Matrix Vol. 12/1992, pp. 221- 232 © 1992 by Gustav Fischer Verlag, Stuttgart

Localization of the Expression of Type I, II, III Collagen, and Aggrecan Core Protein Genes in Developing Human Articular Cartilage ISABELLE TREILLEUX, FREDERIC MALLEIN-GERIN, DOMINIQUE LE GUELLEC and DANIEL HERBAGE Laboratoire de Cytologie Moleculaire CNRS UPR 412, Institut de Biologie et Chimie des Proteines, Universite Claude Bernard Lyon 1,43 boulevard du 11 Novembre 1918, 69 622 Villeurbanne Cedex, France.

Abstract The expression of mRNAs for collagen types I, II, III and for aggrecan core protein was studied in developing human femoral cartilage by in situ hybridization, with special attention given to the cartilage covered by the perichondrium and to the articular surface. In parallel, the synthesis of the related proteins was monitored by immunohistochemistry. The cells metabolically active for type I and type III collagen expression were identified by hybridization using 2 P]-labeled cDNA clones coding for human a1(I) and a1(III), respectively. Type II collagen and core protein mRNAs were detected by hybridization with specific [32 P]-labeled oligonucleotide probes. In the femoral heads of one 22-week old fetus and of one newborn, our in situ hybridization and immunohistochemical analysis revealed that chondrocytes located immediately subjacent to the perichondrium produced collagen types I, II, III as well as aggrecan; whereas only type II collagen and aggrecan gene expression was detected deeper in the cartilage covered by the perichondrium. This observation supports the hypothesis that the inner cell layers of perichondrium are chondrogenic, with a transient state where cells express all the markers studied here. At the articular surface different patterns of expression were observed at the two developmental stages. After 22 weeks of fetal development only collagen types I and III were expressed by the surface zone cells while in the newborn cartilage, these cells expressed all the molecules studied (collagen types I, II, III and cartilage proteoglycan). At both ages the underlying cartilage cells expressed only the cartilage-specific molecules (type II collagen and aggrecan). Thus a progressive transformation of cartilaginous matrix occurs with time from the deep cartilage up to the surface by addition of new components, i. e. aggrecan and type II collagen. These results supplemented by an immunofluorescence analysis on 20-, 26- and 38-week old fetal femoral heads suggest that expression of collagen and aggrecan in the cartilage covered by the perichondrium and in the cartilage at the articular surface are subject to different regulatory mechanisms during development. Furthermore, the appearance of hybridizable core protein and type II collagen mRNAs at the articular surface, closely followed by the appearance of the proteins for which they code, indicates that core protein and type II collagen expression is regulated primarily at the transcriptionaI level in this region. Finally, the similar topography observed for the expression of these two proteins suggests that the genes for these two major constituents of cartilage matrix are coordinately regulated during growth of articular cartilage.

e

Key words: aggrecan, cartilage, collagen, human development, in situ hybridization.

222

I. Treilleux et a1.

Introduction

are subject to different regulatory mechanisms during development.

