Skeletal growth factor is produced by human osteoblast-like cells in culture

Biochirnica et Biophysica Acta 889 (1986) 163-170

163

Elsevier BBA 11870

S k e l e t a l g r o w t h f a c t o r is p r o d u c e d by h u m a n o s t e o b l a s t - l i k e cells in c u l t u r e J o n E. W e r g e d a l a.c,d, S u b b u r a m a n M o h a n b,c.a, A r c h K . T a y l o r d a n d D a v i d J. B a y l i n k a,c,d Departments o f a Biochemisto, ' h pto,siolog v and ' Medicine, Loma Linda Unit,ersiO', and a J e r o , L. Pettis Memorial Veterans' Administration Hospital Loma Linda. CA (U.S.A.)

(Received 24 June 1986)

Key words: Mitogen production; Skeletal growth factor; (Human bone cell)

Human bone cells isolated from femoral heads were cultured in BGJ b medium containing bovine serum albumin (100 ~ g / m i ) , insulin (1 # g / m l ) and epidermal growth factor (10 n g / m l ) , and the conditioned medium collected. The medium was concentrated, chromatographed using HPLC gel filtration (TSK 2000 SW), and assayed for mitogenic activity using [3Hlthymidine incorporation into embryonic chick calvarial cells. The conditioned medium contained mitogenic activity which eluted with a different elution time than insulin or epidermal growth factor. Characterization of this activity suggests that it was due to human skeletal growth factor (SGF), a mitogen which had been previously isolated from human bone matrix. Common properties include: (1) stimulation of DNA synthesis in cultured embryonic chick caivarial cells, (2) competition with human SGF for binding to anti-SGF antibodies, (3) elution from HPLC gel filtration as a large factor ( M r 100000) under native conditions but as a small factor ( M r 10000) under dissociative conditions (4 M guanidine HCI), (4) elution time on HPLC reverse-phase chromatography (small SGF), (5) inactivation by dithiothreitol, (6) stability to heat, acidic or alkaline conditions and (7) inactivation by trypsin and chymotrypsin. These observations provide evidence that human bone cells produce SGF. Conditioned medium from human skin cell cultures also contained mitogenic activity. However, the activity was less than that from bone cells and did not cross-react with the rat anti-SGF antibodies.

Introduction

A bone cell mitogen termed skeletal growth factor (SGF) is present in bone matrices of all species that have been studied [1-3] and appears to be liberated into the medium of bone organ cultures during bone resorption [1]. The source of

Abbreviations: SGF, skeletal growth factor; EGF, epidermal growth factor; ELISA, enzyme-linked immunosorbent assay; Tween-20, polyoxyethylene sorbitol monolaurate. Correspondence: Dr. J.E. Wergedal, Research Service (151), Jerry L. Pettis Memorial Veterans' Hospital, Loma Linda, CA 92357, U.S.A.

SGF present in bone has not been established although there is evidence which indicates that bone cells synthesize the SGF [1]. Because a likely source of SGF is the osteoblast, human bone cell cultures which contain osteoblast-like cells were examined for evidence of human SGF production. In previous studies, we have presented strong evidence that cells isolated from human bone by collagenase digestion are osteoblast like [4]. Therefore, we have examined the medium of human bone cell cultures for evidence of SGF activity. A culture of human skin fibroblasts was included as a control for nonspecific production. The primary characteristic of SGF is stimulation of bone cell proliferation. Therefore, SGF activity was assayed

0167-4889/86/$03.50 © 1986 Elsevier Science Publishers B.V. (Biomedical Division)

164

by the stimulation of [3 H]thymidine incorporation into the D N A of embryonic chick calvarial cells in monolayer cell culture. Materials and Methods

