JOURNAL OF CELLULAR PHYSIOLOGY 146:442450 (1991)

Fibroblasts Inhibit Mineralised Bone Nodule Formation by Rat Bone Marrow Stromal Cells In Vitro BUNNAI OGISO, FRANCIS J.HUGHES, ANTONY H. MELCHER, AND CHRISTOPHER A.G. McCULLOCH* Faculty of Dentistry, University of Toronto, Toronto, Ontario, Canada, M5C I C6

The ability of rat skin fibroblasts (RSF) and human periodontal ligament fibroblasts (HPL) to inhibit the formation of mineralised bone nodules in rat bone marrow stromal cell (BMSC) cultures was studied. Co-culture of HPL or RSF with BMSC resulted in a large reduction of bone nodule formation when compared with controls. Conditioned medium from HPL or RSF cultures inhibited bone nodule formation in a dose-dependent manner. HPL-conditioned medium depressed cell proliferation and alkaline phosphatase expression in BMSC cultures. These effects were not due to increased cytotoxicity or nutrient depletion. Inhibitory activity was recovered in a fraction of less than 1 kD following ultrafiltration and was insensitive to freeze-thawing. The inhibitor activity was blocked when HPL cultures were grown in the presence of 10- M indomethacin. Dose-dependent inhibition of bone nodule formation was also observed in cultures incubated with prostaglandins E, (at 10-6M) or F2a(at lO-’M). The results indicate that fibroblasts may inhibit osteoblast differentiation and function in part by release of soluble factors including prostaglandins.

x .

The regulation of osteogenesis is dependent in part on factors which are produced by bone or which are present in bone matrix. Some of these factors stimulate the expression of the osteoblast phenotype or increase synthetic activity in bone cells (reviewed in Canalis et al., 1988). Regulation may also depend on antagonistic factors that inhibit osteoblastic expression and activity (Antosz et al., 1989; Nicolas et al., 1990). Inhibition of osteogenesis in vivo occurs in cranial sutures during growth and develo ment (Evans, 1978) and in the periodontal ligament ( ignery and Baron, 1980). In both of these situations, cells of the soft connective tissue play a fundamental role in preventing the encroachment of bone into their domain. In pathological situations such as incomplete healing of fractures of long bones, the continuity between the bony fragments is restored by fibrous connective tissue (Ham and Harris, 1971). This phenomenon could be caused by cells originating from extraskeletal connective tissues that migrate into the wound and interfere with osteogenesis (Melcher and Dreyer, 1962). It has been reported that a fraction of < 50 kD of conditioned medium from cultured myeloma cells may inhibit proliferation of human osteoblast-like cells in vitro, and it is possible that this could represent a athological exam le of osteogenic inhibition mediated y the secretion o a soluble factor (Evans et al., 1989) from local cell opulations. To study inhibition of osteogenesis by ibroblasts and their soluble factors, bone nodule formation in vitro by rat bone marrow stromal cells was employed as a model system (Maniatopoulos et al., 1988). The results demonstrate that conditioned medium from fibroblast cultures contains a

1;

E

P

P

0 1991 WILEY-LISS,INC.

factor of less than 1kD molecular size which inhibits proliferation, alkaline phosphatase expression, and bone nodule formation in rat BMSC cultures. MATERIALS AND METHODS Cell cultures Rat bone marrow stromal cells (BMSC) obtained from young adult rats were isolated and cultured as described by Maniatopoulos et al. (1988). Briefly, femora were removed from 120 g Wistar rats which had been killed by cervical dislocation, and the marrow cells were flushed out of the diaphyses. Cells were dispersed by aspiration through a hypodermic needle and plated at an approximate concentration of the content of 1 femurPT75 flask (FalconTM,Becton Dickinson, Oxnard, CA). Cells were cultured in aMEM supplemented with 15% fetal bovine serum (Flow Laboratories, McLean, VA), 50 p,g/ml ascorbic acid, 10 mM sodium P-glycerophosphate, lo-’ dexamethasone (Sigma Chemical Co. St. Louis, MO), 100 pg/ml penicillin, 50 mg/ml gentam cin (Sigma) and 0.3 p.g/ml amphotericin B (fungizoneT’, Flow Labs). The medium was changed every 2 days and subcultured for experimental use after 6 days of primary culture. Human periodontal ligament fibroblasts (HPL) were Received August 28, 1990; accepted December 7, 1990. Francis Hughes’ current address is Dept. of Oral Medicine and Periodontology, The London Hospital Medical College, London E l ZAD, United Kingdom. *To whom reprint requests/correspondence should be addressed a t Faculty of Dentistry, University of Toronto, 124 Edward Street, Toronto, Ontario, Canada M5G 1G6.

