Demineralized Bone Alters Expression of Wnt Network Components During Chondroinduction of Post-Natal ﬁbroblasts Karen E

International Cartilage Repair Society

Demineralized bone alters expression of Wnt network components during chondroinduction of post-natal ﬁbroblasts Karen E. Yates Ph.D* Department of Orthopedic Surgery, Brigham and Women’s Hospital, and Skeletal Biology Research Center, Massachusetts General Hospital, Boston, USA

Summary

Objective: The Wnt family of secreted proteins, their receptors (Fzd proteins) and antagonists (secreted Fzd-related proteins, or Sfrp) regulate chondrocyte differentiation and chrondrogenesis during embryonic development. Here, the hypothesis that the Wnt regulatory network contributes to chondrocyte differentiation of post-natal cells was tested in an in vitro model of chondroinduction by demineralized bone powder (DBP). Design: Human dermal fibroblasts (hDFs) were cultured in porous, three-dimensional (3D) collagen sponges with or without chondroinduc- tive DBP. In some experiments, lithium chloride (LiCl), an agonist of the Wnt/-catenin signaling pathway, was added to the culture media. Sponges were cultured for intervals (0.5–21 days) before processing for molecular, histologic, and biochemical analyses. Expression of wnt, fzd, and sfrp genes was characterized by semi-quantitative RT-PCR. Fibroblasts’ contacts with DBP were documented by histology. Accumulation of proteoglycan in extracellular matrix was evaluated by histology (metachromasia in toluidine blue-stained sections) and quantitative immunoassay (chondroitin 4-sulfate ELISA). Results: Expression of 15 wnt, fzd, and sfrp family members was detected in hDFs by RT-PCR. A subset of those genes (wnt2b, wnt5b, wnt10b, fzd6, fzd7) showed altered expression in hDFs exposed to DBP for 3 days. wnt and fzd gene expression was not altered before hDFs contacted the DBP within the collagen sponge. Human DFs cultured in plain collagen sponges and treated with LiCl accumulated significantly more metachromatic matrix than NaCl-treated controls on day 10, and showed a trend towards increased matrix chondroitin-4 sulfate content. Conclusions: These data suggest that changes in Wnt signaling contribute to chondroinduction of post-natal fibroblasts by DBP. This is the first evidence that Wnt components, which are essential regulators of pre-natal chondrocyte differentiation, may also influence post-natal chondrocyte differentiation induced by DBP. © 2004 OsteoArthritis Research Society International. Published by Elsevier Ltd. All rights reserved.

Key words: Wnt, Demineralized bone, Dermal ﬁbroblast, Chondroinduction.

Introduction particles approximates the architecture of DBP in in vivo implants9. With hDFs used as target cells, the DBP/ The wnt gene family encodes secreted proteins that inter- collagen sponge system models an early step in DBP- act with cellular receptors encoded by the fzd gene family. induced endochondral osteogenesis: the phenotypic Together with secreted, fzd-related proteins (encoded by conversion of mesenchymal cells into chondroblasts10,11. sfrp genes) that act as Wnt antagonists, these proteins Seven days after hDFs are seeded onto DBP/collagen form a regulatory network that profoundly influences sponges, cells that have contacted the DBP show accumu- chondrocyte differentiation and chondrogenesis during em- 12 1–8 lation of metachromatic extracellular matrix and gene bryonic skeletal development . Whether Wnt, Fzd, and expression of the cartilage matrix components, collagen Sfrp (hereafter referred to as Wnt components) similarly type II and aggrecan13. Prior to the expression of cartilage influence post-natal chondrocyte differentiation and chon- phenotypic genes, DBP induces shifts in expression of drogenesis, however, is not well understood. several functional classes of genes in hDFs, including In an experimental system of post-natal chondrocyte matrix proteins, cytoskeletal elements, protein synthesis differentiation, human dermal fibroblasts (hDFs) are cul- machinery, signal transduction proteins, and peptide tured with demineralized bone powder (DBP) in a porous growth/differentiation factors14,15. For some genes, DBP- collagen sponge9. Demineralized bone is used clinically induced changes in expression are transient and their because of its ability to induce skeletal tissue by an mRNAs return to control levels by day 7, when the chondro- endochondral mechanism10,11.Inthein vitro culture de- cyte phenotype is evident14. Altered expression of other vice, a portion of DBP is held in place between two thin genes is sustained during accumulation of cartilage-like layers of collagen. This arrangement of demineralized bone matrix (chondrogenesis) on days 7–21. These patterns of altered gene expression begin to define cellular stages *Address correspondence and reprint requests to: Karen E. of chondroblast differentiation and chondrogenesis in the Yates Ph.D., Orthopedic Research, Brigham and Women’s Hospi- 15 tal, 75 Francis Street, Boston MA 02115, USA. Tel: 617 732-6943; collagen sponge culture system . Fax: 617 732-6705; E-mail: [email protected] The importance of Wnt components in pre-natal Received 22 April 2003; revision accepted 24 February 2004. chondrocyte differentiation suggested the hypothesis that

497 498 K. E. Yates: Alteration of Wnt components by demineralized bone this network also regulates post-natal chondrocyte differen- A preliminary experiment was performed to determine an tiation. The objective of this study was to determine appropriate concentration of LiCl for sponge experiments. whether Wnt signaling contributes to chondroinduction by Monolayer cultures of hDFs exposed to 5 mM LiCl or DBP. Therefore, the DBP/collagen sponge system was control NaCl for 10 days were viable. Greater concen- used to characterize Wnt component gene expression trations of LiCl (10 or 20 mM) were toxic. Therefore, 24 h during chondroblast differentiation and chondrogenesis. after sponges were seeded with hDFs, 4 µl of LiCl or NaCl The contribution of one aspect of Wnt signal transduction stock solution (5 M) was added to each well and mixed into to post-natal chondrocyte differentiation was assessed the culture media by pipetting. Fresh media containing by exposing hDFs in collagen sponges to lithium chloride 5 mM LiCl or NaCl was added 2 days later when the (LiCl), an agonist of the Wnt/-catenin signaling sponges were removed from the seeding chambers, and pathway16. every 3–4 days thereafter. Ten days after seeding, the sponges were harvested for histology and proteoglycan extraction. Methods

