Growth Hormone–Regulated Intracellular Signalingin UMR 106 Osteosarcoma Cells

ORLANDO MORALES,1 URBAN LINDGREN,1 and LARS-ARNE HALDOSE´N2

ABSTRACT

Growth hormone (GH) and insulin-like growth factor 1 (IGF-1) are important growth factors for postnatallongitudinal bone growth. Although many effects of GH on bone growth are mediated by IGF-1, GH candirectly influence bone cells. Limited knowledge exists regarding specific intracellular signaling pathways andgenes activated by GH in bone cells. GH is known to activate several intracellular signaling pathways, amongthem the Janus kinase (JAK)/signal transducers and activators of transcription (STAT) pathway. GH mainlyactivates JAK2 and both isoforms of STAT5, A and B. STAT5 gene deletion experiments have shown theimportance of these transcription factors for growth. To understand the molecular mechanism(s) behind this,different experimental models are needed. The UMR 106 cell line is a rat clonal osteosarcoma cell line withosteoblast-like phenotypic properties, one is the endogenous expression of GH receptor (GHR). The presentstudy focused on whether these cells express a functional GH-responsive JAK2/STAT5 pathway. Analysis ofcell extracts by immunoprecipitation and Western blot showed that physiological concentrations of GHactivated JAK2. Western blot analysis of nuclear extracts from GH-stimulated UMR 106 cells showed thatphysiological concentrations of GH induced nuclear translocation of both STAT5 isoforms, but with STAT5Abeing predominant. Both isoforms displayed similar nuclear turnover after GH stimulation of cells. Gelelectrophoretic mobility shift assay (GEMSA) of nuclear extract revealed that both STAT5A and STAT5Bobtained DNA-binding capacity after GH stimulation. Thus, we have shown, for the first time, the expressionand GH-induced activation of JAK2 and STAT5A/B in UMR 106 osteoblast-like cells. This study also showsthat this cell line is a suitable experimental model to study unique GH effects in osteoblasts mediated bySTAT5. (J Bone Miner Res 2000;15:2284–2290)

Key words: growth hormone, osteoblast, signal transduction, transcription factor, DNA binding

INTRODUCTION

GROWTH HORMONE(GH) and insulin-like growth factor 1(IGF-1) are the most important growth factors for nor-

mal postnatal longitudinal bone growth.(1,2) IGF-1 is pro-duced mainly by the liver under the influence of GH but alsois synthesized locally by many tissues in the body. In bonecells, GH induces expression of IGF-1. Although manyeffects of GH on bone growth are mediated by IGF-1,experimental data indicate that GH can directly influence

bone cell function.(2) The intracellular mechanism involvedin these GH-specific effects on bone cells is at present notfully understood. Furthermore, there is a lack of knowledgeregarding osteoblast genes regulated by GH.

Earlier studies have shown that GH activates differentintracellular signaling pathways.(3) GH binding to amembrane-bound GH receptor (GHR) mediates these cel-lular effects.(3) GHR binding activates receptor-boundmembers of the Janus family of tyrosine kinases (JAK). Inmost cases GH activates JAK2. Activated JAK2 tyrosine

1Department of Orthopedic Surgery, Karolinska Institutet, Huddinge Hospital, Huddinge, Sweden.2Department of Medical Nutrition, Karolinska Institutet, Novum, Huddinge, Sweden.

JOURNAL OF BONE AND MINERAL RESEARCHVolume 15, Number 11, 2000© 2000 American Society for Bone and Mineral Research

2284

phosphorylates and activates several intracellular signal me-diators, including members of the signal transducers andactivators of transcription (STAT) family of transcriptionfactors. After activation, STATs translocate to the nucleus,bind DNA, and regulate gene transcription. At present,seven mammalian STAT genes have been identified.(4,5)

GH mainly activates STAT5 but in experimental modelsit also has been shown to activate STAT1 and STAT3.(6,7)

STAT5 was first identified as a mammary gland factorrequired for prolactin (PRL)-stimulated transcription ofmilk protein genes.(8) STAT5 was later found to be presentin several other tissues and to be activated by other hor-mones including GH, growth factors, and cytokines.(9,10)

