View metadata, citation and similar papers at core.ac.uk brought to you by CORE

provided by Elsevier - Publisher Connector

OsteoArthritis and (2005) 13, 120e128 ª 2004 Research Society International. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.joca.2004.06.015

International Cartilage Repair Society

Inhibition of activity can delay phenotypic modulation of chondrocytes in two-dimensional culture1 J. Farjanel Ph.D.*,S.Se`ve Ph.D., A. Borel Ph.D., P. Sommer Ph.D. and D. J. S. Hulmes Ph.D. Institut de Biologie et Chimie des Prote´ines, CNRS UMR 5086, Universite´ Claude Bernard Lyon I, IFR 128 BioSciences Lyon-Gerland, 69367 Lyon Cedex 07, France

Summary

Objective: Chondrocytes frequently de-differentiate in two-dimensional (2D) culture, especially in the presence of serum. To examine the role of lysyl oxidase (LOX) induced cross-linking in this phenomenon, the effect of the specific LOX inhibitor b-aminopropionitrile (BAPN) was studied in 2D chondrocyte culture. Design: Chick embryo sternal chondrocytes (both proliferative and hypertrophic, from caudal and cranial zones, respectively) were cultured in the presence and absence of BAPN. The production and activities of LOX and LOX-like (LOXL) were assessed by enzyme assay and the use of specific antibodies. Seventeen batches of serum of different origin were compared. Chondrocyte phenotype was assessed both morphologically and biochemically, the latter by quantitative analysis of production of radiolabeled cartilage II, IX, X and XI, and the de-differentiation marker I, for up to 4 weeks in culture. Results: LOX and LOXL were identified, by Western blotting and immunofluorescence, and LO activity was measured in the medium, with both proliferative and hypertrophic chondrocytes. Inhibition of LO activity prevented or delayed chondrocyte de-differentiation, as characterized by changes in cell shape and synthesis of the five different collagen types, from the first days of culture for up to 4 weeks, depending on the origin of the serum added to the culture medium. Conclusion: LO activity may be involved in the control of chondrocyte phenotype, in addition to serum factors. Inhibition of LO activity by BAPN may be useful for the maintenance of the chondrocyte phenotype in 2D culture. Specific variations in the relative proportions of collagens II, IX and XI could be involved in the mechanism underlying these observations. ª 2004 OsteoArthritis Research Society International. Published by Elsevier Ltd. All rights reserved. Key words: Lysyl oxidase, b-Aminopropionitrile, Collagen, Chondrocyte. Abbreviations: LOX, lysyl oxidase; LOXL, lysyl oxidase-like; BAPN, b-aminopropionitrile.

Introduction Accordingly, precise fibrillar forms of collagen, as well as some soluble factors, can induce the phenotypic modulation In normal cartilage, chondrocytes are known to produce of chondrocytes in culture. collagen types II, IX and XI in proportions leading to their 1 The above observations suggest that the physico-chem- association into thin cartilage-specific heterotypic fibrils .A ical properties of the collagenous matrix might influence sub-family of chondrocytes can become hypertrophic and chondrocyte phenotype. To examine this possibility, we progressively switch to production of collagen X, charac- 2 present here the results of experiments carried out in 2D teristic of terminal differentiation . culture designed to test the effects of de novo collagen cross- In two-dimensional (2D) cell culture, de-differentiation of linking on the maintenance of the chondrocyte phenotype. chondrocytes in the presence of serum is a well documented Lysyl oxidases (LOXs) initiate covalent cross-linking in phenomenon. In contrast, in three-dimensional culture, in 3 collagens and (and perhaps other substrates) by agarose gels, the chondrocyte phenotype is maintained . oxidizing certain peptidyl lysine/hydroxylysine residues to Soluble factors present in serum are known to influence 4 peptidyl lysyl/hydroxylysyl aldehydes which then spontane- chondrocyte phenotype . In addition, many effects of ously condense with neighboring amino groups or additional variations on cell differentiation have 8,9 5,6 peptidyl aldehydes . In cartilage, LO activity is known to been reported . In the case of chondrocytes, for example, promote covalent cross-links in heterotypic collagen II/IX/XI we previously showed, in the absence of serum, that de- fibrils10. We show here that inhibition of LO activity using the differentiation can be induced by culture in three-dimensional 7 specific inhibitor b-aminopropionitrile (BAPN) can prevent or gels of collagen I , while in gels of cartilage collagens II, delay de-differentiation of chondrocytes for up to 4 weeks of IX and XI, the differentiated phenotype is maintained. culture, depending on the origin of the serum present in the culture medium. 1This work was supported by the Centre National de la Recherche Scientifique and by the Universite´ Lyon I. AB and SS were recipients of studentships from the French government. Materials and methods *Address correspondence and reprint requests to: J. Farjanel, Institut de Biologie et Chimie des Prote´ines CNRS UMR 5086, 7 CELL CULTURES passage du Vercors, 69367 Lyon Cedex 07, France. Tel: 33- 472722666; Fax: 33-472722604; E-mail: [email protected] Seventeen-day-old chick embryo chondrocytes, prepared Received 22 June 2004; revision accepted 22 October 2004. by collagenase digestion of caudal and cranial regions of