Cell cultures. H u m a n bone cells were obtained from the trabecular bone of femoral head samples obtained during hip replacement surgery. Cells were isolated as previously described [4]. Briefly, diced samples were incubated with crude collagenase (2 m g / m l ) in BGJ b culture medium for 2 h. The cell suspension was rinsed once with BGJ b medium and plated in Fitton-Jackson modified BGJ b culture medium containing 10% fetal calf serum, penicillin (100 u n i t s / m l ) and streptomycin (100 ~ g / m l ) . Cells were grown to confluence and passaged with collagenase (1 m g / m l ) in calcium and magnesium-free Puck's Saline G. Human skin cells were isolated from a human foreskin sample and cultured by the same methods as those described for the human bone cells. Production of conditioned medium. For the collection of conditioned medium, passaged cells were plated in two 75 cm 2 flasks in BGJ b medium with 10% fetal calf serum. When the cell density approached confluence, the cells were rinsed twice with serum-free medium and maintained with serum free medium (10 ml) containing bovine serum albumin (100 /~g/ml), insulin (1 /~g/ml) and epidermal growth factor (EGF) (10 n g / m l ) . Because the human bone cells do not grow well in serum-free conditions the addition of insulin and E G F was necessary to enable the cells to grow and maintain a normal morphology during the collection of conditioned medium. Although these two factors stimulate proliferation in the mitogenic assay used to monitor SGF activity and thus prevents direct assay of conditioned medium for mitogenic activity, the sizes of these two growth factors ( M r 6000) are small enough so that they can be separated from SGF ( M r 10000) by gel filtration chromatography. After 3 days of culture the conditioned medium was collected and frozen ( - 2 0 ° C ) , fresh medium was added to the cells and a second collection was made after another 3 days of culture. The conditioned media were combined and concentrated 10-fold by pressure filtration (Amicon concentrator) using a YM5 mere-

brane (5000 molecular weight cut-off), rinsing twice with distilled water. The concentrate was passed through a 0.45 /~m millipore filter and stored frozen ( - 2 0 ° C ) . Protein content of the concentrates estimated by 280 nm absorption was 4 4 8 / ~ g / m l of which bovine serum albumin should contribute 400/~g. Mitogenic assay. Mitogenic activity on bone cells was assessed by the incorporation of [3H]thymidine into D N A of chick calvarial cells. The method developed by Gospodarowicz et al. [5] was adapted for chick calvarial cells in our laboratory [6]. Briefly, calvaria from 15-day-old embryos were incubated with 1 m g / m l of crude collagenase in calcium and magnesium free Puck's Saline G for 2 h. The cell suspension was rinsed twice with Dulbecco's Modified Eagle Medium and plated in Dulbecco's Modified Eagles Medium containing penicillin (100 u n i t s / m l ) and streptomycin (100 /~g/ml). The cells were plated in multiwell plates (35000 cells in 0.5 m l / 1 0 0 mm z well). 24 h after plating, the test agents (20 t~l) were added. After 18 h, the cells were pulse labeled for 2 h with 0.75 #Ci of [3H]thymidine per well. The medium was drawn off and the cells were rinsed with phosphate-buffered saline, The cells were frozen and then removed from the culture wells with a Q-tip swab dipped in 12.5% trichloroacetic acid. The swabs were rinsed twice with 12.5% trichloroacetic acid and once with 70% ethanol, air dried and the radioactivity was determined. The addition of serum (final concentration 1%) resulted in an 700% increase in thymidine incorporation over controls, Control wells received bovine serum albumin (100 /~g/ml). A 100% increase in [3H]thymidine incorporation over the control level is defined as 1 unit of mitogenic activity. Gel filtration chromatography. The conditioned medium was chromatographed by HPLC gel filtration (Beckman Products Spherogel TSK 2000 SW column, 7.5 × 300 mm) under two different conditions: (a) under native conditions using 30 mM Tris-acetate (pH 7.4) containing 0.15 M NaCl as eluant, and (b) under dissociative conditions using 30 mM Tris-acetate (pH 7.4) containing 4 M guanidine HC1 as an eluant. The conditioned medium concentrate was split into two halves and lyophilized. The lyophilized samples were recon-

