Proc. Nati. Acad. Sci. USAVol. 83, pp. 2228-2232, April 1986Medical Sciences

Human recombinant transforming growth factor a stimulates boneresorption and inhibits formation in vitro

(collagen synthesis/alkaline phosphatase/hypercalcemia/epidermal growth factor)

KENNETH J. IBBOTSON*, JANE HARROD*, MAXINE GOWEN*, SHARYN D'SOUZA*, DONNA D. SMITH*,MARJORIE E. WINKLERt, RIK DERYNCKt, AND GREGORY R. MUNDY**Department of Medicine, Division of Endocrinology and Metabolism, University of Texas Health Science Center, San Antonio, TX 78284; and Departmentsof tProtein Biochemistry and *tMolecular Biology, Genentech, Inc., South San Francisco, CA 94080

Communicated by R. H. Wasserman, December 2, 1985

ABSTRACT Human recombinant transforming growthfactor a (TGFa), which binds to the epidermal growth factor(EGF) receptor and causes several biological effects similar tothose caused by EGF, was compared with murine EGF for itseffects on a number of parameters of bone cell metabolism.TGFa stimulated bone resorption in two organ culture sys-tems, the fetal rat long bone and neonatal mouse calvarialsystems. TGFa stimulated bone resorption at concentrations aslow as 0.1 ng/ml. TGFa effects on bone resorption in mousecalvariae were inhibited by indomethacin, suggesting that, likeEGF, its effects were mediated by prostaglandin synthesis.TGFa had a different time course of action on bone resorptionfrom that of EGF, causing more rapid release of previouslyincorporated 45Ca from bone cultures, suggesting that TGFadoes not function on bone as a simple EGF analogue. TGFaalso caused effects on osteoblast function resembling those ofEGF. It inhibited alkaline phosphatase activity in cultured ratosteosarcoma cells with the osteoblast phenotype and inhibitedcollagen synthesis in fetal rat calvaria at concentrations of 1.0ng/ml. The lowest concentration of TGFa (expressed asnanogram equivalents of EGF per ml) required to produce aresponse in all of the systems tested was about 1/10th of thatneeded for EGF to produce a similar effect. These resultsindicate that TGFa is a potent stimulator of bone resorptionand inhibitor of bone formation as assessed by inhibition ofcollagen synthesis and alkaline phosphatase activity and areconsistent with the hypothesis that TGFa may be responsible,at least in part, for the bone resorption associated with sometumors.

Transforming growth factor alpha (TGFa) is secreted by avariety of human and rodent tumor cells and binds to theepidermal growth factor (EGF) receptor. TGFa causes manyof the known biological effects of EGF (1-3). Multiplepeptides of different molecular weights with TGFa activity(as assessed by inhibition of EGF binding and stimulation ofanchorage-independent growth) have been detected in cul-ture media and extracts of tumor cells. A small species of 50amino acids has been purified (4, 5). It is about 40%homologous with EGF (4, 6). Cloning of the cDNA revealsthat it is initially synthesized as a larger precursor, whichundergoes subsequent posttranslational cleavage. Properengineering of the sequence coding for the 50 amino acidTGFa has allowed the expression of biologically activehuman TGFa in Escherichia coli (6).EGF causes osteoclastic bone resorption (7, 8) and inhibits

bone formation as assessed by collagen synthesis in organcultures of fetal rat calvariae (7-9) and alkaline phosphataseactivity in osteoblast-like cells (10). Since EGF stimulates

osteoclastic bone resorption and TGFa competes with EGFfor the same receptor, it has been suggested that tumorproduction of TGFa could be responsible for the bonedestruction that is associated with some neoplasms (11, 12).Partially purified bone-resorbing activity associated with ananimal tumor contains TGFa activity (11), and this tumor-derived bone-resorbing activity is inhibited by antisera to theEGF receptor (13), which blocks the biological effects ofEGF and TGFa (14), indicating that the bone-resorbing factoris dependent on availability of the EGF receptor for itsbiological effects. In this report, we have compared theeffects of human recombinant TGFa and murine EGF on anumber of parameters of bone cell metabolism. TGFa ap-peared to be more potent than EGF in these systems and hada different time course of action on bone resorption. Theseresults indicate that TGFa does not act on bone as a simplehomologue of EGF.