443

FIBROBLAST EFFECT ON RAT BONE NODULE FORMATION

cultured from explants obtained from freshly extracted partially erupted human third molar teeth extracted in the department of Oral Surgery, Faculty of Dentistry, University of Toronto, as previously described (Melcher et al., 1986). Rat skin fibroblasts (RSF) were cultured from explants of abdominal skin of young adult Wistar rats (similar to those used for BMSC isolation), which had been shaved and cleaned of fat. Inspection of HPL and RSF cultures showed that these ceils were entirely fibroblastic. Cultures were fed every 2 days with aMEM plus supplements as for BMSC cultures above, except that no dexamethasone was present in the culture media. Fibroblast cultures were split 1:2 on reaching confluence using conventional techniques and were used for experiments between their 4th and 9th passage. Conditioned medium (CM)was collected every 2 days from cultures after the 4th passage, and was stored at -20°C until used for experiments. In some experiments CM was stored at 4°C prior to use. Co-culture experiments Rat bone marrow stromal cells were co-cultured with either HPL or RSF cells in 2 separate series of experiments. Each series of experiments was repeated 3 times using different animals in each series. In each of the experiments, BMSC were cultured alone as positive controls and were seeded in 35 mm culture dishes at densities of a ) l x lo5 cells/dish (100% seeding), or b)75% of 1 x lo5 cellddish, or 450% of 1 x lo5 cells/ dish, or d)25% of 1 x lo5 cells/dish. In this series of experiments, BMSC were co-cultured in different proportions with either HPL or RSF, but always at a combined seeding density of 1 x lo5 cells/ dish (Table 1);100%HPL or RSF were cultured alone to provide negative controls. In each experiment, 6 replicate cultures were used for each seeding density. All of the cultures were grown for 16 days in the medium described above containing ascorbic acid, pglycerophosphate and dexamethasone. Conditioned medium experiments Conditioned medium (CM) from HPL and RSF cultures were freshly sup lemented with dexamethasone, ascorbic acid, and p-g cerophosphate, and were combined with fully supp emented freshly prepared medium as before at a proportion of 80% CM to 20% fresh medium. BMSC were again seeded into 35 mm dishes at a density of 1 x lo5 cellddish and cultured in the reconstituted CM or in fresh medium for 16 days. A separate series of ex eriments were carried out using HPL.CM and RSF.C , and each series of experiments was repeated twice, using 6 replicate dishes for each variable. As HPL.CM showed greater inhibitory activity than RSF. CM in these experiments (see results), further experiments were carried out using HPL.CM only. Dose response experiments were conducted using varying concentrations of HPL.CM. The HPL.CM was reconstituted with fresh medium (FM) in the proportions 80% CM: 20% FM, 60% CM: 40% FM, 40% CM: 60% FM, 20% CM: 80% FM, and 100% FM. In these experiments BMSC were suspended in CM at a concentration of 1 x lo5 cells/ml, and 20 p1 droplets of the cell suspension were carefully placed onto the surface of