CELLS RNA ISOLATION Human dermal fibroblasts were isolated by outgrowth Total RNA was prepared from monolayer cultures by from skin explants (neonatal foreskins) and cultured in lysing cells in Trizol reagent (Invitrogen, Carlsbad, CA). monolayer to passage #7 before use12. Basal culture Groups of control collagen or DBP/collagen sponges were pooled and homogenized in Trizol reagent (1 ml per media for all hDF experiments consisted of Dulbecco’s 14 MEM, 10% fetal bovine serum, 100 U/ml penicillin, and 100 sponge) . RNA quality was evaluated by absorbance µg/ml streptomycin. Os Te (osteogenesis imperfecta em- readings at 260 and 280 nm, and by electrophoresis on bryonic human dermal fibroblasts, #CRL-1262) and MG63 formaldehyde-agarose gels. cells (human osteosarcoma, #CRL-1427) were purchased and cultured as recommended by the supplier (American CDNA SYNTHESIS Type Culture Collection, Manassas, VA). Total RNA was treated with DNase I (Roche Molecular Biochemicals, USA) to eliminate any contaminating ge- CHONDROINDUCTION SYSTEM nomic DNA. Aliquots of 5 µg of DNase-treated RNA were Demineralized bone powder was prepared from rat long used in random hexamer-primed cDNA synthesis reactions bones, as described17. Three-dimensional DBP/collagen or (100 µl total volume) according to the manufacturer’s in- plain collagen sponges were prepared as described9.In structions (Invitrogen). One µl of cDNA (the equivalent of brief, to prepare bilaminate sponges, 120 µl of 0.5% type I 50 ng total RNA) was used in each PCR. collagen solution (Cellagen PC-5, ICN Biomedicals, Costa Mesa, CA) was neutralized with 0.01 M HEPES (pH 7.4) PCR PRIMERS and 0.01 M NaHCO3, poured into a mold, and frozen. A spacer of moistened paper was placed over the frozen Primers specific for the housekeeping gene collagen layer and a second aliquot of collagen (130 µl) glyceraldehyde-3-phosphage dehydrogenase (G3PDH) was added. The construct was frozen, lyophilized and were purchased (Clontech, Palo Alto, CA). Published irradiated with ultraviolet light (3 h per side). The spacer primers were used for wnt2b, wnt10b, fzd2, fzd518 and paper was removed and a 3-mg portion of DBP was placed wnt16a, wnt16b19 (Table I). Additional primers (Table I) between the layers of bilaminate sponges. To prepare plain were designed using the Primer3 program20 (code avail- collagen sponges, a single layer of collagen (250 µl) was able at http://www-genome.wi.mit.edu/genome_software/ poured into a mold, frozen, lyophilized and irradiated with other/primer3/html). A previous study showed that a ultraviolet light. specific wnt5a mRNA that has a unique 3′ untranslated region (UTR) is induced in hDFs cultured with DBP21. Because 3′ UTRs are incompletely catalogued, PCR COLLAGEN SPONGE SEEDING AND CULTURE CONDITIONS primers were designed to target protein-coding regions. For cell seeding, each sponge was positioned in a Primers were tested in control PCRs with three different seeding chamber9 and placed into a well of a 12-well cDNAs (Os Te, MG63, or hDF) and conditions for amplifi- culture plate. Human dermal fibroblasts were harvested cation (Table I). For some primer sets, no product was from monolayer cultures and deposited onto the top of the detected under any of those conditions (Table I)orinPCRs sponges (106 cells in 50 µl of basal media). An additional that used cDNA from hDFs cultured in DBP/collagen monolayer culture of cells was harvested on the same day sponges for 3 days. for RNA isolation and measurement of baseline mRNA levels. The sponges were incubated in a humidified cham- SEMI-QUANTITATIVE RT-PCR ber for 1 h at 37°C. A 50-µl aliquot of basal media was added to the top of the sponges, which were returned to the PCRs for each Wnt pathway component were performed incubator for 2 h. Four ml of media was added to each well on the set of cDNA samples from each experiment (DBP/ and the seeding chambers were tilted to remove any air collagen sponges, control collagen sponges, and mono- bubbles from beneath the sponges. Three days after cell layer culture). PCR products were subjected to electro- seeding, the sponges were removed from the seeding phoresis through 2% agarose gels. Photographs of chambers and cultured in 6-well plates (2–4 sponges per ethidium bromide-stained gels were scanned with an well). Media were changed every 3–4 days. Sponges were Epson 1200s Scanner. TIF-formatted images were ana- then harvested at intervals for molecular, histologic, and lyzed with Scion Image for Windows software (Scion biochemical analyses. Corporation, Frederick, MD). PCR product pixel density Osteoarthritis and Cartilage Vol. 12, No. 6 499