Two forms of STAT5 (STAT5A and STAT5B), encodedby two different genes, have been identified and are shownto be expressed in most tissues.(11,12)The functional signif-icance of the presence of the two forms is at present notfully understood. Although overlapping functions seem toexist, recent animal studies have indicated unique functionsfor each STAT isoform. Genetic disruption of the STAT5Agene in mice resulted in impaired terminal differentiation ofthe mammary glands during pregnancy, resulting in lack ofmilk production.(13) A key regulator of these two processesis PRL. Deletion of STAT5B gene resulted in reduced bodygrowth of male mice, a process of which GH is a keyregulator.(14,15) The simultaneous deletion of both genesresulted in the most severe growth and reproductive defects,revealing functional redundancy of the STAT5 proteins inphysiological processes mediated by GHR and PRL.(15)

As described previously, the direct cellular effects of GHon bone cells are not fully understood. In addition, theGH-activated intracellular signaling pathways and GH-activated genes in these cells are not fully described. Rele-vant experimental models are needed to gain insight intothese issues. GH has been shown to stimulate proliferationof fetal chicken and mouse osteoblasts, rat osteoblast-likecells, and osteoblast precursor cells derived from humanbone marrow stroma.(16–19) Functional GHR has been de-scribed in rat and human bone osteoblast-like celllines.(18,20) The recent findings using gene deletion experi-ments motivated us to search for an in vitro experimentalmodel suitable for investigating GH activation of STAT5pathway.

In this study we have investigated the presence of theJAK2/STAT5 signaling system as well as its GH-activatedcharacteristics in the rat osteoblast-like osteosarcoma cellline UMR 106. This cell line shares a number of phenotypicproperties with mature osteoblasts; one of them is the en-dogenous expression of the GHR.

MATERIALS AND METHODS

Cell culture

Rat osteosarcoma cells UMR 106 were routinely grownin monolayer culture at 37°C in 5% CO2-air in modifiedessential medium (MEM; Life Technologies, Ltd., Paisley,Scotland) supplemented with 5% fetal bovine serum (FBS;Life Technologies), 1 mML-glutamine, and 50 U/ml peni-cillin (Life Technologies).

Preparation of nuclear extract

The cells were cultured as described previously. Beforeaddition of bovine GH (bGH), cells were starved of fetalcalf serum for 12–16 h. After treatment with bGH, cellswere cooled on ice and rinsed with ice-cold phosphate-buffered saline (PBS). Cells were then scraped into 10 ml ofextraction buffer (10 mM Tris, pH 7.4, 10 mM NaCl, 6 mMMgCl2, 1 mM dithiothreitol [DTT], and 0.1 mM Na3VO4)and disrupted with a Dounce homogenizer (Go¨tenborgster-mometerfabrik, Gothenburg, Sweden). The nuclear pelletobtained after centrifugation was resuspended in 3 vol oflysis buffer (20% glycerol, 20 mM HEPES, pH 7.9, 420mM NaCl, 1.5 mM MgCl2, 0.2 mM EDTA, 0.2 mM phe-nylmethylsulfonyl fluoride [PMSF], 1 mM DTT, and 0.1mM Na3VO4) and incubated on ice for 30 minutes. Thesupernatant obtained after centrifugation was used as nu-clear extract. Protein concentration was measured by theBradford method. The nuclear extract obtained was furtheranalyzed with Western blotting or gel electrophoretic mo-bility shift assay (GEMSA).