165 stituted in the appropriate ehition buffer for the column chromatography. Mitogenic activity was eluted with a flow rate of 0.5 m l / m i n and 1 min fractions were collected. Individual fractions from the gel filtration chromatography under native conditions were assayed directly for mitogenic activity as described above, while those containing guanidine were first dialyzed against 30 mM Trisacetate for 3 days (three buffer changes) to remove the guanidine before assay. Reverse-phase chromatography. For reversephase HPLC chromatography, the active fractions obtained from H P L C gel filtration chromatography under dissociative conditions were combined, lyophilized and reconstituted in 25% acetonitrile in water (0.1% trifluoroacetic acid). The samples were fractionated using a C4 column (Bio-Rad RP304 4.6 × 250 mm), with an elution gradient (25-60% in 70 min) of acetonitrile in water (0.1% trifluoroacetic acid) and a flow rate of 1 m l / m i n . 1 min fractions were collected and dried by SpeedVac centrifugation, reconstituted in 1 ml of water and the mitogenic activity was determined as described above. Antibody cross-reactivity. The conditioned medium concentrate and the pool of the active fractions from H P L C gel filtration under dissociative conditions were assayed for the ability to displace S G F binding to rat anti-SGF antibodies using competition ELISA. For the assay, microtiter plates were coated with purified human SGF (5 /~g/ml) in 50 mM carbonate buffer (pH 9.6) and then treated with ovalbumin (0.1%) in the carbonate buffer to block nonspecific binding of the antibodies. Samples were mixed with an equal volume of rat anti-SGF antibodies diluted 1 : 300 with antibody buffer (phosphate-buffered saline containing 0.05% Tween 20, 0.05% thimerosol, and 0.1% human serum albumin). After incubating for 2 h at 37°C, the reaction mixture (50 ~1) was added to the SGF coated plates and allowed to incubate for an additional 2 h at 37 ° C. The plates were rinsed with wash buffer (0.9% NaCI, 0.05% Tween-20) and incubated for 2 h at 37°C with secondary antibodies, horseradish peroxidase linked rabbit anti-rat IgG diluted 1:1000 with the antibody buffer containing 10% normal rabbit serum. The plates were rinsed several times with wash buffer, and the horseradish peroxidase activ-

ity was assayed with 0.04% orthophenylenediamine and 0.05% H202 in citrate-phosphate buffer, p H 5.0. The absorbance was read at 570 nm in a Dynatech plate reader.

Treatment with dithiothreitol and proteolytic enzymes. Purified SGF and a sample of the pool of the active fractions from H P L C gel filtration under dissociative conditions were incubated with 5 m M dithiothreitol in 30 mM Tris-acetate buffer (pH 7.0) for 90 min at room temperature. Samples were diluted 10-fold with Dulbecco's Modified Eagles Medium before assaying for mitogenic activity. Final concentration of dithiothreitol in the mitogenic assay medium was less than 0.05 mM. The inclusion of control samples showed that this concentration of dithiothreitol has no effect on the [3H]thymidine incorporation in the mitogenic assay. To determine the effect of proteolytic activity on the mitogenic activity in the conditioned medium, proteinases bound to agarose or CM-cellulose beads were used. Samples were treated with enzymes for 2 h at room temperature and then centrifuged to remove the enzymes. The supernatant was assayed for mitogenic activity. Materials. BGJ b culture medium, Dulbecco's Modified Eagles Medium, penicillin, streptomycin, and crude collagenase were obtained from Grand Island Biological Company. Fetal calf serum was obtained from Flow Research Laboratories. Insulin and E G F were obtained from Collaborative Research. Bovine serum albumin, ovalbumin and myoglobin were obtained from Sigma Chemicals. Purified SGF was extracted from human bone matrix and purified to homogeneity as previously described [7]. Rat anti-SGF antibodies were induced in Lewis rats by injections of partially purified SGF. Horseradish peroxidase linked anti-rat IgG was obtained from Miles Laboratories. Results Since previous studies [1,2,8] revealed that SGF behaves as a large molecular weight factor ( M r 70000-100000) under native conditions and as a low-molecular-weight factor ( M r 10000) under dissociative conditions [7,9], chromatography of the conditioned medium from human bone cell

166

=§

gq

-

•

*

Z 0 t-,<

g~o

200

g

aoo

o

1

D

N o ~O

o o

o o

~oo

~o-

c o

E$ N~ 100

~eooJ

-4

I" i 010

I

i

i 15

i i

i i 20

t

i

i

l i _L 25

ELUTION TIME (Min.)

