0022-1554/92/$3.30

The Journal of Histochemisuy and CytochemiSuy Copyright 0 1992 by The Histochemical Society, Inc.

Vol. 40, No. 11, pp. 1779-1788, 1992 Printed in U5.A.

Original Article

I Nature and Distribution of Mineral-binding, Keratan Sulfate-containing Glycoconjugates in Rat and Rabbit Bone MASAO MAENO, MINORU TAGUCHI, KAZUHIRO KOSUGE, KICH BEE OTSUKA, and MINORU TAKAGI' Departments of Biochemistrj (MM,MT,KO), Periodontology (KK), and Anatomy (MT), Nihon University School of Dentistry, Ekyo 101, Japan. Received for publication April 9, 1992 and in revised form June 22, 1992; accepted June 27, 1992 (2A2650).

The presence of keratan sulfate (KS) and KS proteoglycans in bone has been demonstrated in birds and rabbits but comparison with other animal species has not been investigated. The nature and distribution of mineral-binding, KS-containing glycoconjugates in rat and rabbit bone were investigated with a monoclonal antibody (MAb 5D4) specificfor KS.Mineral-bindingproteins were extracted from the mineralized bone with 0.4 M EDTA without guanidine-HCl (Eextract). On Western blot analysis of SDS-polyacrylamide gel electrophoresis, rat E-extract gave a weak 5D4-reactive band, Mr 66,OOO-68,000, whereas rabbit E-extract produced two major reactive populations of small and large molecular size; one population consisted of two closely spaced bands at Mr 61,000-63,OOO and 66,000-68,000, and the other population consisted of one band at approximately Mr 200,000. The identity of KS chains was further established by the sensitivity of these bands to keratanase 11 (Bacillus

Introduction The predominant proteoglycans (PGs) isolated from bone of several animal species had two different core proteins with a molecular weight (Mr) of 45,000, and one or two attached chondroitin sulfate(s) (average Mr 40,000 each) and 0-linked and N-linked oligosaccharides. These two species, designated PG I (biglycan) and PG I1 (decorin), were different gene products (14,15$3,19). The predominant glycosaminoglycan (GAG) associated with these PGs appeared to be chondroitin-4-sulfate (C4-S) (36). The other GAG was idenin bone, keratan sulfate (KS) and material resembling U, tified in a few species such as Japanese quails (16,17,28), chickens (4,44), and rabbits (2,10,33,51). Kinne and Fisher (30) isolated and partially characterized a KSPG with a Mrof 200,000 from rabbit cortical bone that was rich in sialic acid. In other species, Prince

Correspondence to: Minoru 'I!akagi, DDS, PhD, Dept. of Anatomy, Nihon Univ. School of Dentistry, Kanda-Surugadai,Chiyoda-ku, Tokyo 101, Japan.

sp. Ks 36) and endo-p-galactosidase. Immunocytochemistry with MAb 5D4 showed that, in rat bone, staining associated with the mineral phase was limited to the walls of osteocyticlacunae and bone canaliculi, whereas the remainder of the mineralizedma& lacked staining. In contrast, in rabbit bone the staining was distributed over the enure portion of the mineralized matrix with focal accumulation of staining in the wall of the lacunocanalicular system. These results indicate that rat bone contains a mineral-binding, KScontaining glycoconjugate with preferential localization in the wall of the lacunocanalicular system, whereas rabbit bone containsat least two or possibly three types of KS-containing glycoconjugates distributed over the entire portion of the mineralized matrix. (JHistochemCytochem40:1779-1788, 1992) KEY WORDS: Bone; Keratan sulfate; Proteoglycan; Glycoconjugate.

et al. (38) reported the presence of [ 35S]-sulfate-labeledmaterial sensitive to keratanase in rat bone. However, its amount was so small that it is difficult to judge whether or not KS is a component of rat bone GAG. The unique chain geometry of KS is composed of a number of the repeating disaccharide (N-acetyllactosamine)units, in which sulfate groups are present in the C-6 position of the glucosamine unit and also in some of the galactose residues at the C-6 position (39). KS was found 0-linked to serine and threonine in aggregating cartilage PG (KS II) and N-linked to asparagine in comeal KSPG (KS I) (39). The monoclonal antibody (MAb) 5D4 (5,6,34), which recognizes a minimum of six sugar units containing disulfated N-acetyllactosamineresidues of KS, has been used in a wide range of biochemicalstudies examining KSPG biosynthesis and metabolism in cartilage and cornea (5). Using MAb 5D4, Bartold (1)immunohistochemically localized KS in the immediate periphery of bone lacunae and the mineralized matrix in rabbit alveolar bone; however, this study did not examine the ultrastructural distribution of KS. 1779

Downloaded from jhc.sagepub.com at UCSF LIBRARY & CKM on March 25, 2015

MAENO, TAGUCHI, KOSUGE, OTSUKA, TAKAGI

1780

The present biochemical and immunocytochemical study was undertaken to identlfy the nature and distribution of KS-containing glycoconjugates in rat and rabbit bone, using MAb 5D4 specific for KS (5,6,34).

Materials and Methods Ten Wistar rats weighing approximately 85 g each and seven young rabbits weighing approximately 2.5 kg each were used in this study.

