JOURNAL OF BONE AND MINERAL RESEARCH Volume 5, Number 10, 1990 Mary Ann Liebert, Inc., Publishers

In Vivo Effects of Human Recombinant Transforming Growth Factor 0 on Bone Turnover in Normal Mice CHRISTIAN MARCELLI, A. JOHN YATES, and GREGORY R. MUNDY

ABSTRACT Reports of the effects of TGF-0 on bone cells are conflicting and controversial. Different cell culture and organ culture models for both osteoblasts and osteoclasts have given different responses. In some the effects are dependent on prostaglandin synthesis, and in others they are prostaglandin independent. To determine the effects of TGF-/3 on osteoblasts and osteoclasts in vivo and the role of prostaglandins in mediating these effects, we injected 2.5-5 pg TGF-P into the subcutaneous tissue overlying the calvariae of normal mice for 2-5 days and compared the morphologic responses in underlying calvarial bone with those in mice injected with vehicle alone. TGF-P treatment had no effect on plasma calcium concentration. However, TGF-P caused a marked increase in periosteal thickness (fivefold) and cellularity, morphologic changes in osteoblasts, and new mineralized bone formation. These effects were localized to the site of injection and were partially inhibited by concomitant indomethacin treatment. There was a parallel increase in osteoclast numbers in adjacent marrow spaces, and the osteoclasts formed were unusually large. In contrast, no increase in the numbers of osteoclasts was seen in indomethacin-treated animals. These data show that TGF-P has powerful effects on local bone cell function in vivo and that these effects may be mediated, in part, by prostaglandin generation.

INTRODUCTION ITHIN THE PAST 10 YEARS, it has become clear that skeletal growth and bone remodeling are regulated by local as well as systemic factors.''.') Among the local factors, transforming growth factor /3 (TGF-P) is likely to play an important ole.(^.^) TGF-P is stored within the bone matrix,(') and is released in an active form during bone resorption.(6) Although TGF-fi has potent and powerful effects on bone cells in vitro,"-" reports on its effects on both osteoclasts and osteoblasts have been conflicting, and opposite effects have been observed in different in vitro model system^.('^-*^) In part, these differences can be ascribed t o the capacity of TGF-/3 to stimulate prostaglandin synthesis in some organ cukure s y ~ t e m s . ( " ~ 'In ~) the one in vivo study reported, TGF-P was shown t o have powerful effects on periosteal osteoblasts,('s)but the study was short-term, and the effects on osteoclasts and depen-

dence of the responses on prostaglandin synthesis were not assessed. In this study we have examined the effects of TGF-fi on bone in vivo using a recently described in which the factor is injected in small volumes into the subcutaneous tissue overlying the calvariae of normal mice and its effects on bone turnover examined using bone histomorphometry. The responses were followed for over 1 month, and the role of prostaglandins as mediators of the effect of TGF-P were examined by the use of concomitant treatment with indomethacin.

MATERIALS AND METHODS Human recombinant TGF-f3,, a kind gift of Drs. Daniel Twardzik and Anthony Purchio (Oncogen, Seattle, WA), was dissolved in phosphate-buffered saline (PBS) containing 1 mM HCI and 1% bovine serum albumin (BSA).

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

1087

1088

MARCELLI ET AL.

Indomethacin was purchased from Sigma (St. Louis, MO), stored in 95% ethanol, and suspended in 19’0 polyethylene glycol 8000 in distilled water immediately before each injection. The effects of TGF-(3, were tested in ICR Swiss mice aged 5-6 weeks and weighing 18-27 g. In the first experiment, mice received either 5 pg TGF-(3, in 10 pl or vehicle alone once a day for 5 days (n = 5 for both groups). This dose of TGF-(3, was selected because in previous studies we had found similar doses of other factors, including interleukin-1 (IL-l), parathyroid hormone (PTH), parathyroid hormone-related hormone (PTH-rP), and tumor necrosis factor (TNF), producing significant effects in this model system.(16-’8)Injections were given by Hamilton syringe (Reno, NV) into the subcutaneous tissues overlying the sagittal suture of the calvaria (Fig. 1). Orbital blood was sampled before treatment and at 2, 3, and 5 days thereafter. Plasma ionized calcium was measured in the whole blood using a Ciba Corning 634 ISE Ca2+/pH analyzer (Medfield, MA). Calcium values were adjusted using the internal algorithm of the instrument to a pH of 7.4. The mice were killed 24 h after the last injection and calvariae taken for histology (see later). In a second experiment, to study the time course of TGF-(3, effects, 35 mice were divided into seven groups. Mice in groups 1, 2, 3, and 4 received 2.5 pg TGF-P, in 10 pl once a day injected into the subcutaneous tissue overlying the right side of the calvaria for 5 days from day 1 to day 5 of the experiment and were killed on days 6, 10, 20, and 40 (i.e., a, 5, 15, and 35 days after the last injection), respectively. Animals in groups 5 and 6 received vehicle alone for 5 days and were killed on days 6 and 40 (i.e., 1 and 35 days after the last injection), respectively. Mice in