b

&

35mm culture dishes and cultured for 10 days. The droplet cultures were refed every 2 days. As an additional series of controls, CM was obtained from first passage BMSC cultures and substituted for HPL.CM at the different concentrations. In earlier pilot experiments it was found that these droplet cultures were a convenient and rapid method of producing mineralising bone-like tissue in BMSC cultures, and they permitted the relatively rapid screening of the effects of small volumes of media on osteogenesis in vitro. Analysis of mineralisation Cultures were observed daily using phase contrast microscopy and were photographed at regular intervals. At the end of the culture period the dishes were washed and fixed in 1.5% glutaraldehyde for 2 hours and stained in situ for calcium by the use of 2% aqueous alizarin red S for 5 minutes (Maniatopoulos et al., 1988; Melcher et al., 1986). The amount of mineralised tissue was quantified usin a MacBeth TD502 densitometer (Kollmorgen Gorp, Tfewburgh, NY). This was found to be a more rapid method of quantitation when compared with a previously described method of point counting (Maniatopoulos et al., 1988), although both methods produced similar results (unpublished data). The culture dish was divided into 49 x 4 mm squares by superimposition of a transparent grid. Measurements were taken through each square and the mean value was calculated. Mean values were also calculated for empty dishes and this fi re was subtracted from the readings obtained from t e stained culture dishes. Droplet cultures were observed daily for evidence of nodule formation and for mineralisation of nodules, which was detected by a marked darkening of the nodules when viewed by phase contrast microscopy. The visual assessment of nodules has been found to agree well with the area of mineralisation as assessed by staining with alizarin red or Von Kossa. To assess both the number of nodules formed and their speed of formation, a Nodule Forming Index (NFI) was calculated using the following formula: NFI = C N/d/d, where d = day of culture and Nd = number of mineralised nodules on day d. Thus for example, if 0 nodules were visible on day 8 , 6 visible at day 9, and 8 at day 10, then NFI = 018 + 619 t 8/10 = 1.47. At the termination of the droplet cultures, Von Kossa staining was performed and the area of Von Kossastained material was measured by an automated image analyzer (Bioquant, R & M Biometrics, Nashville, TN). Cell proliferation and alkaline phosphatase expression To assess the effects of HPL.CM on BMSC proliferation and the number of BMSC expressing alkaline phosphatase, BMSC were plated into 35 mm culture dishes at a concentration of 2 x lo4 cells/dish and were cultured in 80%HPL.CM: 20% FM for periods of 2,4,6, and 8 days. Control cultures were grown in FM only, and each test was carried out in triplicate. Cells were removed from the dishes using 0.01% trypsin in citrate saline, and cell numbers were estimated using a Coulter Counter. A cytospin preparation was made from an aliquot of the suspended cells, which was fixed in cold methanol and stained for alkaline phosphatase

r

444

OGISO ET AL.

using the azo-dye coupling method. The proportion of cells exhibiting bright blue stain was computed from counts of 5 randomly chosen fields (x400) of the cytospin preparation. At least 200 cells were counted in each determination. Immediately prior to the termination of cultures, cells were incubated with 10 pg/ml propidium iodide in phosphate buffered saline for 30 min in order to stain dead cells. Cytospin preparations were subsequently examined using epifluorescence (530 nm excitation, 630 nm emission) to determine the proportion of cells in cultures which were stained with propidium iodide. Ultrafiltration of HPL.CM Serum-free HPL.CM was collected from HPL cultures as described above after 2 consecutive periods of 48 hours each. Media were fractionated by ultrafiltration (Amicon) using molecular mass cut off filters of 1, 5, 10, 30, and 100 kD at 4°C under nitrogen gas. All filtrates and associated media retained by the filter were separately restored to their original volume using serum free medium, supplemented with 15% FBS, ascorbic acid, P-glycerophosphate, and dexamethasone, and tested for inhibitory activity on BMSC using droplet cultures as described above. Prostaglandins To examine the possibility that inhibition of osteogenesis may be due in part to prostaglandins (PG), dro let cultures were incubated with PGE2, PG12, PG,! or PGFzu (Sigma) at final concentrations ranging from to 10-7M. PGs were dissolved in 100% absolute ethanol before addition to cultures and were changed with the medium every other day. In each experiment, ethanol vehicle was added to control cultures at the same final concentration as the test cultures. To further explore the possibility of PG inhibition of osteogenesis, confluent cultures of HPL cells were incubated with M indomethacin (Sigma) for up to 16 da s and the medium was collected every 4 days. The PL.CM treated with indomethacin was added to droplet cultures and supplemented with FBS, dexamethasone, ascorbic acid, and sodium-P-glycerophosphate, as described. Control cultures were incubated with either full medium plus equivalent volumes of 100% absolute ethanol or with HPL.CM without indomethacin treatment. Statistical analysis of data Data were analysed for statistical significance using analysis of variance (ANOVA) and the general linear modelling procedure of SAS (SAS Institute, Cary, NC). Means and standard errors of the means were computed (X 2 S.C.M.). Multiple comparisons were carried out using the Tukey Multicomparison test, with significance set at the P < 0.05 level.