Fig. 1. HPLC gel filtration chromatography under native conditions of the mitogenic activity in h u m a n bone cell conditioned medium. The elution buffer was 30 m M Tris-acetate (pH 7.4) with 150 m M NaCI. Bone cell mitogenic activity was assayed in monolayer cultures of embryonic chick calvaria cells. Points represent the mean of four determinations. Molecular weight standards included were phosphorylase B, M~ 98000: ovalbumin, M~ 43000; myoglobin, M r 17500: and insulin, M~ 6(X)0.

cultures was carried out under both conditions to test for evidence of either form of SGF. Fig. 1 shows the profile of bone cell mitogenic activity from the HPLC gel filtration under native conditions of the conditioned medium sample from the human bone cell culture. Bone cell mitogerlic activity was present in the human bone cell conditioned medium in the high-molecular-weight region with a peak corresponding to an approx. M r of 100000. This is similar to the results previously reported for chick calvarial cell conditioned medium [1]. Lesser amounts of activity were also evident in the lower-molecular-weight regions (M~ 30000 and 10000). In the presence of 4 M guanidine HCI, the major peak of bone cell mitogenic activity from the human bone cell conditioned medium was shifted to the low-molecular-weight region (Fig. 2, M r 10000). This M r 10000 peak was not present in medium that had not been conditioned by cells demonstrating that the activity was derived from the cells and was not due to the insulin or E G F present in the medium (Fig. 2). The shift in the molecular weight of the bone cell mitogenic activ-

10

15 ELUTION

20 TIME

25

(MIN.)

Fig. 2. HPL(' gel fihration chromatographer (Spherogel TSK 2000 SW) under dissociative conditions of the milogenic activity in human bone cell conditioned medium (O-e), human skin cell conditioned medium ( O - O) and fresh medium ( × - - × ) . The elution buffer was 30 mM Trisacetate (pH 7.4) with 4 M guanidine HC1. Bone cell mitogenic activity was assayed in monolayer cultures of crabs'ohio chick calvarial cells. Purified S(;F elutes at M, 10000. Points reprc,~cnt the mean of four determinations.

ity with 4 M guanidine HCI treatment is identical to that seen with SGF activity extracted from bone matrix [7]. The mitogenic activity recovered from the column chromatography (from the fractions above M r 6000) appeared to be greater under dissociative conditions than under native conditions (566 units vs. 455 units of mitogenic activity). Further evaluation will be required to determine if this effect is significant. Mitogenic activity was also evident in conditioned medium from skin cells in the low-molecular-weight region (Fig. 2), but the activity was considerably less than that present in the bone cell conditioned medium (95 vs. 490 units of activity). The active fractions from HPLC gel filtration under dissociative conditions were pooled and tested for the presence of SGF characteristics. Dithiothreitol treatment abolished the mitogenic activity of the pooled fractions as it did that of the purified SGF (Table I). The susceptibility of the pooled fractions to inactivation by the proteases was tested using enzymes bound to beads to allow easy removal of the proteinases, Treatment with

167 TABLE I

TABLE III

INACTIVATION OF THE MITOGENIC ACTIVITY BY DITHIOTHREITOL

MITOGENIC ACTIVITY IS STABLE TO TREATMENT WITH HEAT, ACID AND ALKALINE pH

Samples were incubated with or without 5 mM dithiothreitol for 90 rain at room temperature before being diluted 10-fold with Dulbecco's Modified Eagles Medium for the assay of mitogenic activity. Data are the mean+ S.D. of six samples, Control assay wells contain no added factor.