Biochemistry Non-collagenousProtein Extraction. The method for the extraction of non-collagenous proteins was performed with a modified method (12.24) originally described by Termine et al. (47). The fresh midshaft subperiosteal bone of the femora, tibiae, and humeri was obtained from the rats (n = 7). The alveolar bone of the rabbits (n = 5) was collected from the immediate area around the sockets of excised incisors and molar teeth of the mandibular bone. These bone samples were carefully dissected free and cleared of adherent soft connective tissue and reduced in size with a bonecutting forceps. The bone fragments (rat, approximately 1.8 g; rabbit, approximately 3.0 g) were rinsed in PBS containing protease inhibitors (PI) (0.1 M 6-aminohexanoic acid, 0.005 M benzamidine hydrochloride, and 0.001 M phenylmethylsulfonylfluoride), pH 7.3, at 4'C. All extraction and subsequent rinses were performed at 4°C. The ratio of the bone tissue to extractant volume was 1 g:200 ml for each extraction. Extraction to remove organic material bound to the non-mineralized phase was done with constant stirring with 4.0 M guanidine-HC1 (GdnCl), 50 mM %is-HCI containing PI, pH 7.4 (90 or 150 ml/day x 4 days). The supernatant was separated from the residue by centrifugation and recovered (Gl-extract). The residue was subsequently rinsed with PBS-PI (90 or 150 ml/day x 3 days) and then extracted (90 or 150 mllday x 4 days) with 0.4 M EDTA, 50 mM Tris-HU containingPI, pH 7.4. The supernatant, containingnon-collagenous proteins extracted from the mineral phase, was recovered after centrifugation (E-extract). The demineralized collagenous residue was rinsed with PBS-PI (90 or 150 mllday x 3 days) and then re-extractedwith 4.0 M GdnC1, 50 mM Pis-HC1 containing PI, pH 7.4 (90 or 150 ml/day x 4 days) as described above. The supernatant was recovered after centrifugation (G2extract). Each extract was concentrated 10- to 20-fold by ultrafiltration on a PM-IODiaflo membrane filter (Amicon; Danvers, MA), dialyzed exhaustively against 0.1 M ammonium bicarbonatecontaining 0.005% Brij (Sigma; St Louis, MO), and freeze-dried. Enzyme Digestion. Several portions (5 pg) of the lyophilized sample from each extract were digested with either 5 mU of keratanase 11 (from Bucdlus sp. KS 36) in 0.1 M sodium acetate buffer, pH 6.5, or 0.5 mU of endo-P-galactosidase(from Escherichiafreundiz] in 0.1 M sodium acetate buffer, pH 5.8, 2 5 mu of endo-P-N-acetylglucosaminidase (from Fhobucterium sp.) in 0.25 M K-phosphate buffer, pH 6.0, 10 mU of protease-free chondroitinase ABC in 0.1 M %is-HC1, pH 8.0, containing 0.03 M sodium acetate. In each case the appropriate buffer was added to a final volume of 20 pl, These enzymes were obtained from Seikagaku Kogyo (Tokyo,Japan). The enzyme digestion was performed at 37% for 60 min. Reactions were terminated by freezing at - 70'C and freeze-dried for gradient sodium dodccyl sulfate (SDS)-polyacrylamidegel electrophoresis (PAGE). SDS-PAGE. SDS-PAGEwas carried out in 5-20% gradient cross-linked polyacrylamide gel, using the discontinuous %-glycine buffer system as previously described by Laemmli (31). Samples(5 pg each) with or without enzyme digestion were dissolved in 10 pl of sample buffer containing 1% SDS, 2.0 M urea, and bromophenol blue marker to which dithiothreitol 15 mglml (wlv) was added. Subsequently, they were heated at 95°C for 5 min. Six-cm Minislabs (Mighty Small 11; Hoefer Scientific Instruments,

San Francisco, CA) polyacrylamide gels with a thickness of 0.75 mm were used for the separation of proteins. The minislab gradient gels were electrophoresed for 60 min at 150V. After electrophoresisthe gels were washed three times with 25% (vlv) isopropyl alcohol for a total of 60 min. They were stained with 0.25% (wlv) Coomassie Brilliant Blue R-250 (13),0.025% (wlv) silver nitrate (35). or 0.025% (wlv) Stains-all(7). The Stains-all solution was made up fresh before use as follows: 30 mM Eis, pH 8.8, containing 7.5% (vlv) formamide, 25% (v/v) isopropyl alcohol, and volume made to 100 ml, to which 2.5 mg of Stains-allwas added in the dark, with stirring. The Stains-allsolution was then added to the polyacrylamidegel and shaken gently overnight at 22'C. Western Blotting. Full details of the characteristics and specificity of MAb 5D4 have been given elsewhere (5.6.34). MAb 5D4 has been raised against PG core protein isolated after chondroitinaseABC digestion of human articular cartilage PG monomer and specificallyrecognizes disulfated poly-(N-acetyllactosamine)sequences on corneal and skeletal KS oligosaccharides attached to core proteins (5,6). Immunotransfer analysis was performed on a Multiphor I1 Nova Blot system (LKB; Uppsala, Sweden) with 39 mM glycine, 48 mM Tris, 20% (v/v) methanol (continuous buffer system), at constant amperage of 0.8 mA/cm2of the gel, for 60-90 min. At the completion of the transfer, the excessprotein binding sites on transfer membranes (ImmobilonPVDF h s fer Membrane; Millipore, Bedford, MA) were blocked with 10% skim milk (Difco Laboratories; Detroit, MI) in 10 mM %is-buffered saline, 145 mM NaCl, pH 7.4 (TBS), for 18 hr at 4'C. The sheets were then washed three times in TBS containing 0.05% Tween 20 (TBS-Tween). The sheets were incubated for 90 min at 22 "Cwith mouse anti-KS (5D4) (Selkagaku Kogyo) diluted 1:50in distilled water containing 10% (vlv) bloddng reagents (Block Ace; Snow Brand Milk Products, Tokyo, Japan). The sheets were washed three times in TBS-Tween and incubated for 40 min at 22°C with horseradish peroxidase (HRP)-conjugated second antibody (F(ab)2 goat antimouse IgG (H+L) (Zymed Laboratories; San Francisco, CA) diluted 1:400 in distilled water containing 10% blocking reagents. The sheets were washed three times in TBS-Tween and once in TBS, and then incubated in either 0.06% (methanol-dissolved)4-chloro-l-naphthol,0.04% Hz02 in TBS. pH 7.5 (27) or diaminobenzidine (DAB) medium of 50 ml PBS, pH 7.2, containing 25 mg DAB-tetrahydrochloride(Sigma) according to Burns et al. (3) for 5-30 min at room temperature. As controls, the membranes were treated as follows: (a) omission of treatment with MAb 5D4 from the sequence and (b) exposure of the membrane to a solution containing normal mouse serum (diluted in 1:50 in distilled water containing 10% (dv) blocking reagents) instead of MAb 5D4.