sagittal suture

lambdoid suture

I

FIG. 1. Drawing of mouse calvaria. An aliquot of 10 pl solution containing TGF-(3, was injected into the subcutaneous tissues overlying the sagittal suture (experiment 1) or the right Darietal bone (experiments 2 and 3). Dotted line and hatched area on parietal bones indicate the plane of coronal sections and th? area through which sections were cut, respectively.

group 7 received 2.5 pg TGF-P, in 10 pI once a day for 2 days (day 4 and day 5 of the experiment) and were killed 1 day later (day 6). All the animals also received two injections of oxytetracycline (Pfizer, New York, NY), 20 mg/kg body weight, given intraperitoneally at 6 days and at 1 day before they were killed. In a third experiment designed to determine whether the effects of TGF-(3, on bone remodeling were mediated through prostaglandin production, 20 mice were divided into four equal groups. One group received 2.5 pg TGBF(3, in 10 pl once a day for 5 days into the subcutaneous tissue overlying the right side of the calvaria. In addition to this treatment with TGF-(3,, a second group received concomitant indomethacin (50 pg in 100 pl) injected subcutaneously in the flank region every 8 h, the first injection being given 3 h before the first TGF-(3, injection. A third group of mice were injected with indomethacin alone, and the final group acted as untreated controls. All the animals were killed 24 h after the last injection. All groups not given either TGF-(3, or indomethacin were injected with the appropriate vehicle (both vehicles were given to the untreated control group).

Histomorphometry Mice were killed by excess ether inhalation, and the intact calvariae were removed by dissection, fixed in 80% alcohol, dehydrated in graded alcohols, and embedded in methyl methacrylate without decalcification. ( ” ) Nonconsecutive coronal sections (14) were cut from each specimen at the same level of the parietal bone (Fig. 1). Eight 8 pm thick sections were stained with Masson-Goldner trichrome, and six 10 pm thick sections were mounted unstained. Histologic measurements were carried out using a digitizing tablet and the Bioquant IV image analysis system (R & M Biometrics, Nashville, TN) on the injected and noninjected sides of the sections in a standard length of bone between the sagittal suture and the muscle insertion at the lateral borders of each bone. Measurements on each side of the calvariae consisted of (1) thickness of the periosteum (in pm), (2) thickness of the parietal bone between inner and outer periosteal surfaces (in pm), both measured at three separate points, (3) relative bone marrow area (i.e., percentage of total bone area between inner and outer periosteal surfaces occupied by bone marrow), (4) the number of osteoclasts within the marrow cavity (expressed per mm’ total bone area), and ( 5 ) the calcification rate, which was determined by dividing the average distance (in pm) between the two tetracycline labels (measured on unstained sections at six or more separate points) by the time interval between the labeling periods (S days). was therefore expressed in micrometers per day The (pmlday). All measurements were carried out on four sections. Statistical analysis was performed using analysis of variance and the Fisher PLSD test.

1089

EFFECTS OF TGF-o ON BONE IN VIVO

W

n Y

FIG. 2. Photomicrographs of undecalcified coronal sections from parietal bones stained with Masson-Goldner trichrome (A and B, x 95; C and D, x 235). TGF-o,, 5 pg/day for 5 days (A), induced a marked increase in periosteal thickness and production of woven bone in periosteum. An increased number of osteoclasts (arrows) are present within resorption lacunae in bone marrow cavities. In bone from a vehicle-treated mouse (B) the periosteum is thin, no osteoclasts are present, and bone surfaces are smooth. Under higher magnification, TGF-P,-induced proliferation and maturation of the periosteal cells can be seen (C). This is evident especially in the deepest cells, which produced abundant woven bone matrix (m) some of which was already mineralized (b). In contrast, the periosteum of a vehicle-treated mouse (D) is thin and the periosteal cells are flat.