Table I PCR primers for Wnt component genes Gene (Accession No.) Primers Product (bp) Amplification conditions1 Wnt1 (NM_005430) CTGCAGCGACAACATTGACT 399 - GCTGTACGTGCAGAAGTTGG Wnt2 (XM_027186) AAAGGAAAGGATGCCAGAGC 398 60, 35 cycles CCCACAGCACATGACTTCAC Wnt2b (HSWNT13) AAGATGGTGCCAACTTCACCG 320 55, 35 cycles CTGCCTTCTTGGGGGCTTTGC Wnt3 (XM_032187) GACCACATGCACCTCAAATG 401 55, 40 cycles GATGCAGTGGCATTTTTCCT Wnt3a (NM_033131) CAAGATTGGCATCCAGGAGT 289 - GAGACACCATCCCACCAAAC Wnt4 (NM_030761) ACCTGGAAGTCATGGACTCG 399 60, 35 cycles GCCTCATTGTTGTGGAGGTT Wnt5b (NM_032642) CTGTGCCAATTGTACCAGGA 396 60, 35 cycles CCTGCTCCTCTGTCCTTTG Wnt6 (AB059570) AGAAGCTGCCTCCATTTCG 386 - GTCACAGGCAGAGGCTGAG Wnt7a (NM_004624) GAGAAGCAAGGCCAGTACCA 424 60, 35 cycles TAGTTGGGCGACTTCTCGAT Wnt7b (AB062766) GAGCCAACATCATCTGCAAC 391 60, 35 cycles GGAGAAGTCGATGCCGTAAC Wnt8a (NM_031933) GAACTGCCCTGAAAATGCTC 544 - GCTAGGAAGGAAGGCCTCAG Wnt8b (NM_003393) CTGGTGCCCAGAGTGGTATT 408 - TTACACGTGCGTTTCATGGT Wnt10a (NM_025216) CCCAATGACATTCTGGACCT 410 60, 37 cycles TAAGCGGTGCAGCTTCCTAC Wnt10b (XM_047556) GAATGCGAATCCACAACAACAG 195 60, 35 cycles TTGCGGTTGTGGGTATCAATGAA Wnt11 (NM_004626) CGTGTGCTATGGCATCAAGT 509 - CGCATCAGTTTATTGGCTTG Wnt14 (XM_047556) GGGTGTGAAGGTGATCAAGG 396 55, 35 cycles CACCCGGCTCTGTGTGTTAT Wnt15 (NM_003396) CACCTGAAGCAGTGTGACCT 498 - CTGATACGCCATGGCACTTA Wnt16a (NM_016087) CAGAAAGATGGAAAGGCACC 271 - ATCATGCAGTTCCATCTCTC Wnt16b (NM_057168) TGCTCGTGCTGTTCCCCTAC 225 60, 35 cycles ATCATGCAGTTCCATCTCTC Fzd1 (NM_003505) TACGTACCTGGTGGACATGC 497 60, 35 cycles AAGGACGTGCCGATAAACAG Fzd2 (NM_001466) CTAGCGCCGCTCTTCGTGTACCTG 386 60, 35 cycles CAGCGTCTTGCCCGACCAGATCC Fzd3 (NM_017412) GGATCGGTGTTTTCAGCATT 501 60, 35 cycles CCGTGGTAGCTGCTCACTTT Fzd4 (NM_003506) AACTTTCACACCGCTCATCC 602 55, 35 cycles GATATCCTTTCCCGGCCTAC Fzd5 (NM_003468) TTCATGTGCCTGGTGGTGGGC 235 - TACACGTGCGACAGGGACACC Fzd6 (NM_003506) AAAGCAAAAGCTCGACCAGA 595 55, 35 cycles GCCAGGCCAGTGTCAGTAAT Fzd7 (NM_003507) CCAACGGCCTGATGTACTTT 605 60, 35 cycles GCAAGAAGGACGTGCCTATG Fzd8 (NM_031866) CACCGTCTCCACCTTCCTTA 621 - ATGGTGCCGATGAAGAGGTA Fzd10 (NM_007197) AGCCAACAGCAGCTACTTCC 498 - CACCACCAGCAGCATAAAGA Sfrp1 (NM_003012) CCCCTGCTCAACAAGAACTG 508 55, 35 cycles AAGTGGTGGCTGAGGTTGTC Sfrp2 (NM_050625) GCATCGAATACCAGAACATGC 501 - TTGCTCTTGGTCTCCAGGAT Sfrp4 (NM_003014) TTCTTCTGTGCCATGTACGA 490 55, 40 cycles TTGTAATGAGCGGGACTTGA Sfrp5 (NM_003015) CTGGACAACGACCTCTGCAT 440 - GCAGGGGTAGGAGAACATGA Frzb (NM_001463) CCCATCAACCCCTGTAAGTC 394 - ACAGTGTCCCGTGGAATGTT 1 Annealing temperature (°C) and number of amplification cycles. Dashes indicate those primer sets for which no product was obtained from control OsTe, MG63, or hDF cDNAS. 500 K. E. Yates: Alteration of Wnt components by demineralized bone was measured within a user-defined rectangle that circum- incubations, the wells were rinsed four times with scribed the visible band. Background pixel density was phosphate-buffered saline containing 0.05% Tween-20 measured in an adjacent area in the same lane and was (PBS-T). Incubations with chondroitinase ABC (Seikagaku subtracted from the PCR product pixel density. Corporation, Associates of Cape Cod, East Falmouth, MA), A correction factor was calculated for each cDNA by SuperBlock (Pierce Biochemicals, Milwaukee, WI), anti- dividing its G3PDH pixel density by the minimum G3PDH chondroitin-4 sulfate antibody (1:3000 dilution, Seikagaku value in that set of samples. Wnt component pixel densities Corporation), goat anti-mouse IgG+M (Pierce), and were divided by the correction factor and the ratio (r)of streptavidin/alkaline phosphatase (Invitrogen) were per- DBP/collagen to control collagen was calculated. A pre- formed as described12. One hundred µl of 4 mg/ml para- vious study showed that G3PDH mRNA levels are not nitrophenyl phosphate (Invitrogen) in buffer containing greatly changed by DBP in hDFs cultured in collagen 22 mM sodium carbonate, 28 mM sodium bicarbonate, and sponges (r values for 18S:G3PDH were <1.4 in all 1 mM MgCl2, pH 9.8 was added to the wells and the plates samples)15. Therefore, in this study, genes that were were incubated overnight at 25°C. One hundred µl of 1 N changed by DBP on day 3 were defined as those for which NaOH was added to the wells and the plates were read at r was >1.4 or ≤0.6 to account for the known variance 405 nm. All experimental values fell within the linear range inherent in this experimental system15. For statistical analy- of the standard curve. Average values from the three sis, r values were then transformed to a percent difference replicate wells were used to calculate chondroitin-4 sulfate (z1−rz×100). content (pg/sponge).