The development of the cartilage anlage of the long bones of the human skeletal system requires temporally and spaMaterials and Methods tially coordinated phases which include major changes in the composition and organization of the extracellular matHybridization probes rix. Elucidation of the mechanisms involved in the formaThe recombinant DNA clones or the synthetic oligonuction of limb cartilage is fundamental to unterstand normal and abnormal limb development as well as degenerative leotides used in these studies are illustrated in Fig. 1. Double-stranded DNA probes: Two-recombinant phenomena, such as osteoarthrosis occuring at the articular surface of cartilage in adult. A number of changes in the human cDNA clones were used: (i) a 300-bp a1(I) collagen expression of collagen genes is involved in the development eDNA insert cut out of the Hf677 clone (Chu et aI., 1982) and growth of the skeletal tissues. In vertebrates, there are containing the non-translated region, (ii) a 500-bp a1 (lll) at least fourteen types of collagens composed of polypep- collagen eDNA insert cut out of the Hf934 clone (Chu et tides that are the products of more than twenty-five-differ- aI., 1985) containing the non translated region. These DNA ent genes (for a review, see van der Rest and Garrone, fragments were purified by electrophoresis in low-melting 1990). Type I is the most abundant collagen in the human agarose gels and radiolabeled with a[ 32 P]-dCTP (3000 Ci/ body, present in bone, skin, tendon, perichondrium and mmole; Amersham, Cardiff, England) to a specific activity other fibrous tissues. Type II collagen is the major collagen of about 8 X 10 8 cpm/f!g by random oligonucleotideof cartilage whereas type III collagen is generally found in primed DNA labeling (Feinberg and Vogelstein, 1983; 1984). association with type I. Oligonucleotide probes: Two-synthetic oligonucleotide At the seventh week of human embryonic development, the prospective long bones appear as cartilaginous models sequences were used: (i) a 33-bp probe complementary to surrounded by a fibrous perichondrium. Later in develop- mRNA encoding a specific region in the C-propeptide of ment, the perichondrium disappears from the top of the the human a1 (II) chain, located within exon 51 (numbering epiphysis to give rise to articular cartilage. Little is known from Sandell and Boyd, 1990) and 150-bases downstream about the origin of the perichondrium. It has been sug- from the end of the triple helix (Sangiorgi et aI., 1985), (ii) a gested that it may form from superficial chondrocytes 24-bp probe complementary to mRNA encoding a COOHbecause they are exposed to specific environmental terminal region of the core protein of the human aggrecan, influences (Wolpert, 1982). Also, under appropriate condi- located 138-bases upstream from the EGF-like coding tions of in vitro cultivation, the chondrocytes of Meckel's domain (Baldwin et aI., 1989). The domains in type II collacartilage can redifferentiate into perichondral cells and gen and aggrecan encoded by these oligonucleotide probes reform a new fibrous perichondrium (Kavumpurath and are not subject to alternative splicing as far as it is actually Hall, 1989). However, the cartilaginous potential of the known. The oligonucleotide probes were radiolabeled with 2 perichondrium has been demonstrated in vitro (Bulstra et p]-ATP (3000 Ci/mmole, Amersham, Cardiff, Wales) in aI., 1990) and in vivo after perichondral grafts (Skoog and the presence of the T4 Polynucleotide Kinase (Boehringer Johanson, 1976; Amiel et aI., 1988). Moreover, a switch Mannheim, Meylan, France) to a specific activity of 5 X from type I collagen expression to expression of type II col- 10 8 cpm/f!g. lagen occurs during chondrogenesis (von der Mark et aI., Control probes: A 950-bp ScaI/Hindlll restriction frag1976) and from type II to type I in cultured chondrocytes ment of the plasmid PUC9 was routinely used as a control (von der Mark et aI., 1977). Thus, the capability of pre- probe for non specific hybridization with a double stranded chondrogenic or chondrogenic cells to modulate their DNA fragment. Two "sense" oligonucleotides, comphenotype renders the cell lineage of articular superficial plementary to the type II collagen probe or to the aggrecan chondrocytes or perichondral fibroblasts difficult to core protein probe were used on parallel sections for a control of non specific hybridization with oligonucleotide analyze in vivo. In an attempt to localize spatially and temporally the probes. expression of genes coding for collagen types I, II and III as well as aggrecan in human articular cartilage, we analysed In situ hybridization femoral heads obtained from a 22-week old fetus and from a newborn using in situ hybridization and immunohisFemoral heads of a 22-week old fetus obtained from a tochemistry. Our study supplemented by an immunofluor- therapeutic abortion and of a newborn dead after birth escence analysis on 20-, 26- and 38-week old fetal femoral (Laboratory of pathological anatomy, Edouard Herriot heads suggests that the pattern of expression of collagen Hospital, service of Pro Vauzelle, Lyon France) were fixed types I, II, III and of aggrecan in the cartilage covered by the for 2 -4 h in Bouin's fixative and treated as previously perichondrium and in the cartilage at the articular surface described (Treilleux et aI., 1991). Paraplast sections were

ye

In Situ Human Collagen Gene Expression mounted on microscopes slides that had been subbed as described by Hayashi et al. (1986). In experiments in which the hybridization of various probes were being compared, adjacent sections of the same specimen were mounted on separate slides to ensure that the different probes were hybridized to equivalent regions of the femoral head. The sections were dried overnight at 37°C, and usually stored at 4°C in the presence of Drierite (Prolabo, France) until hybridization was performed. The in situ hybridization procedure used in this study was a modification of that described by Hayashi et al. (1986). After deparaffinization, the tissue sections were treated with 10 ~g/ml of proteinase K (Boehringer Mannheim, Meylan, France) in 50 mM Tris-HCI, pH 7.6,5 mM EDTA for 20 min at 3 rc after which they were rinsed in PBS. The sections were then treated with 4% formaldehyde in PBS for 20 min, washed in two changes of PBS for 5 min each, treated with 0.1 M triethanolamine for 5 min, and incubated for 10 min in 0.25% acetic anhydride in a 0.1 M triethanolamine solution that was prepared immediately before use. Following acetylation, the sections were washed twice in 2X Sodium Saline Citrate (SSC) (1 X SSC is 0.15 M NaCl, 0.015 M trisodium citrate, pH 7) for 5 min each, twice in 70% ethanol for 5 min each, once in 95% ethanol for 5 min, and then air-dried prior to addition of the hybridization solution. Double-stranded DNA probes: The hybridization solution contained 50% formamide, 10 mM Tris-HCl pH 7, 0.15 M NaCI, 1 mMEDTA, IX Denhardt's solution, 80 ~g/ ml of denatured salmon sperm DNA, 500 ~g/ml of yeast tRNA, 10% dextran sulfate and the appropriate labeled probe. Hybridization was performed by applying 25 ~l aliquots of the hybridization solution containing 10,000 cpm/ ~l of probe onto the sections, covering them with siliconized coverslips, and sealing the edges with rubber cement. The slides were then incubated at 45°C for about 20 h. Oligonucleotide probes: The hybridization solution contained 40% formamide, 0.6 M NaCl, 50 mM sodium phosphate, pH 7,5 mM EDTA, pH 8,1 X Denhardt's solution, 1 ~g/ml of denatured salmon sperm DNA, and the labeled oligonucleotide probe, as described by Penschow et al. (1986). Hybridization was performed by applying 25 ~l aliquots of the hybridization solution containing about 40 ng/ml of probe onto the sections. After covering the sections with overslips, hybridization was carried out at 37°C for the type II collagen probe and at 32 °C for the aggrecan core protein probe. Following hybridization, the coverslips were removed in 4 X SSC, and the sections primarily hybridized with doublestranded DNA probes were washed twice in 2 X SSC for 10 min each at room temperature, once in 0.5 X SSC for 10 min at 45°C, and three times in 0.1 X SSC for 15 min each at 45°C. The sections primarily hybridized with the oligonucleotide probes were washed twice in 2 X SSC for 10 min each at room temperature, and twice in 1 X SSC at