120 Osteoarthritis and Cartilage Vol. 13, No. 2 121 sternal cartilage, were plated at a density of 2.0 ! 105 cells/ collagen II was observed just below the HMW band (not cm2, unless otherwise mentioned, in 35 mm diameter shown), as previously described11. Amounts of each plastic Petri dishes (P35). The medium was DMEM collagen were then expressed as percentages of total peak supplemented with 25 mg/ml ascorbate, 2 mM glutamine, area. Measured dpm values from pepsinized cell-layer 100 mg/ml streptomycin, 100 units/ml penicillin and 10% extracts and culture media were then used to convert serum (complete medium). Seventeen different batches of percentages of each collagen type into dpm, which were sera from various origins were compared: 12 fetal calf sera then added for each collagen type to determine total (cell (FCS) from Biomedia (B1 and B2), Sigma (S1 to S7), layer C medium) percentages. In this way, results could be Hyclone (H1) and GIBCO/Life Technologies (G1 and G2); compared between collagen types, expressed as percen- two neonatal calf sera (NCS) from Sigma (S8) and GIBCO/ tages (cell layer, medium and total), and between different Life Technologies (G3); two calf sera (CS) from Hyclone experiments. The average variation between duplicates in (H2) and GIBCO/Life Technologies (G4) and one adult each experiment was found to be about 10% of the measured healthy human serum pool (HS). In all, six experiments percentage values shown in Figs. 5e7. were carried out as follows: experiment 1 e FCS B1; It should be noted that, after 14C-proline labeling of experiment 2 e FCS B2; experiment 3 e FCS S1; synthesized collagens, the measured intensities for the experiment 4 e FCS B2, S2, S3; experiment 5 e FCS S2, different collagen types depend on their proline C hydrox- S4, S5, S6, S7 and H1, NCS S8, CS and H2, human adult yproline contents. These proportions are known to be rather serum HS; experiment 6 e FCS S2, S7, G1, G2 and H1, similar among the fibrillar collagens I, II and XI, particularly NCS G3 and CS G4. When FCS was absent, the medium between different a chains of the same collagen type. In was supplemented with 1 mM pyruvate and 1 mM cysteine contrast, due to their smaller triple-helical regions, collagens as protective agents. BAPN was added as indicated at IX and X contain less proline C hydroxyproline than the a concentration of 60 mg/ml. Media including all additives fibrillar collagens12. This was particularly so for collagen IX, (purchased from Sigma) were changed every 2e3 days and where only the HMW fragment was measured, as identified cultures were maintained for up to 4 weeks. by its well characterized differences in migration on SDS- PAGE with and without reduction11, the LMW fragment being negligible in most of the fluorograms. No corrections were COLLAGEN SYNTHESIS made to allow for these differences in proline C hydroxypro- Newly synthesized collagens were metabolically labeled line content, which has no effect when comparisons are made for 24 h with 1 mCi/ml 14C-proline (285 Ci/mmol: Amersham with and without BAPN for each collagen type. International), beginning on days 2, 13, 20 or 27. Culture media were submitted to pepsinization (pepsin from Serva, Heidelberg, FRG) as previously described7 at 0.1 mg/ml for ANTIBODIES 18 h. Cell layers were treated similarly except for pepsin- Two different polyclonal antibodies were used13e15, ization with 0.3 mg/ml pepsin for 48 h. The amounts of total directed against peptides corresponding to residues newly synthesized pepsin-resistant were deter- e 7 228 279 of human preprolysyl oxidase (LOX) and residues mined by liquid scintillation counting as described . 355e416 of human preprolysyl oxidase-like (LOXL). In the Collagen types were analyzed by SDS-polyacrylamide gels e case of LOX, chicken and human forms show 92% (PAGE) on 4.5 15% gradient gels (in non-reducing sequence identity in the region used as antigen. No conditions except when mentioned) followed by fluorogra- sequence data are available for chicken LOXL. phy. Cultures were carried out and analyzed in duplicate.