Preparation of whole cell extract andimmunoprecipitation

UMR 106 cells were grown in 10-cm plates and beforeaddition of bGH, cells were starved of fetal calf serum for12–16 h. After treatment with 10 nM bGH for 5, 10, and 15minutes, cells were rinsed with ice-cold PBS and scrapedinto lysis buffer (50 mM Tris-HCl, pH 7.6, 150 mM NaCl,1mM EDTA, 1mM DTT, 10 mg/ml leupeptin, 10mg/mlpepstatin, 0.2mg/ml aprotinin, 1 mM PMSF, 1 mMNa3VO4, 1 mM NaF, and 1% Triton X-100) and kept on icefor 20 minutes. Then, the cell lysates were clarified bycentrifugation at 4°C for 10 minutes at 13,000 rpm in atable-top centrifuge. Extracts (1 mg total protein) wereimmunoprecipitated overnight at 4°C with 3ml anti-JAK 2antibody (diluted 1:2000; Upstate Biotechnology, LakePlacid, NY, U.S.A.) and 50ml protein A-Sepharose (Am-ersham Pharmacia Biotech, Uppsala, Sweden) in lysisbuffer. The immunoprecipitate was then washed twice with50 mM Tris-HCl, pH 7.5, and 150 mM NaCl (Tris-bufferedsaline [TBS]). Successful JAK2 precipitation and the phos-phorylation status of JAK2 were analyzed by Western blot-ting with anti-JAK2 and antiphosphotyrosine antibodies (di-luted 1:1000; Santa Cruz Biotechnology, Santa Cruz, CA,U.S.A.), respectively, as described in the following.

Western blotting

Sodiumdodecyl sulfate (SDS)–solubilizing buffer wasadded to nuclear extract (50mg) and samples were boiled.Proteins were separated on a 7.5% SDS-polyacrylamide gelelectrophoresis (PAGE) gel and transferred to a polyvinyli-dene difluoride (PVDF) membrane (Millipore, Bedford,MA, U.S.A.) by semidry blotting. The membrane wasblocked for 1 h with a buffer containing 1% milk protein or,in the case of Western Blotting with antiphosphotyrosineantibody, 1% albumin in TBS. After washing, the mem-brane was incubated for 1 h with rabbit anti-STAT5 (diluted

2285GH ACTIVATION OF STAT5 IN OSTEOBLAST-LIKE CELLS

1:400; Oncogene Research product, Calbiochem, Cam-bridge, MA, U.S.A.), rabbit anti-STAT5A (diluted 1:750;Santa Cruz Biotechnology), rabbit anti- STAT5B (diluted1:1000; Upstate Biotechnology, Inc.), rabbit anti-JAK2 (di-luted 1:2000; Upstate Biotechnology), or mouse antiphos-photyrosine (diluted 1:1000; Santa Cruz Biotechnology) in1% milk protein/TBS plus 0.05% Tween 20 (TTBS). Thesecondary antibody, goat anti-mouse immunoglobulin G(IgG) or goat anti-rabbit IgG coupled with horseradishperoxidase, diluted 1:5000 in TTBS, was applied for 1 hafter washing the membrane three times with TTBS. Themembrane was then analyzed with enhanced chemilumines-cence (ECL) method (Amersham).

GEMSA

GEMSA was performed according to standard proto-cols.(21) Nuclear extracts were incubated with32P-labeleddouble-strandedb-casein oligonucleotides (TGCTTCTTG-GAATT) in 15 ml of a buffer containing 4% Ficoll, 12 mMHEPES, pH 7.9, 4 mM Tris-HCl, pH 7.9, 0.1 mM EDTA,1mM DTT, and 5mg poly (dI-dC; Amersham PharmaciaBiotech). For supershift analysis chicken anti-sheepSTAT5,(22) rabbit-anti STAT5A (Santa Cruz Biotechnol-ogy) and rabbit anti-STAT5B (Upstate Biotechnology, Inc.)were added to the incubation mixture before separation onpolyacrylamide gel.

RESULTS

GH-induced activation of JAK2 and nucleartranslocation of STAT5A and STAT5B in UMR 106osteoblast-like osteosarcoma cells

The presence of a GHR has been described earlier inUMR 106 osteoblast-like osteosarcoma cells.(18) To analyzeif these cells also express JAK/STAT-signaling pathway,whether GH activates this system, and to see if bothSTAT5A and STAT5B are present, whole cell and nuclearextracts were prepared from GH-treated UMR 106 osteo-sarcoma cells. Whole cell extracts were immunoprecipitatedwith anti-JAK2 antibody. Immunoprecipitates were ana-lyzed by Western blot technique using antiphosphotyrosineantibody. In Fig. 1 it can be seen that stimulation of UMR106 cells with 10 nM bGH for 5, 10, and 15 minutesresulted in activation of JAK2. Nuclear extracts were ana-lyzed by Western blot technique using antibodies that rec-ognize both isoforms of STAT5 (STAT5A and STAT5B,Fig. 2A) or the unique C-terminal ends of STAT5A andSTAT5B (Figs. 2B–2D). UMR 106 cells were first incu-bated with different concentrations of GH for 10 minutesand nuclear extracts were analyzed by Western blot (Figs.2A–2C, lanes 1–5). This revealed that a 10-minute pulsewith as low a concentration as 1 nM GH induced a clearnuclear translocation of both STAT5A and STAT5B pro-tein. In Fig. 2A, the upper and lower immunoreactive spe-cies represent STAT5A (apparentMr 96,000) and STAT5B(apparentMr 93,000), respectively. Maximal effect wasreached at 10 nM GH with no increase in STAT5 proteinlevels in nuclear extracts from cells treated with 100 nM and