223

37°C for the type II collagen probe or twice in 1 X SSC at 30°C for the core protein probe. After the last wash, the sections were dehydrated in two changes of 70% ethanol for 5 min each, one change of 95% ethanol for 5 min, and air-dried. Autoradiography was performed by dipping the slides in Ilford-K5 nuclear track emulsion which had been diluted with water at 42°C to a final concentration of 30%. After drying vertically at room temperature for at least 1 h, the slides were exposed in the dark at 4°C in the presence of Drierite (Prolabo, France) for 3 days for the doublestranded probes or for 5- 7 days for the oligonucleotide probes. The slides were then developed in KodakD-19 for 3 min at 18°C, fixed with Kodak Rapid Fixer, and washed in running tap water. The slides were then stained with toluidine blue, dehydrated with a graded series of ethyl alcohols, cleared in xylene, and mounted with Depex.

Immunohistochemistry The presence of collagen types I, II and III was detected by indirect immunofluorescence using monospecific polyclonal rabbit anti-human antibodies, a gift from Dr. D. Hartmann (Institut Pasteur, Lyon France). The presence of cartilage proteoglycan was similarly detected using a monoclonal mouse anti-human antibody (6-F-2) previously described (Vilamitjana-Amedee and Harmand, 1990), a gift from Dr. M. F. Harmand. In preparation for immunohistochemical staining, samples of femoral heads of the same 22-week old fetus and of the same newborn used for the in situ hybridization studies were frozen and stored in liquid nitrogen until sectioning. In addition, frozen sections of femoral heads obtained from 20-, 26- and 38-week-old human fetuses were used for immunohistochemistry. Frozen sections (5 ~m) were applied to gelatincoated slides. . Prior to antibody treatment, the sections were incubated for 1 hat 37°C with 0.2% hyaluronidase (Sigma, type Ill, 810 U/mg) in PBS, allowed to dry at 37°C, after which they were washed three times for a total of 30 min with PBS at room temperature. The sections were then incubated for 15 min in 1% bovine serum albumin in PBS, followed by an overnight incubation at 4°C with the collagen type I, II or III polyclonal antiserum diluted 1: 10 in PBS. In another series of experiments and after the same hyaluronidase treatment, sections were incubated for 1 h at room temperature with the aggrecan monoclonal antibody (cell culture supernatant) without dilution. The sections were then washed three times with PBS and incubated with fluorescein-conjugated goat anti-rabbit or sheep anti-mouse secondary antibodies (Institut Pasteur, Paris France). Following several PBS washes, the sections were mounted in glycerol/PBS (1: 1). Observations were made using a Zeiss (Oberkochen, Germany) Universal Microscope equipped for epifluorescence).

224

1. Treilleux et al.

Results In situ hybridization and immunohistochemistry of cartilage covered by the perichondrium In order to optimize in situ hybridization conditions with y[ 32 P]-labeled oligonucleotide probe in localizing type II

collagen mRNA, the probe was hybridized under various conditions to sections of 22-week old fetal femoral head which contained well-differentiated cartilage, as well as perichondrium. Non specific hybridization and emulsion background hybridization signals were routinely estimated by hybridizing adjacent tissue sections with the specific probe, or with a sense probe complementary to the specific probe, or with no probe at all. In these preliminary studies we determined that pretreatment with a minimal concen-

tration of 10 !tg!ml of proteinase K was essential to obtain optimal hybridization and that lower concentrations of the enzyme greatly reduced the hybridization signal. Furthermore, an optimal signal to background ratio was obtained when sections were hybridized at 3rC with about 40 ng! ml of the oligonucleotide type II collagen probe. Under these conditions, the specificity of the type II collagen probe is demonstrated in Fig. 2: Cells in the cartilage of a 22-week old fetal femoral head are intensely labeled, while, in contrast, virtually no detectable hybridization signal above non specific background is observed over the perichondrium (Fig. 2 b). Immunostaining of the equivalent region of a same stage femoral head shows a strong fluorescence in the extracellular matrix of the cartilage, and an absence of fluorescence in the perichondrium (Fig. 2c).

C--

cDNA Probes S' _

S_t_'p

L-

al (I): HI 677

Eco RI

Eco RI

3'

Eco RI

Stop Codon

al (III):

HI 934

------V_I-

S' - _L.-I Eco RI

Eco RI

3'

Eco RI

sao bp

Oligonucleotide Probes Exon 51

at (II):

S' -

-

-

-

------3'

-

,

S' CAG An GAG AGC ATC CGC AGC CCC GAG GGC TCC 3'

EGF-Iike coding domain

CP:

S'- -

)l,P8~1-

". ".,."''/

".

"

_

-

-

3'

................