MATERIALS AND METHODSBone Resorption Assays. The bone resorption assay has

been described (15). Briefly, rats at day 18 of gestation wereinjected subcutaneously with 200 ,Ci (1 Ci = 37 GBq) of45CaC12. The following day the fetuses were removed, and thetubular mineralized shafts of the radii and ulnae were dis-sected free from the cartilagenous ends and surrounding softtissue. The explanted shafts were precultured in BGJbmedium (Irvine Scientific) for 18 hr to allow for the exchangeof loosely complexed 45Ca with stable calcium in the medium.At the beginning of the test culture period, bones were

incubated in wells containing medium alone [BGJb supple-mented with 5% (vol/vol) fetal bovine serum]; KC Biologi-cals, Lenexa, KS) or medium containing TGFa and murineEGF (receptor grade, 99% pure by NaDodSO4/PAGE; Col-laborative Research, Waltham, MA). After a 48-hr incuba-tion, bones were transferred to wells containing either freshmedium alone or medium containing test substances andwere incubated for a further 72 hr. The concentrations ofTGFa and EGF used are given in the legends to the figuresand tables. Bone resorption was measured by the release ofradioactivity from individual bones into the medium duringthe 120-hr test culture period, the percentage of total 45Careleased was calculated, and results were expressed astreated-to-control ratios. Four bones were used in eachgroup, and statistical differences were determined by Stu-dent's t test for unpaired samples. To examine the involve-ment of prostaglandin synthesis in the action of TGFa, somebones were incubated with indomethacin [1 ,uM in 70%(vol/vol) ethanol]. Groups containing TGFa and indometh-acin were compared to controls containing indomethacinalone. Prostaglandin E (PGE) was measured by radio-

Abbreviations: EGF, epidermal growth factor; TGF, transforminggrowth factor; PGE, prostaglandin E.

2228

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertisement"in accordance with 18 U.S.C. §1734 solely to indicate this fact.

Dow

nloa

ded

by g

uest

on

Mar

ch 1

9, 2

020

Proc. Natl. Acad. Sci. USA 83 (1986) 2229

immunoassay (Seragen, Boston, MA) after extraction ofbone culture medium with acidified ethyl acetate. The lowerlimit of detectability of the assay was 0.1 ng of PGE per ml.Another assay measures release of 45Ca from previously

labeled neonatal mouse calvariae (16). Neonatal mice wereeach injected with 1 ,uCi of 45Ca, and 4 days later front andparietal bones were removed by microdissection and splitalong the sagittal suture. After a 24-hr preincubation period,the half calvariae were transferred into fresh medium con-taining test or control substances for a further 72-96 hr, onehalf calvaria being treated and the other acting as control.Four half calvariae were used in each treatment group, andbone-resorbing activity was expressed as treated-to-controlratios as described above. In these experiments, statisticaldifferences were evaluated by using Student's t test for pairedsamples.

Collagen Synthesis. The culture technique for studyingfactors that influence new bone formation has been described(17). The frontal and parietal bones were removed from21-day fetal rats and split along the sagittal suture. The halfcalvariae were precultured for 3-4 hr in BGJb (IrvineScientific) medium at 370C in 5% C02/95% air. The purposeof the preculture is to remove loose fragments of bone andconnective tissue as well as endogenous hormones that are incontact with the bone. After the preculture, each half calvariawas placed individually in a 35-mm-diameter well in six-wellplates (Falcon) containing 2 ml of BGJb medium supplement-ed with bovine serum albumin at 1 mg/ml and was incubatedat 370C for 48 hr. Six half calvariae were used for each group.In 96-hr experiments, the medium and test substances werereplaced after 48 hr.To assess synthesis of collagen and noncollagen proteins,