Aldehyde Fiiiztion and Demineralization of Specimens The midshaft subperiosteal bone of the tibiae was removed from the rats (n = 3) under anesthesia after perfusion-fixationwith modified Kammky's fmuve (29)containing4% paraformaldehyde-0.5% glutaraldehyde in 0.1 M cacodylatebuffer, pH 7.3, for 10 min. The tissues were reduced to approximately 0.3 mm x 2.0 mm x 8.0 mm in size with a razor blade. These specimens were further immersed in the same fixative for 50 min at 4'C. The mandibular bone was obtained from freshly slaughtered rabbits (n = 2). The alveolar bone was collected from the mandibular bone as described above and was reduced to approximately 0.3 mm x 2.0 mm x 8.0 mm in size with a bone-cutting forceps and immersed in the same fixative for 60 min at 4'C. After fixation all specimenswere thoroughly rinsed in 0.1 M cacodylate buffer (pH 7.3) at 4°C. Specimens were dehydrated through a series of graded ethanol (30, 50. 70, and 80%) (11) and then treated for 4-10 weeks at 4'C with 0.2 M triethylammonium EDTA solution in absolute ethanol diluted 4:l in distilled water (ethanolic alkylammoniumEDTA) according to Scott and Kyffin (41). The use of ethanolic alkylammonium

Downloaded from jhc.sagepub.com at UCSF LIBRARY & CKM on March 25, 2015

KERATAN SULFATE IN RAT AND RABBIT BONE

1781

EDTA for demineralization of aldehyde-fixed bone specimens is extremely favorable for obtaining good retention of PGs and glycoproteins in bone tissues without reducing antigenicity (46). After demineralization they were then thoroughly rinsed with 80% ethanol at 4'C.

Light Microscopy Specimens were dehydrated and embedded in paraffin. Sections 6 pm thick were dewaxed in xylene and rehydrated through descending concentrations of ethanol. Some sections were stained with hematoxylin and eosin (HE) for light microscopic examination, whereas others were immunostained as described below. When 5D4 is used prior digestion with the specific glycosidases is nor required; however, in the present study some specimens were evaluated after treatment with chondroitinase ABC.

Immunostaining of KS-containing Glycomnjugates. Tissue sections were thoroughly rinsed in PBS, pH 7.3, at 4°C before and after treatment of these sections in 0.1% glycine in PBS for 30 min to quench free aldehyde groups (49). Some tissues sections were briefly washed in 0.1 M sodium acetate-0.1 M Tris-HC1 buffer (pH 7.3) and were exposed for 1 hr at 37'C to a protease-free chondroitinase ABC (Seikagaku Kogyo) solution [0.2 U/ml in 0.1 M Tris-acetate. pH 7.3, containing 1 mglml bovine serum albumin (BSA)] (9). The tissue sections, with or without chondroitinase digestion, were then permeabilized with 0.03% saponin in PBS (S-PBS)before and during incubation with the antibodies, as well as during the consecutive rinses, as recommended elsewhere (50). BSA was also added to the S-PBS solution at the concentration of 1% (S-PBS and BSA) when indicated. The tissue sections were treated with the S-PBS solution containing 0.3% H202 and 0.1% sodium azide for 30 min at 22°C to eliminate endogenous peroxidase activity (32). Subsequently, these sections were thoroughly rinsed in S-PBS and exposed for 30 min at 22'C to normal goat serum diluted 1:10 in S-PBS and BSA. After a brief rinse with S-PBS and BSA, the specimens were treated with mouse anti-KS (5D4) (Seikagaku Kogyo) diluted 1:50 in S-PBS and BSA for 90 min at 22'C. The specimens were then rinsed in S-PBS and BSA and exposed for 90 min at 22'C to HRP-conjugated second antibody (F(ab')z goat anti-mouse IgG (H+L); Zymed Laboratories) diluted 1:50 in S-PBS and BSA. After thorough rinsing with S-PBS and PBS. the tissue sections were incubated in 3.3'-diaminobenzidine (DAB) substrate solution without H202 for 20 min and the complete DAB substrate solution for 10 min at 22'C as described previously (25). The substrate medium consisted of 100 ml of 0.05 M Tris-HC1 buffer (pH 7.6) in which 50 mg of DAB-tetrahydrochloridewas dissolved and 0.1 ml of 5.2% H202 was added immediately before use. Tissue sections were rinsed with 0.05 M Tris-HCI buffer at 4'C, dehydrated with ethanol, cleared in xylene, and mounted for light microscopic examination. For control preparations, normal mouse serum was substituted for the MAb 5D4.

Electron Microscopy The aldehyde-fixed and demineralized specimens in 80% ethanol at 4'C were cut into 15-wm thick sections by a Microslicer (Dosaka EM; Kyoto. Japan). They were rehydrated through descending concentrations of ethanol (70. 60, 50.40. and 30%) and brought to PBS-PI or 0.1 M cacodylate buffer (pH 7.3) at 4°C.