1000 nM GH (Figs. 2A–2C, lanes 3–5). Weak immunore-activity also was seen in unstimulated cells. At all GHconcentrations tested, a higher amount of STAT5A wasdetected as compared with STAT5B (Fig. 2A, lanes 2–5).

Next, we analyzed the time required for the first appear-ance of STAT5 protein in the nucleus and the duration ofSTAT5 protein presence in the nucleus after a single stim-ulation of UMR 106 with GH. In this experiment the cellswere stimulated with 10 nM GH for different times (Figs.2A and 2B, lanes 6–12, and Fig. 2D, lanes 1–7). After 5minutes both STAT5A and STAT5B were clearly present innuclear extracts. A slight increase in protein levels was seenat 10 minutes, with STAT5A being the most prominentimmunoreactive species (Fig. 2A, lane 8). At 30 minutes, amarked anti-STAT5 protein immunoreactive species with aslightly lower electrophoretic mobility was detected (Fig.2A, lane 9). This could have been caused by an increasedamount of STAT5A protein. Analysis of nuclear extract,taken from the same time point, with anti-STAT5A andanti-STAT5B antibodies revealed a lower electrophoreticmobility for both isoforms (Fig. 2B, lane 9, and Fig. 2D,lane 4). Thus, the increase in signal seen in Fig. 2A alsocould have been caused by too low a resolution of STAT5isoforms. STAT5A and STAT5B protein levels were mark-edly decreased at 60 minutes and had returned to levelscomparable with unstimulated cells at 120 minutes. Also, inthis experiment low but detectable signals were seen forSTAT5A and STAT5B in nuclear extracts from unstimu-lated cells.

STAT5A and STAT5B DNA-binding activity in nuclearextracts from GH-treated UMR 106 cells

The DNA-binding activity of STAT5 in nuclear extractsfrom GH-treated UMR 106 cells was analyzed withGEMSA using a radioactive labeled probe containing theproximal STAT5-binding site in theb-casein promoter.Nuclear extracts also were incubated with STAT5 antibody

FIG. 1. JAK2 was immunoprecipitated from lysates ofcontrol and bGH-treated UMR 106 osteosarcoma cells withanti-JAK2 antibody. The phosphorylation status of the pre-cipitated JAK2 was analyzed by Western blot with an-tiphosphotyrosine antibody. Successful and even precipita-tion of JAK2 was verified by Western blotting of strippedmembrane with anti-JAK2 antibody.

2286 MORALES ET AL.

in order to determine the presence of STAT5 in gel-shiftedprotein–DNA complex.

UMR 106 cells were treated first with different concen-trations of GH for 10 minutes (Fig. 3A). No DNA-bindingactivity was seen in nuclear extracts from untreated cells(Fig. 3A, lanes 1 and 2). GH treatment resulted in theappearance of a gel-shifted complex, which was alreadyseen at 1 nM GH treatment and reached its maximum at 10nM (Fig. 3A, lanes 3–6). A slight decrease in level of thiscomplex was seen in nuclear extracts from cells treated with100 nM and 1000 nM GH. A clear supershift was seen whennuclear extracts were incubated with STAT5 antibodyshowing that the gel-shifted complex contained STAT5(Fig. 3A, lanes 4, 6, 8, and 10).

To study the time course of GH-activated nuclear STAT5with DNA-binding activity, UMR 106 cells were treatedwith 10 nM GH for different times (Fig. 3B). Also, in thisexperiment, no STAT5 DNA-binding activity was seen inuntreated cells (Fig. 3B, lanes 1 and 2). STAT5 with DNA-binding activity was evident after 5 minutes of GH treat-ment and increased at 10 minutes of stimulation after whichit slowly decreased up to 120 minutes of stimulation.