[3,4-3H2]proline (20 ,Ci per well; specific activity, 25-50Ci/mmol; ICN) was added for the last 4 hr of the cultureperiod. The calvariae were extracted, weighed, and homog-enized, and an aliquot was incubated with repurified bacterialcollagenase (Worthington). The enzyme preparation used didnot contain any proteolytic activity on [3H]tryptophan-labeled protein substrates. The labeled proline incorporatedinto collagenase-digestible protein and noncollagen proteinwas measured by the method of Peterkofsky and Dieglmann(18). The percentage of collagen synthesized was calculatedafter multiplying the noncollagen protein by 5.4 to correct forthe relative abundance of proline in collagenase-digestibleand noncollagen proteins (19). Data for collagenase-digest-ible protein were obtained as dpm/,ug of extracted dry weightof calvariae and were expressed as a percentage of totalprotein. Statistical differences were analyzed by using Stu-dent's t test.

Alkaline Phosphatase Activity. Alkaline phosphatase activ-ity is measured in osteoblast-like osteosarcoma cells (UMR-106) by the hydrolysis ofp-nitrophenyl phosphate (20). Cellswere plated at 2.5 x 104 per well in Dulbecco's modifiedEagle's medium (DMEM) (KC Biologicals, Lenexa, KS)containing 5% fetal bovine serum in 16-mm-diametermultiwell plates. After 24 hr the medium was changed to freshserum-free DMEM in the presence or absence of the growthfactors. After a 96-hr incubation, during which the culturemedium was changed at 48 hr, the cell layer was washedextensively with phosphate-buffered saline (pH 7.2, 0.15 M),and then 1 ml of 0.1% Triton X-100 in water was added. Thecells were rapidly frozen and thawed three times, and the celllayers were then scraped. An aliquot of the homogenate wasadded to the substrate solution, which contains 10 mMp-nitrophenyl phosphate in 100 mM diethanolamine buffer(pH 10.5) supplemented with 0.5 mM MgCl2. After incuba-tion at 370C, the reaction was terminated by the addition of1 ml of 0.2 M NaOH, and alkaline phosphatase activity wasdetermined spectrophotometrically (410 nM) by measuringp-nitrophenol released from the substrate. Enzyme activity

was expressed as Amol of substrate cleaved per mg of cellprotein per min. Cell protein was measured by using amodification of the method of Lowry et al. (21).Recombinant TGF. The 50-amino acid TGFa was ex-

pressed in E. coli as described (6), purified by Sephadex G-75gel filtration followed by HPLC, and assessed for biologicalactivity by its capacity to compete with labeled EGF forbinding to the EGF receptors on mink lung cells (CCL 64) andfor the capacity to stimulate growth of NRK (normal ratkidney) indicator cells in soft agar (6). Concentrations ofTGFa in terms of nanogram equivalents of EGF weremeasured by determining the amounts of TGFa required todisplace labeled EGF from its receptors on mink CCL 64 lungfibroblasts (6) or from UMR 106 osteosarcoma cells (11),which were used to verify the equivalence of TGFa andmurine EGF preparations used in the experiments.

RESULTS

Bone Resorption. Fetal rat long bones. Human recombi-nant TGFa stimulated bone resorption in a concentration-dependent manner (Fig. 1). Maximal activity was obtainedwith a concentration of TGFa of approximately 30 ngequivalents of EGF per ml over a 5-day incubation period.Bone resorption was stimulated by concentrations of TGFaas low as 3 ng equivalents of EGF per ml. The maximumvalue obtained with 30 ng of EGF per ml was comparable tothe effect induced by a concentration of TGFa lower by 1order of magnitude. In addition the maximal responses withEGF were always less than those obtained with TGFa (Fig.1). The results shown in Fig. 1 are from a single experimentrepresentative of data obtained on four separate occasions.The time course of action ofTGFa also was compared with

that of EGF (Fig. 2). The results shown are from oneexperiment representative of data obtained from three ex-periments. It was observed that TGFa stimulated boneresorption within 48 hr at concentrations of 30 ng of EGFequivalents per ml and more, whereas EGF had no statisti-cally significant effect over this time period at the concen-trations tested (1-300 ng/ml). In the fetal rat long bone assay,the bone-resorbing activity of TGFa appeared to be inde-

