV. This observation could imply that different functions of basement membranes in various tissues and organs use different mechanisms for the endogenous control of angiogenesis.

Key words: basement membrane/type XV collagen/type XVIII collagen/capillary/smooth muscle cell

The collagen family, which consists of 20 collagen types various pathological processes (Olsen and Ninomiya, 1999). and 38 collagen α-chains (Myllyharju and Kivirikko, 2001), The BM, lying beneath epithelia and around muscle, adi- are known to be major components of connective tissues pose, and Schwann cells, and made visible by the periodic and/or basement membranes (BMs). They not only serve as acid-Schiff reaction under light microscopic observation, is scaffolds for tissues and organs but also are involved in var- composed mainly of type IV collagen, laminin, nidogen, ious biological events such as development, morphogenesis, and heparan sulfate proteoglycan. Multiple components of inflammation, tissue repair, and filtration, as well as in these materials have been reported to comprise BMs in a tissue-specific fashion. For instance, six α(IV) collagen chains *To whom correspondence should be addressed: Ichiro Naito, Ph. D., Divi- have been identified and at least three molecular forms dis- sion of Ultrastructural Biology, Shigei Medical Research Institute, 2117 Ϋ Yamada, Okayama 701-0202, Japan. tribute in various BMs (Hudson et al., 1993; Sado et al., Tel: +81–86–282–3113, Fax: +81–86–282–3115 1998). One form, α3/α4/α5, is specifically present in BMs E-mail: [email protected] where it functions as a permeability selective barrier such as Abbreviations: BM, basement membrane; SDS-PAGE, sodium dodecyl sulfate-polyacrylamide gel electrophoresis; KLH, keyhole limpet hemocy- in the BMs in glomerulus in kidney and alveolus in lung anin; mAb, monoclonal antibody; PBS, phosphate-buffered saline; TBS, (Seki et al., 1998; Saito et al., 2000). Such findings suggest Tris-HCl-buffered saline; TBS-T, Tris-HCl-buffered saline containing the possibility that the composition of collagen IV mole- 0.9% NaCl and 0.05% Tween 20; ELISA, Enzyme-linked immunosorbent cules in various tissues corresponds with specific biological assay.

9 Y. Tom ono et al. functions of specialized BMs. rats (Charles River Japan, Inc., Yokohama, Japan) of 8 weeks of Recently more matrix macromolecules have been report- age (Kishiro et al., 1995; Tomono et al., 1998). ed to be components of BMs. One interesting collagen, type XVIII, which contains an anti-angiogenesis part, endostatin, Production of monoclonal antibodies at its carboxyl end, was reported to be present as the core Three to 4 weeks after the injection, the rats were sacrificed under protein of the third heparan sulfate proteoglycan in BMs of ether anesthesia. Rat lymph node cells obtained from medial iliac various tissues (Muragaki et al., 1994, 1995; Musso et al., lymph nodes were fused to SP2/O mouse myeloma cells by use of 1998; Halfter et al., 1998). Tissue localizations of the dis- polyethylene glycol (Kishiro et al., 1995). The fused cells were tinct variants of the two transcripts, the short and long then screened for reactivity with the synthetic peptides and with forms, and their translation products were demonstrated by the native collagen proteins. The former reaction was examined by in situ hybridization and immunohistochemistry in most enzyme-linked immunosorbent assay (ELISA), and the latter reac- conventional BMs including blood vessels, epithelial struc- tion was analyzed by immunohistochemistry using frozen sections tures and around muscles and in the liver sinusoidal area, of human kidney and placenta. Fused cells positive for both assays (Saarela et al., 1998a, 1998b). Intriguingly, Lin et al. (2001) were cloned by limiting dilution. successfully demonstrated a model in development where differential organ morphogenesis is regulated by an intra- Enzyme-linked immunosorbent assay organ patterning process that involves type XVIII collagen and inductive signals that control gene expression. Compa- After the fused cells were cultured for 5 days, culture fluids were rable type XV collagen, which contains a region called res- screened by ELISA using the peptides without the KLH. ELISA tin (Ramachandran et al., 1999) at its carboxyl end and has plates (Nunc, Roskilde, Denmark) were coated with the peptides (3 similar anti-angiogenic activity, seems to also be one of the µg/ml in phosphate-buffered saline, PBS) for 2 h at 37°C, washed BM components (Myers et al., 1996; Li et al., 2000; Sasaki with PBS, and then blocked with 1% bovine serum albumin for 1 h et al., 2000). Kivirikko and co-workers observed that the at 37°C. Following the washing with PBS, the fluids were applied major producers of type XV collagen were mesenchymal to the plate, and incubation was done for 1 h at 37°C. After having cells such as fibroblasts, muscle cells, and endothelial cells been washed with PBS, the plates were incubated for 1 h at 37°C as well as certain epithelial cells (Kivirikko et al., 1995). with peroxidase-conjugated rabbit antibody to rat immuno- Marked accumulation of type XV collagen was demonstrat- globulins (diluted to 1/1000, Dako Japan, Kyoto, Japan). After a ed in kidney fibrosis (Hägg et al., 1997a). PBS wash, o-phenylenediamine (Wako Pure Chemical, Osaka, Sasaki et al. demonstrated that both endostatin derived Japan) was added to each plate, which was then incubated for from type XVIII collagen and restin from type XV collagen another 30 min at room temperature. The reaction of the o- inhibited chrorioallantoic membrane angiogenesis, but they phenylenediamine was stopped with 3 M H2SO4, and absorbance at showed some differences in structure and functional proper- 490 nm was measured by a Bio-Rad model 450 microplate reader ties as well as tissue distribution, indicating that their reper- (Bio-Rad Laboratories, Hercules, CA, USA). toire of biological interactions is overlapping but not identical. We therefore tried to clarify the precise location of these Immunohistochemistry multiplexin collagens within BMs that had not been clari- Human kidney, placenta, skin, liver, small intestine, lung, skeletal fied yet. We report herein chain-specific antibodies and the muscle, heart and spleen were used for the immunohistochemistry. distribution of newly distinguished multiplexin collagen Renal tissues were obtained from the normal part of kidneys surgi- chains. cally removed from patients suffering from renal cancer (2 males, 56 y and 65 y of age). Placental tissues were obtained at normal Materials and Methods delivery. Skin tissues were obtained from surgically operated normal skin under suspicion of skin cancer (male, 48 y) and Alport Synthetic peptides syndrome (male, 25 y). Normal parts of liver and small intestine were also surgically obtained at operations for liver cancer (male, When the primary structures of type XV and type XVIII collagens 75 y) and caecum cancer (male, 65 y). Lung, skeletal muscle, were compared, we found them to be similar, therefore we selected heart, and spleen tissues were obtained from cadavers (male, 75 y several regions that were different and more specific for each chain of age, and female, 100 y). The experiment using human tissues for use as immunogens. Synthetic peptides used as immunogens was performed according to the Guidelines for Human Tissue were YVSEQDGSVLSVPGP (peptide name: P2), SWEALFSG- Experiments of Shigei Medical Research Institute and with the SEGPLKPGAR (P6) for type XVIII, and PNPISSANYEKPA consent of the patients and the family members. (P12) for type XV. Cysteinyl residues were added to their carboxyl The tissues were embedded in Tissue-Tek O.C.T. compound termini in order to conjugate the peptides with keyhole limpet (Sakura Finetechnical Co., Ltd., Tokyo, Japan), and immediately hemocyanin (KLH) (Fig. 1A, 1B and 1C). Following the conjuga- frozen in liquid nitrogen. Fresh frozen sections of 4-µm thickness tion with KLH, 50 µg of each peptide was emulsified with were fixed with acetone for 10 min. If necessary, the sections were Freund’s complete adjuvant and injected into female WKY/NCrj

