JOURNAL OF BONE AND MINERAL RESEARCH Volume 5, Number 1, 1990 Mary Ann Liebert, lnc., Publishers

Characterization of the Latent Transforming Growth Factor p Complex in Bone JOHANNES PFEILSCHIFTER, LYNDA BONEWALD, and GREGORY R. MUNDY

ABSTRACT Transforming growth factor fl (TGF-B) is a 25 kD multifunctional polypeptide with pronounced effects on the proliferation and differentiation of a variety of cells in vitro. TGF-0 is a potent regulator of the activity of cells with the osteoblast phenotype and of isolated osteoclasts. It is released in increased amounts by bone cultures stimulated to resorb. Organ cultures of neonatal mouse calvaria produce TGF-fl as an inert largemolecular-weight complex that must be dissociated to release biologically active TGF-0 (5-8 ng/ml). We have shown recently that stimulated isolated avian osteoclasts release active TGF-/3 from this bone-derived biologically latent form. In this report we have characterized this bone latent form of TGF-0. Only small amounts of active TGF-0 (< 0.5 ng/ml) and no free binding protein are detectable in conditioned medium from bone cultures. Active TGF-0 can be detected in acid-treated calvarial conditioned media in which none or only minute amounts could previously be detected. Following incubation at 37"C, this activated TGF-0 gradually loses activity. Cross-linking studies using '251-labeledTGF-0 show that this loss of activity is due to TGF-/3 binding to a protein of approximately 300 kD. The TGF-0 latent complex accumulates in a linear manner and is stable in the presence of serum and the protease trypsin. Increases in temperature and pH extremes dissociate the complex to release active TGF-0. Decreases in pH result in an exponential increase in TGF-0 activity. Significant activation of the latent TGF-/3 was detectable at pH values as high as 4 and 5. Since the osteoclastic microenvironment is acidic during bone resorption, these data suggest that this acidic microenvironment may regulate TGF-/3 activity by releasing active TGF-6 from its latent complex.

INTRODUCTION RANSFORMING GROWTH FACTOR 0 (TGF-0) has potent effects on bone cells. TGF-0 affects the expression of alkaline phosphatase, collagen type I, and several other bone proteins in cells of the osteoblast phenotype('-') and can act as either a growth inhibitory factor or a mitogenic factor for this cell type, depending on the cell population and culture TGF-0 has been shown to inhibit bone resorption, to inhibit osteoclast formation,(9) and also to inhibit osteoclast activation.'") Active TGF-0 is released by resorbing bone.'") Bone is the largest tissue source (200 pg/kg of tissue)(12)[although platelets are the

T

most concentrated source'13'] and TGF-0 accumulates during endochondral bone formation just before ossification takes place.(14) Therefore TGF-8 may play a key role in bone remodeling by modulating both bone resorption and bone formation. TGF-0 is present in platelets and culture media from many cell lines in a latent inactive form and can be activated by acidification to pH 2.(15-18)Recent studies in human platelets suggested that a binding protein is responsible for masking TGF-0 activity and that homodimeric TGF-0 is released upon acidificati~n.('~-'~) This or a similar mechanism could play an essential role in the regulation of TGF-0 activity in bone tissue since an acid microen-

University of Texas Health Science Center, Department of Medicine/Endocrinology, 7703 Floyd Curl Drive, San Antonio, TX 78284-7877.

49

PFEILSCHIFTER ET AL.

50 vironment is generated under the ruffled border of the 0steoclast during bone resorption. The present studies were undertaken to determine the nature of the TGF-/3 molecule produced by bone cultures and to determine whether an acid environment enhances the activity of bone-derived TGF-P.

MATERIALS AND METHODS Acid activation of TGF-/3 in bone-conditioned medium In most experiments latent TGF-/3 in bone-conditioned medium and in fractions from gel filtration columns (see later) of bone-conditioned medium was activated by adding 1 N HCl to the sample to a final pH of 2. After 1 h incubation the pH was returned to 7.4 by the addition of 1 N NaOH. In experiments in which acid-activated and nonactivated samples were assayed, equal amounts of 1 N NaCl were added to the nonactivated samples and to all TGF-/3 standard dilutions to avoid differences in volume and 0smolarity.

Bone organ cultures and collection of bone-conditioned medium Frontal and parietal calvaria with periosteum from 4-day-old mice were cultured in serum-free BGJb medium (GIBCO) on steel grids at the interphase between medium and air as described previously.'") In most experiments the medium was buffered with 25 mM Hepes (Sigma) at pH 7.4 without additional bicarbonate buffering to avoid pH changes of the conditioned medium under C0,-free conditions. The medium was originally supplemented with 1 mg/ml of bovine serum albumin (BSA), Sigma. Since the amount of TGF-P activity generated in conditioned medium from bones without BSA supplementation was equivalent to that with BSA, BSA supplementation was omitted in later experiments. The calvaria were preincubated for 24 h under C0,-free conditions. Medium was then changed and conditioned medium collected after an additional 48 h. In one experiment, to test the effects of plasma proteins on TGF-/3 binding proteins and TGF-P activation, BGJ medium was buffered with bicarbonate and supplemented with 10% mouse plasma and calvaria were incubated in a 5% CO, atmosphere for 48 h.