Morphology. Tissue sections were thoroughly rinsed in 0.1 M cacodylate buffer (pH 7.3). They were then post-fixed in 1% os04 in 0.1 M cacodylate buffer (pH 7.3) for 60 min, routinely dehydrated, and embedded in Spurr's (43) low-viscosity resin. Thin sections were either left unstained or stained with uranyl acetate and Sato's (40) triple lead, and were examined with a Hitachi H5OO electron microscope at an accelerating voltage of 75 kV. Immunostaining. Tissue sections were similarly immunostained as de-

GI E

62

Gi E

G2

E

GI

62

200km 116km 93k' 67km 43km 30km 20km 14km

o a Gi E

200k116k93k67k43k30k20k14k-

G2

b GI E

o a

G2

Gi E

G2

+

+

It4

C

c

--t

--c

b

C

Figures 1 and 2. SDS-PAGEand Western blotting of non-collagenousproteins extracted from rat midshaft subperiosteal bone(Figure 1) or rabbit alveolar bone (Figure 2) with a three-step extraction procedure. Proteins (5 vg each) in G1-. E-, and GP-extracts from rat bone or rabbit bone were separated by SDS-PAGE. They werestained with Stains-all (a),or transferred electrophoreticallyto PVDF membranes using the Nova Blot system and analyzed by enzyme-linked immunostaining using MAb 5D4 (b). The bone proteins in each extract were digested with endo-l)-Kacetylglucosaminidase.electrophoresed as described above, and transferred onto the membranes,and the membranes were stained with MAb 5D4 (c). The lanes are as follows: G1. G1-extract; E, EDTA-extract; G2. G2-extract. Arrows indicate reactive bands.

scribed above. Subsequently they were osmicated. routinely dehydrated in ethanol, and embedded as described above. The thin sections were examined with the electron microscope without counterstaining.

Results Biochemistry Aliquots of GI-, E-, and G2-extracts from rat subperiosteal bone and rabbit alveolar bone electrophoresed on 5-20% gradient mini-

Downloaded from jhc.sagepub.com at UCSF LIBRARY & CKM on March 25, 2015

1782

MAENO, TAGUCHI, KOSUGE, OTSUKA. TAKAGI

Figure 3. Light micrograph of 6-pm section of rat tibial midshaft subperiosteal bone. lmmunoperoxidase staining with the MAb 5D4 to KS without chondroitinase ABC digestion. Very weak immunostaining is visible in the bone cell surface and the walls of bone lacunae as well as in vascular canals (VC). whereas no staining is clearly identifiable in other sites. Original magnification x 250. Bar = 10 pm. Figure 4. Light micrograph of 6-pm section of rat tibial midshaft subperiosteal bone. lmmunoperoxidase staining with the MAb 5D4 to KS after chondroitinase ABC digestion. Chondroitinase digestion significantly increases 5D4 immunoreactivity; staining is visible in the cell surfaces of osteocytes. lacuna walls, and bone canaliculi (BC), as well as in vascular canals (VC). However, the rest of the mineralized matrix lacked staining. Original magnification x 250. Bar = 10 pm. Figure 5. Light micrograph of 6-pm section of rabbit alveolar bone. lmmunoperoxidase staining with MAb 5D4 to KS without chondroitinase ABC digestion. In contrast to the rat bone, immunostaining can be seen in the entire portion of the mineralized bone matrix and in vascular canals (VC), and appears to be concentrated in presumed lacuna walls (LW), whereas clearly identifiable staining can not be observed in the osteocyte cell surface and the pericellular matrix. Original magnification x 250. Bar = 10 pm. Figure 6. Light micrograph of 6-vm section of rabbit alveolar bone. lmmunoperoxidase staining with MAb 5D4 to KS after chondroitinase ABC digestion. Prior chondroitinase digestion of tissue sections increases immunostaining in the mineralized bone matrix. Original magnification x 250. Bar = 10 pm.

slab gels and stained with Stains-all (Figures la and 2a) showed that E-extract produced a number of bands which stained blue, while proteins in the other two extracts stained pink with the exception ofafew blue-stained bands in G2-extracts (Figures la and 2a). Antibody Reaction to KS-containing Glycoconjugates. When

peroxidase-labeledmembranes were stained with the method using 4-chloro-l-naphthol(27),the reactivity of rat extracts was very weak on wet membranes and faded when the membrane was dried (not illustrated), whereas antigen in rabbit extracts could be clearly detected (Figure 2b). When rat extracts were stained with the DAB method ( 3 ) (Figure lb), it increased the sensitivity as previously

Figure 7. Pre-embedment immunoperoxidase staining of rat midshaft subperiosteal bone with 5D4 specific for KS after chondroitinase ABC digestion. MAb 5D4 stains weakly to moderately the surface of an osteocyte, the pericellular matrix, and the walls of bone canaliculi (BC; enlarged in lower right inset), and stains moderately to strongly the lacuna wall (LW); enlarged in upper left inset). In the pericellular matrix, staining (arrows) is localized between and at the periphery of collagen fibrils (CO). In the lacuna wall, the most intensely reactive site is a boundary between the pericellular matrix and the lacuna wall. The space between osteocytic processes and the lacuna wall does not show definitive staining. Original magnification x 17,500; Insets x 35,000. Bars = 0.5 pm. Figure 8. Pre-embedment immunoperoxidase staining of vascular canals (VC) in rat subperiosteal bone with 5D4 specific for KS after chondroitinase ABC digestion. MAb 5D4 stains not only the extracellular matrix surrounding vascular vessels (V) but also the unmineralized matrix (M). which often lines the wall of the vascular canal. Original magnification x 7500. Bar = 1.0 pm.