10 Monoclonal Antibodies for Multiplexin

Fig. 1. Location of the synthetic peptides used for the antigens within the α1 (XVIII) and α1 (XV) collagen chains and their amino acid sequences. Schematic representations of the human α1 (XVIII) and α1 (XV) chains are drawn with non-collagenous (NC) domains (thin bars) and collagenous (COL) domains (white boxes). The 2 thick bars at the carboxyl ends indicate the location of restin and endostatin. A: Peptide 12 (P12) for α1 (XV) chain is within the NC1 domain located at the carboxyl end of the α1 (XV) chain. B: Locations of the peptides 2 (P2) and 6 (P6) for α1 (XVIII) are indicated by short gray bars, and their amino acid sequences used for antigens are shown in C. denatured with 0.2M KCl-HCl buffer, pH 1.4. After having been washed with PBS, the sections were blocked with normal goat se- Double staining for type XVIII and type XV collagens rum (diluted to 1/15 by PBS) for 10 min and then incubated with Comparative distribution of type XV and type XVIII collagens the monoclonal antibodies (mAbs) for 1 h. An anti-α2(IV) chain was assessed by double staining using anti-α1(XV) and anti- mAb, H22, was used for demonstrating BM. After a wash with α1(XVIII) mAbs. Type XV collagen was stained with CH15A, fol- PBS, goat anti-rat IgG antibody labeled with fluorescein isothio- lowed by incubation with secondary anti-rat IgG antibody labeled cyanate (FITC), or Texas Red (TXRD) (diluted to 1/100 with PBS, with TXRD as mentioned above. Type XVIII collagen was stained ICN Pharmaceuticals Inc.-Cappel Product, Aurora, OH, USA) was with CH18D, which was labeled with FITC prior to the immuno- applied to the sections. Following the washing with PBS, the sec- staining. Fresh frozen sections of the tissues were prepared by the tions were mounted with Perma-fluor (Immunotech, Marseilles, method described above. Without the denaturation step, the sec- France) and examined with a Zeiss Axiophoto microscope. All tions were incubated with CH15A for 45 min, following by 10-min procedures for the immunostaining were performed at room tem- blocking with normal goat serum. This mAb was then detected perature. with TXRD-labeled secondary antibody (described previously).