Soft agar transformation assay TGF-/3 activity was measured through its effects on colony formation of NRK fibroblasts (clone 49F) in soft agar as described before.(1L)With each assay, a TGF-P standard curve was constructed using serial dilutions of known amounts of TGF-/3. TGF-P purified from bovine bone for these standards was generously provided by Dr. S. Seyedin (Collagen Corporation, Palo Alto, CA).

1251-labeled TGF-P binding assay All cell lines used for the binding assays, NRK (clone 49F), AKR-2B, and A-459, have been described previously for the binding of TGF-/3 to r e ~ e p t o r s . ( ~ The ~ . ~ binding ~-~~) procedure of Frolik et a1.(24)was used with several modifications as follows. Cells were plated in 24-well plates and allowed to grow to subconfluency in medium supplemented (for NRK cells DMEM; for ARK 2B cells McCoy's 5A medium; for A-549 cells Ham's Nutrient Mixture F12; all from Hazelton, Lenexa, KS) with 10% fetal calf serum (FCS, GIBCO). The cell layer was then washed and incubated for 1 h in serum-free medium containing 1 mg/ml of BSA and 24 mM Hepes. The monolayer was then washed again and then incubated with serial dilutions of each sample plus 180 pl of a 500 pM 1251-labeledTGF-/3 solution (Biomedical Technologies Inc., Stoughton, MA) in phosphate-buffered saline (PBS). After a 2-3 h incubation at room temperature with gentle mixing on a rocker platform, the incubation was terminated by aspirating the sample and washing the monolayer four times with BSAcontaining medium appropriate for each cell line. The cells were then solubilized by incubation with 600 pl lysing buffer (20 mM Hepes, 1% Triton X-100, 10% glycerol, and 0.01% BSA, pH 7.4) for 30 minutes at room temperature. Bound 1z51-labeledTGF-/3 was determined using a LKB 1274 Quatro Gamma counter.

Gel filtration Conditioned medium was dialyzed against 0.05 M ammonium bicarbonate buffer using a 3500 molecular weight cutoff membrane (Spectropor), lyophilized, and reconstituted in PBS and 25 mM Hepes, pH 7.4. The neutral reconstituted conditioned medium was applied to a Sephadex (3-200 column (1.5 x 75 cm; Pharmacia, Piscataway, NJ) that had been equilibrated with PBS, and fractions were eluted with PBS at a flow rate of 5 ml/h. Fractions of 5.1 ml were collected and sterilized by filtration through a 0.22 pm filter preflushed with 10% BSA to prevent protein binding (Millex-GV, Millipore Corp., BedfordJ MA), and then aliquots of each fraction were tested in the soft agar assay with acid activation and without prior acid activation. The column was calibrated using BSA (68 kD) and ribonuclease (14 kD;Bio-Rad). A second batch of dialyzed lyophilized conditioned medium was reconstituted in 10% acetic acid and was applied to the same column after re-equilibration of the column with 10% acetic acid. Fractions of 4.3 ml were collected at a flow rate of 5 ml/h. Aliquots of each fraction were lyophilized, reconstituted in DMEM medium, sterilized by filtration, and tested in the soft agar assay.

Cross-linkingprotocol A 500 pM lzsI-labeledTGF-0 solution (50 pl) was added to 50 p1 of (1) neutral conditioned medium, (2) acid-acti-

TGF-/3 IN BONE vated conditioned medium, (3) BGJ medium, or (4) acidactivated aliquots from the peak fraction (fraction 13, Fig. 2) of latent TGF-/3 after Sephadex (3-200 gel filtration using PBS. Samples were either incubated with the I2'-labeled TGF-P for 24 h at 37OC or the '2sI-labeled TGF-P was added without incubation. Disuccinimidyl suberate (10 81; DSS, Pierce, Rockford, IL) in dimethylsulfoxide (Sigma) was added to each sample for 1 h at 4"C, at which time the samples were then subjected to SDS-polyacrylamide gel electrophoresis.

Electrophoresis and autoradiography Sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) of the cross-linked samples was performed according to the procedure of Laem~nli'~'~ on 6% polyacrylamide gels using a Mini-Protean electrophoresis apparatus (Bio-Rad, Richmond, CA). Gels were stained for protein with Coomassie blue, dried, and subjected to autoradiography using Kodak X-OMAT film (Picker Corporation, Irving, TX). The molecular weight markers used were 45 kD ovalbumin, 68 kD bovine serum albumin, 94 kD phosphorylase b, 116 kD 0-galactosidase, 200 kD myosin (Bio-Rad), and 400 kD apoferritin (Sigma).

p f f , temperature, and enzyme studies To examine the influence of pH on activation of latent TGF-/3 in conditioned medium, conditioned medium was dialyzed against ammonium bicarbonate, lyophilized, and reconstituted in PBS and 25 mM Hepes, pH 7.4, as previously described, followed by dialysis of 1 ml aliquots against different pH solutions 2, 3,4, 5,6, or 7 in PBS and 25 mM Hepes using a 3500 molecular weight cutoff membrane (Spectrapor). After 24 h of dialysis and three medium changes, the samples were dialyzed back to pH 7.4, sterilized by filtration and tested for TGF-P activity in the soft agar assay. To examine the influence of temperature on activation of latent TGF-P in conditioned medium, conditioned medium was incubated in tightly sealed tubes at 37, 54, 65, or 100°C in a waterbath (Versa-bath, Fischer). The incubation was stopped by immediate freezing of the samples at -20°C. After completion of the incubation periods the samples were thawed at room temperature and assayed for active TGF-P in the soft agar assay. For enzyme studies, 50 pl of enzyme carrier 324A immobilized trypsin (Calbiochem, San Diego, CA) suspension was added to 1 ml conditioned medium to a final trypsin concentration of 6 U/ml. The bone-conditioned medium had been either acid activated or NaCl supplemented (as described in the first paragraph of Materials and Methods) before the trypsin was added. Samples were incubated at 37OC under constant agitation for various time periods,

51

as shown in Fig. 7, and the reaction was stopped by centrifugation and removal of supernatant from the immobilized trypsin. After the incubation, half of each sample was again acid activated or supplemented with NaCl.