Downloaded from jhc.sagepub.com at UCSF LIBRARY & CKM on March 25, 2015

KERATAN SULFATE IN RAT AND RABBIT BONE

Downloaded from jhc.sagepub.com at UCSF LIBRARY & CKM on March 25, 2015

1783

MAENO, TAGUCHI, KOSUGE, OTSUKA, TAKAGI

1784

Y Y

1

P

Downloaded from jhc.sagepub.com at UCSF LIBRARY & CKM on March 25, 2015

1785

KERATAN SULFATE IN RAT AND RABBIT BONE

reported (8). Therefore, the latter method was used for staining rat extracts and the former method was used for staining rabbit extracts. Westem blot analyses of rat extracts revealed that MAb 5D4 did not produce any reactive bands in GI- and G2-extracts, but weakly stained only a single band in E-extract at Mr 66,000-68,000 (estimated from the gradient gel) (Figure Ib). Analysis of rabbit extracts showed that MAb 5D4 stained three bands in E-extract at Mr 61,000-63,000, Mr 66,000-68,000, and Mr approximately 200,000, the lower band being weakly reactive and the higher band strongly reactive, indicating that the larger was more predominant than the smaller. MAb 5D4 also weakly stained only a single band in G2-extract at Mr 66,000-68,000, whereas G1-extract failed to react with MAb 5D4 (Figure 2b). These reactive bands were further confirmed to be attributed to KS by subjecting them to a variety of well-defined enzymes and visualizing the results on Westem blotting. The epitope in all bands recognized by 5D4 was destroyed by keratanase I1 and endo-0galactosidase(data not shown). Prior digestion of the extract with endo-P-N-acetylglucosaminidase eliminated 5D4 reactivity of all bands except for a band at Mr approximately 200,000 (Figures IC and 2C). In addition, digestion of the extracts with chondroitinase ABC had no apparent effect on the electrophoretic mobility and Western blot, thus eliminating the possibility that the band arose from oligosaccharidestubs attached to the core protein after chondroitinase ABC digestion of PGs (data not shown). In all controls either exposed to normal mouse serum instead of MAb 5D4 or not exposed to MAb 5D4,no reactive bands were observed (data not shown).

Light Microscopy Immunohistochemistry. In rat bone specimensnot exposed to chondroitinase ABC, MAb 5D4 very weakly stained the cell surface of osteocytes and the walls of bone lacunae, and weakly stained vascular canals, with no identifiable staining in the rest of the bone matrix (Figure 3). After chondroitinase ABC pre-treatment, a significant increase in MAb 5D4 antigenicity was observed in the cell surfaces of osteocytes, lacuna walls, and bone canaliculi, as well as in the walls of vascular canals, but the rest of the bone matrix lacked staining (Figure 4). In rabbit bone specimens not exposed to chondroitinase ABC, MAb 5D4 staining was distributed over the entire portion of the mineralized matrix and present in vascular canals; focal accumulation of staining was observed in presumed lacuna walls. However, the cell surfaces of osteocytes and the unmineralized matrix sur-

rounding osteocytes (pericellular matrix) did not show clearly identifiable staining at the light microscopic level (Figure 5). When chondroitinase ABC digestion was used a significant increase in staining was observed in the aforementioned sites, whereas the distribution of the stained material was very similar in the two specimens (Figure 6). Similar staining of the aforementioned sites of rat and rabbit bone specimens was not observed in all control specimens (not illustrated).

Electron Microscopy Immunocytochemistry. We evaluated only MAb 5 D 4 staining in tissue sections treated with chondroitinase at the electron microscopic level. In rat bone, MAb 5D4 staining was observed in the unmineralized portion, such as the cell surfaces of osteocytes, the pericellular matrix surroundingosteocytes, and the unmineralized matrix, which often lined the inside of the wall of the vascular canal, as well as in the walls of bone lacunae and bone canaliculi,which were mineralized (Figures 7 and 8). However, the rest of the mineralized bone matrix lacked staining. In rabbit bone, MAb 5D4 staining appeared to be especially prominent in the walls of bone lacunae and bone canaliculi. The rest of the mineralized bone matrix demonstrated relatively homogeneousstaining, which was observed on collagen fibrils (Figures 9-11). The narrow space between the canalicular wall and the osteocytic process did not demonstrate definitive staining. Immunoreactivity of the pericellular matrix surrounding osteocytes varied with bone lacunae, ranging from very weak to moderate. The matrix material appeared to be attached to the cell surface of an osteocyte (Figures 9 and 10).The vascular canals stained similarly to those of the rat subperiosteal bone (not illustrated). Similar staining of the aforementioned sites was not observed in all control specimens (not illustrated).

Discussion This study demonstrates the nature and distribution of KScontaining glycoconjugates associated with the mineral phase of rat midshaft subperiosteal bone at the biochemical and immunocytochemical level, using MAb 5D4 which recognizes epitopes within KS, consisting of a minimum of six sugar units containing disulfated N-acetyllactosamineresidues (5,6,34).Furthermore, the results obtained from the rat bone were compared with those of rabbit alveolar bone. On the basis of Western blot analysis of

Figure 9. Pre-embedment immunoperoxidase staining of rabbit alveolar bone with 5D4 specific for KS after chondroitinase ABC digestion. MAb 5D4 stains the pericellularmatrix (thick arrows) very weakly, the lacunawall (LW enlarged in inset) moderately, and the wall of bone canaliculi (BC) moderately to strongly,whereas no clearly identifiablestaining is visible on the surface of an osteocyte; some pericellular matrix material appears to be attached to the cell surface. Weak staining (thin arrows) is seen in the rest of the mineralized bone matrix. Original magnification x 17,500; Inset x 37,000. Bars = 1 pm. Figure 10. Pre-embedment immunoperoxidase staining of rabbit alveolar bone with 5D4 specific for KS after chondroitinase ABC digestion. In this instance, MAb 5D4 moderately stained the pericellular matrix surrounding an osteocyte (0)and strongly stained the lacuna wall (LW). Original magnification x 37,500. Bar = 1.o pm. Figure 11. Pre-embedment immunoperoxidase staining of the mineralized bone matrix among bone lacunae in rabbit alveolar bone with 5D4 specific for KS after chondroitinase ABC digestion. Stain deposits (thin arrows) are observed on the peripheryof collagen fibrils. The walls of bone canaliculi(BC) stain strongly. Original magnification x 37,500.Bar = 1 m.