QUANTITATION OF COLLAGEN TYPES SEPARATED IMMUNOFLUORESCENCE STAINING BY SDS-PAGE Cultured cells were seeded at 1.5 ! 105 cells/cm2. After Fluorograms were subjected to densitometry and the 2 to 5 days of culture in the presence of FCS with or without data analyzed using the program Image Quant 4 (Molecular BAPN, non-confluent cell layers were washed with phos- Dynamics). The linearity of the photographic response of phate buffered saline (PBS), then fixed in 4% paraformal- the fluorograms was verified and calibrated using a standard dehyde for 30 min followed by washing with 50 mM NH4Cl/ series of radiolabeled collagen chains. Where bands PBS. Intracellular staining was performed by permeabilizing displayed very low intensities, several fluorograms with cells with 1% Triton X-100 in PBS for 10 min. Each step was different exposure times were compared to confirm the followed by washing with PBS. Treated cell layers were absence of these bands, such as that of the a2 type I (a2(I)) incubated for 90 min at room temperature with primary collagen chain. antibodies diluted in PBS/1% serum albumin (BSA) at the The amounts of each of the five different collagen types (I, following dilutions: 1:50 for anti-LOX, 1:200 for anti-LOXL. II, IX, X and XI; each made up of three a chains) were Samples were then washed thoroughly with PBS, incubated determined as follows. Collagen X was measured directly in cyanine dye-conjugated donkey anti-rabbit IgG (from from the area under the a1(X) peak. Collagen I, with two a1(I) Jackson Immuno Research Laboratories), which showed and one a2(I) chains, was determined as three times the area here no cross reaction with chicken proteins. Samples were under the a2(I) peak (the a1(I) chain co-migrates with a1(II)). mounted in Dako fluorescent mounting medium and Collagen XI was determined as 1.5 times the total area under examined using a Zeiss Universal microscope equipped the a1(XI) and a2(XI) peaks (the a3(XI) chain co-migrates for epifluorescence. with a1(II)). Collagen IX was measured from the peak area of the major, pepsinized fragment, HMW. Collagen II was determined from the total area under all collagen peaks EXTRACTION FOR IMMUNOBLOTTING OF LOX ISOFORMS minus the amounts of all other collagens (as determined above). With some cell-layer extracts from BAPN-free Cell seeding was performed at 2 ! 105 cells/cm2. After 1 cultures, a feint band of b11(II) cross-linked chains of or 2 weeks of culture, the medium was discarded and cell 122 J. Farjanel et al.: Chondrocyte phenotype and lysyl oxidase layers were thoroughly washed with PBS. Cell lysis was polymerized or unprocessed forms. Overnight incubation at performed by gentle stirring for 2 h at 4(C with 0.3 ml/P35 4(C of both antibodies with the corresponding peptide of the following lysis buffer: 16 mM Na2HPO4 C 6M antigens adsorbed onto nitrocellulose membranes pre- urea C 0.5% Nonidet NP-40 C protease inhibitors (Com- vented subsequent detection of protein bands by Western plete mini C EDTA from Roche), pH 8.0. From each dish, blotting (data not shown). the lysate was collected, and then the P35 was washed with By immunofluorescence, both antibodies revealed the 3 ! 0.3 ml 16 mM Na2HPO4 pH 8.0. The combined presence, on day 5 of culture, of LOX and LOXL, in both volumes (lysate C wash) obtained for each P35 dish were cranial and caudal chondrocytes. Unlike LOXL, immunos- centrifuged at 20,000g for 20 min at 4(C. Supernatants taining of LOX was relatively intense in the nucleus, in the were submitted to precipitation with trichloracetic acid at presence or absence of BAPN (Fig. 2), as previously found a final concentration of 10% for 20 min at 4(C. Protein with 3T6 myofibroblast-like cells15. In control experiments, pellets obtained after centrifugation at 10,000g for 10 min immunostaining was negligible when antibodies were pre- were washed with cold acetone, then dried before adding absorbed on the corresponding antigenic peptides (data not SDS-PAGE sample buffer. shown). The presence of BAPN had no obvious effect on the amounts of each form of LOX present, but only on their activities. ELECTROBLOTTING Protein extracts, each corresponding to about one P35, EFFECT OF BAPN ON CHONDROCYTE MORPHOLOGY were submitted to electrophoresis in 10% SDS-PAGE. Electrophoretically resolved bands were then transferred to Chondrocyte morphology and biochemically-defined phe- PVDF membranes (Immobilon, Millipore Corp.) by electro- notype in culture were studied in six different experiments, blotting. Proteins were immunodetected using an alkaline using 17 different batches of sera (see Materials and phosphatase conjugate substrate kit (Biorad) at the methods). When cultured at initial densities of following dilutions: 1:100 for anti-LOX, 1:300 for anti-LOXL. 2.0 ! 105 cells/cm2 for up to 4 weeks in the presence of Protein mass markers were purchased from Biorad. 10% serum, cells from the cranial region became pro- gressively hypertrophic in appearance and some lost their characteristic rounded morphology adopting a spread, LO ACTIVITY fibroblast-like shape [Fig. 3(C)]. In serum-free medium LO assays were performed with a tritiated elastin supplemented with cysteine and pyruvate, cells multiplied substrate as previously described16. dpm released after relatively slowly appearing less hypertrophic but again 12 h incubation were measured in the presence and becoming progressively fibroblast-like in shape [Fig. 3(D)]. absence of BAPN. Specific LO activity was measured by When cultured in the presence of BAPN however, in the the difference in dpm obtained without and with BAPN. presence or absence of serum, cranial chondrocytes One day prior to collecting conditioned media for LO retained almost exclusively their rounded morphology, assay, complete culture medium containing FCS was whether hypertrophic or not [Fig. 3(E,F)]. The protective replaced by serum-free medium supplemented with 1 mM effect of BAPN on cell morphology was similar with caudal pyruvate with (day 2) or without (days 6, 13) 1 mM cysteine. chondrocytes, which never became hypertrophic (data not Cysteine and pyruvate were used as protective agents shown). against oxidative metabolites. Cysteine (1 mM) was found to partially inhibit LO activity (by approximately 50%) when added to assays using purified aortic LOX (not shown). EFFECT OF BAPN ON BIOCHEMICALLY-DEFINED PHENOTYPE Cleared conditioned media were maintained at 4(C and immediately concentrated by ultrafiltration with 5000 NMWL BAPN was found to prevent changes in chondrocyte cellulose Ultrafree-MC centrifugal filter devices (Millipore), phenotype with respect to synthesis of different collagen then 100 ml aliquots were used for the LO assay. types. As shown in Fig. 4, cranial chondrocytes, when cultured in the presence of FCS, produced in the culture medium on day 14, in addition to cartilage collagens II, IX, X and XI. In contrast, in the presence of Results both FCS and BAPN, type I collagen production was not EXPRESSION OF LOX VARIANTS observed. Similar results were obtained with caudal chondrocytes, which neither became hypertrophic nor Chondrocytes were isolated from both cranial (becoming produced collagen X. In contrast, for both caudal and hypertrophic after 1 week in culture) and caudal (pro- cranial chondrocytes, using serum-free medium supple- liferative) regions of chick embryo sterna. As shown in mented with 1 mM of cysteine and pyruvate as protective Table I, LO activity, as inhibited by the specific LOX inhibitor agents, synthesis of collagen I was not observed, and BAPN, was detected on days 3, 7 and 13, with both caudal BAPN did not affect the production of cartilage collagens. and cranial chondrocytes. Compared to day 7 however, The extent to which BAPN prevented chondrocyte de- activity on day 14 in the culture medium was diminished, differentiation depended on the duration of culture and the despite the increased cell number. type/batch of serum used. On day 14, for example, only Both cranial and caudal chondrocytes expressed at least some batches of FCS tested led to de-differentiation of two variants of LOX, LOX and LOXL, in the presence and cranial chondrocytes, as measured by production of absence of BAPN, as shown by Western blotting of cell- collagen I (Fig. 5, middle column), whether in the combined layer extracts on day 7 (Fig. 1), using specific antibodies. cell layer plus medium [total, Fig. 5(A)], the cell layer Similar results were obtained on day 14 (data not shown). [Fig. 5(B)] or the medium alone [Fig. 5(C)]. With those cells The expected molecular weights of active processed forms that did de-differentiate, however, this effect was completely of LOX17 and LOXL14 were observed, along with higher prevented by the presence of BAPN, in both cell layer and molecular weight bands, which probably correspond to medium. In contrast, by days 21 and 28 of culture, almost all Osteoarthritis and Cartilage Vol. 13, No. 2 123