Samples of the pooled fractions from the active peak of the HPLC gel filtration under dissociative conditions were treated as indicated for 120 rain and assayed for stimulation of mitogenie activity. Buffers used were Mcllvaine's citrate (pH 2.5), Sorensen's gtycine II (pH 10.0) and 30 mM tris-acetate (pH 7.4). Data are the mean_+ S.D. of six samples.

Factor

None HPLC pool ~ SGF ~

Mitogenic activity percent of control

P <

- dithiothreitol

+ dithiothreitol

100+ 7 228 + 15 436 + 55

118_+ 12 108 -+ 16 115 + 15

nsh 0.001 0.001

" Pooled fractions from the active peak of the HPLC gel filtration chromatography under dissociative conditions. ~' Comparison of +dithiothreitol samples with dithiothreitol samples. ~ Concentration of SGF was 20 ng/ml.

trypsin or chymotrypsin significantly reduced the mitogenic activity of the pooled fractions as it did for purified SGF (Table II). The mitogenic activity was stable to heat (70°C), acid (pH 2.5) and alkali (pH 10) (Table III). Thus, the mitogenic activity released by human bone cells into the culture medium shared many physical characteristics with the purified human SGF.

Treatment

Mitogenic activity (percent of control)

None pH 2.5, 25°C pH 10.0, 25 o C pH 7.4, 4 ° C pH 7.4, 70°C

274 ± 65 ~ 307±66 267 ± 48 298_+52 268_+36

P <

ns ns ns ns

h

Basal incorporation was 415 +50 cpm, incorporation with untreated active fraction was 1137 + 270 cpm. h Comparison with the no treatment group. "

An aliquot of the active pool was chromatographed on reverse phase HPLC using a C4 column and a gradient of acetonitrile in 0.1% trifluoroacetic acid. The majority of the mitogenic activity eluted in the same position as purified SGF l

~

3so

~ 8

45

wz

z

300

t

t

,"

t

O

/J

tz. /

40

A I 35 ~

TABLE II INACTIVATION OF THE MITOGENIC ACTIVITY BY TRYPSIN A N D CHYMOTRYPSIN Samples were incubated with proteinase bound to beads for 0 or I20 min before being centrifuged and assayed for mitogenic activity. Data are the mean ± SD. of six samples.

0 0 2oc On" Zl--z L , u o 150 Zo r",~ >.. ae -rv 1ts

Proteinase

Factor

Mitogenic activity percent of control 0 rain

_

25 ~~

_

100 50

p < a

I 0

l 10

I

ELUTION

120 min

Trypsin

HPLC pool h SGF ~

258+46 215+_37

153+32 103_+12

0.001 0.001

Chymotrypsin

HPLC pool SGF c

266+37 218+ 15

142+28 85+18

0.001 0.00l

" 120 min value vs. 0 min value. h Pooled fractions from the active peak of the HPLC gel filtration under dissociative conditions. Concentration of SGF was 10 ng/ml.

I 20

I

TIME

I 30

I

~j20 40

(Min.)

Fig. 3. HPLC reverse-phase chromatography (Bio-Rad C4 column 4.6×250 ram) of the active fraction from HPLC gel filtration chromatography under dissociative conditions of human bone cell conditioned medium. The elution solvent was a gradient of acetonitrile from 25 to 60% in 70 min in 0.1% trifluoroacetic acid. Bone cell mitogenic activity was determined in monolaver cultures of embryonic chick calvarial cells. Points represent the mean of six determinations. Purified SGF, insulin and EGF were run under the same conditions as standards.