HISTOLOGY STATISTICAL ANALYSES Sponges were fixed in 2% paraformaldehyde, 0.1 M The Mann–Whitney rank sum test was performed with sodium cacodylate for 24 h. Sponges from each exper- Sigma Stat version 2.03 (SPSS Inc., San Raphael, CA). iment were processed as a group and embedded in paraffin The Mann–Whitney U statistic and one-way ANOVA with or methylmethacrylate (JB-4), sectioned, and stained. To Bonferroni multiple comparisons were performed with In- evaluate hDF’s migration into collagen sponges and inter- Stat Version 3.05 (GraphPad Software Inc., San Diego, action with DBP, 4 µm-thick, paraffin sections were stained CA). with Safranin O. To evaluate metachromatic extracellular matrix, 20 µm-thick JB-4 sections were stained with toluid- Results ine blue (pH 4.0)9. Two independent experiments were performed to evalu- HUMAN DERMAL FIBROBLASTS CULTURED WITH DBP FOR 3 DAYS ate metachromatic extracellular matrix in LiCl- or NaCl- SHOW ALTERED EXPRESSION OF WNT COMPONENTS treated sponges (N=3 per group in each experiment). Wnt component mRNA levels were measured by semi- Toluidine blue-stained slides were viewed in random order quantitative RT-PCR. Altered expression of five genes (i.e. with their identifiers masked. The slides were grouped relative expression ≥1.4 or ≤0.6) was detected in hDFs according to the amount of matrix present and assigned a cultured in collagen sponges with DBP for 3 days. Human matrix score (0=low; 1=moderate; 2=abundant). Matrix DFs cultured in DBP/collagen sponges expressed 1.8-fold scores from the two experiments were pooled for statistical more wnt5b and 1.7-fold more wnt10b mRNAs than hDFs analysis. cultured in control collagen sponges, whereas the wnt2b mRNA level was 40% of control, fzd6 was 20% of control, and fzd7 was 60% of control (Fig. 1, upper panel). Altered PROTEOGLYCAN EXTRACTION AND CHONDROITIN-4 SULFATE expression was not found on day 3 for the 10 other Wnt ELISA component mRNAs that were detected in hDFs (wnt2, Proteoglycan was extracted from cultured sponges as wnt3, wnt4, wnt7b, fzd1, fzd2, fzd3, fzd4, sfrp1, and sfrp4) described12. Each sponge was incubated in 1 ml of extrac- (Fig. 1, lower panel). tion buffer (4 M guanidine-HCl, 0.05 M sodium acetate, The results from day 3 showed that gene expression of 0.01 M sodium EDTA, 0.1 M aminohexanoic acid, 0.005 M some Wnt components was altered in DBP/collagen benzamidine HCl, pH 5.8) containing protease inhibitors sponges during chondroblast differentiation. To determine (Complete Tablets, Roche Molecular Biochemicals, USA) whether altered expression of Wnt components persisted at 4°C with shaking. After 48 h, the sponges were centri- during chondrogenesis, mRNA levels were measured on fuged for 5 min at 10,000×g. Three hundred µl of super- days 7, 14, and 21. Those Wnt components whose expres- natant was removed and diluted with an equal volume of sion was found to be altered on day 3 were analyzed as a water. Proteoglycan was precipitated with three volumes of group. Some of those genes also showed altered expres- cold absolute ethanol containing 1.3% potassium acetate. sion on days 7 and 14 (Fig. 1, upper panel). Expression The ethanol precipitation was repeated and the samples levels appeared to return to control levels (i.e. relative were dissolved in 300 µl of carbonate buffer (0.035 M expression values were close to 1) on day 21. mRNA levels

NaHCO3, 0.018 M Na2CO3, pH 9.8). were also measured on days 7, 14, and 21 for the 10 Wnt For measurement of chondroitin-4 sulfate content by components whose expression was not altered on day 3. ELISA, each sample was diluted 1:5 in carbonate buffer. A Gene expression for that group of Wnt components was not standard curve was prepared by diluting proteoglycan altered on day 7 (mean percent difference in DBP/collagen monomer from bovine nasal cartilage (ICN Biochemicals, vs control sponges 22%±16%) or day 21 (23%±13%) Irvine, CA) in carbonate buffer (2-fold serial dilutions begin- (P>0.05 compared to day 3, one-way ANOVA with Bonfer- ning with 200 pg). One hundred µl of each sample was roni multiple comparisons (Fig. 1, lower panel). A small but aliquoted into each of three replicate 96-well ELISA plates statistically signiﬁcant change in expression was found for (Pro-Bind™, Becton Dickinson, Franklin Lakes, NJ), which this group of genes on day 14 (mean percent differences were incubated overnight at 4°C. After that and subsequent 54%±28%, P=0.001 compared to day 3) (Fig. 1, lower Osteoarthritis and Cartilage Vol. 12, No. 6 501