"

S' CAG CAG ACC CAG CGC CCT GCA GAG 3'

Fig. 1. Recombinant eDNA or synthetic oligonucleotide probes used for in situ hybridization. On the top, the a1(1) collagen probe is a 300-bp EcoRI fragment cut our of the Hf 677 clone (Chu et aI., 1982) containing the non translated region. The a1 (III) collagen probe is a 500-bp EcoRI fragment cut our the Hf934 clone (Chu et aI., 1985) containing the non translated region. EcoRI sites are cloning sites for Hf 677 and Hf 934 and black boxes represent the fragments used as probes. At the bottom, the type II collagen probe is a 33-oligomer probe complementary to mRNA encoding a specific a1(II) C-propeptide located within exon 51 (from the 5' end) of the human a1(II) collagen gene and ISO-bases downstream from the end of the triple helix (Sangiorgi et aI., 1985). Exon 51 is represented by a grey box. The aggrecan core protein probe is a 24-bp probe complementary to mRNA encoding a COOH-terminal region of the core protein of the human aggrecan, located 138-bases upstream from the EGF-like coding domain (Baldwin et aI., 1989). The EGF-like domain is represented by a dashed box on the eDNA. The written sequences are the coding sequences, i. e. the oligonucleotide probes are complementary to these sequences.

In Situ Human Collagen Gene Expression

.

.

.

•

a

225

._

Fig. 2. Longitudinal sections of epiphyseal cartilage covered by the perichondrium from a femoral head of a 22-week old fetus: Bright light (A) and corresponding dark field image (B) of a section hybridized with the [32 PJ-labeled oligonucleotide type II collagen probe. Note the intense accumulation of silver grains over the chondrocytes in the cartilage (C) up to the perichondrium (P). (C): Equivalent section of the same region stained with a monospecific anti-human type II collagen antibody by fluorescence. Specific staining is limited to the well differentiated cartilage in the extracellular matrix. (A - C) X 170.

e

2 PJ-Iabeled doubleHybridization conditions with stranded DNA probes were performed as described previously (Mallein-Gerin et aI., 1988; 1990), except that hybridization was carried out at 45°C instead of 50 °C because of the smaller sizes of the probes used in the present study. After 22 weeks of development, cells producing high amounts of collagen types I and III were found in perichondrium (Fig. 3). The cells in cartilage bordering the epiphyseal perichondrium were repeatedly observed to contain al(I) and al(III) collagen mRNAs (Fig. 3 band 3 e). At the same stage, extracellular matrix of the cartilage of an equivalent region contained collagen types I and III proteins as detected by immunofluorescence (Fig. 3 c and 3 f.). It should be noted that our in situ hybridization and immunofluorescence analysis performed on a femoral head of a newborn showed a topographic pattern of expression of collagen types I, II and III in the cartilage covered by the perichondrium similar to the one observed after 22 weeks of fetal development (data not shown).

In situ hybridization and immunohistochemistry of articular cartilage 22-week old fetus: A specific signal of hybridization with the al(I) and al(III) collagen DNA probes was clearly evident over the chondrocytes constituting the superficial cell layers at the articular surface of the femoral epiphysis (Fig. 4 band 4 e). Collagen types I and III were also synthe-

sized and deposited in the extracellular matrix of this region (Fig. 4 c and 4 f). In contrast, the expression of the cartilage characteristic type II collagen was not detected in the superficial cell layers of the articular surface, but was confined to the chondrocytes of the deep cartilage as judged by our in situ hybridization analysis with a type II collagen oligomer probe (Fig. 4 h). In parallel, the extracellular matrix of the superficial cartilage did not contain immunodetectable type II collagen, while the deeper zone of the cartilage intensely stained with the anti-type II collagen antibody (Fig.4i). In order to compare the pattern of distribution of type II collagen expression with that of another cartilage specific protein, sections were hybridized with a 24-oligomer probe complementary to a sequence encoding a COOH-terminal region of the core protein of the human aggrecan (Baldwin et aI., 1989). We used the same hybridization protocol as for the type II collagen oligomer probe but with a hybridization temperature of 32°C instead of 37°C because of the smaller size of the probe. Thus, in situ hybridizations performed on parallel sections with the type II collagen and core protein probes indicated that type II collagen and core protein mRNAs were colocalized in the deep zone of epiphyseal cartilage but were both absent from the superficial chondrocytes of the articular surface at this stage (Fig. 4 b and Fig. 5 b). Furthermore, while cartilage stained with the anti-core protein antibody up to the perichondrium (Fig. 5 a), the articular surface of the 22-week old fetal

226

I. Treilleux et a1.

o

Fig. 3. High magnifications of longitudinal sections of epiphyseal cartilage covered by th~ perichondrium from a femoral head of a 22week old fetus. Bright light and corresponding dark field images of sections hybridized with the 2 P]-labeled double-stranded type 1collagen probe (A, B) or type III collagen probe (D, E). The perichondrium (P) is intensely labeled by either probe (B, E) and note the presence of a1(1) mRNAs or a1(III) mRNAs within the chondrocytes in the cartilage (C) just underneath the perichondrium (B and E, arrows). (C, F): Equivalent sections of the same region stained with an anti-human type 1collagen monospecific antibody (C) or with an anti-human type III collagen monospecific antibody (F). Type 1 and III collagens are detected in the perichondrium and underneath to about 2-cell layers within the cartilage (C and F, arrows). (A-F) X 260.