Downloaded from jhc.sagepub.com at UCSF LIBRARY & CKM on March 25, 2015

1786

MAENO, TAGUCHI, KOSUGE, OTSUKA, TAKAGI

E-extract from the rat bone with MAb 5D4, we demonstrate that KS-containing glycoconjugates having an average Mr of 66,00068,000from SDS-PAGE are present in the mineral phase. Furthermore, this glycoconjugate contains P-N-acetylglucosaminidiclinkage susceptible to enzyme digestion with endo-P-N-acetylglucosaminidase. The immunocytochemicallocalization of this glycoconjugate to the wall of the bone lacuna and canaliculus in this study was identical with that observed for chondroitin sulfate PGs in rat bone (45), which suggests that KS-containingglycoconjugates co-localize with chondroitin sulfate PGs. Previous biochemical studies (38) have shown that [ 35S]-sulfate-labeledmaterial sensitive to keratanase is present in mineralized matrix of rat bone; however, its amount is so small that it is difficult to judge whether or not KS is a component of rat bone GAGs. The present study is the first to demonstrate the presence of KS-containingglycoconjugates associated with the mineralized bone matrix of rat midshaft subperiosteal bone and also the presence of a similar type of the glycoconjugate having an MI of66,OOO-68,000 in the rabbit alveolar bone. The presence of KSPG- and KS-resembling material in rabbit bone has been demonstrated biochemically (2,10,33,51).Kinne and Fisher (30) have isolated and partially characterized one of these molecules, demonstrating the presence of a KSPG with an Mr of approximately 200,000 in rabbit cortical bone. KS chains are 0-linked to a core protein (MI approximately 80,000), the aminoterminal protein sequence of which is nearly identical to human bone sialoprotein I1 (30). Based on our Western blot analyses of E-extract from rabbit bone, two major bands were seen, the smaller giving the doublet at Mr 61.000-63,000 and Mr 66,000-68,000 and the larger being more predominant than the smaller and having an Mr of approximately 200,000 on 5-20% SDS-PAGE. The identity of KS chains was further established by the sensitivity of these bands to keratanase I1 and endo-P-galactosidase.Keratanase I1 hydrolyzes 1,3-~-glucosaminidiclinkage to galactose in KS, but on cleavage it requires the sulfate at C-6 position of the participating glucosamine (26,37),whereas endo-P-galactosidasehydrolyzes specifically the P-galactosidic linkages of non-sulfated galactosyl residues in KS and various substrates(23,37).To determine the presence of an asparaginelN-acetylglucosaminelinkage serving as an digestion attachment site for KS, endo-P-N-acetylglucosaminidase was similarlyperformed. The enzyme hydrolytically cleaves the N,Mdiacetylchitobiose moiety of asparagine-linkedoligosaccharidesof various glycoproteins (52). Either the susceptibility of the smaller or the insensitivity of the larger to endo-P-N-acetylglucosaminidase suggests that the major distinction between the two forms of KScontaining glycoconjugates is their mode of attachment to the core protein; the smaller contains KS linked to protein through the P-N-acetylglucosaminidiclinkage between N-acetylglucosamine and the amide group of asparagine as observed in corneal KS (39), and the larger appears to lack this linkage. Therefore, the larger population may represent the KSPG previously identified by Kinne and Fisher (30), whereas the smaller population has not been previously demonstrated in rabbit bone. However, quail medullary bone contains KSPGs of similar size (Mr 40,000-70,000) (28). The present study demonstrates that prior chondroitinase ABC digestion of tissue sections increases immunohistochemical 5D4 staining of rat and rabbit bone. This observation may indicate that the enzyme pre-treatment can facilitate penetration of staining reagents by the elimination of chondroitinaseABC-digestibleGAGs,

since chondroitin sulfate chains, being longer than KS chains, provide steric hindrance (42) and are known to be co-localized in reactive sites of rat subperiosteal bone (45). The validity of this interpretation is supported by the results of previous studies (42) demonstrating increased 5D4 antigenicity in PGs purified from chick and bovine cartilage after chondroitinasepre-treatment, using enzyme-linked immunosorbent assays. Previous immunohistochemicalstudies by Bartold (1) have used MAb 5D4 to localize KS staining in the immediate periphery of the osteocyte lacunae and the mineralized matrix of rabbit alveolar bone. The present study confirms and extends these observations to the electron microscopic level and also compares two different species. The present immunocytochemical staining of mineral-binding, KS-containingglycoconjugates was observed in the wall of the lacunocanalicularsystem and the rest of the mineralized bone matrix of rabbit bone, but was confined only to the former site in rat bone. Differences between immunocytochemical localization in these two specimens could be explained by our results of Western blot analysis, showing the presence of the larger population having Mrof approximately 200,000 and presumed to be a KSPG rich in sialic acid or bone sialoprotein I1 in rabbit bone (30),but the absence of this population in rat bone. Bone sialoprotein 11, which has been identified in human, bovine, and rat bone (20-22) does not appear to have any KS chains (30). Therefore, 5D4 staining in the rest of the mineralized matrix of rabbit bone is presumably attributed to the KSPG. Furthermore, the present ultrastructural immunocytochemistry localized this staining on collagen fibrils. This observation presumably indicates that KS-containing glycoconjugates bound to collagen fibrils are binding with calcium phosphate or hydroxyapatite in vivo. This assumption is similar to the postulated mechanism of the collagen-osteonectin complexmediated bone mineralization (48). Finally, the larger population of mineral-binding,U-containing glycoconjugates (Mr approximately 200,000) is considered unique to rabbit bone, whereas identification of the smaller population, especially the Mr 66,000-68,000 glycoconjugate in both the rat and rabbit bone, suggests that this molecule appears to be a widely distributed component in the bone matrix of a wide variety of animal species. The function of these molecules in bone is not known at present. Further biochemical studies of these molecules may result in a better understanding of their role in biological functions such as bone matrix assembly, bone metabolism, and possibly mineralization.