11 Y. Tom ono et al.

After a wash with PBS, normal rat serum (diluted to 1/15 with analysis of type XV antibodies: for CH15A and CH15B, 19 PBS) was applied to the sections, followed by incubation with sequences of 9-residue-long peptides for PHQLLPPPNPIS- FITC-conjugated CH18D for 45 min. Finally, coverslips were SANYEKPALHLAALN were analyzed to detect the epitope re- mounted with the mounting media, following a wash with PBS. gion for antibodies against the PNPISSANYEKPA antigen (P12). Tissues used for the double staining were kidney, small intestine, All the materials used for these assays were purchased from Kura- and skin. bo Industries Ltd, Neyagawa, Japan. MAbs were incubated in 96- well micro-titer plates (Nalge Nunc International, Roskilde, Den- SDS-polyacrylamide gel electrophoresis and Western mark) with the peptides attached to the rod heads. The reactivity of blot analysis the mAbs with the peptides was analyzed by the ELISA method descried previously. BM fraction was obtained from human kidney and placenta by the method previously described (Sado et al., 1997). Briefly, the kidney and placenta were sliced into specimens of 1-mm thickness. Results After homogenization, they were washed with PBS, and centrifuged at 1,500 g for 10 min. The precipitate was dissolved in 1 M Monoclonal antibodies NaCl, sonicated with a SONIFIER/CELL DISRUPTOR 185 By the primary screening using ELISA with the synthetic (Branson Sonic Power, Co., Danbury, CT, USA), and centrifuged peptide P12 for α1 (XV) chain without KLH, we found 44 at 1,500 g for 10 min. These procedures were repeated 3 times positive clones out of 384 wells. Six clones out of these 44 until the precipitate turned white. The precipitate was washed with showed the expected staining pattern for BM collagen in distilled water, centrifuged at 12,000 g for 10 min, lyophilized placental tissues; i.e., positive staining was found along the (BM fraction), and kept at –80°C until further processed. BM zone of the trophoblasts and capillaries of the placental The BM fraction was dissolved in 0.5 M acetic acid (pH 2). Col- villi, and the cytoplasm and nucleus were negative. The lagen components were extracted from the fractions by adding 2 M positive hybridoma clones were subcloned twice by limiting urea containing 0.2% SDS and by heating at 95°C for 5 min. If dilution to obtain single monoclonal hybridomas. As a re- necessary, 10% 2-mercaptoethanol was added to the extract. SDS- sult, CH15A and CH15B were selected and established to polyacrylamide gel electrophoresis (SDS-PAGE) was performed be of different subclasses, IgG2a and IgG2b, respectively on 4–20% gradient polyacrylamide gels. (Fig. 1C).A similar screening method was used for mAbs Proteins were transferred from the gels after the SDS-PAGE to for P2 and P6 for the α1 (XVIII) chain, and we obtained GVHP filter membranes (Nihon Millipore, Yonezawa, Japan), similar efficiency and established 4 mAbs, CH18A, which were then washed with 10 mM Tris-HCl buffer (pH 7.4) CH18B, CH18C, and CH18D (Fig. 1C). containing 0.9% NaCl and 0.05% Tween 20 (TBS-T), and blocked Specificity of the mAbs was confirmed by immuno- with 5% skim milk in TBS-T. After a TBS-T wash, each mem- histochemistry on human placental and renal sections (Fig. brane was incubated with a given mAb for 60 min at 37°C and 3). CH15A and CH15B showed an identical staining pattern then washed with TBS-T. The mAbs bound to the membranes were on placental sections (Fig. 3A) as well as on renal sections. detected by incubation with affinity-purified peroxidase-conjugat- CH18A, CH18B, CH18C, and CH18D also showed the ed goat anti-rat IgGs (American Quolex, San Clemente, CA, USA) identical staining pattern in the renal tissue (Fig. 3B). The diluted to 1/1000 with TBS-T. The incubation was done for 60 min specificity was also confirmed by epitope-mapping using at room temperature. After the washing, peroxidase bound to the the multipin method as described previously (Ninomiya et sections was developed by using the diaminobenzidine tablet al., 1995). As shown in Fig. 2, all mAbs thus cloned were reagent (Wako Pure Chemical, Osaka, Japan). demonstrated to have epitope within the amino acid sequences of the antigens. The epitopes for CH15A and Epitope analysis CH15B were both identified to be NPI within the antigen Multipin assay was performed for epitope analysis. A set of consequence (Fig. 2a and 2b), despite of their different sub- secutive overlapping peptides including the amino acid sequences classes. The epitopes of the 2 clones for the α1 (XVIII) used for immunogens were synthesized, and fixed to the heads of chain, CH18B and CH18D, were narrowed down to the four the rods (Kurabo Industries Ltd, Neyagawa, Japan). For epitope amino acid residues of ALFS and VLSV (Fig. 2c and 2d), analysis of type XVIII antibodies CH18A and CH18B, 23 consec- respectively. utive 10 amino acid-residue-long peptides within the KDELLF- The antibodies were tested by Western blotting to deter- PSWEALFSGSEGPLKPGARIFSFDGK sequence, which in- mine if they reacted with denatured materials. Bacterial col- cluded the antigen sequence of P6 (SWEALFSGSEGPLKPGAR, lagenase digests of the BM fractions from human placenta underlined), were synthesized and tested for epitope activity. For and kidney were applied to SDS-PAGE, blotted onto mem- CH18C and CH18D antibodies, 20 consecutive 10 amino acid-res- branes, and analyzed with the antibodies. As shown in Fig. idue-long sequences were analyzed for the GPPGPVVYVSE- 4, both antibodies CH15A and CH15B recognized 3 major QDGSVLSVPGPEGRPGFA sequence (for P2 antigen of YVSE- bands of 37, 27 and 22 kDa, which could be related to the QDGSVLSVPGP). The same method was used for epitope NC1 domain of type XV. Type XVIII antibody, CH18B, re-