168

(Fig. 3). However, a small amount (23%) of the activity eluted in the same position as EGF and insulin suggesting that the gel filtration did not completely separate the bone cell mitogenic activity from these two mitogens or that another mitoden was present. The total amount of activity recovered from the column was 64% of the amount added. The conditioned medium activity was tested to determine if it cross-reacted with antibodies against SGF. The assay was competition ELISA in which rat anti-SGF antibodies were reacted with the sample and then with authentic purified human SGF bound to microtiter plates to determine the amount of the unreacted antibodies. Samples of human bone cell conditioned medium reduced the amount of antibodies bound to the plate and therefore displaced authentic SGF from binding to the antibody (Fig. 4). This observation provides strong evidence for the presence of SGF

,00

~

....

~ - ~ --o-_

6o IIl

40 0

_m 20

in the conditioned medium as suggested by the physical and chemical characteristics of the bone cell mitogenic activity. A sample of the active fraction from HPLC gel filtration chromatography under dissociative conditions, also displaced SGF binding to the antibody. This indicates that the mitogenic activity and antibody binding activity co-migrate on HPLC gel filtration chromatography. The specificity of this binding assay is dependent upon the purity of the SGF bound to the microtiter plates because the antibodies used were induced by injections of partially purified SGF. Tests of the specificity showed that neither insulin nor EGF nor the culture medium showed any displacement (Fig. 4). Conditioned medium from human skin cells also did not cross-react with the antibody suggesting that the mitogenic activity in the skin cell conditioned medium was not SGF. The concentration of SGF present in the HPLC active pool as estimated by the antibody binding was 53 n g / m l . The mitogenic activity of the active pool as assessed by [3H]thymidine incorporation of DNA into chick calvarial cells was equivalent to 95 n g / m l of SGF. This indicates that SGF was responsible for about half of the mitogenic activity in the active pool. The reverse-phase chromatography indicated that part of the mitogenic activity of the active pool may be due to insulin or EGF that was not completely separated by the gel filtration chromatography. Thus SGF may be responsible for a greater proportion of the mitogenic activity secreted by the cell than the analysis of the active pool indicates.

SAMPLE DILUTION

Fig. 4. Cross-reactivity of human hone ce]l conditioned medium for rat anIJ-SGF antibodies determined by competition ELISA.

Discussion

Samples were reacted with anti-SGF antibodies and the unrcacted antibodies were determined by incubation with purified human SGF bound to microtiter plates, The antibodies bound to the S(iF on the microtiter plates were assayed with horseradish peroxidase linked sheep anti-rat IgG. B,, was the amount of antibody bound when a blank sample was reacted with the anti-SGF. Samples tested include: partially purified human SGF (equivalent to 600 n g / m l purified SGF) (O), I0 × concentrate of human bone cell conditioned medium (mean of two samples) (O). 10× concentrate of fresh medium ( × ) , 10× concentrate of human skin cell conditioned medium (IlL the active fraction from the HPLC gel filtration under dissociative conditions (v), insulin 110 # g / m l ) (zx), and EGF (1 /*g/ml) (UI). Points represent the mean of two determinations. The difference between samples was always less than 10g..

These studies demonstrate that human bone cells, like a number of other cell types [10-14] including bone cells from other species [1,15,16] release mitogenic activity into the culture medium. A substantial part of the mitogenic activity released by the human bone cells appeared to be due to a mitogen that, by several criteria, was identical to SGF isolated and purified from human bone. Each of the tests applied during this study gave results consistent with the presence of SGF. Although it was necessary that the mitogenic activity