RESULTS Latent and active TGF-0 in conditioned medium from mouse calvaria The TGF-j3 activity present in media harvested from bone organ cultures was less than 0.5 ng/ml but increased more than 10-fold to 5-8 ng/ml after acidification of the conditioned medium to pH 2. TGF-P activity in media harvested after 48 h of bone culture was measured in the soft agar assay and in the TGF-P receptor binding assay (Fig. 1). Although active TGF-P levels in neutral (non-acid-activated) conditioned medium were below the detection limit in the receptor binding assay (0.5 ng/ml), 0.2-0.3 ng/ml of TGF-P activity was measurable in the soft agar assay. The specificity of the TGF-P measurement in conditioned medium was examined by adding purified TGF-/3 to either conditioned medium or nonconditioned medium, and samples were compared for activity in the soft agar and radioreceptor assays (Fig. 1B). The receptor binding assay using A-549 cells showed a significant shift in the TGF-/3 binding curve when conditioned medium was used. This resulted in an overestimate of TGF-f3 activity. Only slight over- or underestimations occurred when NRK cells or AKR-ZB cells (data not shown) were used for the receptor binding assay. Conditidned medium did not significantly alter colony numbers in the soft agar transformation assay. Since the soft agar assay proved to be the most sensitive assay, it was used to measure TGF-P activity in the following experiments.

Evidence for latent TGF-0 complexes When the conditioned medium was eluted at neutral pH on Sephadex (3-200 gel filtration none of the fractions showed measurable TGF-P activity. After acid activation of these fractions TGF-P activity was detectable with a peak of TGF-/3 activity corresponding to a molecular weight > 100 kD (Fig. 2A). When the lyophilized conditioned medium was resuspended in 10% acetic acid and applied to the same column equilibrated with 10% acetic acid, TGF-/3 activity was detectable in fractions corresponding to a molecular weight of approximately 15 kD (Fig. 2B). This is the same position at which purified TGF/3 elutes. Cross-linking experiments were performed using acid-activated conditioned medium or the acid-activated peak fraction 13 from gel filtration experiments performed at neutral pH. After SDS-PAGE and autoradiography, a band migrating at approximately 300 kD was seen when the samples were incubated with the '%labeled TGF-/3 for 24 h at 37°C (Fig. 3, lanes 7 and 9). This band was not de-

PFEILSCHIFTER ET AL.

52

A

B A 549 - Receptor Binding Assay 2

c

3 =;'

0

&;

m P

.-

z 9

' X

1

I

I

92

2,_ -.-----

ul

L

1

I

I

2 2

4 4

8 8

0 1 16 6

1 0.1

1

10

100

1000

NRK - Receptor Binding Assay &

"1

61

-I 1

2

4

8

16

crn (x fold diluted)

NRK

1

2

- Soft Agar

4

8

16

crn (x fold diluted)

FIG. 1. Measurement of TGF-P activity in conditioned medium. (A) TGF-P activity in 48 h acid-activated murine calvaria conditioned medium (CM). Conditioned medium was diluted from 1:l to 1:32 and activity measured with the soft agar assay or the TGF-0 receptor binding assay using binding to NRK cells or A-549 cells. TGF-0 activity was calculated from serial dilutions of known amounts of purified TGF-P in BGJ medium. Less than 0.5 ng/ml of TGF-/3 activity was detectable in all systems without acid activation. (B) Influence of conditioned medium with the addition of purified TGF-0 on the measurement of TGF-/3 in the systems described in A. Serial dilutions of TGF-/3 were added to BGJ medium ).( or neutral BGJ medium incubated for 48 h with mouse calvaria (0).

tectable in conditioned medium that had not been acid activated (Fig. 3, lanes 4 and 9, suggesting no free binding protein. Another band greater than 400 kD, which also entered the gel, was detected in the acid-activated conditioned medium sample (Fig. 3, lanes 6 and 7). This band was also seen with the acid-activated peak fraction from the neutral gel filtration column, which appeared to decrease or convert to the lower molecular weight band upon incubation over time (Fig. 3, lanes 8 and 9).

Incubation of dissociated TGF-0 results in reassociation Incubation of acid-activated conditioned medium at 37°C also led to a 10-50% loss of TGF-0 activity that increased with time over a 24 h incubation period (Fig. 4, lane 2 and Fig. S), which could be due to degradation of the active TGF-0. This loss of TGF-0 activity (Fig. 4, lane 2) was completely abrogated by neutralization before incubation of the conditioned medium (Fig. 4, lane 3).