12 Monoclonal Antibodies for Multiplexin

Fig. 2. Epitope analysis of the mAbs by use of the multipin assay. For epitope analysis of the CH15A (a) raised from the PNPISSANYEKPA sequence as antigen, the 19 consecutive overlapping peptides of 9 amino acid residues covering PHQLLPPPNPISSANYEKPALHLAALN sequence were assayed if they react with CH15A. As indicated by underlines of individual amino acid sequences, the overlap area, NPI, was judged as the epitope. The results of the same analyses for CH15B, CH18B, and CH18D are shown in b, c, and d, and the epitopes were defined as NPI, ALFS, and VLSV, respectively. acted with bands with 28 kDa or less, which were probably characterized. These collagens were found along the BMs, derived from the NC1 domain of type XVIII that contains and their distribution is described separately below as being endostatin. The other antibody, CH18D, did not recognize in (1) subepithelial BM, (2) BMs surrounding muscular any bands of 28 kDa or less, indicating that there existed no tissues, and (3) subendothelial and vascular BMs. obvious metabolic fragments that contained NC7. Thus, these antibodies were considered useful for Western blot Subepithelial BM analysis. Kidney: As shown in Fig. 5a–5c, type XV was limited to Immunohistochemical distribution of type XV and the interstitial area of the kidney sections, whereas type XVIII XVIII was distributed widely in the BM zone. Bowman’s capsular BM was positive for both collagens; however, type The distribution of type XV and XVIII collagens in human XVIII was found in the entire layer of the membrane, but kidney, placenta, skin, and other tissues was immuno- type XV was only in the outer half (Fig. 6g–6i). In the inter- histochemically examined by using the specific mAbs thus tubular capillary, BMs of both collagens were present.

13 Y. Tom ono et al.

Fig. 3. Specificity of mAbs raised against α1 (XV) chains in placenta (A) and against α1 (XVIII) chain in kidney (B). A: Two mAbs, CH15A and CH15B, stain intensely BMs of blood vessels, and moderately BM areas delineating cytotrophoblasts. Mild staining is also recognized in mesenchyme of the chorion. a, CH15A; b, CH15B; c, negative control using culture medium containing 10% of fetal bovine serum; d, Azan staining. B: Four different mAbs were applied to serial sections of the kidney. CH18A (a), CH18B (b), CH18C (c), and CH18D (d) reveal a similar staining pattern in the kidney; i.e., strongly positive BMs in Bowman’s capsule and tubules, and considerably positive glomerular BMs and mesangial matrix. In addition, BMs surrounding smooth muscle cells and capillaries of blood vessels are also positive with all 4 different mAbs.

Placenta: Type XV was widely distributed along the BM 5d–5f). Pili muscles in the dermis contained both type zone as well as in the mesenchymal area of the chorion (Fig. XVIII and type XV (data not shown). Interestingly, BMs in 3A). BM beneath cytotrophoblasts was reactive with mAbs neurons (Fig. 6a–6c) and surrounding fat cells (data not against type XV. A strongly positive reaction with CH15A shown) were positive with α1 (XV) antibody but negative was found in the BM of capillaries and in areas enclosing with α 1(XVIII) antibody. arteries, whereas a weak reaction was seen in areas sur- Small intestine: Type XVIII was stained moderately rounding veins. On the contrary, type XVIII was present along the BM delineating mucous membranes of villi and along the capillaries and both arteries and veins (data not crypts, whereas type XV was stained moderately in the BM shown). of the crypt but weakly in that of the villi (data not shown). Lung: H22, the mAb for α2(IV), distinguished 2 different Both collagens were also found in capillary BM and smooth types of BMs in the alveolar walls (Fig. 5l); i.e., the sub- muscle cells in the lamina propria and muscularis mucosae epithelial BM and capillary subendothelial BM. Type XV (data not shown). was absent in both BMs, whereas type XVIII was detected Liver: BMs in the perisinusoidal space, which were visu- mildly on the subendothelial BM and weakly on the sub- alized with H22 (Fig. 5i), were stained linearly by mAbs epithelial BM (Fig. 5j and 5k). Bronchiolar subepithelial against type XVIII collagen (Fig. 5h), but not by those BM contained type XVIII but not type XV. against type XV (Fig. 5g). A striking contrast of the locali- Skin: A striking contrast of the localization of the 2 col- zation of the two collagens was also found in the portal ar- lagen types was found in skin epidermal BM; i.e., type eas: BMs surrounding bile duct and arterial smooth muscle XVIII was strongly expressed in the epidermal BM (Fig. cells contained type XVIII, whereas small vessels in the 5e), whereas type XV was weakly present, if at all (Fig. 5d). area contained type XV (data not shown). Arterial smooth muscle in the dermis also contained type XVIII but very little of type XV (Fig. 6a–6c). However, both types were found in BMs of capillaries and veins (Fig.

14 Monoclonal Antibodies for Multiplexin

sels, too, where double staining of both collagens revealed the differential localization of the two collagens. The smooth muscle cells in veins and venules expressed both collagens, whereas those in arteries and arterioles produced mainly type XVIII (Fig. 6a–6c, in skin; Fig. 6d–6f, in small intestine).