3.4

3.2-O 3.0-

2.8V) 2.6-

0~

2.2-

2.0-

°1.8.

4-1.6-*1.~l4- <1.2,

1 .0 3.0 10 30 100Growth factor, ng/ml

FIG. 1. Comparison of dose-response curves of human recom-binantTGFa and EGF. The experiment was carried out as described.Doses are expressed as ng/ml for EGF or as ng equivalents of EGFper ml for TGFa. Data shown here are treated-to-control ratios aftera 5-day culture period, expressed as means ± 1 SEM (n = fourbones). A, Bovine parathyroid hormone-1-84 at 400 ng/ml; e, TGFa;o, EGF; ***, P < 0.005; **, P < 0.025; *, P < 0.05.

Medical Sciences: Ibbotson et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 1

9, 2

020

2230 Medical Sciences: Ibbotson et al.

Time of incubation, days

FIG. 2. Time course of action of human recombinant TGFa andEGF. The experiment was carried out as described. The data shownhere are treated-to-control ratios of 45Ca release. The values areexpressed as means 1 SEM (n = four bones) at 2 days and 5 days.A, Parathyroid hormone at 400 ng/ml; *, TGFa at 30 ng equivalentsof EGF per ml; o, EGF at 30 ng/ml; ***, P < 0.005; **, P < 0.025.

pendent of prostaglandin production by the bone because theprostaglandin synthesis inhibitor indomethacin (1 ,uM) had noeffect on the bone-resorbing activity ofTGFa in two separateexperiments (data not shown). Indomethacin at this concen-tration totally inhibited PGE production as measured byradioimmunoassay. PGE concentrations were 1.4 0.01ng/ml in control cultures and 1.1 0.01 ng/ml in TGFa-treated cultures. Treatment with indomethacin reduced me-dium PGE concentrations to <0.01 ng/ml. Similar resultshave been described for EGF in this assay system (8).Neonatal mouse calvariae. TGFa and EGF also stimulated

bone resorption in the calvariae bioassay system (Tables 1and 2). The concentrations of both TGFa and EGF requiredwere lower than those required to stimulate resorption in thefetal rat long bone system. The lowest concentration ofTGFarequired to produce a significant effect were at most 1/10thof the concentration of EGF. There was, however, nodifference in the maximal responses obtained with TGFa andEGF. In the calvariae assay, we found that bone resorptionstimulated by TGFa and EGF was strongly inhibited byindomethacin (1 ,uM) (Table 2). Data shown are from one

experiment and representative of data obtained from threeexperiments.Because the TGFa was produced in E. coli, the possibility

existed that the TGFa preparation might contain endotoxin,

Table 1. Effects of recombinant human TGFa on boneresorption in neonatal mouse calvariae

Bone-resorbing activity,treated/control ratios of 45Ca release

TGFa,ng/ml Exp. 1 Exp. 2 Exp. 3 Exp. 4

100 1.66 ± 0.15* - 1.88 ± 0.21t33 1.41 ± 0.2310 1.47 ± 0.11* 1.20 ± 0.07* 1.72 ± 0.11t 1.53 ± 0.13*3 1.50 ± 0.10t1 1.37 ± 0.06t 1.24 ± 0.OO4t 1.69 ± 0.02t 1.63 ± 0.11*0.3 1.16 ± 0.03* -0.1 1.01 ± 0.03 1.59 ± 0.08t 1.45 ± 0.20*0.01 1.28 ± 0.10*