RESULTS Short-term effects of 5 &day of TGF-0, on bone morphology In this first experiment daily injections of TGF-P], 5 &day for 5 days, over the calvariae of mice induced the proliferation of periosteal cells and a marked increase in periosteal thickness along the entire outer surface of the bone (Fig. 2A). The periosteum over the outer surface of the calvariae in vehicle-treated animals was very thin and

periosteal cells were flat (Fig. 2B). However, the deepest cells lining the bone surface were larger and surrounded by matrix proteins that stained red with Goldner trichrome (Fig. 2D). Many of those cells were active osteoblasts since we observed a continuous mineralization front, labeled with tetracycline, along the outer periosteal surface in vehicle-treated animals. As shown in Fig. 2C, the increased thickness of the periosteum in the TGF-P,-treated mice was mainly due to the hypertrophy of the inner layer of cells, which became cuboidal and produced abundant ma-

1090

MARCELLI ET AL.

trix, some of which had already mineralized by day 6, forming new woven bone. In addition, TGF-0, stimulated bone resorption at the endosteal surface (Fig. 2A). Giant multinucleated osteoclasts were seen in enlarged bone marrow cavities or along the inner periosteal surface of the calvariae. Few osteoclasts were observed along the outer periosteal surface. N o statistical difference in plasma ionized calcium levels was found between TGF-P,-treated and vehicle-treated mice at any time during the study (Table 1). No differences were observed histologically between sections from the spine or femur of TGF-P,-treated mice and those from controls. An increase in the weight of the animals during the experiment was observed without any statistical difference between animals treated with TGF-0,and vehicle-treated animals (Table 1).

ization and lined by active osteoblasts (Fig. 3E). At day 40, the woven bone was covered by a layer of new lamellar bone and the appearance of periosteal cells returned to that of control (Fig. 3F). Vehicle-treated animals killed at day 40 (group 6) showed an increase in their calvarial thickness by approximately 75% in comparison with the

periosteal thickness (vm) 250

1

calvarial thickness (vm)

Time course of the effects of 2.5 p/day of TGF-0,: Bone histomorphometry These studies were performed to examine both the earlier events and the longer term consequences of the changes in calvarial bones induced by TGF-0,. Figure 3 illustrates the time course of the effects of TGF-PI, 2.5 pg/ day, injected over the right side of the calvaria, and Fig. 4 gives the results of histomorphometric analysis. After 2 injections of TGF-#?,, a significant increase in periosteal thickness was observed as well as changes in shape and size of the deepest cells of the periosteum (Fig. 3B). After 5 days of treatment (on day 6 ) , a fivefold increase in the periosteal thickness was observed on the treated side whereas the increase was threefold on the nontreated side (Fig. 4A). At this time almost all the periosteal surface on the treated side was covered by new bone matrix, which had already mineralized in some areas (Fig. 3C). At later time points a progressive decrease in periosteal thickness was observed (Fig. 3D and E). However, 35 days after the end of the treatment the periosteal thickness was still significantly increased on both the treated and nontreated sides compared to control mice (Fig. 4A). Osteoblasts in the depth of the periosteum were still active 5 days after the last injection (day lo), forming a layer of new woven bone (Fig. 3D). After a further 10 days (day 20), the periosteal surface had become quite smooth and the new woven bone was completely mineralized except for a thin layer of osteoid tissue undergoing mineral-

20,

osteoclast number

bone marrow area (%)

I treated side untreated side

-14 U 0

0

0

FIG. 4. Time course of the effects of TGF-0,. 2.5 pg/ day, on periosteal (A) and calvarial (B) thickness and on the number of osteoclasts (C) and the relative bone marrow area (D). TGF-0, or vehicle alone (Cont.) was injected over the right side of the calvaria for 2 days (TGF-0 2d) or 5 days (other groups). Mice were killed 1 day after the last injection in the following treatment groups: TGF-0 2d, Cont. 5d, and TGF-0 5d. All other treatment groups were killed on the day indicated from the start of the 5 day injection period. Results (mean + SEM) obtained in each group for each side of the calvaria were compared (Fischer PLSD test) with values of the same side in the respective control group (*p < 0.05; **p < 0.01; ***p < 0.001). No statistical analysis was performed for groups TGF-P 1Od and TGF-0 20d in the absence of appropriate control groups.

TABLE1. EFFECTS ON PLASMA IONIZED CALCIUM (Ca”) AND BODYWEIGHT (BW) OF TGF-0, (5 pg/DAY 5 DAYS)INJECTEDINTO THE SUBCUTANEOUS TISSUES OVERLYING THE CALVARIAE OF M I C E ~

Ca”, mmol/liter BW, g aResults (mean study.