Subendothelial BMs in capillaries Distribution of type XVIII and XV collagens in capillary BMs was investigated in a wide range of tissues, and the findings are summarized in Table I. In general, most of the capillary BMs commonly contained both type XVIII and XV collagens, but there was some heterogeneity in their distribution. Capillary BMs could be classified into three groups in terms of the distribution of the two collagen types: 1) Both types were observed in many tissues such as kidney except for glomeruli, skin, and small intestine. 2) Capillary BMs with type XVIII predominant were in lung alveolus, liver sinusoid, splenic sinus and kidney glomerulus, which are composed of capillaries where endothelial cells are well differentiated in each organ. 3) Capillary BMs with predominantly type XV were found in heart, skeletal muscle, mus- Fig. 4. Western blot analysis of extracts from human placenta and cle layer of small intestine, and placenta. But we did not kidney. Human placenta (lane 1, 2) and kidney (lane 3, 4, 5) BM fraction find any capillary BMs where both type XVIII and type XV after collagenase digestion (see Materials and Methods). Samples extracted from 0.66 mg of dry weight of BM fraction were loaded onto each lane were absent. and detected by specific mAbs (lane 1: CH15A, lane 2: CH15B, lane 3: CH18B, lane 4: CH18D, and lane 5: H22). The α2(IV)-specific antibody Discussion H22 is against peptide and recognizes the α2 (IV) NC1 domain (Sado et al., 1995). Positions of molecular weight markers are shown by arrows at the right side of the panel. Specificity and reliability of the peptide antibodies Type XV and type XVIII collagens consist of homologous polypeptides, hence their expression and function can, and Muscular tissues will overlap. Thus, it is essential to have antibodies specific for the individual polypeptides for the investigation of their Muscular tissues are divided into 3 different types of mus- distribution. Since we earlier developed a rat lymph node cular cells: skeletal muscle, cardiac muscle, and smooth method to raise mAbs that was ten times more efficient than muscle cells. conventional methodsዄKishiro et al., 1995; Sakamoto et al., Skeletal muscle: Skeletal muscle obtained from the ab- 1999), we applied it to obtain mAbs specific for the 2 dominal wall contained a moderate amount of type XV, but polypeptides, using synthetic peptides against specific very little of type XVIII in the surrounding BM (Fig. 5m regions of the individual polypeptides. We confirmed the and 5n). Capillaries among the muscular cells contained a reliability of the antibodies by producing two clones for one moderate amount of type XV and a small amount of type peptide that demonstrated the same staining patterns. The XVIII. peptide antibodies showed different staining patterns from Cardiac muscle: Cardiac muscle revealed a collagen those of type XVIII despite the homology in their primary composition similar to that of the skeletal muscles, i.e., a structure, indicating their specificities of the two poly- moderate amount of type XV but very little of type XVIII in peptides. Further, all the clones displayed reaction with dis- the surrounding BM (data not shown). Capillaries among tinct epitopes within the corresponding primary sequences the cardiac muscle cells also displayed a similar pattern as of the synthetic peptides, demonstrating the reliability of the those in the skeletal muscle tissue. antibodies. Smooth muscle: Smooth muscle cells found in the muscular layer of the small intestine also revealed the same pattern Distribution of the type XV and type XVIII collagens on of distribution as those of skeletal and cardiac muscles; i.e., subepithelial BM type XV predominated (Figures 5p and 5q). On the contrary, skin pili muscle cells produced both type XV and XVIII. In previous reports, type XVIII collagen was shown to be Smooth muscle cells were distributed widely in blood ves- distributed in sinusoidal areas in liver, BMs in Bowman’s

15 Y. Tom ono et al.

Table I. DISTRIBUTION OF TYPE XV AND TYPE XVIII COLLAGENS IN CAPILLARY BMs FROM VARIOUS TISSUES

Tissues α1(XV) α1(XVIII) α2(IV) Continuous capillaries Kidney Intertubular capillary ++ ++ +++ Skin Capillary in papillary layer ++ ++ +++ Small intestine Capillary in mucosa ++ ++ +++ Placenta Capillary ++ + +++ Muscular tissues Capillary in cardiac muscle ++ + +++ Capillary in skeletal muscle ++ + +++ Capillary in smooth muscle ++ + +++