Table I Lysyl oxidase activities in chondrocyte culture media Day Cells/sample Original volume of dpm/100 ml dpm/100 ml Average LO medium (ml) concentrated concentrated activity of original before concentration medium ÿ BAPN medium C BAPN medium (dpm/ml) 3 Caudal chondrocytes 0.6 930, 1002 674, 718 481 7 Caudal chondrocytes 1 8158, 8464 5296, 6210 2558 7 Cranial chondrocytes 4 10,932, 10,510 5800, 6536 1138 14 Caudal chondrocytes 3 5010, 5322 4344, 4702 215 14 Cranial chondrocytes 4 5038, 5840 4580, 4618 210 ee e dpm ÿ BAPN dpm C BAPN Average LO activity (dpm) e Standard LOX from e 8540, 8810 4340, 4762 4130 bovine aorta Chondrocytes were cultured in Petri dishes (9.5 cm2) seeded with 2 ! 106 caudal or cranial cells in the presence of 10% FCS. After a further 24 h incubation at 37(C in serum-free medium, supplemented with cysteine and pyruvate (day 3) or pyruvate alone (days 7, 14), conditioned medium was concentrated and assayed for LO activity on a 3H-elastin substrate. Samples (100 ml) were measured in duplicate. The original volume of medium corresponding to each sample, prior to concentration, is indicated. Activity was determined by net dpm measured in the absence and presence of BAPN, then expressed per ml of medium collected prior to concentration. Two different batches of elastin substrate were used for the assays on day 3 and days 7/14 media. batches of FCS led to chondrocyte de-differentiation (Fig. 5, FCS) was less pronounced (see Supplementary material). right hand column), as shown by the production of collagen I. With NCS and CS, BAPN only partially inhibited total This effect was eliminated or considerably reduced with all collagen I synthesis by cranial chondrocytes on day 14, and batches of FCS, with the exception of the two obtained from hardly at all on day 21. With the HS, relatively low levels of GIBCO/Life Technologies (G1, G2). This variability in the collagen I were synthesized, and BAPN had no effect. onset and extent of de-differentiation suggests the presence In contrast to the results obtained with collagen I, BAPN of different amounts of factors in the different batches of was found to have no systematic effect on the percentage serum that lead, either directly or indirectly, to chondrocyte of collagen II production by cranial chondrocytes (see de-differentiation. Supplementary material). BAPN also appeared to have no BAPN also prevented or reduced de-differentiation of caudal chondrocytes (Fig. 6), in the presence of FCS, where again the reduction in collagen I synthesis was caudal cranial dependent on the duration of culture and the batch of serum used. AB The effect of BAPN on cranial chondrocyte de-differen- tiation in the presence of other types of serum (other than

AB LOX kDa 1234 1 2 3 4

97 66

45 CD

31 LOXL LOX LOXL

BAPN -+-+ - + - +

Fig. 1. Western blotting analysis reveals similar expression levels of each lysyl oxidase variant in sternal chondrocytes of both caudal and cranial origin, in the presence or absence of BAPN. Caudal (1, Fig. 2. Indirect immunofluorescence staining of 5-day-old chon- 2) and cranial (3, 4) chondrocytes were seeded at 2 ! 105 cells/ drocytes in culture reveals the expression of LOX and LOXL by cm2, then cultured in the presence of FCS with or without BAPN. chondrocytes of both caudal and cranial origin. Cells were plated at Protein extracts from the cell layer obtained after 1 week in culture 1.5 ! 105 cells/cm2 and cultured with FCS in the presence or were separated by electrophoresis on 10% SDS-polyacrylamide absence of BAPN. A, B: Anti-LOX (228e279). C, D: Anti-LOXL gels. After transfer, PVDF membranes were probed using (A) anti- (355e416) (respective dilutions: 1:50, 1:200). A, C: Caudal LOX (228e279) or (B) anti-LOXL (355e416), at dilutions of 1:100 chondrocytes. B, D: Cranial chondrocytes. Unlike LOXL, LOX and 1:300, respectively. Arrows indicate the mature forms of LOX staining was relatively intense in the nuclei of many cells. Bar: and LOXL. 20 mm. 124 J. Farjanel et al.: Chondrocyte phenotype and lysyl oxidase

+FCS -FCS caudal cranial AB 1234 5678 910

HMW Day3 -BAPN α1(XI) α2(XI) α1(I),α1(II),α3(XI) α2(I)

CD α1(X)

Day14 -BAPN FCS ++- - ++-- ++ BAPN -+-+ -+-+ -+

Fig. 4. Prevention of collagen I biosynthesis by sternal chondro- cytes in the presence of BAPN following 2 weeks of culture. Cells EF were seeded at 2 ! 105 cells/cm2, then cultured with or without FCS, in the presence or absence of BAPN. Radiolabeling of the cells (24 h) was begun on day 13. On day 14, culture media were e e Day14 collected from caudal (lanes 1 4) or cranial (lanes 5 8) cells then +BAPN subjected to limited digestion with pepsin, or cell-layer pepsin extracts from cranial cells (lanes 9e10), were analyzed by SDS- PAGE (in non-reducing conditions) and fluorography. Migration positions of different collagen chains are indicated. HMW denotes the high molecular weight pepsin-resistant fragment of collagen IX.