Mice

Day 0

TGF-6, Control

1.396 + 0.018 1.382 + 0.017

TGF-0, Control

11.9 + 0.3 13.2 + 0.3

Day 2

1.396 1.358

f 0.006

+ 0.001

14.7 f 0.2 16.6 f 0.4

Day 3

1.4 1.386

f

15.9 17.7

f

f

f

0.005 0.018 0.2 0.4

FOR

Day 5

1.382 1.384

f

19.7

f

+ 18.0 +

0.019 0.12 0.1 0.5

SEM) were not statistically different between TGF-P,-treated and control mice at any time during the

D

FIG. 3. Photomicrographs of undecalcified coronal sections from parietal bones stained with Masson-Goldner trichrome ( x 90). In vehicle-treated animals (A) the periosteum is thin, whereas 2 days of treatment with TGF-0, (2.5 pg/day) induced proliferation of the deepest cells of the periosteum (B). After 5 days ( C ) of treatment the periosteal thickness is increased and bone surface is covered by new bone matrix (m). Osteoblasts over the periosteal surface were still active 5 days after the last TGF-0, injection (day lo), forming woven bone (D). On day 20 the woven bone was almost completely mineralized (E). By day 40 the woven bone (wb) was covered by a layer of new lamellar bone (Ib) and the appearance of periosteal cells had returned to that of control (F).

4 FIG. 4.

See left.

FIG. 5. Photomicrographs of undecalcified coronal sections from parietal bones stained with Masson-Goldner trichrome ( x 100). Mice received TGF-P,, 2.5 &day for 5 days, injected over the right side of the calvaria. Active osteoblasts are forming abundant bone matrix in the periosteum (A). Concomitant treatment with indomethacin, 50 p g every 8 h for 5 days, has partially inhibited periosteal bone formation and has reduced the mineralization lag time since almost all the matrix has already mineralized (B).

1092

MARCELLI ET AL.

calvarial thickness in controls at day 6 (Fig. 4B). This increase is related to the growth of the animals. At day 40 in the TGF-P,-treated mice the calvarial thickness was significantly increased relative to either the untreated side or to the bones of animals treated with vehicle alone. TGF-PI, 2.5 pg/day for 5 days, induced a marked and significant increase in osteoclast numbers on both the treated and nontreated sides of the calvaria (Fig. 4C). As in the first experiment, these osteoclasts were found to be much larger than those seen in control mice injected with vehicle alone. No osteoclasts were seen along the outer periosteal surface of the calvariae. The number of osteoclasts fell quickly after the last injection. A modest, transient increase in the bone marrow area was also observed (Fig. 4D), reflecting a stimulation of bone resorption along the endosteal surfaces of the marrow cavities. The measurement of mineralization rate under fluorescence was complicated by a diffuse tetracycline labeling of areas of new woven bone, which indicates an active mineralization process. However, the mineralization rate on the treated side of the calvariae was found to be greater in TGF-/3,treated animals (2.5 pg/day for 5 days) than in vehicletreated controls (1.87 f 0.12 versus 1.34 f 0.08 &day, respectively; p < 0.02, mean f SEM).

Effects of indomethacin on bone formation induced by TGF-0, The dose of indomethacin (50 pg injected subcutaneously every 8 h for 5 days) used in this study was the same as that previously used by Boyce et a1.(I6)Unfortunately, three of five mice treated with both TGF-0, and indomethacin and two of five mice treated with indomethacin alone died during the last 24 h of this experiment. The dissection showed abdominal hemorrhage in all these mice. No morphologic differences in bone were seen between mice treated with indomethacin alone and untreated controls (data not shown).

t

-a --E

The effects of a concomitant treatment with indomethacin on the bone formation induced by TGF-0, (2.5 pg/day for 5 days) injected over the right side of the calvariae are illustrated in Fig. 5. Indomethacin appears to reduce bone matrix formation (Fig. 5B). However, indomethacin did not completely inhibit the formation of woven bone in the periosteum. As shown in Fig. 6, the increase in periosteal thickness observed on the treated side of the calvaria of TGF-P,-treated mice was not inhibited by indomethacin. Osteoclasts were absent on the treated side of the calvariae of TGF-0, plus indomethacin-treated mice, whereas 3.6 f 3.0 osteoclasts per mm2 total bone area were counted in mice treated with TGF-0, alone.