Specialized capillaries Kidney Glomerular capillary – ++ + Liver Sinusoid – ++ +++ Lung Capillary in alveolar wall – + +++ Spleen Sinus – ++ +++ *Symbols are: +++, strongly positive; ++, moderately positive; +, weakly positive; –, negative. capsules, dermal BMs, and arterial walls and capillary BMs contain primarily type XV. This situation implies that the in human (Saarela et al., 1998a, 1998b) and mouse (Sasaki presence or type XVIII is associated with cellular polarity. et al., 1998, Muragaki et al., 1995) tissues. Another member of the multiplexin family, type XV collagen, was found in Type XV and type XVIII collagens in muscular BMs almost the same areas as type XVIII; i.e., in BMs surround- Our findings indicated that the BMs of cardiac and skeletal ing skeletal muscles, cardiac muscles, and vascular systems muscles contained primarily type XV. Those of smooth both in human (Kivirikko et al., 1995; Myers et al., 1996; muscle cells demonstrated, interestingly, three different Hägg et al., 1997a, 1997b) and mouse (Sasaki et al., 2000) staining patterns; smooth muscle cells in the muscular layer organs. We compared the distribution of the two homolo- of the small intestine expressed predominantly type XV col- gous collagen types belonging to the same subfamily and, in lagen, whereas those in small arteries expressed type XVIII the present report, clarified which BMs contained only one collagen and those in pili muscles and veins contained both of them or both. We concluded that nonvascular BMs collagens. This heterogeneity in distribution of the two contained predominantly one of the two types; in general, types in BMs in various smooth muscle cells could be relat- muscles predominantly contained type XV, whereas sub- ed to the preference for sympathetic or parasympathetic epithelial BMs harbored primarily type XVIII. neurons that control smooth muscle contraction. Despite Epithelial cells cover the inner and outer surfaces of the their complete lack of type XV collagen, mutant mice de- body, or form tubular organs, and have functions of excre- veloped and reproduced normally; however, they showed tion, filtration, permeation, mechanical protection, and so progressive histological changes after 3 months of age and on. One of the characteristic features of epithelia is that they were more vulnerable than controls to exercise-induced always have cellular polarity. BMs attach to epithelia only damage (Eklund et al., 2001). These vulnerable changes at the basal side. In other words, epithelial cells “sit” on could be due to the lack of type XV, preventing a linkage BMs, which contain primarily type XVIII. Vascular epi- between skeletal muscle cell BMs and surrounding fibrillar thelial cells do not have such polarity, but they always sit on matrix. BMs at one side. BMs underneath endothelia always contain type XVIII, whereas muscle cells and adipocytes do not have polarity at all and are encapsulated by BMs that

16 Monoclonal Antibodies for Multiplexin

Fig. 5. Immunohistochemical detection of α1 (XV) and α1 (XVIII) chains in human tissues. Kidney (a–c), skin (d–f), liver (g–i), lung (j–l), skeletal muscle (m–o), and smooth muscle of the intestine (p–r) were stained with CH15A (mAb specific for α1 (XV); a, d, g, j, m, and p), with CH18B (mAb for α1 (XVIII); b, e, h, k, n, and q), and with H22 (mAb for α2 (IV); c, f, i, l, o, and r). Localization of both α1 (XV) and α1 (XVIII) chains are rather limited to BMs among the various tissues. H22 was used as a control since it stained all BMs. Note that the subepithelial BM in the kidney, the epidermal BMs of the skin, the subendothelial BM in the liver, and the lung alveolar BM contain predominantly α1 (XVIII) chain but not α1 (XV) chain. On the contrary, those surrounding the skeletal muscles and the smooth muscles are strongly stained only with CH15A. Brown staining found in the lung (j, k, l) is non-specific fluorescence of elastic fibers in the tissue. Scales: 100 µm.

17 Y. Tom ono et al.

Fig. 6. Double immunohistochemical staining of α1 (XV) and α1 (XVIII) chains in human blood vessels. Localization of α1 (XV) and α1 (XVIII) chains in various blood vessels is demonstrated using CH15A indirectly labeled with Texas Red (a, d, g) and CH18D labeled with FITC (c, f, i), respectively. Merged figures are shown in b, e, and h. Both α1 (XV) and α1 (XVIII) chains were defined as various BM components of blood vessels, but the double staining patterns demonstrate the characteristic differences in their local distribution. Venous (V) smooth muscle cells contain both chains, whereas arterial (A) ones contain only α1 (XVIII) chain (a, b, and c). Capillary BMs of inter-tubular areas (arrows in h) of the kidney possess both, whereas glomerular BMs (G) are positive for only α1 (XVIII) chain. BM in peripheral neuron (N) contains type XV but not type XVIII. a, b, and c, skin epidermis; d, e, and f, muscular layer of small intestine; g, h, and i, renal cortex. A, small artery; Al, arteriolus; V, small vein; Vl, venule; SMC, smooth muscle cells; G, glomerular BM; T, tubular BM. Scale bars indicate 50 µm.

The splenic sinusoid permits red blood cells to pass through Subendothelial basement membranes containing type the gap between the rod-shaped endothelial cells, which XVIII, type XV, or both have a orientation parallel and lack typical intercellular Blood fluid containing nutrients and oxygen is secreted junctions. Thus, they are distinguished from the other capi- through capillary walls to the surrounding tissues, and that llary walls by having windows in their walls or gaps between containing waste materials and carbon dioxide in turn are the cells. One possible consequence of this compositional absorbed into capillaries via the capillary wall. The capi- difference in the collagens is that type XV and type XVIII llary is composed of a few endothelial cells, subendothelial may provide a differential scaffold on which endothelial BMs and surrounding tissues. Some capillaries are devel- cells are able to behave in a local situation in each tissue. oped into specialized forms according to their functions in BMs are thought to function not only as a scaffold for the different tissues. The glomerular endothelial cells having a epithelial cells but also as a reservoir of the regulators of number of small pores with no diaphragm allows excretion proliferation and signaling of such cells. In the present study of a great amount of the blood fluid as urine. The alveolar we observed tissue localization of the two collagens type capillary composes the alveolar wall, which is well known XV and type XVIII in the subendothelial BMs. The contin- to be a blood-gas barrier providing for the intake of oxygen uous capillaries, found in all kinds of muscular tissues and and the elimination of carbon dioxide. The liver sinusoid is connective tissues, had BMs containing both type XV and covered with highly attenuated endothelial cells, through type XVIII. However, the fenestrated capillaries including which the nutrients and other components are transported. the liver sinusoid, the splenic sinus, and the glomerular and