Fig. 3. Morphological alterations of sternal chondrocytes from the cranial zone are prevented by the presence of BAPN. Phase ! 5 2 cells (by a factor of 1.31 G 0.46 (SD), n Z 50), as measured contrast micrographs of cells seeded at 2 10 cells/cm and 14 cultured for 3 days (A, B) or 2 weeks (CeF) with or without BAPN by incorporation of C-proline, thereby excluding any and FCS. A, C, E: With FCS. B, D, F: Without FCS. E, F: With possible toxic effects of BAPN. BAPN. A, B, C, D: Without BAPN. At day 3 (A, B), similar morphologies were seen with BAPN (not shown). Bar: 20 mm. Discussion systematic effect on collagen X production, which appeared The main observation reported here is that inhibition of to be inversely related to that of collagen II in the case of LOX activity by BAPN can prevent or delay de-differentia- cranial chondrocytes. This is illustrated in Fig. 7, which tion of chick embryo chondrocytes in 2D culture. BAPN was shows the amount of collagens II and X expressed as ratios found to preserve the characteristic rounded morphology of of percentages in the presence and absence of BAPN, for chondrocytes and reduce or eliminate the synthesis of cranial chondrocytes, in different batches of FCS. These collagen I for up to 4 weeks in culture. ratios are centred around 1, thus indicating little effect of The onset and extent of chondrocyte de-differentiation BAPN on collagens II and X production. Similar results were was found to be strongly dependent on the type/batch of obtained with caudal chondrocytes (see Supplementary serum used in the cultures. For this reason, we decided to material). carry out an exhaustive study of the effects of different sera. In contrast to the inhibition of collagen I production by Some batches of FCS were found to induce de-differenti- BAPN, and the lack of systematic effect on collagens II and ation by day 7 of culture, as detected by the appearance of X, BAPN appeared to increase the percentages of collagen I, while with others the effect was not observed collagens IX and XI. This is also illustrated in Fig. 7, which until days 14 or 21. From the onset of de-differentiation, the shows a significant increase in collagens IX and XI in the percentage of collagen I increased with time. Regardless of presence of BAPN with cranial chondrocytes, during the the onset and extent of chondrocyte de-differentiation, first 2 weeks of culture, in the presence of different batches however, inhibition of LO activity by BAPN was found to of FCS, as found in both total [Fig. 7(A)] and cell layer either prevent or lower the percentage of collagen I, for all [Fig. 7(B)] collagens. This indicated a preferential stimula- batches of FCS tested, with the exception of the two tion of collagens IX and XI production by BAPN. obtained from GIBCO/Life Technologies. With the other As reported elsewhere18, de-differentiation as assessed types of serum tested, this effect of BAPN was either much by the above biochemical alterations was minimized when reduced (NCS or CS) or not seen (adult HS). These data cells were seeded at high densities (5 to 8 ! 105 cells/cm2 suggest that different types/batches of serum contain instead of 2 ! 105 cells/cm2; data not shown). In addition, variable amounts of different factors that lead to phenotypic the presence of BAPN was found to slightly increase the modulation, the effects of at least some of which can be amount of total pepsin-resistant protein produced by these effected by inhibition of LO activity. Osteoarthritis and Cartilage Vol. 13, No. 2 125

A day 3 day 14 day 21 30 day 7 30 30 day 28

20 20 20

percentage 10 percentage 10 percentage 10

0 0 0

- BAPN + - BAPN + - BAPN + B 10 10 10

8 8 8

6 6 6

4 4 4

percentage percentage percentage

2 2 2

0 0 0

- BAPN + - BAPN + - BAPN + C 50 50 50

40 40 40

30 30 30

20 20 20

percentage percentage

10 10 percentage 10

0 0 0

- BAPN + - BAPN + - BAPN +

Fig. 5. Effects of long-term added BAPN on collagen I production by sternal chondrocytes of cranial origin in the presence of different batches of FCS. Cells were seeded at 2 ! 105 cells/cm2, then cultured with FCS (from 12 different batches), in the presence or absence of 60 mg/ml BAPN. Radiolabeling of the cells (24 h) was begun on days 2, 6, 13, 20 and 27. Pepsinized culture media and cell-layer extracts were analyzed by SDS-PAGE and fluorography, then collagen I, expressed as a percentage of total collagens, was quantitated by densitometry. (A) Total, i.e., cell layer C medium, (B) cell layer, (C) medium. Batches of serum are indicated in color, as follows: B1, B1#, B2, B2# (purple); G1, G2 (blue); H1 (green); S1 (orange), S2eS7 (black). All points shown represent the average of two separate experiments performed at the same time. Experiments indicated by a prime (e.g., B1#) refer to those carried out on a different occasion to those previously performed using a particular batch of serum. Overlapping points are offset horizontally.

Four possible mechanisms can be proposed for the determined by ascorbate, would influence the stability observed effects of LOX on the modulation of chondrocyte and hence mechanical properties of the fibrils. phenotype: (ii) Several different types of cell surface collagen receptors, acting either individually or synergistically 22,23 (i) The extent of cross-linking, initiated by LOXs, may across different signaling pathways , might be alter the mechanical properties of the extracellular involved in the control of the differentiated phenotype matrix and, consequently, cellular responses to by the structure of the matrix, through differential mechanical stimuli involving integrin receptors19,20. binding to different fibrillar forms of collagen cross- This may also account for the observation that linked to different extents. These receptors include: (1) ascorbate leads to enhanced chondrocyte de-differ- b1 integrins (native collagens) or bv integrin (dena- 7,24 entiation21, since the extent of prolyl hydroxylation, tured collagens) , (2) discoidin domain receptors 126 J. Farjanel et al.: Chondrocyte phenotype and lysyl oxidase

30 day 3 30 day 1430 day 21 day 7

20 20 20

10 10 10 percentage percentage percentage

0 0 0 - BAPN + -BAPN+ - BAPN +

Fig. 6. Effects of long-term added BAPN on production of collagen type I by sternal chondrocytes of caudal origin in the presence of different batches of FCS. Cells were seeded at 2 ! 105 cells/cm2, then cultured with FCS (of four different origins), in the presence or absence of 60 mg/ml BAPN. Radiolabeling of the cells (24 h) was begun on days 2, 6, 13 and 20. Pepsinized culture media and cell-layer extracts were analyzed by SDS-PAGE and fluorography, then total collagen type I (cell layer C medium), expressed as a percentage of total collagens, was quantitated by densitometry. Overlapping points are offset horizontally.