53

TGF-/3 IN BONE VO

64 kD

14 kD

I

I

I

Preservation of TGF-6 in conditioned medium: A potential protective role for the latent complex

n

&,

r

0

0

5

5

15

10

10

15

20

20

30

25

25

30

35

Fraction Number

FIG.2. Sephadex (3-200 gel filtration of conditioned medium. Approximately 100 ml conditioned medium was first dialyzed agAinst 0.05 M ammonium bicarbonate, concentrated by lyophilization, and resuspended in 1 ml PGS and 25 mM Hepes. Neutral (A) or acid-activated (B) concentrated conditioned medium was eluted on a Sephadex G-200 column (1.5 x 75 cm) with either PBS, pH 7.4 (A), or with 10% acetic acid (B) at 100 drops per tube. TGF-/3 activity in the fractions was measured in the soft agar assay with prior acid activation of the fractions. Vo, void volume; BSA, bovine serum albumin; RNA, ribonuclease.

Latent TGF-/3 amounts in conditioned medium increased linearly with time of incubation of the mouse calvaria. Latent activity was tested by acid activation of aliquots retrieved at sequential time periods and tested in the soft agar transformation assay (Fig. 6). When calvaria were incubated with medium containing 10% mouse plasma, the same linear increase was observed but starting from a higher baseline since mouse plasma itself contained approximately 40-50 ng/ml of latent TGF-/3. No increase in active TGF-/3 was observed with time in culture with or without mouse plasma or when 10% mouse plasma was incubated with conditioned medium harvested from the calvaria. This suggests the absence of active proteolytic enzymes in mouse plasma that can activate the complex. Addition of trypsin to bone-conditioned medium in concentrations that rapidly degraded active TGF-P in acid-activated bone-conditioned medium did not affect latent (complexed) TGF-8 that had not been acid activated (Fig.

7). Bone-conditioned medium incubated at 37°C up to 7 days displayed no spontaneous activation of TGF-P and no loss of latent TGF-/3 activity (data not shown).

kD 400 -

200

-

116-

92-

1

2

3

4

5

6

7

8

9

FIG. 3. Autoradiography of an SDS-PAGE in which 12SI-labeledTGF-P was incubated with conditioned medium, which was acid activated or untreated. All samples, the conditioned medium, and fraction 13 from the PBS Sephadex G-200 column were incubated for 0 or 24 h with 250 pM l*SI-labeledTGF-0 at 37°C. Cross-linking was then performed with the addition of 0.025 mM disuccinimidyl suberate (DSS) and the samples incubated for 1 h at 4"C, after which time all samples were subjected to a 6% SDS-PAGE under nonreducing conditions followed by gel drying and autoradiography. The position of molecular size markers are indicated: (1) 12SI-labeledTGF-P in PBS; (2) BGJ medium acid activated, no incubation; (3) BGJ medium acid activated, neutralized, 24 h incubation; (4) conditioned medium neutral, no incubation; (5) conditioned medium neutral, 24 h incubation; (6) conditioned medium acid activated, neutralized, no incubation; (7) conditioned medium acid activated, neutralized, 24 h incubation; (8) peak fraction (13) of latent TGF-P (Sephadex G-200 gel filtration) acid activated, neutralized, no incubation; (9) same sample as 8, neutralized, incubated for 24 h.

54

PFEILSCHIFTER ET AL.

3 2o

'1

1

T

0

1

3

2

5

4

8

16

24

32

40

40

Hours

6

FIG. 6. Linear accumulation with time of latent TGF-P in medium conditioned by calvaria. Calvaria were incuFIG. 4. Loss of acid-activated TGF-fl activity after incu- bated in BGJ medium containing 24 mM Hepes and 1 mg/ bation at 37°C. Conditioned medium was acid activated or ml BSA under C0,-free conditions ( 0 )or in BGJ medium left untreated and then was either directly assayed for containing 1.6 g/liter of bicarbonate and 10% mouse TGF-P activity or incubated for 24 h at 37°C and then as- plasma under 5% C o t conditions (0).TGF-P activity was sayed. Part of the incubated samples were acid activated measured in the soft agar assay after acid activation of the again before testing in the soft agar assay: (1) conditioned samples. TGF-/3 activity without acid activation was less medium acid activated, neutralized, no incubation; (2) than 0.5 ng/ml in all samples. conditioned medium acid activated, incubated for 24 h, neutralized just before testing; (3) conditioned medium acid activated, neutralized, incubated for 24 h, and then acid activated and neutralized again just before testing; (4) conditioned medium neutral; ( 5 ) conditioned medium neutral, incubated for 24 h; (6) conditioned medium neutral, incubated for 24 h, and then acid activated and neutralized just before testing.

Activation of latent TGF-6 in conditioned media by p H and temperature Increases in temperature led to activation of latent TGF-

fl followed by loss of TGF-P activity presumably due to

0J ,

I

1

I

I

I

1

0

4

8

12

16

20

24

thermal denaturation. Activation of TGF-fl was observed at 65°C at 30 minutes while thermal denaturation was occurring and complete activation at 100°C at 3 minutes followed by thermal denaturation (Fig. 8A). At 54"C, activation of latent TGF-fl was not observed. However, there was 50% loss of latent TGF-/3 activity when tested in the soft agar assay following acid activation of samples taken at the times indicated (Fig. 8B). An exponential activation of latent TGF-P could also be observed with decreasing pH (Table 1). Although the majority of the latent TGF-fl was activated at pH 2, significant increases in TGF-fl activity were observed at pH 4 (2.3 f 0.04 ng/ml) and 5 (3.0 f 0.6 ng/ml).