Effects of TGF-0 on cortical bone in the femur TGF-PI (5 pg in 10 pl) was injected directly into the left femur of two rats via PE 10 polyethylene tubing daily for 3 days as illustrated in Fig. 7. The right femora of both rats were similarly injected with the vehicle alone (PBS, 1 mM HCI, and 1 To BSA). The rats were sacrificed 6 days following the first injection and the femora removed for histology. In both rats an extensive layer of woven bone was seen in the cortical bone adjacent to the TGF-0 injection site, with greatly increased numbers of large osteoclasts present (Fig. 8B). Furthermore, on the endosteal bone surface opposite the site of TGF-/3 injection new woven bone without osteoclastic activity was seen in both rats (Fig. 8D). In contrast, the area of new bone formation adjacent to the injection hole was much smaller on the vehicle-injected side of both rats (Fig. 8A), and no new bone formation could be seen on the opposite endosteal surface (Fig. 8C). Thus these effects could not be ascribed to periosteal injury or injection.

treated side non-treated side

T

60

Q,

C

.t Q y"o

40

Q,

.-

5

20 0

control

TGFD

irldo TGFD+indo

FIG. 6. Effects of daily local injections of TGF-PI, 2.5 pg/day for 5 days, with and without indomethacin on periosteal thickness in calvariae of mice. Values (mean + SEM) were compared to values in controls (*p < 0.05). No statistical analysis was performed for groups treated with indomethacin alone or indomethacin plus TGF-0, because of the low numbers ( n = 2 and 3, respectively).

FIG. 7. Drawing of rat femur. In two rats a hole was made with a 21 gauge needle in the lateral cortical bone of the femoral diaphysis oin both sides. TGF-0, (5 pg in 10 pl) was injected directly into the hole of the left femur via polyethylene tubing daily for 3 days, whereas vehicle alone was injected into the right side. A and B in Fig. 8 show sections taken from the area corresponding to a, and C and D (Fig. 8) show sections corresponding to area b.

EFFECTS OF TGF-0 ON BONE IN VIVO

1093

FIG. 8. Photomicrographs of undecalcified sections from rat femora stained with Masson-Goldner trichrome (A and B x 40; C and D, x 100). TGF-@,(5 pg/day for 3 days) was injected directly into the hole of the left femur, whereas vehicle alone was injected into the right femur. On the vehicle-injected side (A) an inflammatory reaction with modest bone resorption was observed around bone debris in the area of the cortical defect. On the TGF-@,-injectedside (B) a thicker, more extensive layer of woven bone with greatly increased numbers of large osteoclasts (arrows) was present around the area of the hole. The cortical endosteal surface opposite the hole was quiescent on the vehicle-injected side ( C ) ,whereas on the TGF-P,-injected side (D) formation of woven bone was observed in the absence of osteoclasts.

MARCELLI ET AL.

1094

DISCUSSION In this study, the short- and long-term effects of subcutaneous injections of recombinant human TGF-fi, over the calvariae of mice were analyzed using bone histomorphometry. TGF-PI, 5 pg/day for 5 days, markedly stimulated the periosteal formation of woven bone as well as endosteal osteoclastic bone resorption. After only 2 days of TGF-& (2.5 pg/day), periosteal osteoblast precursors had proliferated, and by day 5 these cells had produced a large amount of new bone matrix on the injected periosteal surface. The new bone became almost completely mineralized within 2 weeks, forming a thick layer of woven bone. Indomethacin partially inhibited bone matrix production, suggesting that some of TGF-PI effects were mediated by endogenous prostaglandin production. Noda and Ca~nilliere,('~) in their recent study, have observed similar effects of TGF-PI injected at doses of 50 ng, 200 ng, or 1 pg/day for 12 days over the calvariae of neonatal rats. However, when they stopped the TGF-PI injections after 8 days and studied the calvariae 8 days later, they did not find any difference between treated and control animals. This could be due to the lower dose of TGFPI used by these investigators or to the higher level of bone turnover in neonatal rats in comparison with older mice. The same reasons might explain the absence of any histologic change found by these investigators on the noninjected side of the TGF-P,-treated calvariae. The changes we observed on the noninjected side of the bone were likely due to the spread of the TGF-P, solution into the subcutaneous tissue overlying the nontreated side during injections. The focal nature of the response to injected TGF-0, was apparent in our studies. Thus, not only were there marked differences in both periosteal thickness and thickness of new woven bone between the treated and untreated sides of the calvariae, but the magnitude of the response on the treated side varied considerably with the depth of section studied. Furthermore, new bone formation was restricted to the outer periosteal bone surface, none being observed at the endosteum or inner periosteum. The absence of endosteal new bone formation could be due to intrinsic differences between TGF-0, responses at the endosteum versus periosteum. Human studies'20)with infusions of prostaglandin El have shown marked new periosteal bone formation without endosteal bone formation, indicating that differences in responses may occur at these two sites. However, in studies in which we injected TGF-6, in the marrow cavity of rats, we have observed new bone formation (Fig. 8) both at the site of injection and on the opposite endosteal surface. This indicates that endosteal bone, like periosteal bone, is capable of responding to TGF-/3, with an increase in bone formation. There are clearly site-specific differences in the response to TGF-/3, injections. Thus in our own study and that of Noda and Camilliere,(15)TGF-0, was injected over rodent calvariae whereas Joyce et al.(211in a preliminary report utilized a model in which TGF-PI was injected into the subperiosteum of the femur of newborn rats. In the latter model, the initial effect was an induction of cartilage for-