18 Monoclonal Antibodies for Multiplexin alveolar capillaries rested on BMs containing type XVIII Peltoketo, H., Pihlajaniemi, T., and Vainio, S. 2001. Induced repattern- but lacking type XV. Thus, the differential distribution of ing of type XVIII collagen expression in ureter bud from kidney to lung the two collagens could be related to the function of such type: association with sonic hedgehog and ectopic surfactant protein C. Development, 128: 1573–1585. specialized capillaries in various organs. In other words, Muragaki, Y., Abe, N., Ninomiya, Y., Olsen, B.R., and Ooshima, A. 1994. functional differences between the capillaries in various or- The human α1 (XV) collagen chain contains a large amino-terminal gans could be attributed to the underlying BMs which are non-triple helical domain with a tandem repeat structure and homology structured from different molecular compositions. to α1 (XVIII) collagen. J. Biol. Chem., 269: 4042–4046. The endothelial cells maintain their stability by attaching Muragaki, Y., Timmons, S., Griffith, C.M., Oh, S.P., Fadel, B., themselves to the BMs underneath. Obvious examples for Quertermous, T., and Olsen, B.R. 1995. Mouse Col18a1 is expressed in such an idea are keratinocytes and spermatogonia that main- a tissue-specific manner as three alternative variants and is localized in basement membrane zones. Proc. Natl. Acad. Sci. USA, 92: 8763–8767. tain their adherence to neighboring BMs to prevent their Musso, O., Rein, M., Saarela, J., Thereto, N., Liétard, J., Hintikka, E., differentiation; i.e., as long as the keratinocytes or sper- Lotrian, D., Campion, J.-P., Pihlajaniemi, T., and Clement, B. 1998. matogonia adhere to nearby BMs, they do not differentiate Collagen XVIII is localized in sinusoids and BM zones and expressed or divide (Junqueira et al., 1992). The endothelial cells by hepatocytes and activated stellate cells in fibrotic human liver. might be controlled with various BM components and espe- Hepatology, 8: 98–107. cially with these two collagens, since endostatin and restin, Myers, J.C., Kivirikko, S., Gordon, M.K., Dion, A.S., and Pihlajaniemi, T. which are derived from type XVIII and type XV, respective- 1992. Identification of a previously unknown human collagen chain, α1(XV), characterized by extensive interruptions in the triple-helical ly, are known to have the potential of inhibiting the prolifer- region. Proc. Natl. Acad. Sci. USA, 89: 10144–10148. ation and/or migration of the endothelial cells. Myers, J.C., Dion, A.S., Abraham, V., and Amenta, P.S. 1996. Type XV collagen exhibits a widespread distribution in human tissues but a dis- Acknowledgments. The authors would like to thank S. Inoue and C. Taka- tinct localization in basement membrane zone. Cell Tissue Res., 286: hashi for their technical assistance, and Dr. F. Shigei for continuous support 496–505. of the work. This work was supported in part by a Grant-in-Aid from the Myllyharju, J. and Kivirikko, K.I. 2001. Collagens and collagen-related Ministry of Education, Science, Sports, and Culture of Japan. diseases. Ann. Med., 33: 7–21. Ninomiya, Y., Kagawa, M., Iyama, K., Naito, I., Kishiro, Y., Seyer, J. M., References Sugimoto, M., Oohashi, T., and Sado, Y. 1995. Differential expression of two basement membrane collagen genes, COL4A6 and COL4A5, Eklund, L., Piuhola, J., Komulainen, J., Sormunen, R., Ongvarrasopone, demonstrated by immunofluorescence staining using peptide-specific C., Fåssler, R., Muona, A., Ilvas, M., Ruskoaho, H., Takala, T.E.S., and monoclonal antibodies. J. Cell Biol., 130: 1219–1229. Pihlajaniemi, T. 2001. Lack of type XV collagen causes a skeletal Olsen, B.R. and Ninomiya, Y. 1999. Collagens. In Guidebook to the myopathy and cardiovascular defects in mice. Proc. Natl. Acad. Sci. Extracellular Matrix, Anchor, and Adhesion proteins (Thomas Kreis and USA, 98: 1194–1199. Ron Vale, eds.). CRC Press, Boca Raton, Florida. pp.380–408. Hägg, P.M., Hägg, P.O., Peltonen, S., Auto-Harmainen, H., and Ramachandran, R., Dhanabal, M., Volk, R., Waterman, M.J.F., Segal, M., Pihlajaniemi, T. 1997a. Location of type XV collagen in human tissues Lu, H., Knebelmann, B., and Sukhatme, V.P. 1999. Antiandiogenic and its accumulation in the interstitial matrix of the fibrotic kidney. Am. activity of restin, NC10 domain of human collagen XV: comparison to J. Pathol., 150: 2075–2086. endostatin. Biochem. Biophys. Res. Commum., 255: 735–739. Hägg, P.M., Horelli-Kuitunen, N., Eklund, L., Palotie, A., and Saarela, J., Ylikäppä, R., Rehn, M., Purmonen, S., and Pihlajaniemi, T. Pihlajaniemi, T. 1997b. Cloning of the mouse type XV collagen se- 1998a. Complete primary structure of two variant forms of human type quences and mapping of the corresponding gene to 4B1-3. Genomics, XVIII collagen and tissue-specific differences in the expression of the 45: 31–41. corresponding transcripts. Matrix Biology, 16: 319–328. Halfter, W., Dong, S., Schurer, B., and Cole, G.J. 1998. Collagen XVIII is Saarela, J., Rehn, M., Oikarinen, A., Autio-Harmainen, H., and a basement membrane heparan sulfate proteoglycan. J. Biol. Chem., 273: Pihlajaniemi T. 1998b. The short and long forms of type XVIII collagen 25404–25412. show clear tissue specificities in their expression and location in base- Hudson, B.G., Reeders, S.T., and Tryggvason, K. 1993. Type IV collagen: ment membrane zones in humans. Am. J. Pathol., 153: 611–626. structure, gene organization, and role in human diseases. Molecular Sado, Y., Kagawa, M., Kishiro, Y., Sugihara, K., Naito, I., Seyer, J.M., basis of Goodpasture and Alport syndromes and diffuse leiomyomatosis. Sugimoto, M., Oohashi, T., and Ninomiya, Y. 1995. Establishment by J. Biol. Chem., 268: 26033–26036. the rat lymph node method of epitope-defined monoclonal antibodies α Junqueira, L.C., Carneiro, J., Long, J.A. 1992. Basic Histology, Lange recognizing the six different chains of human type IV collagen. Medical Publications, Los Altos, California. Histochem. Cell Biol., 104: 267–275. Kishiro, Y., Kagawa, M., Naito, I., and Sado, Y. 1995. A novel method of Sado, Y., Kagawa, M., Kishiro, Y, Naito, I., Joh, K., and Ninomiya, Y. preparing rat-monoclonal antibody-producing hybridomas by using rat 1997. Purification and characterization of human nephritogenic antigen medial iliac lymph node cells. Cell Struct. Funct., 20: 151–156. that induces anti-GBM nephritis in rats. J. Pathol., 182: 225–232. Kivirikko, S., Saarela, J., Myers, J.C., Autio-Harmainen, H., and Sado, Y., Kagawa, M., Naito, I., Ueki, Y., Seki, T., Momota, R., Oohashi, Pihlajaniemi, T. 1995. Distribution of type XV collagen transcripts in T., Ninomiya, Y. 1998. Organization and expression of basement mem- human tissue and their production by muscle cells and fibroblasts. Am. J. brane collagen IV genes and their roles in human disorders. J. Biochem., Pathol., 147: 1500–1509. 123: 767–776. Li, D., Clark, C.C., and Myers, J.C. 2000. Basement membrane zone type Saito, K., Naito, I., Seki, T., Oohashi, T., Kimura, E., Momota, R., Kishiro, XV collagen is a disulfide-bonded chondroitin sulfate proteoglycan in Y,, Sado, Y., Yoshioka, H., and Ninomiya, Y. 2000. Differential expres- human tissues and cultured cells. J. Biol. Chem., 275: 22339ᯝ22347. sion of mouse alpha 5 (IV) and alpha 6 (IV) collagen genes in epithelial Lin, Y., Zhang, S., Rehn, M., Itaranta, P., Tuukkanen, J., Heljasvaara, R., basement membranes. J. Biochem., 128: 427–434.