all genes showed similar expression levels in DBP/collagen and control collagen sponges 12 h after seeding, before hDFs interacted with DBP (relative expression values 0.9– 1.2; note the overlap or proximity of open and closed symbols at the 12 h time point). A decrease in wnt2b gene expression (50% of control level) was detected 24 h after hDFs were seeded onto DBP/collagen sponges. At 72 h after cell seeding, wnt2b mRNA levels were 35% of control. This was a ∼50% decrease from the initial 12 h level in the DBP/collagen sponge. In contrast, in control collagen sponges, wnt2b mRNA levels at 72 h were 60% greater than the 12 h level. Therefore, both decreased expression in the DBP/collagen sponge and increased expression in the control collagen sponge contributed to the difference in relative expression for this gene at 72 h. Changes in other Wnt component mRNAs were less striking at early time points (Fig. 2B). Wnt10b mRNA levels decreased slightly in both DBP/collagen and control collagen sponges. At 72 h, hDFs in DBP/collagen sponges expressed 50% more wnt10b than control. Fzd7 expression remained fairly constant in control collagen sponges; Fig. 1. Wnt component gene expression in hDFs cultured in the lower relative expression (64% of control at 72 h) was collagen sponges. Cells were cultured in DBP/collagen or control due to decreased levels in DBP/collagen sponges. In this collagen sponges for 3 days (12 sponges per group) or 7, 14, 21 experiment, wnt5b and fzd6 relative expression at 72 h days (3 sponges per group). Groups of sponges were pooled at (Fig. 2) were not in concordance with the data from the each time point for RNA isolation. Gene expression levels were 21-day experiment (Fig. 1). Differences in the measured measured by semi-quantitative RT-PCR. The relative expression for each gene was calculated as the ratio of DBP/collagen to gene expression levels at early time points may reflect control collagen sponges (DBP:Collagen). Relative expression differences in the donor fibroblasts used for induction. To values were plotted according to whether they met the criteria for test this hypothesis, ‘baseline’ mRNA levels of the five Wnt change on day 3 (DBP:Collagen ratio ≥1.4 or ≤0.6; upper panel) or components that were found to be changed in the 21-day not (lower panel). experiment were measured in monolayer cultures of the hDFs that were used in each experiment (Fig. 3). Strikingly, genes that showed similar results in the 21-day and 72-h experiments had equivalent baseline mRNA levels (wnt2b, panel). These results show that gene expression of a wnt10b, and fzd7), whereas genes that showed disparate subset of Wnt components is altered during chondroinduc- results had >2-fold differences in baseline mRNA levels tion (day 3) and early chondrogenesis (days 7–14) in the (wnt5b and fzd6). DBP/collagen sponge.

ACTIVATION OF THE -CATENIN WNT SIGNALING PATHWAY IN THE WNT COMPONENT EXPRESSION IS ALTERED AFTER HDFS ABSENCE OF DBP INCREASES PROTEOGLYCAN SYNTHESIS BY INTERACT WITH DBP HDFS CULTURED IN COLLAGEN SPONGES A kinetic analysis was performed to characterize the DBPs’ effects on Wnt component gene expression sug- temporal relationships between fibroblasts’ interactions gested the hypothesis that Wnt signaling mediates with the inductive DBP and changes in Wnt component chondrocyte differentiation in this system. Wnt signal trans- expression. Human DFs were seeded onto DBP/collagen duction is complex and at least three pathways are and control collagen sponges and harvested at 12 h inter- known22–24. Several of the Wnt components that were vals. Histologic analysis of control collagen sponges found to be affected by DBP have been implicated in showed that hDFs had migrated into the top third of the signaling via the -catenin pathway25–27. Therefore, the sponge at 12 h, into the next third at 24 h, and had reached contribution of -catenin signaling to post-natal chondro- the bottom of the sponge by 36 h (data not shown). In cyte differentiation was assessed by exposing hDFs in plain DBP/collagen sponges, a few hDFs were attached to DBP collagen sponges (without DBP) to LiCl, an agonist of this contained in the center of the sponge at 12 h (Fig. 2A). At pathway16. 24 h, many more cells were visible attached to and sur- LiCl or NaCl was added to cultures (5 mM final concen- rounding the DBP (Fig. 2A). Many hDFs were attached to tration) 24 h after sponges were seeded with hDFs. This the fibers of the collagen layers that contained the packet of time point was selected because the gene expression DBP. Cells were similarly distributed at 12 h intervals up to analyses showed DBP-altered expression of Wnt compo- 72 h (not shown). DBP/collagen sponges cultured for 7 nents was first evident at that time (Fig. 2). Sponges were days showed metachromatic extracellular matrix (not harvested 10 days after seeding, when accumulation of shown), confirming chondroinduction of the hDFs used in sulfated glycosaminoglycans in DBP/collagen sponges this experiment. In sum, hDFs rapidly migrated into DBP/ reaches a plateau12. Histologic analysis showed that hDFs collagen and control collagen sponges. Abundant fibro- cultured in NaCl-treated sponges were surrounded by low blasts’ contacts with DBP were evident by 24 h after or moderate amounts of metachromatic matrix (Fig. 4A, seeding. Table II). In contrast, hDFs cultured in LiCl-treated sponges Gene expression levels for those Wnt components that were surrounded by significantly more metachromatix ma- were changed on day 3 were measured at earlier intervals, trix (Table II; P=0.041, Mann–Whitney rank sum test). The also by semi-quantitative RT-PCR (Fig. 2B). As expected, LiCl effect on metachromatic matrix was reproduced in an 502 K. E. Yates: Alteration of Wnt components by demineralized bone