Acknowledgments We thank Dr Richard T. Parmley (University of Exas Health Science Center at San Antonio, Zxas) for critical4 reading the manuscript and he&fui comments. We ah0 thank Drs Miasaaki Toda andMasato Kageyamafor their technicai assistance.

Literature Cited 1. Bartold PM:A biochemical and immunohistochemical study of the proteoglycans of alveolar bone. J Dent Res 69:7, 1990 2. Burckard J, Degand P: Isolement et Etude des mucopolysaccharides sulfates de l’os compact de lapin. CR SOCBiol Paris 164432, 1970

3. BurnsJ, Chan VTW, JonassonJA, Fleming KA, Eylor S. McGee JO’D: Sensitive system for visualizing biotinylated DNA probes hybrydized in situ: rapid sex determination of intact cells. J Clin Path01 38:1085, 1985

Downloaded from jhc.sagepub.com at UCSF LIBRARY & CKM on March 25, 2015

K E R A W SULFATE IN RAT AND RABBIT BONE

4. CandlishJK, Holt F The proteoglycansof fowl cortical and medullary bone. Comp Biochem Physiol 40B:283, 1971 5 . Caterson B, Calabro T, Hampton A: Monoclonal antibodies as probes

for elucidating proteoglycan structure and function. In Wight TN, Mecham RP, eds. Biology of proteoglycans. Orlando, Academic Press,

1787

24. Goldberg HA, Domenicucci C, Pringle GA, SodekJ: Mineral-binding proteoglycans of fetal porcine calvarial bone. J Biol Chem 263:12092, 1988 25. Graham RC Jr, Kamovsky MJ: The early stages of absorption of injected horseradish peroxidase in the proximal tubules of mouse kidney: ul-

trastructuralcytochemistry by a new technique.J Histochem Cytochem

1987, 1 6. Caterson B, ChristnerJE, Baker J R Identification of a monoclonal an-

tibody that specificallyrecognizes corneal and skeletal keratan sulfate. Monoclonal antibodies to cartilageproteoglycan. J Biol Chem 258:8848, 1983 7. Champbell KP, Maclennan DH, Jorgensen AO: Staining of the Ca2’binding proteins, calsequestrin, calmodulin, troponin C, and S-100, with the cationic carbocyaninedye “Stains-all.”J Biol Chem 258:11267, 1983

14:291, 1966 26. Hashimoto N, Morikawa K, Kikuchi H, Yoshida K, Tokuyasu K: Keratanase 11: A novel enzyme acting on skeletal type keratan sulfate (KS11). XVth International Carbohydrate Symposium Abstracts 1990,271 27. Hawkes R, Niday E, Gordon J: A dot-immunobinding assay for monoclonal and other antibodies. Anal Biochem 119:142, 1982

8. Chu N-M, Janckila AJ, WallaceJH, Yam LT Assessment of a method

28. Hunter SJ, Schraer H: In vitro synthesis of proteoglycans associated with medullary bone inJapanese quail. Arch Biochem Biophys 220272, 1983

for immunochemicaldetection of antigen on nitrocellulosemembranes. J Histochem Cytochem 37257, 1989

29. Karnovsky MJ: A formaldehyde-glutaraldehyde fixativeof high osmolality for use in electron microscopy. J Cell Biol 27:137A, 1961

9. CouchmanJR. Caterson B, ChristnerJE, BakerJ R Mapping by monoclonal antibody detection of glycosaminoglycans in connective tissue. Nature 307:650, 1984 10. Diamond AG, Triffitt JT, Herring GM: The acidic macromoleculesin rabbit cortical bone tissue. Arch Oral Biol 27:337, 1982 11. Dickson IR, Jande SS: Effects of demineralization in a ethanolic solution of triethylammonium EDTA on solubility of bone matrix components and on ultrastructural presenation. CalcifTissue Int 32175, 1980

30. Kinne RW, Fisher LW Keratan sulfate proteoglycan in rabbit compact bone is bone sialoprotein 11. J Biol Chem 262:10206, 1987

12. Domenicucci C, Goldberg HA, Hofmann T, Isenman D, Wasi S, Sodek J: Characterizationof porcine osteonectin extracted from fetal calvariae. Biochem J 253:139, 1988 13. Fairbanks G, Steck TL, Wallach DFH: Electrophoretic analysis of ma-

jor polypeptides of the human erythrocyte membrane. Biochemistry 10:2606, 1971

ed. The chemistry and biology of mineralized tissues. Birmingham, AL. Ebsco Media, 1985, 188 15. Fisher LW, Hawkins GR, Tuross N, TermineJD: Purification and partial characterizationof small proteoglycans I and 11, bone sialoprotein

17. 18.

19.

20. 21. 22.

23.