e

Fig. 4. Localization of types I, II and III collagen mRNAs in paraffin sections containing a portion of femoral articular cartilage of a 22week old fetus. Serial sections were hybridized with cDNA probes for type I (B) and type III (E), and with an oligonucleotide type II probe (H). In (A), the photomicrograph shows the articular surface and corresponds to the dark field image seen in (B). In (D) an adjacent section was hybridized with a non specific double-stranded DNA probe, i.e. a 950-bp restriction fragment of plasmid PUC9. In (G), a section was hybridized was a "sense" oligonucleotide type II collagen probe, i. e. complementary to the type II oligonucleotide probe for a control for non specific hybridization. (5) represents the articular surface. The presence of a1(I) mRNAs and a1(III) mRNAs is detected within the chondrocytes at the articular surface (B, E); whereas type II collagen mRNA sequences remain located in the chondrocytes of the deeper zone (H). (C), (F) and (I) are frozen sections of regions equivalent to (B), (E) and (H), respectively, that were incubated with anti-type I (C), anti-type III (F), or anti-type II (I) collagen antibodies. Note that the staining is localized in the same regions as the corresponding in situ hybridization signal. (A-I) X 170.

In Situ Human Collagen Gene Expression femoral head did not show any detectable aggrecan in the extracellular matrix by immunofluorescence (Fig. 5 c). Newborn: Cells expressing significative amounts of al (I) and al(III) collagen mRNAs were identified within the superficial zone of the articular surface in a pattern similar to that seen in a 22-week old fetal femoral head (Fig. 6 a and

227

6b). At the same time, collagen types 1 and III were identified in the extracellular matrix of the same region (Fig. 6 d and 6e). However, type II collagen mRNA sequences were present not only in the deep cartilage, but also up to the articular surface as detected by our in situ hybridization analysis (Fig. 6 c). Type II collagen synthesis was also

228

I. Treilleux et al.

detected by immunofluorescence at the articular surface (Fig. 6 f). Concomittantly, aggrecan core protein gene expression occured within chondrocytes from the deep zone of the cartilage up to the articular surface, as detected by in situ hybridization (Fig. 7 b), and synthesis of the corresponding gene product followed the same tissue distribution (Fig. 7 c).

Discussion Human femoral heads of a 22-week old fetus and of a newborn were studied by in situ hybridization in order to identify cells expressing collagen types I, II and III as well as the aggrecan gene, with special attention given to the articular surface and the cartilage underneath the perichondrium. The results of our studies indicate that the technique of in situ hybridization with [32 P]-labeled oligonucleotide probes can be successfully applied to human cartilage fixed in Bouin's liquid and embedded in paraffin (Treilleux et aI., 1991). For example, after 22 weeks of fetal development, an intense hybridization signal is observed with the type II collagen oligonucleotide probe over the differentiated cartilage covered by the perichondrium, whereas no detectable signal above non specific background occurs over the perichondrium. The hybridization signal closely corresponds to the localization of type II collagen protein detectable by immunohistochemical staining with a monospecific antibody. Our observations clearly illustrate that a collagen oligomer probe labeled with 2 p] is quite suitable for in situ

e

hybridization in that it provides good resolution at the tissue level, with low background. Our in situ hybridization and immunohistochemistry analysis of the distribution of collagen types I, II and III reveal that for both developmental stages studied here, chondrocytes immediately adjacent to the perichondrium are able to express the three types of collagen. This observation very likely reflects a cartilaginous potential of the perichondrium with its inner cell layers able to undergo chondrogenesis as development progresses. Thus, this differentiation could contribute to the growth of epiphyseal cartilage in addition to the chondrogenesis developing from the cartilage canals that are perichondral invaginations within the epiphysis (Chappard et aI., 1986). This is supported by the demonstration in vivo of formation of articular cartilage from free perichondrial grafts in man (Skoog and Johansson, 1976) and in rabbit (Amiel et aI., 1988). Furthermore, during human fetal development, growth of epiphyseal cartilage is observed but the thickness of the perichondrium decreases with time (Chappard et aI., 1986). Thus, the in situ hybridization signal observed over the chondrocytes bordering the perichondrium with the collagen types I, II and III DNA probes and the presence within this region of the related proteins and aggrecan core protein very likely represent a transition state of chondrogenic differentiation from the inner side of perichondrium with cell layers continuing to express collagen types I and III while initiating expression of type II collagen and aggrecan. This is in agreement with the study of von der Mark et a1. (1976) who showed that, in the prospective long bones of

Fig. 5. Sections of a 22-week old fetal femoral head. (A): Immunostaining with the antibody against aggrecan core protein shows fluorescence within the cartilage (C) but not in the perichondrium (P). (B): Dark field image of a section hybridized with a [32 PJ-labeled core protein oligonucleotide probe. Note that mRNAs encoding the core protein are present in the deep zone of the cartilage, but not at the articular surface (S). (e): Equivalent section to (B) stained with the anti core protein antibody showing fluorescence in the same region as the in situ hybridization signal. (A-C) X 200.