mation that only later became bone by endochondral ossification, whereas neither we nor Noda and Camilliere('S) observed cartilage formation following injections over the cavariae. This difference may occur because the femoral periosteum contains precursor cells for both osteoblasts and chondroblasts,(z2)whereas chondroblast precursors are absent in calvarial periosteum. Although in vitro studies have produced variable results that are highly dependent on the system used,(8-10)some parallels between in vitro findings and our data are striking. Centrella et a1.(l4)tested the effects of TGF-PI on fetal rat calvariae in organ cultures. They observed that continuous exposure of cultured calvariae to human platelet TGF-P, in serum-free cultures for 96 h enhanced calvarial DNA synthesis but produced no increase in the synthesis of either collagen or noncollagen protein or in alkaline phosphatase activity. Prostaglandin E, (PGE,) levels in culture medium were significantly enhanced by TGF-PI, but only after 48 h, and addition of indomethacin to the cultures did not inhibit the increase in DNA synthesis induced by TGF-&, suggesting that PGE, does not account for this mitogenic effect. Indomethacin alone decreased collagen and noncollagen protein synthesis, and addition of TGF-P, abolished this inhibitory effect. These results suggest that, in this organ culture system, the primary effect of TGF-6, is the stimulation of osteoblast replication and a secondary effect is the stimulation of endogenous production of PGE,. In a more recent paper, Hock et al.',') have studied effects of TGF-0 on cultured fetal rat calvariae using histomorphometry and autoradiography. TGF-fi increased, after 24 h, the cell replication in all cell zones of the pericranial periosteum and, after 48 h, also increased the total number of osteoblasts over the bone surface, the matrix apposition rate and new matrix formation, and the rate of collagen and noncollagen protein synthesis. The doses of TGF-8 needed to stimulate matrix apposition were lower than those needed to increase cell replication, and inhibition of cell replication only partially blocked the stimulatory effect of TGF-6 on matrix formation. Taken together, these data suggest that the increase in matrix protein synthesis induced by TGF-6 is partly dependent on increased cell replication and partly the consequence of an increase in differentiated cell activity. In addition, TGF-0 significantly decreased the number of osteoclasts in bone sections. A biphasic effect of PGE, on collagen synthesis in cultured fetal rat calvariae was observed by Raisz and colleagues. (24.25) At high concentrations ( M), PGE, inhibited collagen production,(24)whereas at low concentration (lo-' M) it stimulated both DNA and collagen synthesis. (25) However, collagen synthesis was stimulated later (96 h) than DNA synthesis (24 h). The proliferation of periosteal osteoblast precursors we observed after 2 days of treatment may have been due to TGF-PI, whereas a secondary endogenous production of prostaglandins could explain the large amount of bone matrix formed later by osteoblasts. Because indomethacin did not totally abolish the TGF-(3, stimulation of bone formation, it seems likely that TGF-0, also stimulated both the proliferation and the