d

Fig. 1. Cocultures of 50%RBMC and 50%HPL illustrating bone-like nodules (B) and the surrounding cell multilayer (M).A: Bone-like nodule after 10 days in culture. B: A different and larger bone-like nodule from that illustrated in a, also after 10 days in culture. Its periphery was clearly-defined in part but is more diffuse in one area (arrow).C: A larger bone-like nodule after 13 days in culture. Note the diffuse periphery. Phase contrast microscopy x 100.

RESULTS Co-culture experiments When cultured alone, BMSC formed mineralised bone nodules as previously described by Maniatopoulos bone nodules were also seen in co-cultures, but were et al. (1988). As expected, neither HPL or RSF cells consistently observed to appear earlier when the BMSC formed raised nodules of any description. Mineralised were cultured alone, compared with when they were

445

FIBROBLAST EFFECT ON RAT BONE NODULE FORMATION

TABLE 1. Mineralization of co-cultures of rat bone marrow stromal cells (BMSC) with either human periodontal ligament fibroblast (HPL) or rat skin fibroblasts (RSF)'

Initial concentration of cells when plated 100%1BMSC 75%)BMSC 75'E BMSC 50% BMSC 5o!h BMSC 2581 BMSC 25% RMSC 100% HPL

+ 25% HPL + 5Ooii1HPL + 75% HPL

Densitometric measurements of mineralized tissue (means; n = 6 per group) Expt. 1 Expt. 2 Expt. 3

.46 50 .50 .43 .48

.33

64 59 .64 .51 .49 .46

.25*

.16**

.04

Expt. 1 100%RMSC 75'11BMSC 75% BMSC 50% BMSC 50'11BMSC 2,593BMSC 25% BMSC

100'11RSF Fig. 2. Photographs of 16 day cultures seeded at different proportions

of cells as indicated on the prints and stained in situ with alizarin red. The mineralised bone-like tissue was of dark appearance. Top row: Co-cultures of BMC and RSF.Middle row: Cultures of BMC. Bottom row: Co-cultures of BMC and HPL.

co-cultured with either of the other 2 cell types. Observations on newly developed nodules in co-cultures comprising a pre onderance of the soft connective tissue cells showe them to have a strikingly circumscribed appearance; a few days later the periphery was much less clearly defined (Fig. la-c). Examination of alizarin red-stained cultures suggested that co-culture depressed bone nodule formation and that the effect of RSF cells was greater than HPL cells (Fig. 2). Densitometric measurements showed that all co-cultures of BMSC and RSF produced significantly less mineralised bone-like tissue at all seeding densities of BMSC alone (Table 1). In addition, in co-cultures with 75% HPL cells and 25% BMSC the amount of mineralised tissue was less than when 25% BMSC cells were cultured alone. Histological examination of sections from 16 day co-cultures showed that bone nodules were elevated above the cell multilayer, and were mushroom shaped (Fig. 3). Conditioned medium experiments Conditioned medium from HPL and RSF significantly reduced bone nodule formation compared with controls. HPL.CM showed a greater inhibitory effect than that seen with RSF.CM. In experiments with droplet cultures, HPL.CM inhibited nodule formation and the nodule forming index in a dose-dependent manner (Fig. 4). Repeated freeze-thawing of the HPL.CM produced no significant decrease in the inhibition of osteogenesis (Table 2). BMSC.CM obtained from cultures grown with dexamethasone produced about a 50% increase in nodule forming index compared with FM controls. Similar results have been obtained in 3 subsequent series of experiments (Hughes and McCulloch, in press).

ts

+ 25% R S F + 50%R S F

+ 75% RSF

.ll

Expt. 2

55

.41

59 25' .47 .ll*

.40

.32 .11**

.08

.13* .38 .12* .22

.11** .05

.51 .40

.44 .33 .42 .22 .04** .04 Expt. 3

54 53

.16* .48 .08* .41 .lo**

.07

'Co-cultures of BMSC and either HPL or RSF were grown for 16 days and then mineralization assessed by densitometric measurement of alizarin-red stained material. All comparisons were made within a single concentration of cells. The pooled standard error computed for ANOVA did nolexceed0.02densilyunits in any experimental gmup. *Differences are significant at P < 0.01. **Differences are significant at P < 0.001.