In Situ Human Collagen Gene Expression

229

Fig. 6. Sections of femoral head of a newborn showing the articular surface (S). In situ hybridizations with [32P]-labeled DNA probes for collagen types I (A), III (B) and II (C). Equivalent sections incubated with antibodies against collagen types I (D), III (E) and II (F). Note that al(l) and al(III) collagen mRNAs are present within the chondrocytes at the articular surface but not deeper in the cartilage (A, B) and that type II collagen mRNA sequences are present in the deep cartilage as well as in the chondrocytes of the articular surface (C). Also note that the staining pattern of the related proteins follows the in situ hybridization signal (D, E, F). the chick embryo, type I and type II collagen are present to a depth of one cell layer at the perichondral border. These authors suggested that the type I collagen seen in the cartilage matrix could be a remnant of perichondral type I collagen, overgrown by cartilage (von der Mark et a1., 1976), but our in situ hybridization studies clearly show that the cells bordering the perichondrium are still metabolically active for collagen types I and III production. Moreover, since a1(I) and a1(III) transcripts as well as the corresponding proteins are detected in the same region of a newborn but not deeper in the cartilage below the perichondrium (data not shown), our observations suggest that these chondrocytes stop expressing collagen types I and III and that these proteins are degraded in the extracellular matrix under the perichondrium (about 2-3 cell layers apart) as cartilage grows with development.

In parallel, and for the two developmental stages studied here, different patterns of expression were observed at the articular surface after in situ hybridization and immunohistochemistry. After 22 weeks of fetal development, the presence of collagen types I and III expression at the articular surface was concomitant with the absence of expression of two cartilage molecules, i. e. type II collagen and aggrecan. However, collagen types I, II, III as well as aggrecan expression was detected at the articular surface of the newborn, revealing that a progressive transformation of cartilaginous matrix occurs with time from the deep cartilage up to the surface by addition of new components, i. e. aggrecan and type II collagen. Thus, our results suggest that expression of collagens and aggrecan in the cartilage covered by the perichondrium and in the cartilage at the articular surface are under different regulatory mechanisms during develop-

230

l. Treilleux et ai.

Fig. 7. Sections of femoral head of a newborn showing the articular surface (A): Bright light (B): Dark field image of (A). Autoradiography of this section hybridized with the oligonucleotide core protein probe reveals that core protein mRNAs are present in the chondrocytes of the deep cartilage as well as in the chondrocytes located at the articular surface. (C): Staining with the antibody against the core protein demonstrates the presence of the protein in the extracellular matrix of the cartilage up to the surface. (A -C) X 240.

ment. Interestingly, examination of the articular surface under the light microscope reveals a classical cartilage morphology although type II collagen and aggrecan are not present after 22 weeks (see Fig. 4 a). Comparative biochemical and immunoelectron microscopic studies of the articular surface and of a deeper zone in the cartilage could provide a new insight into this extracellular architecture in development. Sandberg et ai. (1989) have also detected type III collagen expression in the cells of the (pre)articular surfaces of digit joints of 16-week human fetuses. However, the absence of type I collagen expression in these regions (Sandberg et aI., 1989) and the presence of type I collagen expression detected in this study at the femoral articular surface could be related to the earlier fetal stage used by Sandberg et al. (1989) (16 weeks vs. 22 weeks). Type III collagen is found in relative abundance in fetal tissues: for instance the ratio of al(III)/al(l) collagens in fetal skin decreases as development progresses (Epstein, 1974; Sykes et aI., 1976). Thus, it is possible that type III collagen during development is present at the articular surface before type I collagen, if we consider that articular chondrocytes are derived from perichondral fibroblasts, as it has been shown in birds (Fell, 1925; von der Mark et aI., 1976). Although the number of cell layers expressing type I or III collagen was difficult to estimate at the articular surface because the metabolically active cells were increasingly scattered with increasing depth of the cartilage, it was clear that this number was higher at the articular surface than in the cartilage underneath the perichondrium. Since during

development, the perichondrium that initially covers all the cartilage model is replaced in some areas by articular surfaces, this suggests that differentiation of articular surface eventually proceeds through all the cell layers of perichondrium. The presence of type I and type III collagen expression within several cell layers at the articular surface could then represent remnant expression of the outside, fibrous layers of perichondrium, that, unlike the inner layers are not chondrogenic. These outside cell layers would give rise initially to the fibrocartilage described by Fell (1925) and von der Mark et al. (1976) and later in development to the articular cartilage containing collagen types I and III overlapped by type II collagen and aggrecan, as seen in the newborn stage in the present study. During chick embryo development, the articular surface of the prospective long bones is observed primarily with a matrix containing type I collagen which is progressively overlapped by type II collagen (von der Mark et aI., 1976). It should be noted also that Sandberg et al. (1988) detected very low levels of type II collagen mRNAs in the cells of articular surfaces of 16-week old human fetuses. Furthermore, although in situ hybridization and immunohistochemistry were performed on tissues of only one 22-week old specimen and of one newborn, complementary immunofluorescence studies were done on 20-, 26- and 38-week old fetal femoral heads with collagen types I, II and III antibodies. The same patterns of collagen expression were observed for the 20- and 22-week old fetuses: But the 26- and 38-week old fetuses showed a