Hours

FIG. 5. Loss of TGF-P activity in acid-activated conditioned medium. Conditioned medium was acid activated DISCUSSION and incubated at 37°C for the lengths of time indicated in the figure. Each time point represents four samples that A high degree of homology exists between the human were neutralized before measurement in the soft agar and murine forms of TGF-P, with only a single amino acid assay. substitution. ( 2 6 ) The molecule is identical in the human, simian,'27)bovine,(zB) porcine,(29'and avian(3o)species. By sequence data, it appears that TGF-P is synthesized as a 390-amino-acid inactive precursor that must be processed via cleavage and dissociation to yield the 112-amino-acid

55

TGF-8 IN BONE

-5

5

r

4

-

% a

2

Qa L

P

0

I

I

I

I

I

I

I

0

4

8

12

16

20

24

Hours FIG. 7. TGF-/3 activity in conditioned medium after incubation with low concentrations of immobilized trypsin. Sepharose beaded trypsin (6 U/ml) was added to conditioned medium, which was either acid activated (B)or untreated (0) before the addition of trypsin. The protease was removed after incubation by centrifugation. After trypsin treatment aliquots of the acid-activated media were reactivated (0)and aliquots of the neutral media were activated ( 0 )immediately before the soft agar assay was performed.

6l A

'1

curring form of latent TGF-P has been characterized. Several laboratories have shown acid activation of latent TGF/3(15-1s)and recently that the proteases plasmin and cathepsin D activae kidney fibroblast latent TGF-P whereas elastase and thrombin do n ~ t . ' ~We ~ . have ~ ~ )recently shown that isolated osteoclasts stimulated with retinol or bone particles activate bone latent TGF-P.(36)Since pH 4 has been found in osteoclast l a c ~ n a e , ( ~we ' ~ sought ~ ~ ) to investigate the relationship of acid conditions with activation of bone-derived latent TGF-P and to characterize this bone latent form in this report. Our study shows that bone organ cultures release TGF-P predominantly as an inactive large-molecular-weight complex. This complex is composed of homodimeric TGF-b and an unknown binding protein(s). TGF-P, although present as a latent complex, is activated by an increase in temperature and also by changes in pH compatible with those observed under the ruffled border of resorbing osteoclasts. The TGF-@-binding protein complex is, however, stable in the presence of serum and the protease trypsin. Cross-linking studies showed that 1a51-labeledTGF-P specifically bound to a large-molecular-weight protein in acid-treated bone-conditioned medium that appeared to convert to a 300 kD complex. Binding of the '251-labeled TGF-fi to this large protein was only observed after neutralization followed by incubation at 37OC for 24 h. We found that under these same conditions TGF-P can be fully recovered upon a second acid activation. This suggests noncovalent binding of TGF-P to this 300 kD complex. The initial latent TGF-P complex in bone-conditioned media is not necessarily the same as the latent complex formed after acidification followed by neutralization and further incubation. However, this secondary reassociation form can also be reactivated to give previous amounts of TGF-P activity. It may be that bone latent TGF-P is initially produced by bone in a complex similar or related to that of the platelet c ~ m p l e x ~ and ~ ~ ~upon ' ~ ) acidification either rebinds or binds to a second component, such as a2macroglobulin. f32,33) If the latter is the case, then acidification can also release free TGF-/3 from binding to this serum protein. The binding protein(s) for TGF-P may have several important functions. Binding proteins for other growth factors have been shown to serve as a transport mechan i s m ~ , ' ~may ~ . ~ protect ~' growth factors from enzymatic d i g e ~ t i o n , ( ~may ' . ~ ~aid ) in activation of the enhance its effects,'") or inhibit its activity,(") or may serve as a storage site for the growth factor.(*6) Some growth factors have different binding proteins to serve several of these functions. In the case of TGF-0, the binding protein may be responsible for determining the activity of TGF-P at a local site rather than its production by the tissue. Since the binding protein also seems to protect TGF-P and forms as very stable complex with TGF-8, latent and protected TGF-P levels may be maintained in a tissue and then later the active molecule of TGF-P may be released under appropriate and regulated conditions. We found, for instance, that concentrations of trypsin that rapidly inacti-

2huoc .

'I

Minutes

/

,

!

,

,

,

I

0

8

16

24

32

40

48

Hours

FIG. 8. Effects of temperature on activation of TGF-P in neutral (nonactivated) conditioned medium. (A) Samples were incubated at 4, 37, 54, 65, or 100°C for 60 minutes and TGF-/3 activity was measured without further treatment of the samples. (B) The sample was incubated at 54°C for the amount of time indicated. TGF-/3 activity was then measured in the soft agar assay with ( 0 ) or without (0)acid activation.

subunit of the TGF-P h~rnodimer.'~') Human platelet latent TGF-b is the most intensely studied form of latent TGF-P. This molecule is comprised of the TGF-P homodimer of 25 kD and the homodimer precursor form, which is covalently attached to a single binding protein with a molecular weight of 135 kD.('O) Platelet-derived TGF-P also appears to bind a-macroglobulin in serum to produce a secondary latent complex.("'.33)No other naturally oc-