(DDRs)25, (3) transmembrane heparan sulphate pro- in any cell type. This is consistent with a recent report46 that teoglycans, e.g., syndecans26, (4) transmembrane the anti-tumor effects of LOX are associated with the chondroitin sulphate , e.g., CD4427 and propeptide domain of the LOX precursor, independent of NG228 and (5) annexin V/anchorin C II29. It should be LO activity. noted that other forms of cross-linking activity, not In addition to preventing/delaying the appearance of sensitive to BAPN, could also modulate matrix collagen I, BAPN appeared to stimulate the production of structure, and thereby cell phenotype, which might cartilage collagens IX and XI. This was observed at the account for the reduced effect of BAPN after 3 or 4 beginning of the culture period, sometimes before the weeks in culture. Tissue transglutaminases30,31, for production of collagen I, which suggests that it was more example, are known to be present in the calcifying a stimulatory effect than a simple consequence of the drop growth plate of cartilage during late differentiation. in percentage of collagen I resulting in increased percen- (iii) Several soluble factors present in serum, including tages of the other collagen types. Such a stimulation in the BMPs, FGF, IGF and TGF-b, can directly modulate the production of collagen IX and, more importantly, collagen chondrocyte phenotype2,32. Furthermore, XI, compared to the lack of effect on collagen II, particularly TGF-b33, FGF234 and also PDGF35 and TNF36 can in the cell layer, would probably modify the morphology and modulate LOX synthesis and/or activity, in different physical properties of the heterotypic collagen II/IX/XI fibrils. cell lines, while oxidation of lysine residues by LOX It is well documented1,47 that pure collagen II molecules can inhibit bFGF activity37. Thus, in addition to their form large diameter fibril-like tactoids, quite different from general anabolic effects on protein synthesis, serum the normal, small diameter, flexible heterotypic II/IX/XI factors might act synergistically with LOXs to regulate fibrils. The low levels of collagens IX and XI produced chondrocyte phenotype. Differences in the amounts of when LOX is active could lead to abnormal fibrils which, in such factors would account for the observed variation turn, could induce phenotype modulation and de novo in the timing and extent of de-differentiation between synthesis of collagen I. In this regard, collagen XI has also FCS and other types of serum. The nature, amounts recently been reported to limit the diameters of collagen I and synergistic effects of the specific factors involved fibrils48. In contrast, the presence of BAPN has been shown remain to be determined. to increase collagen fibril diameters in osteoblast cultures49. (iv) Finally, the demonstration of LOX in an active BAPN- This may be a direct consequence of reduced cross-linking dependent form within the cell nucleus38 and associ- on fibril structure, in the absence of any effects of collagens ated with the cytoskeleton39 strongly suggests possi- IX and XI, which would not be expected to be present in ble intracellular targets of the enzyme, such as osteoblast cultures. histones40. In this regard, we note that LOX was Changes in LOX expression have also been linked to observed here inside the nuclei of the sternal chick phenotypic changes in other cell types. For example, embryo chondrocytes. down-regulation of LOX occurs prior to retinoic acid induced differentiation of pre-adipocytes to adipocytes50, Whichever mechanisms are involved, it is clear that while enhanced secretion of LOX precedes differentiation inhibition of LO activity, rather than any changes in of fibroblasts to myofibroblasts in cholestatic fibrosis51.In expression of the different forms of LOX, is required for both cases, however, the existence of a causal relation- the observed effects on chondrocyte differentiation. Both ship between LO activity and cell differentiation, as LOX and LOXL were detected in the cultures, but only their reported here for chondrocytes, was not demonstrated. activities, not their expression, were inhibited by BAPN. In In contrast, daily treatment with BAPN has been shown52 the specific context of tumor cell differentiation, LOX has to accelerate the recovery of mice after spinal cord injury, been identified as the product of the ras recision gene possibly by preventing unwanted modulation of ECM. The (rrg)41 whereby up-regulation of LOX gene expression results presented here suggest that this effect might be results in reversion of the ras induced phenotype, while due, at least in part, to BAPN induced changes in cell down-regulation using anti-sense LOX mRNA can be phenotype. Such effects of LO inhibition on the mainte- tumorigenic42. The tumorigenic phenotype is generally, nance of cell phenotype could be envisaged for other cell but not always, associated with a low level of expression of types interacting with an extracellular matrix, since this LOXs13,43e45. In contrast, inhibition of LO activity by BAPN ubiquitous enzyme is present in practically all solid has never been reported to induce a tumorigenic phenotype tissues. Osteoarthritis and Cartilage Vol. 13, No. 2 127

100 A 3. Benya PD, Schaffer JD. Dedifferentiated chondrocytes * * re-express the differentiated collagen phenotype when cultured in agarose gels. Cell 1982;30:215e24. 4. Tschan T, Bohme K, Conscience-Egli M, Zenke G, Winterhalter KH, Bruckner P. Autocrine or paracrine 10 transforming growth factor-beta modulates the phe- notype of chick embryo sternal chondrocytes in serum-free agarose culture. J Biol Chem 1993;268: 5156e61.