Experiments were carried out as described. TGFa concentrationsare expressed as ng equivalents ofEGF per ml. In experiment 4, thebones were incubated for 96 hr. In the other experiments, they wereincubated for 72 hr. Each result is expressed as the mean ± 1 SEM(four half calvariae per group).*Significantly greater than 1.0; P < 0.05.tSignificantly greater than 1.0; P < 0.01.

which has previously been shown to stimulate bone resorp-tion (22). For this reason, we tested control medium andmedium containing maximally active concentrations ofTGFa for the presence of endotoxin (23). We found nodifference in the concentration of endotoxin between controland TGFa-containing medium.

Alkaline Phosphatase Content in Osteoblast-Like Cells.TGFa and EGF caused a concentration-dependent inhibitionof alkaline phosphatase activity in cultures of rat osteosar-coma cells that have the osteoblast phenotype (Table 3). Datashown are from one experiment representative of resultsfound on three separate occasions. An inhibition of alkalinephosphatase activity could be obtained with both TGFa andEGF. However, the minimum concentration of the growthfactor needed for this effect in the case of TGFa is clearlylower (perhaps by factors of 3-10) than for EGF.

Collagen Synthesis in Fetal Rat Calvaria. In the collagensynthesis assay, TGFa was also significantly more effectivein inhibiting collagen biosynthesis than was EGF (Table 4).Data shown are from one experiment representative of dataobtained in three experiments. The lowest concentration atwhich TGFa clearly inhibited collagen synthesis was at 1ng/ml. In contrast, EGF at 1 ng/ml had no significant effecton collagen synthesis. The lowest concentration tested atwhich EGF caused significant inhibition of bone collagensynthesis was 10 ng/ml. TGFa and EGF had no significanteffect on noncollagen protein synthesis. Over longer incuba-tion periods (96 hr), TGFa continued to inhibit collagensynthesis and was effective at doses as low as 1.0 ng/ml(Table 4). Inhibition of collagen synthesis does not appear to

Table 2. Effect of indomethacin on bone resorption stimulated by TGFaand EGF

Bone-resorbing activity,treated/control ratios of 45Ca release

Growth factor, TGFa + EGF +ng/ml TGFa Indo (1 A.tM) EGF Indo (1 ,uM)100 1.46 ± 0.19* 0.96 ± 0.12 1.42 ± 0.18* 1.10 ± 0.0610 1.63 ± 0.12* 1.04 ± 0.08 1.19 ± 0.06 1.13 ± 0.061.0 1.41 ± 0.02* 1.16 ± 0.14 1.09 ± 0.02 1.09 ± 0.12

Experiments were carried out as described. EGF concentration is expressed in ng/ml. TGFaconcentration is expressed in ng equivalents of EGF per ml. Results are expressed as means ± 1 SEM(n = four half calvariae per group). Indo, indomethacin.*Significantly greater than corresponding control; P < 0.05.

Proc. Natl. Acad. Sci. USA 83 (1986)

Dow

nloa

ded

by g

uest

on

Mar

ch 1

9, 2

020

Proc. Natl. Acad. Sci. USA 83 (1986) 2231

Table 3. Effects of human recombinant TGFa and mouse EGFon alkaline phosphatase activity in osteosarcoma cells with theosteoblast phenotype

Growth factor, Alkaline phosphataseTreatment ng/ml content*

Control medium 0.355 ± 0.015TGFa 30t 0.185 ± 0.016t

lot 0.190 ± 0.004tlt 0.249 ± 0.035t

EGF 30 0.230 ± 0.010t10 0.225 ± 0.010t1 0.388 ± 0.028

Experiments were carried out as described. Values are means +SEM for triplicate cultures.*,umol of substrate cleaved per mg of protein per min.tng equivalents of EGF per ml.tSignificantly different from control; P < 0.05.

be a result of increased cell proliferation and loss of differ-entiated function. Concentrations ofTGFa between 30 and 1ng/ml inhibited collagen synthesis by 40% (Table 4) but hadno significant effect on thymidine incorporation into thecalvariae (data not shown).