1095

EFFECTS OF TGF-B ON BONE IN VIVO differentiation of osteoblast precursors by a nonprostaglandin-mediated mechanism. However, an incomplete blockade of prostaglandin synthesis cannot be excluded as an explanation for the incomplete inhibition of TGF-PImediated effects by indomethacin. In addition to its stimulatory effects on bone formation, TGF-P, induced a marked increase in endosteal bone resorption reflected by a significant increase in osteoclast numbers and an increase in relative bone marrow area. However, TGF-PI treatment had no effect on plasma calcium concentrations. This contrasts with previous observations following subcutaneous injections of known resorbing agents in similar gravimetric doses, including IL-1,(16) PTH and PTHrP,‘18’ lymphotoxin (LT),(26)and TNF-a (Yates, unpublished observations). No osteoclasts were seen along the outer periosteal surface of the calvariae in animals treated with TGF-PI, 2.5 pg for 5 days, in contrast to the effects of IL-1 injected in the same manner.(I6) When animals were concomitantly treated with TGF-P, and indomethacin in the third experiment, no osteoclasts were observed. These observations taken together suggest that TGF-PI did not stimulate bone resorption directly but rather enhanced endogenous prostaglandin production, which in turn stimulated bone resorption. Consistent with these observations, Tashjian et and Pfeilschifter et a1.(12)have reported stimulation of bone resorption by TGF-0, in neonatal mouse calvariae in organ culture. These reports showed a parallel increase in PGE, levels in the culture medium, and when indomethacin was added to the culture medium, both bone resorption and the PGE, production were inhibited. Boyce et a1.(’6)found a different response to IL-1 injections in the same in vivo model as used in the present studies. They found that active osteoclasts appeared focally in large numbers on the periosteal surface with an associated inflammatory reaction. Treatment with indomethacin inhibited the development of this IL-1-induced bone resorption, also suggesting a prostaglandin-mediated mechanism. As discussed earlier, local prostaglandin production was probably involved in the periosteal bone formation induced by TGF-P,, since concomitant treatment with indomethacin attenuated the production of new bone matrix. Recent studies indicate that TGF-PI inhibits both the formation of osteoclastlike cells in long-term human marrow and osteoclastic bone resorption in fetal rat long bone cultures.(12)Therefore, the absence of osteoclasts along the periosteal surface, observed in this study despite a presumably increase in local prostaglandin production, could be due to a direct inhibitory effect of TGFPI on osteoclast formation and bone resorption at this site. Because the studies presented here were conducted with relatively high doses of TGF-PI, the relevance of our observations to the physiologic role of TGF-PI remains to be determined. However, our findings may have important therapeutic implications. For instance, local administration of TGF-PI might be used to accelerate fracture repair. TGF-6, is indeed expressed and synthesized during fracture healing,”’) acting as a regulator of cartilage and bone formation by suppressing the cartilage phenotype and enhancing bone development.

In conclusion, we have documented striking changes in local bone turnover with predominant new bone formation at the site of injection of TGF-0, in intact mice and have shown these effects to be partially mediated by prostaglandin production. Although the precise mechanisms by which TGF-0, alters bone metabolism are not known, TGF-PI is likely to have an important role in normal bone turnover and our studies support possible therapeutic potential for this cytokine.

ACKNOWLEDGMENTS This work was supported by grants AR-28149, CA40035, DE-08569, and AR-39357 and by 1’AcadCmie Francaise. We are grateful to Thelma Barrios for her expert secretarial help.

REFERENCES 1. Baron R, Vignery A, Horowitz M 1983 Lymphocytes, mac-

2.

3.

4. 5.

6.

7.

8.

9.

10.

11.

12.

13.

rophages and the regulation of bone remodeling. In: Peck WA (ed.) Bone and Mineral Research, Annual 2. Elsevier, Amsterdam, pp. 175-243. Canalis E, McCarthy T, Centrella M 1988 Growth factors and the regulation of bone remodeling. J Clin Invest 81:277281. Sporn MB, Roberts AB, Wakefield LM, decrombrugghe B 1987 Some recent advances in the chemistry and biology of transforming growth factor-beta. J Cell Biol 105: 1039- 1045. Centrella M, McCarthy TL, Canalis E 1988 Skeletal tissue and transforming growth factor 0. FASEB J 2:3066-3073. Seyedin S, Thompson AY, Bentz H, Rosen DM, McPherson JM, Conti A, Siege1 NR, Gallupi GR, Piez KA 1986 Cartilage-inducing factor-A; apparent identity to transforming growth factor-@. J Biol Chem 2615693-5695. Pfeilschifter J, Mundy GR 1987 Modulation of type p transforming growth factor activity in bone cultures by osteotropic hormones. Proc Natl Acad Sci USA 84:2024-2028. Robey PG, Young MF, Flanders KC, Roche NS, Kondaiah R, Reddi AH, Termine JD, Sporn MB, Roberts AB 1987 Osteoblasts synthesize and respond to transforming growth factor type-p in vitro. J Cell Biol 109457-463. Pfeilschifter J, D’Souza SM, Mundy GR 1987 Effects of transforming growth factor p on osteoblastic osteosarcoma cells. Endocrinology 121:212-218. Ibbotson KJ, Orcutt CM, Anglin A, D’Souza SM 1989 Effects of transforming growth factors /3, and S I on a mouse clonal, osteoblast-like cell line MC3T3-E,. J Bone Min Res 4: 37-45. Centrella M , McCarthy TL, Canalis E 1987 Transforming growth factor p is a bifunctional regulator of replication and collagen synthesis in osteoblast-enriched cell cultures from fetal rat bone. J Biol Chem 262:2869-2874. Tashjian AJ Jr, Voelkel EF, Lazzaro M, Singer FR, Roberts AB, Derynck R, Winkler ME, Levine L 1985 a And p human transforming growth factors stimulate prostaglandin production and bone resorption in cultured mouse calvaria. Proc Natl Acad Sci USA 82:4535-4538. Pfeilschifter J, Seyedin SM, Mundy GR 1988 Transforming growth factor beta inhibits bone resorption in fetal rat long bone cultures. J Clin Invest 82:680-685. Chenu C, Pfeilschifter J, Mundy GR, Roodman GD 1988

10%

14.