ratio +/- BAPN 1 5. Blaschke RJ, Howlett AR, Desprez P-Y, Petersen OW, Bissell MJ. Cell differentiation by extracellular matrix components. In: Ruoslahti E, Engvall E, Eds. Methods in Enzymology. San Diego: Academic Press 1994; Volume 245: 535e56. 0.1 6. Ramirez F, Rifkin DB. Cell signaling events: a view from II IX X XI the matrix. Matrix Biol 2003;22:101e7. 7. Farjanel J, Schurmann G, Bruckner P. Contacts with 100 B * * fibrils containing collagen I, but not collagens II, IX, and XI, can destabilize the cartilage phenotype of chondrocytes. Osteoarthritis Cartilage 2001;9(Suppl A):S55e63. 8. Kagan HM, Li W. Lysyl oxidase: properties, specificity, 10 and biological roles inside and outside of the cell. J Cell Biochem 2003;88:660e72. 9. Molnar J, Fong KS, He QP, Hayashi K, Kim Y, Fong SF, et al. Structural and functional diversity of lysyl oxidase ratio +/- BAPN 1 and the LOX-like proteins. Biochim Biophys Acta 2003;1647:220e4. 10. Eyre DR, Pietka T, Weis MA, Wu JJ. Covalent cross- linking of the NC1 domain of collagen type IX to collagen type II in cartilage. J Biol Chem 2004;279: 0.1 2568e74. II IX X XI 11. Bruckner P, Horler I, Mendler M, Houze Y, Winterhalter KH, Eich-Bender SG, et al. Induction and prevention of Fig. 7. BAPN selectively stimulates production of collagens IX and chondrocyte hypertrophy in culture. J Cell Biol 1989; XI. Amounts of collagens II, IX, X and XI produced by cranial cells e cultured with FCS (12 different batches) are expressed as ratios of 109:2537 45. percentages in the presence and absence of BAPN. (A) Total (cell 12. Fukai N, Apte SS, Olsen BR. Nonfibrillar collagens. In: layer C medium). (B) Cell-layer extracts. Data for days 7 and 14 Ruoslahti E, Engvall E, Eds. Methods in Enzymology. have been combined. To improve clarity, data are plotted on San Diego: Academic Press 1994; Volume 245: 3e28. a logarithmic scale. Data that are significantly different from the 13. Decitre M, Gleyzal C, Raccurt M, Peyrol S, Aubert- ratios for collagen II (P ! 0.05 by paired t test, after logarithmic Foucher E, Csiszar K, et al. Lysyl oxidase-like protein transformation) are indicated by an asterisk (*). localizes to sites of de novo fibrinogenesis in fibrosis and in the early stromal reaction of ductal breast e In conclusion, 2D culture with 10% serum could represent carcinomas. Lab Invest 1998;78:143 51. relatively hostile conditions, highly different from those 14. Borel A, Eichenberger D, Farjanel J, Kessler E, Gleyzal in situ. In such conditions, active LOXs, essential to produce C, Hulmes DJS, et al. Lysyl oxidase-like protein from normal cartilage, could lead to inappropriate effects. This bovine aorta. Isolation and maturation to an active form by bone morphogenetic protein-1. J Biol Chem latter property could be involved in some processes e occurring in pathological cartilage. LO activity appears to 2001;276:48944 9. be involved in the phenotype control of chondrocytes, giving 15. Seve S, Decitre M, Gleyzal C, Farjanel J, Sergeant A, a novel biological role to the LOX family. Ricard-Blum S, et al. Expression analysis of recombi- nant lysyl oxidase (LOX) in myofibroblastlike cells. Connect Tissue Res 2002;43:613e9. Appendix. Supplementary material 16. Shackleton DR, Hulmes DJS. An ultrafiltration assay for e Supplementary material associated with this article can lysyl oxidase. Anal Biochem 1990;185:359 62. be found, in the online version, at doi:10.1016/ 17. Trackman PC, Bedell-Hogan D, Tang J, Kagan HM. j.joca.2004.06.015. Post-translational glycosylation and proteolytic pro- cessing of a lysyl oxidase precursor. J Biol Chem 1992;267:8666e71. References 18. Watt FM. Effect of seeding density on stability of the differentiated phenotype of pig articular chondrocytes 1. Blaschke UK, Eikenberry EF, Hulmes DJS, Galla HJ, in culture. J Cell Sci 1988;89:373e8. Bruckner P. Collagen XI nucleates self-assembly and 19. Chiquet M. Regulation of extracellular matrix gene limits lateral growth of cartilage fibrils. J Biol Chem expression by mechanical stress. Matrix Biol 1999;18: 2000;275:10370e8. 417e26. 2. Cancedda R, Cancedda FD, Castagnola P. Chondro- 20. Chiquet M, Renedo AS, Huber F, Fluck M. How do cyte differentiation. Int Rev Cytol 1995;159:265e358. fibroblasts translate mechanical signals into changes in 128 J. Farjanel et al.: Chondrocyte phenotype and lysyl oxidase