DISCUSSIONThe results of these experiments show clearly that TGFa isa potent osteotropic agent. TGFa stimulates bone resorptionas measured by the 45Ca release and inhibits bone formationin vitro as assessed by inhibition of collagen synthesis inorgan cultures and alkaline phosphatase activity of cells withthe osteoblast phenotype, which contain EGF receptors. Ingeneral, the concentrations of TGFa (as expressed innanogram equivalents ofEGF) required to cause these effectswere 1 order of magnitude less than that required for EGF toproduce similar effects.One of the interesting facets of this study is the comparison

between TGFa and EGF effects on bone. EGF and TGFashare approximately 40% homology (4), and TGFa hasbiological properties ofEGF in vitro. TGFa binds to the EGFreceptor, and its biological effects on indicator cells in softagar can be inhibited by specific antisera that block the EGFbinding site on the EGF receptor (14). Our data indicate thatthe effects of TGFa are qualitatively similar to EGF on bone

Table 4. Effects of human recombinant TGFa and purifiedmouse EGF on collagen synthesis in organ cultures of fetalrat calvariae

Collagen synthesis,Sample added to Concentration, % CDPbone culture ng/ml 48 hr 96 hr

Control medium 13.2 ± 0.8 17.5 ± 0.6TGFa 30* 8.6 ± 0.2t 4.2 ± 0.7t

10* 8.4 ± 0.4t 5.8 ± 1.ot1* 8.1 ± 0.4t 11.1 ± 1.8t0.1* 11.6 ± 0.6 17.3 ± 0.90.01* 13.9 ± 0.5 NT0.001* 14.4

±0.3 NT

EGF 30 8.3 ± 0.5t10 9.1 ± 0.7t1 11.4 ± 0.40.1 11.8 ± 1.1

cells, but there are quantitative differences. TGFa is clearly10-fold more potent in these assay systems and has adistinctly different time course on bone resorption in vitro.The quantities of TGFa and EGF added to the cultures arebased on their capacity to displace 1251I-labeled EGF in aradioreceptor assay. It is possible that the TGFa preparationcontains more protein that does not displace EGF in thisassay. Thus, a much larger amount of TGFa protein may bepresent. If this is so, this protein would be unlikely to bebiologically active, since TGFa appears to exert its effectthrough EGF receptors (14). We believe these data representa real difference in activity, since in other studies we foundthat purified rat synthetic TGFa caused biological effects atsimilar concentrations (24).Thus, these two factors have some structural homology

and share the same receptor but have quantitatively differenteffects on target cells. There are several possible explana-tions. One possibility is that TGFa has one or more additionalreceptors, as was suggested earlier by Massague et al. (5).However, TGFa does not inhibit alkaline phosphatase ac-tivity in osteoblast-like cells that do not have EGF receptors.Rat osteosarcoma cells with the osteoblast phenotype de-rived from the osteosarcoma line ROS 17/2.8 do not possessEGF receptors and do not decrease their alkaline phospha-tase content after incubation with TGFa (J.H., unpublisheddata), in contrast to UMR 106 cells, a related rat osteosar-coma cell line with the osteoblast phenotype that does haveEGF receptors. These observations suggest that binding tothe EGF receptor may be necessary for TGFa to exert itseffects on osteoblast-like cells. It is also possible that thereare two distinct populations of receptors with differentaffinities for TGFa and EGF. This is consistent with theobservations of Gregoriou and Rees (25), who suggested theexistence of two types of EGF receptors. Another possibilityis that there are two binding domains on the EGF receptor forTGFa. One of these domains could bind both EGF and TGFawith equal affinity, but the second could have a higher affinityfor TGFa. Alternatively, the differences in biological effectsbetween TGFa and EGF could be related to differences inintracellular processing of the receptor-ligand complex fol-lowing binding of TGFa or EGF to their receptors on theirtarget cells. It is still unknown which, if any, of the importantbiological effects of EGF or TGFa are due to events thatoccur as a consequence of internalization of the ligand andbound-receptor complex. Finally, differences in the rates ofdegradation of TGFa and EGF in culture could explain thedifferences observed in the potencies of these factors onbone, although in vivo studies using EGF and TGFa did notsuggest any pharmacokinetic differences between the twofactors (26). Differences in degradation could not easilyaccount for either the differences in the time course of actionor the maximal responses obtainable between TGFa andEGF in the fetal rat long bone-resorption system.The results of this study emphasize the apparent difference