Because both STAT5A and STAT5B were shown byWestern blot to be present in UMR 106 nuclear extracts, weinvestigated whether both isoforms had DNA-binding ac-tivity (Fig. 4). Nuclear extracts from UMR 106 cells, un-treated or treated with 10 nM GH for 10 minutes, wereincubated with a radiolabeled probe and with or withoutanti-STAT5, anti-STAT5A, or anti-STAT5B antibodies andthereafter analyzed by GEMSA. No DNA-binding activitywas seen in untreated cells (Fig. 4A, lanes 1–3, and Fig. 4B,lanes 1 and 2). GH treatment induced a band, which wassupershifted by the three antibodies (Fig. 4A, lanes 5 and 6,and Fig. 4B, lanes 4 and 5). This showed that GH activatedboth STAT5A and STAT5B in UMR 106 cells to DNA-binding form.

DISCUSSION

The recently reported findings from STAT5 gene deletionexperiments showing the importance of this signaling path-way for growth motivated us to study the molecular detailsof GH signaling in bone cells. The UMR 106 is a rat clonal

FIG. 2. GH induced nuclear transloca-tion of STAT5 in UMR 106 osteoblast-like cells. Cells were grown in serum-containing medium and before addition ofGH, the cells were starved of fetal calfserum for 12–16 h. The cells were treatedwith (A–C, lanes 1–5) GH at differentconcentrations for 10 minutes or with (Aand B, lanes 6–12, and D, lanes 1–7) 10nM GH for different time periods. Nuclearextracts were prepared as described in theMaterials and Methods section, and pro-teins were separated by SDS-PAGE andanalyzed by Western blotting with anti-bodies that recognize both isoforms of (A)STAT5, (B) STAT5A, or (C and D)STAT5B.

2287GH ACTIVATION OF STAT5 IN OSTEOBLAST-LIKE CELLS

cell line with well-characterized osteoblast-like phenotypicproperties and has been shown to be a good in vitro cellularsystem for studies of the physiology of osteoblasts. It sharesa number of phenotypic properties with mature osteoblasts.These similarities include an osteoblast-like morphologicalappearance, responsiveness to calciotropic agents such asparathyroid hormone(23,24) and vitamin D3,

(25) and a rela-tively high level of expression of the cell surface alkalinephosphatase activity.(23) Additionally, UMR cells synthe-size several matrix proteins expressed by normal osteoblastsincluding type 1 collagen(24) and proteoglycans.(26) Othercharacterized phenotypic properties are the expression ofGHR; production of IGF-1; and response to the proliferativeeffects of GH, IGF-1, and insulin.(18,27,28) In the presentstudy, we have analyzed if these cells express JAK2 andSTAT5 and also, if present, if these signal mediators areactivated by GH. Our experiments showed that UMR 106cells express JAK2 and both isoforms of STAT5 (STAT5Aand STAT5B) and that these are activated by GH.

Maximal nuclear translocation of both STAT5A andSTAT5B and maximal activation of STAT5 to DNA-

binding form in UMR 106 cells were obtained with 10 nMof GH, which is equivalent to 220 ng/ml and which is withinthe physiological range of GH concentration in vivo.(29)

Higher concentrations of GH did not further increase nordecrease nuclear translocation or activation of STAT5 to aDNA-binding form. Thus, a biphasic “bell-shaped” dose-response curve for GHR activation, which has been de-scribed in some earlier studies,(30) could not be seen withUMR 106 cells. This experiment also showed that STAT5Awas the predominant isoform in nuclear extracts from GH-treated UMR 106 cells.

The GH secretory pattern in adult male rats is character-ized by plasma GH pulses of 1-h duration every 3.5–4 h,

FIG. 3. Gel electrophoretic shift analysis of nuclear ex-tracts of GH-treated UMR 106 osteoblast-like cells. UMR106 cells were grown as described in the Material andMethods section and in Fig. 2. The cells were treated with(A) GH at different concentrations for 10 minutes or with(B) 10 nM GH for different time periods. Nuclear extractswere prepared as described in the Materials and Methodssection and incubated with [32P]-labeledb-casein probe andwith or without anti-STAT5 and thereafter analyzed byGEMSA.