extracellular matrix production? Matrix Biol 2003;22: factor and inactivates its mitogenic potential. J Cell 73e80. Biochem 2003;88:152e64. 21. Ronziere MC, Farjanel J, Freyria AM, Hartmann DJ, 38. Li W, Nellaiappan K, Strassmaier T, Graham L, Herbage D. Analysis of types I, II, III, IX and XI Thomas KM, Kagan HM. Localization and activity of collagens synthesized by fetal bovine chondrocytes in lysyl oxidase within nuclei of fibrogenic cells. Proc Natl high-density culture. Osteoarthritis Cartilage 1997;5: Acad Sci USA 1997;94:12817e22. 205e14. 39. Wakasaki H, Ooshima A. Immunohistochemical local- 22. Couchman JR, Woods A. Syndecan-4 and integrins: ization of lysyl oxidase with monoclonal antibodies. combinatorial signaling in cell adhesion. J Cell Sci Lab Invest 1990;63:377e84. 1999;112:3415e20. 40. Giampuzzi M, Oleggini R, Di Donato A. Demonstration of 23. San Antonio JD, Karnovsky MJ, Gay S, Sanderson RD, in vitro interaction between tumor suppressor lysyl Lander AD. Interactions of syndecan-1 and heparin with oxidase and histones H1 and H2: definition of the human collagens. Glycobiology 1994;4:327e32. regions involved. Biochim Biophys Acta 2003;1647: 24. Loeser RF, Sadiev S, Tan L, Goldring MB. Integrin 245e51. expression by primary and immortalized human 41. Mariani TJ, Trackman PC, Kagan HM, Eddy RL, Shows chondrocytes: evidence of a differential role for TB, Boyd CD, et al. The complete derived amino acid alpha1beta1 and alpha2beta1 integrins in mediating sequence of human lysyl oxidase and assignment of chondrocyte adhesion to types II and VI collagen. the gene to chromosome 5. Matrix 1992;12:242e8. Osteoarthritis Cartilage 2000;8:96e105. 42. Giampuzzi M, Botti G, Cilli M, Gusmano R, Borel A, 25. Vogel W. Discoidin domain receptors: structural rela- Sommer P, et al. Down-regulation of lysyl oxidase- tions and functional implications. FASEB J 1999; induced tumorigenic transformation in NRK-49F cells 13(Suppl):S77e82. characterized by constitutive activation of ras proto- 26. David G. Integral membrane heparan sulfate proteo- oncogene. J Biol Chem 2001;276:29226e32. glycans. FASEB J 1993;7:1023e30. 43. Hamalainen ER, Kemppainen R, Kuivaniemi H, Tromp 27. Salter DM, Godolphin JL, Gourlay MS, Lawson MF, G, Vaheri A, Pihlajaniemi T, et al. Quantitative poly- Hughes DE, Dunne E. Analysis of human articular merase chain reaction of lysyl oxidase mRNA in chondrocyte CD44 isoform expression and function in malignantly transformed human cell lines demon- health and disease. J Pathol 1996;179:396e402. strates that their low lysyl oxidase activity is due to 28. Burg MA, Tillet E, Timpl R, Stallcup WB. Binding of the low quantities of its mRNA and low levels of NG2 to type VI collagen and other transcription of the respective gene. J Biol Chem extracellular matrix molecules. J Biol Chem 1996; 1995;270:21590e3. 271:26110e6. 44. Peyrol S, Galateau-Salle F, Raccurt M, Gleyzal C, 29. Mollenhauer J, Von der Mark K. Isolation and charac- Sommer P. Selective expression of lysyl oxidase terization of a collagen-binding glycoprotein from (LOX) in the stromal reactions of broncho-pulmonary chondrocyte membranes. EMBO J 1983;2:45e50. carcinomas. Histol Histopathol 2000;15:1127e35. 30. Aeschlimann D, Wetterwald A, Fleisch H, Paulsson M. 45. Kirschmann DA, Seftor EA, Fong SF, Nieva DR, Expression of tissue transglutaminase in skeletal Sullivan CM, Edwards EM, et al. A molecular role for tissues correlates with events of terminal differentiation lysyl oxidase in breast cancer invasion. Cancer Res of chondrocytes. J Cell Biol 1993;120:1461e70. 2002;62:4478e83. 31. Aeschlimann D, Thomazy V. Protein crosslinking in 46. Palamakumbura AH, Jeay S, Guo Y, Pischon N, Sommer assembly and remodelling of extracellular matrices: P, Sonenshein GE, et al. The pro-peptide domain of the role of transglutaminases. Connect Tissue Res lysyl oxidase induces phenotypic reversion of Ras- 2000;41:1e27. transformed cells. J Biol Chem 2004;62:40593e600. 32. Valcourt U, Gouttenoire J, Moustakas A, Herbage D, 47. Lee SL, Piez KA. Type II collagen from lathyritic rat Mallein-Gerin F. Functions of transforming growth chondrosarcoma: preparation and in vitro fibril forma- factor-beta family type I receptors and SMAD proteins tion. Collagen Rel Res 1983;3:89e103. in the hypertrophic maturation and osteoblastic differ- 48. Hansen U, Bruckner P. Macromolecular specificity of entiation of chondrocytes. J Biol Chem 2002;277: collagen fibrillogenesis: fibrils of collagens I and XI 33545e58. contain a heterotypic alloyed core and a collagen I 33. Feres EJ, Choi YJ, Han XY, Takala TES, Trackman PC. sheath. J Biol Chem 2003;278:37352e9. Pre- and post-translational regulation of lysyl oxidase by 49. Hong HH, Pischon N, Santana RB, Palamakumbura transforming growth factor-beta 1 in osteoblastic AH, Chase HB, Gantz D, et al. A role for lysyl oxidase MC3T3-E1 cells. J Biol Chem 1995;270:30797e803. regulation in the control of normal collagen deposition 34. Palamakumbura AH, Sommer P, Trackman PC. Auto- in differentiating osteoblast cultures. J Cell Physiol crine growth factor regulation of lysyl oxidase expres- 2004;200:53e62. sion in transformed fibroblasts. J Biol Chem 2003;278: 50. Dimaculangan DD, Chawla A, Boak A, Kagan HM, 30781e7. Lazar MA. Retinoic acid prevents downregulation of 35. Giampuzzi M, Oleggini R, Di Donato A. Altered adhesion RAS recision gene/lysyl oxidase early in adipocyte features and signal transduction in NRK-49F cells differentiation. Differentiation 1994;58:47e52. transformed by down-regulation of lysyl oxidase. Bio- 51. Desmouliere A, Darby I, Costa AMA, Raccurt M, chim Biophys Acta 2003;1647:239e44. Tuchweber B, Sommer P, et al. Extracellular matrix 36. Pischon N, Darbois LM, Palamakumbura AH, Kessler deposition, lysyl oxidase expression, and myofibro- E, Trackman PC. Regulation of collagen deposition blastic differentiation during the initial stages of chole- and lysyl oxidase by tumor necrosis factor-a in static fibrosis in the rat. Lab Invest 1997;76:765e78. osteoblasts. J Biol Chem 2004;279:30060e5. 52. Gilad GM, Gilad VH. Beta-aminopropionitrile treatment 37. Li W, Nugent MA, Zhao Y, Chau AN, Li SJ, Chou IN, can accelerate recovery of mice after spinal cord et al. Lysyl oxidase oxidizes basic fibroblast growth injury. Eur J Pharmacol 2001;430:69e72.