Fig. 2. Analysis of temporal events in hDFs’ interactions with DBP and changes in Wnt component gene expression. Human DFs were seeded onto DBP/collagen and control collagen sponges. Sponges were removed from culture at 12 h intervals for histologic and gene expression analyses. (A) Photomicrographs (4 µm-thick Safranin O-stained sections) of DBP/collagen sponges at 12 and 24 h after cell seeding. Part of the upper layer of collagen is visible at the top of each image. Arrowheads indicate cells that have migrated into the DBP contained in the center of the sponge. (B) Expression kinetics of Wnt components in hDFs cultured in sponges. Cells were seeded onto DBP/collagen and control collagen sponges. Sponges were cultured and harvested at 12-h intervals (3 sponges per group) and pooled for measurement of Wnt component mRNA levels by semi-quantitative RT-PCR. independent experiment that used cells from a different poor capacity for repair. A better understanding of the donor (Table II). Quantitative analysis of matrix proteogly- cellular mechanisms that regulate chondrocyte differenti- can content (Fig. 4B) showed a trend towards increased ation may aid in designing therapies for damaged cartilage. chondroitin 4-sulfate content in LiCl-treated sponges In this study, the potential for the Wnt signaling pathway to (median of 115 pg/sponge vs 93 pg/sponge for NaCl regulate post-natal chondrocyte differentiation was as- sponges, P=0.14, Mann–Whitney U statistic). sessed in an in vitro model of chondroinduction by DBP. Altered expression several Wnt components was found during the cellular stages of chondroblast differentiation Discussion and chondrogenesis induced by DBP. Activation of Wnt signaling by the -catenin pathway agonist, LiCl, was Strategies to stimulate cartilage synthesis and chondro- sufficient to increase proteoglycan synthesis by hDFs cul- cyte differentiation are desired because cartilage has a tured in collagen sponges. These results implicate the Wnt Osteoarthritis and Cartilage Vol. 12, No. 6 503

Fig. 3. Baseline mRNA levels of Wnt components in hDFs used for collagen sponge experiments. Gene expression levels in monolayer cultures were measured by semi-quantitative RT-PCR.

regulatory network as mediators of post-natal chondroinduction of dermal fibroblasts by demineralized bone. Some of the Wnt components that were found to be changed in this study are known to regulate chondrogenesis during embryonic skeletal development. In the chick, chondrogenesis in embryonic mesenchymal cells is enhanced by ectopic expression of wnt5b6 and inhibited by fzd75,6,28. The DBP-induced changes in expression of these Wnt component genes (decrease in fzd7 and increase in wnt5b) are consistent with these developmental B roles. In contrast, wnt2b, which is required for limb development in zebrafish29, was found to be downregulated by DBP in the collagen sponge system. DBP-induced changes in expression of other Wnt components that regulate embryonic skeletal development, such as wnt111, wnt142, and frzb8 were not detected in this study. Wnt regulation of post-natal chondrocyte differentiation may differ from the pre-natal process because the target cells are different (undifferentiated mesenchyme vs dermal fibroblast). Sup- porting that hypothesis, a different set of Wnt components’ expression is altered occur during chondrogenesis in an- other experimental system: human bone marrow stromal cells induced with TGF- show decreased fzd2 and increased wnt11 mRNAs30. Wnt signal transduction mechanisms are complex and at least three pathways are known: the -catenin pathway22, the planar cell polarity pathway31, and the calcium path- way24,32. The -catenin pathway has been well studied. Fig. 4. Effects of Wnt agonist (5 mM LiCl) on extracellular matrix Cytosolic -catenin is marked for proteosomal degradation produced by human dermal fibroblasts cultured in plain collagen via phosphorylation by glycogen synthase kinase 3 (GSK- sponges (day 10). (A) Photomicrographs of representative NaCl- accumulates in the cytosol and is subsequently trans- or LiCl-treated sponges (20 µm-thick Toluidine blue-stained sec- located to the nucleus and interacts with transcriptional tions). Human DFs are darkly stained. Empty pores within the 16 sponges are indicated (p). The matrix scores for these sponges machinery. LiCl inhibits GSK-3 activity , thereby mimick- were 1 (NaCl) and 2 (LiCl). (B) Proteoglycan content in collagen ing Wnt signaling via the -catenin pathway. sponges treated with LiCl (N=8) or NaCl (N=8). Solid dashes Several of the Wnt components whose expression was indicate the median chondroitin 4-sulfate content (pg per sponge) found to be altered by DBP in this study activate signal for each group. transduction via the -catenin pathway25–27. Some of the DBP-induced changes (i.e. increased wnt5b) suggested that DBP activates the -catenin pathway in target hDFs, whereas others (decreased wnt2b) suggested the oppo- system, hDFs were cultured in collagen sponges and site. Moreover, the Xenopus Frizzled 7 protein has been exposed to the Wnt/-catenin agonist, LiCl. Lithium ap- shown to activate both the -catenin and calcium path- peared to promote differentiation of hDFs although some ways25. Therefore, to determine the overall effect of variability in response was observed. Wnt proteins also -catenin signaling on chondrogenesis in this experimental affect proteoglycan synthesis in chondrogenic precursor 504 K. E. Yates: Alteration of Wnt components by demineralized bone