in sensitivity between the two bioassay systems for boneresorption that we used. Although we found that both TGFaand EGF resorb bone in each of the models, the mousecalvarial system was approximately 1 order of magnitudemore sensitive to both growth factors than was the rat longbone system. This had been suggested previously by theindependent studies of Tashjian and Levine (7), who usedmouse calvaria, and Raisz et al. (8), who used the fetal ratlong bone system, with EGF. There were other differencesnoted between these models. Tashjian and Levine (7) foundthat EGF-mediated resorption of bone was dependent onprostaglandin generation by the bone cultures, whereas Raiszand colleagues (8) using the rat long bone system found thatbone resorption occurred independent of prostaglandin syn-thesis. These two assay systems have been the major in vitrobone resorption systems studied over the past 20 yr, and

Experiments were carried out as described. % CDP, percentcollagenase-digestible protein. Values are means ± SEM for sixhalf-calvariae cultures.*ng equivalents of EGF per ml.tSignificantly different from control; P < 0.05.

Medical Sciences: Ibbotson et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 1

9, 2

020

2232 Medical Sciences: Ibbotson et al.

differences between them clearly need to be clarified. Ingeneral, the similarity of responses of the two systems hasbeen remarkable. Differences between them are probablydue to variations in cell populations that comprise thecalvariae ofthe neonatal mouse and the long bones ofthe fetalrat at day 19 of gestation.

Increased bone resorption associated with inhibition ofbone formation is the histologic change in bones usually seenin animal models as well as patients with humoral hypercal-cemia of malignancy (12). Quantitative bone histomorphom-etry shows increased osteoclastic bone resorption and inhi-bition of bone formation in this syndrome (27). Theseobservations are consistent with the hypothesis that thetumor-derived osteotropic product responsible for hypercal-cemia of malignancy in some tumors is TGFa or a factor thathas identical biological effects on bone to those ofTGFa. Wehave convincing evidence in one animal model of humoralhypercalcemia that TGFa itself is the mediator of boneresorption. In this case, the biological activity of TGFacopurified with bone-resorbing activity (11) and the tumor-derived bone-resorbing activity can be blocked by antisera tothe EGF receptor, which block the binding ofEGF as well asthe biological effects of EGF and TGFa (13).

Note. Since the original submission ofthis manuscript, Tashjian et al.(28) have reported that human recombinant TGFa stimulates resorp-tion of neonatal mouse calvariae by a prostaglandin-dependentmechanism.

The authors thank Nancy Garrett for preparing this manuscript.This work was supported by Grants CA29537 and AM28149 from theNational Institutes of Health.