169 of bone satisfy all these test, a number of the tests are not specific for SGF, i.e., (1) stimulation of DNA synthesis in cultured bone cells, (2) inactivation by dithiothreitol, (3) stability to heat acid and alkaline pH and (4) inactivation by trypsin and chymotrypsin. The best evidence that the activity is SGF and not some other known mitogen is given by the HPLC chromatography and antibody binding studies. Both the reverse phase chromatography and the antibody binding studies demonstrated that the conditioned medium contains activity different from the insulin and EGF initially present in the medium. These analyses indicated that SGF is responsible for 50-75% of the mitogenic activity in the HPLC gel filtration pool. These observations provide evidence that human bone cells released SGF in vitro. It would be desirable to confirm that the gel filtration pool contains a protein that migrates with SGF on SDS-polyacrylamide gel electrophoresis. This was not achieved because the concentration of SGF in the active pool was too low to detect on electrophoresis gels with the staining methods available. SGF stains poorly with the silver staining technique which is the most sensitive method of protein detection for electrophoresis gels. Previous studies have shown that cultured embryonic chick calvarial cells also release mitogenic activity into the culture medium [1]. The chick bone cell activity appeared to be SGF on the basis of gel filtration under native conditions and stability to heat, acid pH and dithiothreitol treatment [1]. Thus, it seems very likely that bone cells produce the SGF found in bone matrix in vivo. At present, there is no evidence as to whether all cells in the cell cultures are producing the mitogen, or whether only a selected population of cells is producing the mitogen. However, there is heterogeneity within the cell population as shown by cytochemical staining reactions for the enzyme alkaline phosphatase [4]. Therefore, it is possible that only a subpopulation of the human bone cells is producing the mitogen. Osteoblasts are one of the most likely sources of the SGF. The human bone cells used for this study have been shown [4] to have characteristics consistent with their identification as cells of the osteoblast line i.e., (1) the presence of high levels of alkaline phosphatase, (2)

responsiveness to SGF, (3) presence of la-hydroxylase for 25-hydroxyvitamin D-3 and (4) response to growth factors similar to embryonic chick cells of the osteoblast line. Thus, our results are consistent with the hypothesis that it is the osteoblasts that produce the SGF. Further work will be required to identify more precisely the bone cell type which produces SGF. There may have been other mitogens besides SGF produced by human bone cells and present in the conditioned medium. There was evidence for a second small peak of mitogenic activity ( M r -~ 30 000) in the gel filtration chromatography under dissociative conditions, and the second mitogenic peak on the reverse-phase chromatography may have been a mitogen produced by bone cells rather than residual insulin or EGF. Also, any guanidine sensitive factor would have been inactivated prior to the characterization studies. Nevertheless, if SGF was not the only mitogen present, it still appeared to be the major bone cell mitogen present in the human bone cell conditioned medium. Production of mitogens by bone cells has previously been reported by Canalis et al. [15]. The mitogenic activity identified by Canalis in fetal rat calvaria cultures appeared to have a higher molecular-weight (25000) than that found for the small-molecular-weight human SGF. They also reported a stimulator of collagen production that appears to be closer in molecular weight (1020000) to the small molecular weight SGF. More recently these workers have purified a bone-derived growth factor that stimulates DNA synthesis and the incorporation of proline into collagen in rat calvaria [16] and have identified a B-transforming growth factor activity in calvaria conditioned medium [17]. The relationship of these factors to SGF remains to be determined. A human osteosarcoma cell line (U2-OS) has been shown to produce a growth factor immunologically related to platelet-derived growth factor [18]. However the apparent molecular weight of this factor was 30 000 even in the presence of 4 M guanidine HCI, and thus, it appears to be different from the mitogen released by the human bone cell cultures. Mitogenic activity was also evident in the culture medium from human skin cells in our study.

170 However, in the presence of 4 M guanidine, only a small a m o u n t of activity was evident in the region where S G F elutes during gel filtration chromatography, and the skin cell conditioned m e d i u m did not cross-react with the a n t i - S G F antibodies. Although we can not rule out the possibility that the skin cells are producing low levels of SCF, the skin cells clearly do not produce S G F in the same q u a n t i t y as bone cells do. These results also suggest the possibility that S G F p r o d u c t i o n may be limited to skeletal cells. N o r m a l h u m a n skin fibroblasts have previously been reported to release growth-promoting activities into the culture m e d i u m [11-14] i n c l u d i n g s o m a t o m e d i n - l i k e peptides. The mitogenic activity we observed in conditioned m e d i u m from h u m a n skin cell cultures may be related to these activities, since insulin-like growth factor 1 does stimulate mitogenesis in the assay used in this study [4]. The activities in the skin cell conditioned m e d i u m have not been sufficiently characterized to draw any further conclusions. The a p p a r e n t molecular weight of the majority of the mitogenic activity in the h u m a n bone cell conditioned m e d i u m decreased from an M r value of about 100000 to one of 10000 u n d e r dissociative conditions (4 M g u a n i d i n e HC1). This behavior was previously observed with the activity extracted directly from h u m a n b o n e matrix [7]. It suggests that S G F in the native state may be b o u n d to a n o t h e r larger-molecular-weight protein(s). If so, these other proteins are also being produced by the cultured b o n e cells. The significance of such interactions between S G F a n d other proteins is not clear at this time. We can speculate that they may play a role in the incorporation of S G F into bone matrix or in stabilizing SGF.