Table II bone. These results are a step towards defining cellular Histologic scores1 of extracellular matrix produced by human mechanisms that regulate post-natal chondrocyte differen- dermal fibroblasts cultured in collagen sponges with Wnt agonist tiation, and have important implications for cartilage repair (5 mM LiCl) or control (5 mM NaCl) (day 10) and tissue engineering strategies. Treatment Matrix score Experiment 1 Experiment 2 Acknowledgements LiCl 2, 2, 1 2, 2, 1 NaCl 1, 1, 1, 1, 1, 0 I am grateful for the technical assistance provided by John D. Clark III and Rachael Fiske, and the advice given by Drs 1 0=low amount of matrix; 1=moderate; 2=abundant. Julie Glowacki, Jeffrey Lyczak, and Shuichi Mizuno. This project was supported by grants from the Aircast Foun- dation (#SA-501) and the Oral and Maxillofacial Surgery Research Foundation. cells: Wnt5a and Wnt11 are stimulatory, and Wnt1, -3a, -4, -7a, and -7b are inhibitory33. In other experimental systems of chondrogenesis, Wnt References signaling via the -catenin pathway has been shown to have both stimulatory and inhibitory effects. In human 1. Lako M, Strachan T, Bullen P, Wilson DI, Robson SC, bone marrow stromal cells, LiCl-treatment increases the Lindsay S. Isolation, characterisation and embryonic chondrogenic response to TGF-34. In the embryonic expression of WNT11, a gene which maps to 11q13.5 mouse limb, chondrogenic cells do not express Dkk-1, an and has possible roles in the development of skel- inhibitor of the Wnt/-catenin pathway35. In the murine eton, kidney and lung. Gene 1998;219:101–10. mesenchymal cell line, C3H10T1/2, overexpression of 2. Hartmann C, Tabin CJ. Wnt-14 plays a pivotal role in wnt3a increases nuclear -catenin levels and these cells’ inducing synovial joint formation in the developing chondrogenic response to bone morphogenetic protein 2 appendicular skeleton. Cell 2001;104:341–51. (BMP-2)36. Curiously, exposure of C3H10T1/2 cells to 3. Chimal-Monroy J, Montero JA, Ganan Y, Macias D, LiCl inhibits BMP-2-induced chondrogenesis37. Chick em- Garcia-Porrero JA, Hurle JM. Comparative analysis bryonic limb mesenchymal cells exposed to LiCl also of the expression and regulation of Wnt5a, Fz4 and show reduced chondrogenesis38. Finally, -catenin ex- Frzb1 during digit formation and in micromass pression inversely correlates with expression of the carti- cultures. Dev Dyn 2002;224:314–20. lage matrix protein, collagen type II, during in vitro 4. Church V, Nohno T, Linker C, Marcelle C, chondrogenesis of chick embryonic mesenchymal Francis-West P. Wnt regulation of chondrocyte cells39. differentiation. J Cell Sci 2002;115:4809–18. Recent data showing that Wnt signaling via the calcium 5. Tufan AC, Daumer KM, Tuan RS. Frizzled-7 and limb pathway can antagonize the -catenin pathway40–43 may mesenchymal chondrogenesis: effect of misexpres- explain the observed duality of -catenin signaling in sion and involvement of N-cadherin. Dev Dyn 2002; chondrogenesis. During mouse limb development, loss of 223:241–53. calcium-pathway-signaling Wnt5a increases -catenin42 6. Church VL, Francis-West P. Wnt signalling during limb and inhibits chondrogenesis27,42. This suggests that development. Int J Dev Biol 2002;46:927–36. Wnt effects on chondrogenesis reflect a balance of path- 7. de Crombrugghe B, Lefebvre V, Nakashima K. Regu- ways activated by the specific components present. The latory mechanisms in the pathways of cartilage and net effect of DBP-induced changes in Wnt component bone formation. Curr Opin Cell Biol 2001;13:721–7. expression on signaling pathways in the collagen sponge 8. Wada N, Kawakami Y, Ladher R, Francis-West PH, system is yet to be determined, as well as the contri- Nohno T. Involvement of Frzb-1 in mesenchymal bution of accessory proteins such as Lrp/arrow co- condensation and cartilage differentiation in the chick receptors44,45. limb bud. Int J Dev Biol 1999;43:495–500. The precise mechanism of how DBP affects Wnt compo- 9. Mizuno S, Glowacki J. Three-dimensional composite nent gene expression is not known. Previous studies found of demineralized bone powder and collagen for in that DBP induces shifts in expression of several functional vitro analysis of chondroinduction of human dermal classes of genes, before the chondroblast phenotype is fibroblasts. Biomaterials 1996;17:1819–25. evident14,15. These shifts are specific to cellular interac- 10. Reddi AH, Huggins C. Biochemical sequences in the tions with DBP and may relate, in part, to changes in cell transformation of normal fibroblasts in adolescent shape induced by cell attachment to the particles of dem- rats. Proc Natl Acad Sci USA 1972;69:1601–5. ineralized bone14. Moreover, the gene expression re- 11. Glowacki J, Altobelli D, Mulliken JB. Fate of mineral- sponse elicited in hDFs by DBP differs from the response ized and demineralized osseous implants in cranial elicited by BMP-246. defects. Calcif Tissue Int 1981;33:71–6. In this study, RT-PCR was used because it is a highly 12. Mizuno S, Glowacki J. Chondroinduction of human sensitive method to detect gene expression. Some Wnt dermal fibroblasts by demineralized bone in three- components may have been undetected for technical rea- dimensional culture. Exp Cell Res 1996;227:89–97. sons. Also, the sequence-specific primers used for PCR 13. Glowacki J, Yakes K, Little G, Mizuno S. Induced impose limitations on the type of transcripts that may be chondroblastic differentiation of human fibroblasts by detected: DBP may induce alternatively processed mRNA three-dimensional culture with demineralized bone transcripts that would not be detected by primers that target matrix. Mat Sci Eng C 1998;6:199–203. protein-coding regions21. Nevertheless, the data presented 14. Yates KE, Mizuno S, Glowacki J. Early shifts in gene here suggest that Wnt signaling contributes to post-natal expression during chondroinduction of human dermal chondroblast differentiation induced by demineralized fibroblasts. Exp Cell Res 2001;265:203–11. Osteoarthritis and Cartilage Vol. 12, No. 6 505