1. Sporn, M. G. & Todaro, G. J. (1980) N. Engl. J. Med. 303,878-880.

2. DeLarco, J. E. & Todaro, G. J. (1982) J. Cell Physiol. 102,267-277.

3. Todaro, G. J., Fryling, C. & DeLarco, J. E. (1980) Proc. Natl.Acad. Sci. USA 77, 5258-5262.

4. Marquardt, H., Hunkapiller, M. W., Hood, L. E. & Todaro,G. J. (1984) Science 223, 1079-1082.

5. Massague, J., Czech, M. P., Iwata, K., DeLarco, J. E. &Todaro, G. J. (1982) Proc. Natl. Acad. Sci. USA 79,6822-6826.

6. Derynck, R., Roberts, A. B., Winkler, M. E., Chen, E. Y. &

Goeddel, D. V. (1984) Cell 38, 287-297.7. Tashjian, A. H. & Levine, L. (1978) Biochem. Biophys. Res.

Commun. 85, 966-975.8. Raisz, L. G., Simmons, H. A., Sandberg, A. L. & Canalis, E.

(1980) Endocrinology 107, 270-273.9. Canalis, E. (1983) Endocrol. Rev. 4, 62-77.

10. Kumegawa, M., Hiramatsu, M., Hatakeyama, K., Yajima, T.,Kodama, H., Osaki, T. & Kurisu, K. (1983) Calcif. Tissue Intl.35, 542-548.

11. Ibbotson, K. J., D'Souza, S. M., Ng, K. W., Osborne, C. K.,Niall, M., Martin, T. J. & Mundy, G. R. (1983) Science 221,1292-1294.

12. Mundy, G. R., Ibbotson, K. J., D'Souza, S. M., Simpson,E. L., Jacobs, J. W. & Martin, T. J. (1984) N. Engl. J. Med.310, 1718-1727.

13. Ibbotson, K. J., D'Souza, S. M., Carpenter, G. & Mundy,G. R. (1985) Endocrinology 116, 469-471.

14. Carpenter, G. R., Stoscheck, C. M., Preston, Y. A. &DeLarco, J. E. (1983) Proc. Natl. Acad. Sci. USA 80,5627-5630.

15. Raisz, L. G. (1965) J. Clin. Invest. 44, 103-116.16. Gowen, M., Meikle, M. C. & Reynolds, J. J. (1983) Biochem.

Biophys. Acta 762, 471-474.17. Dietrich, J. W., Canalis, E. M., Mauna, D. M. & Raisz, L. G.

(1976) Endocrinology 98, 943-959.18. Peterkofsky, B. & Dieglmann, R. (1971) Biochemistry 10,

988-994.19. Peterkofsky, B. (1972) Arch. Biochem. Biophys. 152, 318-328.20. Lowry, 0. H. (1957) Methods Enzymol. 4, 366-381.21. Lowry, 0. H., Rosebrough, N. J., Farr, A. L. & Randall,

R. J. (1951) J. Biol. Chem. 193, 265-275.22. Haussman, E., Raisz, L. G. & Miller, W. A. (1970) Science

168, 862-864.23. Levin, J., Tomasulo, P. A. & Oser, R. S. (1970) J. Lab. Clin.

Med. 75, 903-911.24. Ibbotson, K. J., Twardzik, D. R., D'Souza, S. M.,

Hargreaves, W. R., Todaro, G. J. & Mundy, G. R. (1985)Science 228, 1007-1009.

25. Gregoriou, M. & Rees, A. R. (1984) Embo. J. 3, 929-937.26. Smith, J. M., Sporn, M. B., Roberts, A. B., Derynck, R.,

Winkler, M. E. & Gregory, H. (1985) Nature (London) 315,515-516.

27. Stewart, A. F., Vignery, A., Silverglate, A., Ravin, N. D.,Livolsi, V., Broadus, A. E. & Baron, R. (1982) J. Clin.Endocrinol. Metab. 55, 219-227.

28. Tashjian, A. H., Voelkel, E. F., Lazzaro, M., Singer, F. R.,Roberts, A. B., Derynck, R., Winkler, M. E. & Levine, L.(1985) Proc. Natl. Acad. Sci. USA 82, 4535-4538.

Proc. Natl. Acad. Sci. USA 83 (1986)

Dow

nloa

ded

by g

uest

on

Mar

ch 1

9, 2

020