56

PFEILSCHIFTER ET AL. TABLE1. EFFECTS OF PH ON ACTIVATION OF TGF-0 IN CONDITIONED MEDIUM^ TGF-P activity

(ns/ml) 63 f 0.lb 25 f 0.2b 2.3 f 0.4b 3.0 f 0.6b 1.2 f. 0.4 0.6 f 0.1 aconditioned medium (200 ml) was concentrated 30 times after dialysis against ammonium bicarbonate by lyophilization and then resuspended in PBS and 25 mM Hepes. Aliquots ( I ml) of the resuspended material were dialyzed against PBS, pH 2, 3 , 4 , 5 , 6, or 7, for 24 h and then all aliquots dialyzed back to pH 7.4 and TGF-@activity was measured in the soft agar assay. Each sample was measured in triplicate. bSignificantlydifferent from pH 7, p < 0.001 (Student's t-test).

vate free TGF-P do not affect TGF-/3 when it is part of the latent complex. If the release of TGF-/3 from a latent complex is the mechanism by which active TGF-P is made available for its local biologic effects, then the conditions under which this release occurs are vital for understanding the control mechanisms for regulation of TGF-/3 activity. We have used as our model system for examining the activation of TGF-/3 an organ culture system that represents a competent, complex tissue composed of the heterogeneous cell types present in normal bone. Under these conditions, however, only a relatively small percentage of the total TGF-/3 produced by the tissue was in an active form. Likewise, excess free binding protein could not be detected in our cross-linking experiments. Most of the TGF-/3 produced by bone organ cultures persists as the latent TGF-(3 complex and accumulates linearly in this latent form with time in bone-conditioned medium. To determine if plasma components could convert the latent TGF-P complex to active TGF-P, we incubated the latent TGF-P complex with mouse plasma in one experiment and in another cultured the calvaria in the presence of mouse plasma. Neither alone nor in combination with the bone cultures was mouse plasma able to increase the amount of active TGF-P present in the bone-conditioned medium. Interestingly, mouse plasma itself contained large amounts of the latent TGF-P complex, suggesting the absence of active proteolytic enzymes in mouse plasma necessary to activate the latent plasma complex.

Although the bone latent TGF-0 complex seems to be quite stable under culture conditions that resemble a physiologic tissue environment, dissociation occurred with changes in pH or increases in temperature. Nevertheless, the complex remained quite resistant to the effects of temperature. The latent complex is quite stable, as shown by the fact that there was no detectable dissociation up to 2 days at 54°C. Activation by temperature and pH indicates noncovalent binding of TGF-P to the latent complex, presumably through hydrogen bonds and van der Waals forces. Latent TGF-/3 is produced by many tissues and cell types, including platelets, fibroblasts, fetal rat calvaria, and osteoblasts, as detected by acidification of conditioned media to pH 2.(6,34,47,48) Here we show that a significant percentage of the total material in conditioned media from bone cultures can be activated at pH 5. Lyons et al. also found significant activation of fibroblast latent TGF-/3 at pH 5.5-4.5, the same latent TGF-/3 pool that was activated by plasmin. However, pH may play a more significant role in TGF-P activation in bone. Baron et al. used acridine orange to determine that there was an acidic zone beneath osteoclasts in chick Silver et al. have demonstrated that adherent osteoclasts can reduce the pH of the contact zone with the culture dish to pH 3,14') which would be more than sufficient for activation of significant amounts of active TGF-0. We found that dissociation of the TGF-/3 complex occurred gradually but exponentially with decreasing pH. Presumably the complex completely dissociated at pH 2, and a significant increase in active TGF-P was also observed at pH 5. Although this increase at pH 5 comprised only 4.7% of the total latent TGF-P in bone-conditioned medium, this increase may still be very significant. Conceivably, even calvarial conditioned media could contain biologically significant amounts of active TGF-P. TGF-P is known to have potent effects in concentrations as low as 0.1 ng/ml. Maximal effects may be observed in some assays with as little as 1 ng/ml.'3) TGF-0 is chemotactic at the picogram per milliliter or femtomole level.'5o) Local activation of only a few percentage of latent TGF-/3 may therefore be sufficient for achieving optimal TGF-P activity in tissue. One niche in the bone cell microenvironment that would offer ideal conditions for TGF-0 activation is under the ruffled border of the osteoclast during bone resorption. Low pH at the surface of resorbing bone related to a proton pump on the ruffled border of the osteoclast has recently been d e m ~ n s t r a t e d . ' ~ ' .In ~ ~a) previous study we have shown that there is an increase in active TGF-/3 in bone-conditioned medium harvested from bone cultures during bone resorption,'") and we have shown that isolated osteoclasts activate bone latent TGF-P.(io)It is therefore possible that latent TGF-P from surrounding bone tissue or stored in bone matrix becomes activated and released at this site during the bone resorption process. Since TGF-P has been shown to have stimulatory effects on osteoblastic cells in a variety of in vitro systems and inhibitory effects on osteoclast formation and function, activa-

TGF-0 IN BONE

57

tion of TGF-0 during bone resorption may therefore represent one of the mechanisms that link bone resorption to new bone formation.

12. Seyedin SM, Thompson AY, Bentz H, Rosen DM, McPherson JM, Conti A, Siege1 NR, Galluppi GR, Piez KA 1986

13.

ACKNOWLEDGMENTS We are grateful to Thelma Barrios and Nancy Garrett for expert secretarial assistance. This work was performed with support from Grants CA40035 from the National Cancer Institute and AR28149 from the National Institutes of Health. Pfeilschifter is a recipient of a fellowship from the Deutsche Forschungsgemeinschaft Bonn-Bad Godesberg, Federal Republic of Germany.