Cell proliferation and alkaline phosphatase (AP) expression HPL.CM significantly inhibited cell proliferation at all time points as determined by total cell numbers in cultures (Fig. 5 ) . When the effect of HPL.CM on the numbers of AP positive and AP negative cells was examined separately, it was found that the CM significantly reduced the numbers of AP positive cells by about 50%at days 6 and 8 (Fig. 6a). The numbers of AP negative cells was not significantly reduced in HPL.CM cultures (Fig. 6b). Further, the proportion of AP positive cells was also reduced in HPL.CM cultures at day 6 and 8 (Fig. 7). There were no signficant differences between test and controls in the number of dead cells in BMSC cultures, which was always less than 1.5% after 4 or more days in culture (data not shown). Ultrafiltration experiments Following ultrafiltration, bone nodule inhibitory activity was recovered in a fraction of less than 1 kD molecular mass (Table 2). Nodule formation was not inhibited by fractions containing factors of greater than 1kD and indeed tended to be somewhat higher than control values. Prostaglandins Inhibition of cyclo-oxygenase by indomethacin in HPL cultures resulted in no significant difference (P > 0.5) in area of nodule formation between positive control cultures (5.06 ? 0.81 x lo4 pm2) and in cultures that had been incubated with HPL.CM originally treated with 10-5M indomethacin (5.63 & 1.19 X lo4 pm2) while cultures treated with HPL.CM exhibited

446

OGISO ET AL.

Fig. 3. A bone-like nodule (B) and surrounding multilayer in a coculture of 50% RBMC and 50% HPL after 16 days. Note that the bone-like nodule is in contact, or almost in contact, with the floor of the dish centrally (arrow), but that, peripherally, it extended over cells of the multilayer that frequently appeared to be necrotic (crossed-arrow).The bone-like tissue was covered by a layer of cells, presumably osteoblasts and pre-osteoblasts. Epon section. Stevenel's Blue ~ 3 0 0 .

0

20

60

40

Concentration of CM

80

(%)

Fig. 4. Histogram of nodule forming index (NFI) in rat bone marrow stromal cell cultures incubated with combinations of human periodontal ligament-conditioned medium (HPL.CM) and bone marrowconditioned medium (BMC.CM). NFI, a measure of the speed of nodule formation, was inhibited in a dose-dependent fashion by HPLCM.

significant (P < 0.05) reduction in mineralized tissue area (2.50 2 0.56 X lo4 Fm?. BMSC treated with HPL.CM previously incubated with 10P5Mindomethacin showed no significant difference in nodule formation compared with ethanol-treated controls. In contrast, HPL.CM depressed nodule formation, consistent with findings described above. The osteogenic inhibitory effect of HPL.CM was also blocked when HPL cells were incubated with loP5 or 10-7M dexamethasone. All PG's exhibited inhibition of nodule formation, however, both PG-E2 and PG-FZo,exhibited dose-dependent inhibition of nodule formation with significant inhibition detectable at 10P6Mand 10p7M,respectively (Fig. 8).

DISCUSSION The mineralised bone nodules produced by the BMSC cultures in these experiments were identical to those that previously have been described and extensively characterised by Maniatopoulos et al. (1988).The application of the nodule-forming assay to study the

TABLE 2. Molecular mass of osteogenic inhibitory factors' NF1

Molecular mass cut-offs

< 10 kd < 5kd < Ikd > 10 kd > 6kd > 1kd

Unfractionated Control

Fresh HPL.CM

Frozen HPLCM

0.61

0.60

1.13 ~. 0.92 ~

3.00 2.77 2.89 0.20 1.38

n zn 0.40 1.56 2.05 1.80

-

~~

~

INodule forming index (NFI) of bone marrow stromal cells cultures incubated with ultrafiltrates of different molecular mass. Filters of 1, 5, and 10 kd were used to separate inhibitory species into fractions above (>) and below (