IS.

16.

17.

18.

19.

20.

21.

22.

Transforming growth factor /3 inhibits formation of osteoclast-like cells in long term human marrow cultures. Proc Natl Acad Sci USA 855683-5687. Centrella M, Massague J, Canalis E 1986 Human platelet-derived transforming growth factor /3 stimulates parameters of bone growth in fetal rat calvariae. Endocrinology 119:23062312. Noda M, Camilliere JJ 1989 In vivo stimulation of bone formation by transforming growth factor b. Endocrinology 124: 2991-2994. Boyce BF, Aufdemorte TB, Garrett IR, Yates AJP, Mundy GR 1989 Effects of interleukin-1 on bone turnover in normal mice. Endocrinology l25:1142-1150. Sabatini M, Boyce B, Aufdemorte T, Bonewald L, Mundy GR 1988 Infusions of recombinant human interleukin-1 (Y and 0 cause hypercalcemia in normal mice. Proc Natl Acad Sci USA 855235-5239. Yates AJP, Gutierrez GE, Smolens P, Travis PS, Katz MS, Aufclemorte TB, Boyce BF, Hymer TK, Poser JW, Mundy GR 1988 Effects of a synthetic peptide of a parathyroid hormone-related protein on calcium homeostasis, renal tubular calcium reabsorption, and bone metabolism in vivo and in vitro in rodents. J Clin Invest 81:932-938. Meunier PJ 1983 Histomorphometry of the skeleton. In: Peck WA (ed.) Bone and Mineral Research, Annual 1 . Excerpta Medica, Amsterdam, pp. 28-59. Ueda K, Saito A, Nakano H, Aoshima M, Yokota M, Muraoka R, Iwaya T 1980 Cortical hyperostosis following longterm administration of prostaglandin E, in infants with cyanotic congenital heart disease. J pediatr 975334436. Joyce ME, Jingushi S, Roberts AB, Sporn MB, Bolander ME 1989 Transforming growth factor 5 initiates cartilage and bone formation in vivo (abstract). J Bone Min Res qSuppl 1):566. Ogden JA 1980 Chondro-osseous development and growth. In: Urist MR (ed.) Fundamental and Clinical Bone Physiology. J.B. Lippincott, Philadelphia, pp. 108-171.

MARCELLI ET AL. 23. Hock JM, Canalis E, Centrella M 1990 Transforming growth factor 0 stimulates bone matrix opposition and bone cell replication in cultured fetal rat calvariae. Endocrinology 126: 421 -426. 24. Fall PM, Raisz LG 1989 Mechanism of the biphasic effects of prostaglandin E, on bone formation in cultured fetal rat calvariae (abstract). J Bone Min Res 4(Suppl 1):661. 25. Chyun YS, Raisz LG 1984 Stimulation of bone formation by prostaglandin E. Prostaglandins 27:97-103. 26. Garrett IR, Durie BGM, Nedwin GEN, Gillespie A, Bringman T, Sabatini M, Bertolini DR, Mundy GR 1987 Production of lymphotoxin, a bone-resorbing cytokine, by cultured human myeloma cells. N Engl J Med 317526-532. 27. Joyce MD, Jingushi S, Flanders KC, Roberts AB, Sporn MB, Bolander ME 1989 Transforming growth factor is expressed and synthesized during fracture healing (abstract). In: Transforming Growth Factor bs: Chemistry, Biology and Therapeutics. Conference of the New York Academy of Sciences, May 18-20, p. 11-12. 28. Bolander ME, Joyce ME, Jinguishi S, Terek RM 1989 Role of TGFP in fracture repair (abstract). In: Transforming Growth Factor 0s: Chemistry, Biology and Therapeutics. Conference of the New York Academy of Sciences, May 18-20, p. 10.

a,

Address reprint requests to: Dr. Gregory R. Mundy University of Texas Health Science Center at San Antonio Department of Medicine Division of Endocrinology and Metabolism 7703 Floyd Curl Dr. San Antonio, TX 78284-7877 Received for publication March 16, 1990; in revised form May 1 1 , 1990; accepted June 18, 1990.