14.

15.

16.

Cartilage-inducing factor-A. Apparent identity to transforming growth factor-0. J Biol Chem 2615693-569s. Assoian RK, Komoriya A, Meyers CA, Miller DM, Sporn MB 1983 Transforming growth factor-beta in human platelets. Identification of a major storage site, purification, and characterization. J Biol Chem 258:7155-7160. Carrington JL, Roberts AB, Flanders KC, Roche NS, Reddi AH 1988 Accumulation, localization, and compartmentation of transforming growth factor 0 during endochondral bone development. J Cell Biol 107:1969-1975. Coffey RJ Jr, Shipley GD, Moses HL 1986 Production of transforming growth factors by human colon cancer lines. Cancer Res 46:1164-1169. Lawrence DA, Pircher R, Jullien P 1985 Conversion of a high molecular weight latent beta-TGF from chicken embryo fibroblasts into a low molecular weight active beta-TGF under acidic conditions. Biochem Biophys Res Commun 133:

1. Noda M 1989 Transcriptional regulation of osteocalcin pro-

1026-1034. 17. Lawrence DA, Pircher R, Kryckve-Martinerie C, Jullien P 1984 Normal embryo fibroblasts release transforming growth factors in a latent form. J Cell Physiol 121:184-188. 18. Sporn MB, Roberts AB, Wakefield LM, Assoian RK 1986

duction by transforming growth factor 0 in rat osteoblastlike cells. Endocrinology 124:612-617. Centrella M, McCarthy TL, Canalis E 1987 Transforming growth factor beta is a bifunctional regulator of replication and collagen synthesis in osteoblast-enriched cell cultures from fetal rat bone. J Biol Chem 262:2869-2874. Pfeilschifter J, D’Souza SM, Mundy GR 1987 Effects of transforming growth factor-beta on osteoblastic osteosarcoma cells. Endocrinology 121:212-218. Noda M, Yoon K, Prince CW, Butler WT, Rodan GA 1988 Transcriptional regulation of osteopontin production in rat osteosarcoma cells by type 0 transforming growth factor. J Biol Chem 263:13916-13921. Gehron-Robey PG, Young MF, Flanders KC, Roche NS, Kondaiah P, Reddi AH, Termine JD, Sporn MB, Roberts AB 1987 Osteoblasts synthesize and respond to transforming growth factor-type 0 (TGF-0) in vitro. J Cell Biol 105457-

Transforming growth factor-beta: Biological function and chemical structure. Science 233532-534. Miyazono K, Hellman U, Wernstedt C, Heldin CH 1988 Latent high molecular weight complex of transforming growth factor 01. Purification from human platelets and structural characterization. J Biol Chem 2635407-6415. Wakefield LM, Smith DM, Flanders KC, Sporn MB 1988 Latent transforming growth factor+ from human platelets. A high molecular weight complex containing precursor sequences. J Biol Chem 263:7646-7654. Tucker RF, Branum EL, Shipley GD, Ryan RJ, Moses HL 1984 Specific binding to cultured cells of 1251-labeled type beta transforming growth factor from human platelets. Proc Natl Acad Sci USA 81:6757-6761. Massague J, Like B 1985 Cellular receptors for type beta transforming growth factor. Ligand binding and affinity labeling in human and rodent cell lines. J Biol Chem Mo:

REFERENCES

2.

3.

4.

5.

19.

20.

21.

22.

463. 6. Rosen DM, Stempien SA, Thompson AY, Seyedin SM 1988

2636-2645. 23. Ruff E, Rizzino A 1986 Preparation and binding of radioac-

Transforming growth factor-beta modulates the expression of osteoblast and chondroblast phenotypes in vitro. J Cell Physiol 134:337-346. 7. Centrella M, Massague J, Canalis E 1986 Human platelet-derived transforming growth factor-beta stimulates parameters of bone growth in fetal rat calvariae. Endocrinology 1 1 9

tively labeled porcine transforming growth factor type beta. Biochem Biophys Res Commun 138:714-719. 24. Frolik CA, Wakefield LM, Smith DM, Sporn MB 1984 Characterization of a membrane receptor for transforming growth factor-beta in normal rat kidney fibroblasts. J Biol Chem 259:10995-1 IOOO. 25. Laemmli UK 1970 Cleavage of structural protein during the assembly of the head of bacteriophage T4. Nature 227:680-

2306-2312. 8. Pfeilschifter JP, Seyedin SM, Mundy GR 1988 Transforming growth factor beta inhibits bone resorption in fetal rat long bone cultures. J Clin Invest 82680-685. 9. Chenu C, Pfeilschifter J, Mundy GR, Roodman GD 1988 Transforming growth factor beta inhibits formation of osteoclast-like cells in long-term human marrow cultures. Proc Natl Acad Sci USA 855683-5687. 10. Oreffo ROC, Bonewald L, Garrett IR, Seyedin S, Mundy GR 1988 Transforming growth factors 0 1 and 11 inhibit osteoclast activity (abstract). J Bone Min Res 439:S178. 11. Pfeilschifter J, Mundy GR 1987 Modulation of type beta transforming growth factor activity in bone cultures by osteotropic hormones. Proc Natl Acad Sci USA 84:20242028.