I and 11, and osteonectin from the mineral compartment of developing human bone. J Biol Chem 2629702, 1987 Fisher LW, Schraer H: Keratan sulfate proteoglycan isolated from the estrogen-induced medullary bone in Japanese quail. Comp Biochem Physiol 72B:227, 1982 Fisher LW, Schraer H: The glycosaminoglycans of estrogen-induced medullary bone in Japanese quail. Arch Biochem Biophys 205:396, 1980 Fisher LW,Termine JD, Dejter SW Jr, Whitson SW, Ymagishita M, Kimura JH, Hascall VC, Kleinman HK, Hassell JR, Nilsson B Proteoglycans of developing bone. J Biol Chem 258:6588, 1983 Fisher LW, TermineJD, Young MF Deduced protein sequence of bone small proteoglycan I (biglycan) shows homology with proteoglycan I1 (decorin) and several nonconnective tissue proteins in a variety of species. J Biol Chem 264:4571, 1989 Fisher LW, Whitson SW, Avioli LV, Termine JD: Matrix sialoprotein of developing bone. J Biol Chem 258:12723, 1983 FranzCn A, Heinegird D: Isolation and characterization of two sialoproteins present only in bone calcifiedmatrix. BiochemJ 232:715, 1985 FranzCn A, Heinegird D: Proteoglycans and proteins of rat bone. Purification and biosynthesis of major noncollagenousmacromolecules. In Butler WT, ed. The chemistry and biology of mineralized tissues. Birmingham, AL, Ebsco Media, 1985, 132 Fukuda MN, M a m u ” G Endo-P-galactosidaseofficherichiafreundti. Purification and endoglycosidic action on keratan sulfates, oligosaccharides, and blood group active glycoprotein.J Biol Chem 251:6218, 1976

32. Li CY, Ziesmer SC, Lazcano-Villareal 0: Use of azide and hydrogen peroxide as an inhibitor for endogenousperoxidasein the immunoperoxidase method. J Histochem Cytochem 35:1457, 1987 33. Masubuchi M, Miura S, Yoshizawa A: Glycosaminoglycans and glycoproteins isolated from rabbit femur. J Biochem 77:617, 1975 34. Mehmet H, Scudder P, Tang PW, Hounsell EF, Caterson B, Feizi T

The antigenic determinants recognized by three monoclonal antibodies to keratan sulphate involve sulphated hepta- or larger oligosaccharides of the poly (N-acetyllactosamine)series. EurJ Biochem 157:385, 1986

14. Fisher LW The nature of the proteoglycans of bone. In Butler WT,

16.

31. Laemmli UK Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature 227:680, 1970

35, Merril CR, Goldman D, Van Keuren M L Simplified silver protein de-

tection and image enhancement method in polyacrylamidegels. Elcctrophoresis 3:17, 1982 36. Meyer K, Davidson E, Linker A, Hoffman P: The acid mucopolysaccharides of connective tissue. Biochim Biophys Acta 21:506, 1956 37. Nakazawa K, Ito M, Yamagata T, Suzuki S: Substrate specificity of ker-

atan sulphate-degradingenzymes (endo-P-galactosidase,keratanse and keratanse 11) from microorganisms. In Greiling H, ScottJE, eds. Keratan sulphate. Chemistry, biology, chemical pathology. London, The Biochemical Society, 1989, 99 38. Prince CW, Rahemtulla F, Butler WT Incorporation of (35S)sulphate

into glycosaminoglycans by mineralized tissues in vivo. Biochem J 224:941, 1984 39. RodCn L Structure and metabolism of connectivetissue proteoglycans. In Lennarz WJ, ed. The biochemistry of glycoproteins and proteoglycans. New York, London, Plenum Press, 1980, 267 40. Sato T A modified method for lead staining of the sections.J Electron Microsc 17:158, 1968 41. ScottJE, Kyffin TW: Demineralization in organic solvents by alkylammonium salts of ethylenediaminetetra-aceticacid. BiochemJ 169:697, 1978 42. Sorrel1JM, Caterson B: Monoclonal antibodies specific for kcratan sulfate detect epithelial-associatedcarbohydrates.Histochemistry94269, 1990 43. Spurr AR: A low-viscosity epoxy resin embedding medium for electron microscopy. J Ultrastruct Res 26:31, 1969 44. Sugahara K, Ho P-L, Dorfman A: Chemical and immunologicalchar-

acterization of proteoglycans of embryonic chick calvaria. Dev Biol 85:180, 1981 45. %gi

M, Maeno M, Kagami A, Takahashi Y, Otsuka K Biochemical

Downloaded from jhc.sagepub.com at UCSF LIBRARY & CKM on March 25, 2015

1788

MAENO, TAGUCHI, KOSUGE, OTSUKA, TAKAGI

and immunocytochemicalcharacterizationof mineral binding proteoglycans in rat bone. J Histochem Cytochem 39:41, 1991 46. W g i M, Maeno M, %!ashashi Y,Otsuka K: Biochemical and immunoand lectin-histochemical studies of solubility and retention of bone matrix proteins during EDTA demineralization. HistochemJ 24:78, 1992 47. Termine JD, Belcourt AB, Conn KM, Kleinman HK: Mineral and collagen-bindingproteins of fetal calf bone.J Biol Chem 256:10403, 1981 48. TermineJD, Kleinman HK, Whitson SW, Conn KM, McGarvey ML, Martin GR: Osteonectin, a bone-specific protein linking mineral to collagen. Cell 26:99, 1981

49. Tokuyasu KT: Application of cryoultramicrotomy to immunocytochemistry. J Microsc 143:139, 1986 50. Vila-Porcile E, Picart R, Tixier-Vidal A, Tougard C: Cellular and subcellular distribution of laminin in adult rat anterior pituitary. J Histochem Cytochem 35:287, 1987 51. Waddington RJ, Embery G, Last KS: The glycosaminoglycanconstitu-

ents of alveolar and basal bone of the rabbit. Connect Tissue Res 17:171, 1988 52. Yamamoto K, Takegawa K, Fan J, Kumagai H, Tochikura T The release

of oligosaccharidesfrom glycoproteins by endo-P-N-acetylglucosaminidase of Flavobacterium sp. J Ferment Techno1 64:397, 1986

Downloaded from jhc.sagepub.com at UCSF LIBRARY & CKM on March 25, 2015