In Situ Human Collagen Gene Expression pattern of collagen expression identical to the one observed in the newborn (data not shown). Thus, the overlapping of types I and III collagen expression by type II collagen expression at the articular surface would occur between 22 and 26 weeks of development. The present report shows that the appearance of detectable core protein and type II collagen mRNAs at the articular surface is closely followed by the appearance of their translated products as detected by immunohistochemical staining, indicating that aggrecan core protein and type II collagen expression are regulated primarily at the transcriptionallevel in this region. Moreover, the expression of these two proteins show a similar topography for both stages studied here suggesting that the genes for these two major constituents of cartilage matrix may be coordinately regulated during growth of cartilage. These observations are consistent with previous studies in which it was found that the type II collagen and core protein mRNAs become detectable by in situ hybridization in the prechondrogenic condensation of mesenchymal cells that characterizes the onset of cartilage differentiation during chick limb formation (Mallein-Gerin et aI., 1988; Nah et aI., 1988). Our study shows that chondrocytes express different phenotypes depending on their anatomical location and on the stage of development. It is worth noting that collagen types I and III are still synthesized by chondrocytes at the articular surface of a newborn. Other studies have shown that subpopulations of chondrocytes isolated from different depths of articular cartilage display metabolic differences that mayor may not persist in culture (Aydelotte and Kuettner, 1988; Archer et aI., 1990; Siczkowski and Watt, 1990). In supplement of such studies, in situ hybridization appears to be a valuable tool to characterize the phenotypic markers actively expressed by the cells in different zones of the tissue at a given stage. Furthermore, an extension of the work presented here to include different ages of adult cartilage may lead to a clearer understanding of the roles played by chondrocytes of different zones in both normal and pathological conditions. For instance, in osteoarthritis, the initial damage seen in the superficial layers may induce (or be the result of) altered expression of the particular chondrocyte subpopulations of the surface zone (Ronziere et aI., 1990). Finally, the differential expression of an additional 69-amino acid domain within the NH 2 propeptide of the human type II collagen has been recently demonstrated (Ryan and Sandell, 1990); Further studies are needed to determine a possible involvement of this domain during articular cartilage formation. Acknowledgements We are grateful to Dr. Francesco Ramirez (Mount Sinai Hospital, New York city, USA) for generously providing the al(l) and al(III) collagen probes. We thank Dr. Daniel Hartmann (Centre de Radioanalyse, Institut Pasteur Lyon, France) for kindly supplying the anti-human collagen antibodies and to Dr. M. F. Harmand

231

(Bordeaux France) for her gift with the anti-human core protein antibody. We also thank Dr. Raymonde Bouvier (Laboratoire d'Anatomie Pathologique, Service du Pro Vauzelle, Hopital Edouard Herriot, Lyon France) for· supplying the human specimens. Finally, the expert assistance of Alain Bosch in photographic printing is gratefully acknowledged. This work was financially supported by CNRS UPR 412 and by an INSERM grant (880004).

References Amiel, D., Coutts, R. D., Harwood, F. L., Ishizue, K. K. and Kleiner,J. B.: The chondrogenesis of rib perichondrial grafts for repair of full thickness articular cartilage defects in a rabbit model: a one year postoperative assessment. Connect. Tissue Res. 18: 27-39, 1988. Archer, C. W., McDowell,J., Bayliss, M. T., Stephens, M. D. and Bentley, G.: Phenotypic modulation in sub-populations of human articular chondrocytes in vitro. ]. Cell Sci. 97: 361-371,1990. Aydelotte, M. B. and Kuettner, K.: Differences between sub-populations of cultured bovine articular chondrocytes: morphology and cartilage matrix production. Connect. Tissue Res. 18: 205-222,1988. Baldwin, C. T., Reginato, A. M. and Prockop, D.J.: A new epidermal growth factor-like domain in the human core protein for the large cartilage-specific proteoglycan. ]. Bioi. Chern. 264: 15747-15750,1989. Bulstra, S. K., Homminga, G. N., Buurman, W. A., TerwindtRouwenhorst, E. and Van der Linden, A. J.: The potential of adult human perichondrium to form hyaline cartilage in vitro.]. Orthop. Res. 8: 328-335, 1990. Chappard, D., Alexandre, C. and Riffat, G.: Uncalcified cartilage resorption in human fetal cartilage canals. Tissue Cell 18: 701-707, 1986. Chu,M.L., Myers,J.c., Bernard, M.P., Ding,J.F. and Ramirez, F.: Cloning and characterization of five overlapping cDNAs specific for the human pro-al(l) collagen chain. Nucl. Acid Res. 10: 5925 - 5934, 1982. Chu, M. L., Wei!, D., De Wet, W., Bernard, M., Sippola, M. and Ramirez, F.: Isolation of eDNA and genomic clones encoding human pro-al(III) collagen: partial characterization of the 3' end region of the gene.]. Bioi. Chern. 260: 4357 -4363, 1985. Epstein, E.H.: [al(III)h human skin collagen.]. BioI. Chern. 249: 3225-3231,1974. Feinberg, A. P. and Vogelstein, B.: A technique for radiolabeling DNA restriction endonuclease fragments to high specific activity. Anal. Biochern. 132: 6-13, 1983. Feinberg, A. P. and Vogelstein, B.: A technique for radiolabeling DNA restriction endonuclease fragments to high specific activity: addendum. Anal. Biochern. 137: 266-267, 1984. Fell, H. B.: The histogenesis of cartilage and bone in the long bones of the embryonic fowl. ]. Morphol. Physiol. 40: 417-459, 1925. Hayashi,M. Y., Ninomiya,J., Parsons,J., Hayashi,K., Olsen, B. R. and Trelstad, R. L.: Differential localization of mRNAs of collagen types I and II in chick fibroblasts, chondrocytes, and corneal cells by in situ hybridization using eDNA probes.]. Cell Bioi. 102: 2302-2309, 1986. Kavumpurath,S. and Hall,B.K.: In vitro reformation of the perichondrium from perichondral-free Meckel's cartilages of the embryonic chick.]. Cran. Genet. Dev. Bioi. 9: 173-184, 1989.