683. 26. Derynck RJ, Jarrett JA, Chen EY, Goeddel DV 1986 The

murine transforming growth factor-beta precursor. J Biol Chem 261:4377-4379. 27. Sharples K, Plowman GD, Rose TM, Twardzik DR, Purchio AF 1987 Cloning and sequence analysis of simian transforming growth factor-0 cDNA. DNA 6:239-244. 28. Van Obberghen-Schilling E, Kondaiah P, Ludwig RL, Sporn MDB, Baker CC 1987 Complementary deoxyribonucleic acid cloning of bovine transforming growth factor-pl. Mol Endocrinol 1:693-698. 29. Derynck R, Rhee L 1987 Sequence of the porcine transforming growth factor-0 precursor. Nucleic Acids Res 15:3187.

PFEILSCHIFTER ET AL.

58 30. Jakowlew SB, Dillard PJ, Sporn MB, Roberts AB 1988 Nu-

42. Cohen KL, Nissley SP 1976 The serum half-life of somato-

cleotide sequence of chicken transforming growth factor-beta I (TGFPI). Nucleic Acids Res 168730. 31. Derynck R, Jarrett JA, Chen EY, Eaton DH, Bell JR, Assoian RK, Roberts AB, Sporn MB, Goeddel DV 1985 Human transforming growth factor-0 cDNA sequence and expression in tumor cell lines. Nature 316:701-705. 32. O’Connor-McCourt MD, Wakefield LM 1987 Latent transforming growth factor @ in serum: A specific complex with alpha, macroglobulin. J Biol Chem 262:14090-14099. 33. Huang SS, O’Grady P, Huang JS 1988 Human transforming growth factor p a,-macroglobulin complex is a latent form of transforming growth factor p. J Biol Chem 263:1535-

medin activity: Evidence for growth hormone dependence. Acta Endocrinol (Copenh) 83:243-258. Isackson PJ, Silverman RE, Blaber M, Server AC, Nichols RA, Shooter EM, Bradshaw RA 1987 Epidermal growth factor binding protein: Identification of a different protein. Biochemistry 26:2082-2085. Elgin RG, Busby WH, Clemmons DR 1987 An insulin-like growth factor (IGF) binding protein enhances the biologic response to IGF-I. Proc Natl Acad Sci USA 84:3254-3258. Knauer D, Smith G 1980 Inhibition of biological activity of multiplication-stimulating activity by binding to its carrier protein. Proc Natl Acad Sci USA 77:7252-7256. Smith GL 1984 Somatomedin carrier proteins. Mol Cell Endocrinol 37:83-89. Centrella M, Canalis E 1985 Transforming and nontransforming growth factors are present in medium conditioned by fetal rat calvariae. Proc Natl Acad Sci USA

43.

44.

45.

1541. 34. Lyons RM, Keski-Oja J, Moses HL 1988 Proteolytic activa-

46.

tion of latent transforming growth factor-p in fibroblast-conditioned medium. J Cell Biol. 106:1659-1665. Keski-Oja J, Lyons RM, Moses HL 1987 Inactive secreted form(s) of transforming growth factor-0 (TGFP): Activation by proteolysis. J Cell Biochem (Suppl) llA:60. Oreffo ROC, Mundy GR, Seyedin SM, Bonewald LF 1989 Activation of the bone-derived latent TGF beta complex by isolated osteoclasts. Biochem Biophys Res Commun 158: 817-823. Fallon MD 1984 Bone resorbing fluid from osteoclasts is acidic: An in vitro micropuncture study. In: Cohn CV, Fujita T, Potts JT Jr, Talmadge RV (eds) Endocrine Control of Bone and Calcium Metabolism, Vol 8A. Elsevier Science Publishers, Amsterdam, pp. 144-146. Baron R, Neff L, Louvard I, Courtoy PJ 1985 Cell-mediated extracellular acidification and bone resorption: Evidence for a low pH in resorbing lacunae and localization of a 100-kD lysosomal membrane protein at the osteoclast ruffled border. J Cell Biol 101:2210-2222. White RM, Nissley SP, Moses AC, Rechler MM, Johnsonbaugh RE 1981 The growth hormone dependence of a somatomedin-binding protein in human serum. J Clin Endocrinol Metab 53:49-57. Balfour WE, Tunnicliffe HE 1960 Thyroxine binding by serum proteins. J Physiol (Lond) 153:179-198. Oppenheimer JH 1968 Role of plasma proteins in the binding, distribution, and metabolism of the thyroid hormones. N Engl J Med 278:1153-1162.

47.

35.

36.

37.

38.

39.

40. 41.

82:7335-7339. 48. Pircher R, Jullien P, Lawrence DA 1986 &transforming

growth factor is stored in human blood platelets as a latent high molecular weight complex. Biochem Biophys Res Commun 136:30-37. 49. Silver IA, Murrills RJ, Etherington OJ 1988 Microelectrode studies on the acid microenvironment beneath adherent macrophages and osteoclasts. Exp Cell Res 175266-276. 50. Wiseman DM, Polverini PJ, Kamp DW, Leibovich SJ 1988 Transforming growth factor-beta (TGFP) is chemotactic for human monocytes and induces their expression of angiogenic activity. Biochem Biophys Res Commun 157:793-800.

Address reprint requests to:

Dr. Lynda Bonewald University of Texas Health Science CentPr Department of Medicine/Endocrinology 7703 Floyd Curl Drive San Antonio, TX 78284-7877

Received for publication March 7, 1989; in revised form July 17, 1989; accepted July 18, 1989.