FIG. 4. Supershift analysis of DNA-protein complexes innuclear extracts from GH-treated UMR 106 osteoblast-likecells. UMR 106 cells were grown as described in the Ma-terial and Methods section and in Fig. 2. (A and B) Cellswere treated with 10 nM GH for 10 minutes. Nuclearextracts were prepared as described in the Materials andMethods section and incubated with [32P]-labeledb-caseinprobe and with or without anti-STAT5, anti-STAT5A, oranti-STAT5B antibodies and thereafter analyzed byGEMSA.

2288 MORALES ET AL.

with a typical peak plasma GH concentration of 200 ng/ml,while females exhibit more continuous plasma GH levels ofabout 20–40 ng/ml.(29) Pulsatile GH is more effective thancontinuous GH in promoting weight gain and longitudinalbone growth. The mechanism by which a temporal plasmaprofile of GH regulates this physiological response is notwell understood. The STAT5 gene deletion experimentsdescribed previously showed that lack of STAT5B in malemice resulted in reduced body growth. From our experi-ments, it can be concluded that stimulation of osteoblastcells with one pulse of GH, with a concentration seen inmales, activates both STAT5A and STAT5B. This sameconcentration also induced similar nuclear entry and nucleardisappearance of STAT5A and STAT5B protein. Thus, it isnot likely that selective activation of STAT5 isoforms couldexplain the effect of sex-differentiated GH secretory pat-terns on bone growth. In male mice with both STAT5A andSTAT5B genes deleted body growth was reduced further.Furthermore, in female mice with both STAT5 genes de-leted body growth was also reduced. This is an indicationthat both STAT5 isoforms probably participate in regulatinggenes important for growth. Further studies are needed toidentify these genes and to describe the role of each STAT5isoform for activation of these genes.

To obtain maximal transcriptional activity, some STATsrequire serine phosphorylation, in addition to tyrosine phos-phorylation.(31) In the case of STAT5A, a recent study hasshown GH-induced activation of mitogen-activated proteinkinase (MAPK), which results in phosphorylation of aserine residue in the C-terminal activation domain ofSTAT5A.(32) STAT5B also has been shown to be serinephosphorylated after stimulation of cells with PRL.(22,33) Inour time-course study (Fig. 2), both STAT5A and STAT5Bhad a lower electrophoretic mobility after 30–60 minutes oftreatment of UMR 106 cells with GH. This experimentalfinding could be an indication of GH-induced serine phos-phorylation of STAT5A and STAT5B in UMR 106 cells.Further studies are needed to resolve this issue.

Both STAT5A and STAT5B proteins were shown to bepresent in nuclear extracts after GH stimulation of UMR106 cells. To determine if both isoforms displayed DNA-binding activity and if formation of homodimers and het-erodimers and their relative levels could be determined,supershift experiments with antibodies recognizing bothisoforms or each isoform individually were performed (Fig.4). Addition of these antibodies resulted in each case in twosupershifted complexes. The complexes supershifted withisoform-specific antibodies had slightly different electro-phoretic mobility as compared with the complexes super-shifted with STAT5 antibody. This experiment clearlyshowed that both isoforms were activated to DNA-bindingform by GH. The appearance of two supershifted complexeswith each antibody could be an indication of STAT5A andSTAT5B homo- and heterodimers with different electro-phoretic mobility. It cannot be excluded that faster andslower migrating DNA-protein complexes could have beencaused by binding of one and two antibodies, respectively.Thus from these data it is not possible to determine thepresence of STAT5A and STAT5B heterodimers or thelevels of homodimers and heterodimers in nuclear extracts

from GH-treated UMR 106 cells. More experiments andother tools are needed to do this.

In conclusion, we have shown that the rat osteoblast-likeosteosarcoma cell line UMR 106 expresses functional GH-JAK2/STAT5 signaling pathway. Furthermore, these cellsexpress both isoforms of STAT5. The recent finding of theimportance of STAT5 for body growth together with thefact that, at present, limited knowledge exists regardingintracellular pathways and genes activated directly by GH inosteoblasts calls for intensified research. To resolve theseissues, different experimental models are needed. In thisstudy, we have shown that the osteoblast-like osteosarcomacell line UMR 106 is a suitable experimental model to studyunique GH effects in osteoblasts mediated by STAT5.