232

1. Treilleux et al.

Mallein-Gerin, F., Kosher, R.A., Upholt, W. B. and Tanzer,M.L.: Temporal and spatial analysis of cartilage proteoglycan core protein gene expression during limb development by in situ hybridization. Dev. BioI. 126: 337-345, 1988. Mallein-Gerin, F., Ruggiero, F. and Garrone, R.: Proteoglycan core protein and type II collagen gene expressions are not correlated with cell shape changes during low density chondrocyte cultures. Differentiation 43: 204-221, 1990. Penshow,J. D., Haralambidis,J., Aldred, P., Tregear, G. W. and Coghlan,]. P.: Location of gene expression in CNS using in situ hybridization histochemistry. In: Methods in Enzymology, Vol. 124, Academic Press, Inc., 1986, pp. 534-548. Ronziere, M. c., Ricard-Blum, S., Tiollier,J., Hartmann, D., Garrone, R. and Herbage, D.: Comparative analysis of collagen solubilized from human foetal, and normal and osteoarthritic adult articular cartilage, with emphasis on type VI collagen. Biochim. Biophys. Acta 1038: 222-230, 1990. Ryan, M. C. and Sandell, L.J.: Differential expression of a cysteinrich domain in the amino-terminal propeptide of type II (cartilage) procollagen by alternative splicing of mRNA. J. BioI. Chem. 265: 10334-10339,1990. Sandberg, M., Vuorio, T., Hirvonen, H., Alitalo, K. and Vuorio, E.: Enhanced expression ofTGF-~ and c-fos mRNAs in the growth plates of developing human long bones. Development 102: 461-470, 1988. Sandberg, M., Makela,]. K., Multimaki, P., Vuorio, T. and Vuorio, E.: Construction of a human proa1(III) collagen cDNA done and localization of type III collagen expression in human fetal tissues. Matrix 9: 82-91, 1989. Sandell, L.J. and Boyd, C. D.: Sequence convergence in collagen genes. In: Extracellular matrix genes, ed. by Sandell, L.]. and Boyd, C. D., Academic Press, 1990. Sangiorgi, F. 0., Benson-Chanda, B., de Wet, W.J., Sobel, M.E., Tsipouras, P. and Ramirez, F.: Isolation and partial characterization of the entire human pro-a1(II) collagen gene. Nucleic Acid Res. 13: 2207-225, 1985. Siczkowski, M. and Watt, F. M.: Sub-populations of chondrocytes

from different zones of pig articular cartilage: isolation, growth and proteoglycan synthesis in culture. J. Cell Sci. 97: 349-360, 1990. Skoog, T. and Johansson, S. H.: The formation of articular cartilage from free perichondrial grafts. Plast. Reconstr. Surg. 57: 1-6,1976. Sykes, B., Puddle, B., Francis, M. and Smith, R.: The estimation of two collagens from human dermis by interrupted gel electrophoresis. Biochem. Biophys. Res. Comm. 72: 1472-1480, 1976. Treilleux, I., Mallein-Gerin, F., Le Guellec, D., Bouvier, R., Berger, F. and Herbage, D.: Detection des ARNm par hybridation in situ sur des tissus fixes fans Ie liquide de Bouin et indus en paraffine. Ann. Pathol. 11: 47 -53,1991. van der Rest, M. and Garrone, R.: Collagens as multidomain proteins. Biochimie 72: 473-484,1990. Vilamitjana-Amedee,]. and Harmand, M. F.: Osteoarthritisrelated changes in human articular cartilage. An immunologic study. Arthritis Rheum. 33: 219-227, 1990. von der Mark, K., von der Mark, H. and Gay, S.: Study of differential collagen synthesis during development of the chick embryo by immunoflorescence II: localization of type I and type II collagen during long bone development. Dev. Bioi. 53: 153-170,1976. von der Mark, K., Gauss, V., von der Mark, H. and Muller, P.: Relationship between cell shape and type of collagen synthesized as chondrocytes lose their cartilage phenotype in culture. Nature 267: 531-532, 1977. Wolpert, L.: Cartilage morphogenesis in the limb. In: Cell behaviour, ed. by Bellairs, R., Curtis, A. and Dunn, G., Cambridge University Press, Cambridge, 1982, pp. 359-372. Dr. Daniel Herbage, Laboratofre de Cytologie Moleculaire CNRS UPR 412, Institut de Biologie et Chimie des Proteines, Universite Claude Bernard Lyon 1, 43 boulevard du 11 Novembre 1918,69622 Villeurbanne Cedex, France.