15. Yates KE, Glowacki J. Altered expression of connec- 31. Strutt DI, Weber U, Mlodzik M. The role of RhoA in tive tissue genes in postnatal chondroinduction hu- tissue polarity and Frizzled signalling. Nature 1997; man dermal fibroblasts. Connect Tissue Res 2003; 387:292–5. 44:121–7. 32. Slusarski DC, Corces VG, Moon RT. Interaction of 16. Hedgepeth CM, Conrad LJ, Zhang J, Huang H-C, Lee Wnt and a Frizzled homologue triggers G-protein- VMY. Activation of the Wnt signalling pathway: a linked phospatidylinositol signalling. Nature 1997; molecular mechanism for lithium action. Dev Biol 390:410–3. 1997;185:82–91. 33. Bergwitz C, Wendlandt T, Kispert A, Brabant G. Wnts 17. Glowacki J. Cellular reactions to bone-derived differentially regulate colony growth and differenti- material. Clin Orthop 1996;0:47–54. ation of chondrogenic rat calcavaria cells. Biochim 18. Sen M, Lauterbach K, El-Gabalawy H, Firestein GS, Biophys Acta 2001;1538:129–40. Corr M, Carson DA. Expression and function of 34. Zhou S, Eid K, Glowacki J. Cooperation between wingless and frizzled homologs in rheumatoid TGF-b and Wnt pathways during chondrocyte and arthritis. Proc Natl Acad Sci USA 2000;97:2791–6. adipocyte differentiation of human marrow stromal cells. J Bone Miner Res 2004;19:463–70. 19. Fear MW, Kelsell DP, Spurr NK, Barnes MR. Wnt-16a, 35. Grotewold L, Ruther U. Bmp, Fgf and Wnt signalling in a novel Wnt-16 isoform, which shows differential programmed cell death and chondrogenesis during expression in adult human tissues. Biochem Biophys vertebrate limb development: the role of Dickkopf-1. Res Commun 2000;278:814–20. Int J Dev Biol 2002;46:943–7. 20. Rozen S, Skaletsky H. Primer3 on the WWW for 36. Fischer L, Boland G, Tuan RS. Wnt-3A enhances bone general users and for biologist programmers. morphogenetic protein-2-mediated chondrogenesis Methods Mol Biol 2000;132:365–86. of murine C3H10T1/2 mesenchymal cells. J Biol 21. Yates KE, Glowacki J. Gene expression changes in an Chem 2002;277:30870–8. in vitro model of chondroinduction: a comparison of 37. Fischer L, Boland G, Tuan RS. Wnt signaling during two methods. Wound Repair Regen 2003;11:386–92. BMP-2 stimulation of mesenchymal chondrogenesis. 22. Wodarz A, Nusse R. Mechanisms of Wnt signaling in J Cell Biochem 2002;84:816–31. development. Annu Rev Cell Dev Biol 1998; 38. Ryu JH, Kim SJ, Kim SH, Oh CD, Hwang SG, Chun 14:59–88. CH, et al. Regulation of the chondrocyte phenotype 23. Boutros M, Paricio N, Strutt DI, Mlodzik M. Dishevelled by beta-catenin. Development 2002;129:5541–50. activates JNK and discriminates between JNK path- 39. Yoon YM, Kim SJ, Oh CD, Ju JW, Song WK, Yoo YJ, et ways in planar polarity and wingless signaling. Cell al. Maintenance of differentiated phenotype of articu- 1998;94:109–18. lar chondrocytes by protein kinase C and extracellu- 24. Kuhl M, Sheldahl LC, Park M, Miller JR, Moon RT. The lar signal-regulated protein kinase. J Biol Chem Wnt/Ca2+ pathway: a new vertebrate Wnt signaling 2002;277:8412–20. pathway takes shape. Trends Genet 2000;16: 40. Olson DJ, Gibo DM. Antisense wnt-5a mimics wnt-1- 279–83. mediated C57MG mammary epithelial cell trans- 25. Medina A, Reintsch W, Steinbeisser H. Xenopus friz- formation. Exp Cell Res 1998;241:134–41. zled 7 can act in canonical and non-canonical Wnt 41. Ishitani T, Kishida S, Hyodo-Miura J, Ueno N, Yasuda signaling pathways: implications on early patterning J, Waterman M, et al. The TAK1-NLK mitogen- and morphogenesis. Mech Dev 2000;92:227–37. activated protein kinase cascade functions in the Wnt-5a/Ca(2+) pathway to antagonize Wnt/beta- 26. Katoh M, Kirikoshi H, Terasaki H, Shiokawa K. catenin signaling. Mol Cell Bol 2003;23:131–9. WNT2B2 mRNA, up-regulated in primary gastric can- 42. Topol L, Jiang X, Choi H, Garrett-Beal L, Carolan PJ, cer, is a positive regulator of the WNT-beta-catenin- Yang Y. Wnt-5a inhibits the canonical Wnt pathway TCF signaling pathway. Biochem Biophys Res by promoting GSK-3-independent beta-catenin Commun 2001;289:1093–8. degradation. J Cell Biol 2003;162:899–908. 27. Yang Y, Topol L, Lee H, Wu J. Wnt5a and Wnt5b 43. Westfall TA, Brimeyer R, Twedt J, Gladon J, Olberding exhibit distinct activities in coordinating chondrocyte A, Furutani-Seiki M, Slusarski DC. Wnt-5/pipetail proliferation and differentiation. Development 2003; functions in vertebrate axis formation as a negative 130:1003–15. regulator of Wnt/beta-catenin activity. J Cell Biol 28. Rudnicki JA, Brown AM. Inhibition of chondrogenesis 2003;162:889–98. by Wnt gene expression in vivo and in vitro. Dev Biol 44. Pinson KI, Brennan J, Monkley S, Avery BJ, Skarnes 1997;185:104–18. WC. An LDL-receptor-related protein mediates Wnt 29. Ng JK, Kawakami Y, Buscher D, Raya A, Itoh T, Koth signalling in mice. Nature 2000;407:535–8. CM, et al. The limb identity gene Tbx5 promotes limb 45. Wehrli M, Dougan ST, Caldwell K, O’Keefe L, initiation by interacting with Wnt2b and Fgf10. Schwartz S, Vaizel-Ohayon D, et al. Arrow encodes Development 2002;129:5161–70. an LDL-receptor-related protein essential for Wing- 30. Sekiya I, Vuoristo JT, Larson BL, Prockop DJ. In vitro less signalling. Nature 2000;407:527–30. cartilage formation by human adult stem cells from 46. Glowacki J, Yates KE, Zhou S. Comparison of TGF-b/ bone marrow stroma defines the sequence of cellular BMP pathways signaled by BMP-2 and demineral- and molecular events during chondrogenesis. Proc ized bone powder in human dermal fibroblasts. Natl Acad Sci USA 2002;99:4397–402. J Bone Miner Res 2003;18:S184.