Acknowledgments The authors acknowledge the excellent technical assistance of David Olson a n d Barbara Fiori. This work was supported by N I H grant AM31062,

Veterans A d m i n i s t r a t i o n Research Support, and by Loma Linda University Research F o u n d a t i o n .

References 1 Farley, J.R. and Baylink, D.J. (1984) in Osteoporosis (Christiansen, C., Arnaud, C.D, Nordin, B.E.C., Parfit, A.M., Peck, W.A. and Riggs, Bi., eds.), Vol. I, pp. 423 440, Department of Clinical Chemistry, Glostrup Hospital, Denmark 2 Linkhart, TA., Mohan, S., Jennings, J.C., Farley, J.R. and Baylink, D.J. (1984) in Hormonal Proteins and Peptides (Li, C.H., ed.), Vol. 12, pp. 279 297, Academic Press, New York 3 Mohan, S., Linkhart, T., Farley, J. and Baylink, D. (1984) Caleif. Tissue Int. 36, $139-$145 4 Wergedal, J.E. and Baylink, D.J. (1984) Proc. Soc. Exp. Biol. Med. 176, 27 31 5 Gospodarowicz, D., Bialecki, H. and Greenburg, G. (1978) J. Biol. Chem. 253, 3736-3743 6 Puzas, J.E., Drivdahl, R.H., Howard, G.A. and Baylink, D.J. (1981) Proe. Soc. Exp. Biol. Med. 166, 113-122 7 Mohan, S., Jennings, J_ Linkhart, T. and Baylink, D. (1986) Biochim. Biophys. Acta 884, 234-242 8 Farley, J.R. and Baylink, D.J. (1982) Biochem. 21, 3502-3507 9 Jennings, J.C. and Baylink, D.J. (1985) in The Chemistry and Biology of Mineralized Tissues (Butler, W.T., ed.), pp. 48-53, Ebsco Media Inc., Birmingham, AL 10 Gospodarowicz, D. and Moran. J.S. (1976) Ann. Rev. Biothem 45, 511-558 11 D'Ercole, A.J., Applewhite, G.T .and Underwood, L.E. (198(/) Dev. Biol. 75, 315-378 12 Atkinson, P.R., Weidman, E.R., Bhaumick, B. and Bala, R.M. (1980) Endocrinology 106, 2006-2012 13 Clemmons, D.R.. Underwood, L.E. and Van Wick, J.J. (1981) J. Clin. Invest. 67, 10-17 14 Clemmons, D.R. and Van Wick, J.J. (1985) J. Clin. Invest. 75, 1914 1918 15 Canalis, E., Peck, W.A. and Raisz, L.G. (1980) Science 210, 1021 1023 16 Canalis, E. and Centrella, M. (1985) in The Chemistry' and Biology of Mineralized Tissues (Butler, W.T., ed.), pp. 7(t-74, Ebsco Media Inc., Birmingham. AL 17 Centella, M. and Canalis, E. (1985) Proc. Natl. Acad. Sci, USA82,7335 7339 18 Stracke, H., Schulz, A., Moeller, P., Rossol, S. and Schatz. H. (1984) Acta Endocrinol. 107, 16-24 19 Heldin, C., Westermark, B. and Wasteson, A. (198(t) J. Cell Physiol. 105, 235-246