19 Y. Tom ono et al.

Sakamoto, H., Sado, Y., Naito, I., Kwon, T.H., Inoue, S., Endo, K., angiogenic activity. J. Mol. Biol., 301: 1179–1190. Kawasaki, M., Uchida, S., Nielsen, S., Sasaki, S., and Marumo, F. 1999. Seki, T., Naito, I., Oohashi, T., Sado, Y., and Ninomiya, Y. 1998. Differ- Cellular and subcellular immunolocalization of ClC-5 channel in mouse ential expression of type IV collagen isoforms, alpha 5 (IV) and alpha 6 kidney: colocalization with H (+)-ATPase. Am. J. Physiol., 277: F957– (IV) chains, in basement membranes surrounding smooth muscle cells. F965. Histochem. Cell Biol., 110: 359–366. Sasaki, T., Fukai, N., Mann, K., Göhring, W., Olsen, B. R., and Timpl, R. Tomono, Y., Sado, Y., Naito, I., Taki, M., Ninomiya, Y., and Okigaki, T. 1998. Structure, function and tissue forms of the C-terminal globular 1998. Localization of type XVIII collagen in mouse and human kidney domain of collagen XVIII containing the angiogenesis inhibitor detected by monoclonal antibody. Cell Struct. Funct., 23: 564 (a). endostatin. EMBO J., 17: 4249–4256. Sasaki, T., Larsson, H., Tisi, D., Claesson-Welsh, L., Hohenester, E., and (Received for publication, September 27, 2001 Timpl, R. 2000. Endostatins derived from collagens XV and XVIII dif- and in revised form, November 27, 2001) fer in structural and binding properties, tissue distribution and anti-