REFERENCES

1. Boullion R 1991 Diabetic bone disease. Calcif Tissue Int49:155–160.

2. Ohlsson C, Bengtsson BA, Isaksson OG, Andreassen TT,Slootweg MC 1998 Growth hormone and bone. Endocr Rev19:55–79.

3. Carter-Su C, King AP, Argetsinger LS, Smit LS, VanderkuurJ, Campbell GS 1996 Signalling pathway of GH. Endocr J43:S65–S70.

4. Ihle JN 1996 Stat: Signal transducer of activator of transcrip-tion. Cell 84:3331–3345.

5. Leonard WJ, O’Shea JJ 1998 Jaks and STATs: Biologicalimplications. Annu Rev Immunol16:293–322.

6. Benbassat C, Shoba LN, Newman M, Adamo ML, Frank SJ,Lowe WL Jr 1999 Growth hormone-mediated regulation ofinsulin-like growth factor I promoter activity in C6 gliomacells. Endocrinology140:3073–3081.

7. Smit LS, Meyer DJ, Billestrup N, Norstedt G, Schwartz J,Carter-Su C 1996 The role of the growth hormone (GH)receptor and JAK1 and JAK2 kinases in the activation of Stats1, 3, and 5 by GH. Mol Endocrinol10:519–533.

8. Wakao H, Gouilleux F, Groner B 1995 Mammary gland factor(MGF) is a novel member of the cytokine regulated transcrip-tion factor gene family and confers the prolactin response.EMBO J 14:854–855.

9. Wood TJ, Sliva D, Lobie PE, Pircher TJ, Gouilleux F, WakaoH, Gustafsson JA, Groner B, Norstedt G, Haldosen LA 1995Mediation of growth hormone dependent transcriptional acti-vation by mammary gland factor/STAT5. J Biol Chem270:9448–9453.

10. Illeux F, Pallard C, Dusanter-Fourt L, Wakao H, Haldosen LA,Norstedt G, Levy D, Groner B 1995 Prolactin, growth hor-mone, erythropoietin and granulocyte-macrophage colonystimulating factor induce MGF-stat5 DNA binding activity.EMBO J 14:2005–2013.

11. Mui AL, Wakao H, O’Farrel A, Harada N, Mirajima A 1995Interleukine-5 transduce signals through two STAT5 ho-mologs. EMBO J14:1166–1175.

12. Liu X, Robinson GW, Gouilleux F, Groner B, Hennighausen L1995 Cloning and expression of Stat5 and an additional ho-mologue (Stat5b) involved in prolactin signal transduction inmouse mammary tissue. Proc Natl Acad Sci USA92:8831–8835.

13. Liu X, Robinson GW, Wagner KU, Garret L, Wynshaw-BorisA, and Hennighausman L 1997 Stat5a is mandatory for adultmammary gland development and lactogenesis. Genes Dev11:179–186.

14. Udy GB, Towers RP, Snell RG, Wilkins RJ, Park SH, RamPA, Waxman DJ, and Davey HW 1997 Requirement of Stat5b

2289GH ACTIVATION OF STAT5 IN OSTEOBLAST-LIKE CELLS

for sexual dimorphism of body growth rates and liver geneexpression. Proc Natl Acad Sci USA94:7239–7244.

15. Teglund S, McKay C, Schuetz E, van Deursen JM, Stravopo-dis D, Wang D, Brown M, Bodner S, Grosveld G, Ihle JN 1998STAT5a and STAT5b proteins have essential and nonessen-tial, or redundant roles in cytokine responses. Cell93:841–850.

16. Slootweg MC, van Buul-Offers SC, Herrmann-Erlee MP,Duursma SA 1988 Direct stimulatory effect of growth hor-mone on DNA synthesis of fetal chicken osteoblasts in culture.Acta Endocrinol118:294–300.

17. Slootweg MC, van Buul-Offers SC, Herrmann-Erlee MP, vander Meer JM, Duursma SA 1988 Growth hormone is mito-genic for fetal mouse osteoblasts but not for undifferentiatedbone cells. J Endocrinol116:R11–R13.

18. Barnard R, Ng KW, Martin TJ, Waters MJ 1991 Growthhormone (GH) receptors in clonal osteoblast-like cells mediatea mitogenic response to GH. Endocrinology128:1459–1464.

19. Kassem M, Mosekilde L, Eriksen EF, Kassem M, MosekildeL, Eriksen EF 1994 Growth hormone stimulates proliferationof normal human bone marrow stromal osteoblast precursorcells in vitro. Growth Regul4:131–135.

20. Nilsson A, Swolin D, Enerback S, Ohlsson C 1995 Expressionof functional growth hormone receptors in cultured humanosteoblast-like cells. J Clin Endocrinol Metab80:3483–3488.

21. Sliva D, Wood TJ, Schindler C, Lobie PE, Norstedt G 1994Growth hormone specifically regulates serine protease inhibi-tor gene transcription via gamma activated sequence likeDNA-elements. J Biol Chem269:26208–26214.

22. Petersen H, Haldosen LA 1998 EGF modulates expression ofSTAT5 in mammary epithelial cells. Exp Cell Res243:347–358.

23. Partridge NC, Alcorn D, Michelangeli VP, Ryan G, Martin TJ1983 Morphological and biochemical characterization of fourclonal osteogenic sarcoma cell lines of rat origin. Cancer Res43:4308–4314.

24. Partridge NC, Dickson CA, Kopp K, Teitelbaum SL, CrouchEC, Kahn AJ 1989 Parathyroid hormone inhibits collagensynthesis at both ribonucleic acid and protein levels in ratosteogenic sarcoma cells. Mol Endocrinol3:232–239.

25. Forrest SM, Ng KW, Findlay DM, Michelangeli VP, LiveseySA, Partridge NC, Zajac JD, Martin TJ 1985 Characterizationof an osteoblast-like clonal cell line which responds to both

parathyroid hormone and calcitonin. Calcif Tissue Int37:51–56.

26. McQuillan DJ, Findlay DM, Hocking AM, Yanagishita M,Midura RJ, Hascall VC 1991 Proteoglycans synthesized by anosteoblast-like cell line (UMR 106–01). Biochem J277:199–206.

27. Lakatos P, Caplice MD, Khanna V, Stern PH 1993 Thyroidhormones increase insulin-like growth factor I content in themedium of rat bone tissue. J Bone Miner Res8:1475–1481.

28. Hickman J, McElduff A 1989 Insulin promotes growth of thecultured rat osteosarcoma cell line UMR-106–01: Anosteoblast-like cell. Endocrinology124:701–706.

29. Jaffe CA, Ocampo-Lim B, Guo W, Krueger K, Sugahara I,DeMott-Friberg R, Bermann M, Barkan AL 1998 Regulatorymechanisms of growth hormone secretion are sexually dimor-phic. J Clin Invest102:153–164.

30. Fuh G, Cunningham BC, Fukunaga R, Nagata S, Goeddel DV,Wells JA 1992 Rational design of potent antagonists to thehuman growth hormone receptor. Science256:1677–1680.

31. Horvalt CM, Darnell JE 1997 The state of the STATs: Recentdevelopments in the study of signal transduction to the nu-cleus. Curr Opin Cell Biol9:233–239.

32. Pircher TJ, Petersen H, Gustafsson JA, Haldosen LA 1999Extracellular signal-regulated kinase (ERK) interacts with sig-nal transducer and activator of transcription (STAT) 5a. MolEndocrinol13:555–565.

33. Yamashita H, Xu J, Erwin RA, Farrar WL, Kirken RA, Rui H1998 Differential control of the phosphorylation state ofproline-juxtaposed serine residues Ser725 of Stat5a andSer730 of Stat5b in prolactin-sensitive cells. Biol Chem273:30218–30224.

Address reprint requests to:Orlando Morales

Department of Orthopedic SurgeryHuddinge University Hospital K54

S-141 86 Huddinge, Sweden

Received in original form January 18, 2000; in revised form May25, 2000; accepted July 5, 2000.

2290 MORALES ET AL.