Growth Hormone Specifically Regulates Serine Protease Inhibitor ...

THEJOURNAL OF BIOLOGICAL CHEMISTRY

Vol. 269, No. 42, Issue of October 21, pp. 26208-26214, 1994 Printed in U.S.A.

0 1994 by The American Society for Biochemistry and Molecular Biology, Inc.

Growth Hormone Specifically Regulates Serine Protease Inhibitor Gene Transcription via yActivated Sequence-likeDNA Elements* (Received for publication, May 6, 1994, and in revised form, July 12, 1994)

Daniel SlivaSO, Timothy J. J. Wood+, Chris Schindlea, PeterE. Lobie, and Gunnar Norstedtll From the Center for Biotechnology a n d the Department of Medical Nutrition, KarolinskaInstitute, NOWM, 141 57 Huddinge, Sweden a n d the Wepartment of Molecular Medicine, College of Physicians a n d Surgeons of Columbia University, New York, New York 10032

Growth hormone activates gene transcriptionof the ropoietin, IL-3, IL-6, leukemia inhibitory factor, oncostatinM, serine protease inhibitors (SPI) 2.1 and2.2by an un- ciliary neurotrophic factor, and interferon-y have been demonknown mechanism. In order to define the promoterre- strated to interact with tyrosine kinasesof the Janus kinase gions responsible for this effect and to characterize the family (7-11). These shared properties maybe fundamental to transcription factors involved, we have performed gel the biology of the cytokine receptors and, as such, probably on nuclear extracts dictate that the receptors share similar mechanisms of signal electrophoresis mobility shift assays from cells lines transfected with growth hormone receptransduction. However, some mechanisms directing hormone tor cDNk We have identified a 9-base pair DNA element, specificity are likely to exist. In this paper, we have investithe SPI-GLE 1, which forms a complex with nuclear pro-gated the mechanism and specificity of regulation of transcripteinsfollowingactivation bygrowthhormoneand tion from the serine protease inhibitor (SPI) 2.1 and 2.2 prowhich, when placed upstream of a minimal thymidine moters following GH stimulation. We havedefined a 9-bp kinase promoter,driveschloramphenicolacetyltransregion of t h e SPI growth hormone response element (GHRE) ferase expression in a growth hormone-dependent fashion. This elementis similar to thosefrom several genes described by Yoon et al. (12), which shares homology with eleregulated by other cytokines including interferon. The ments that recognize the interferon-y-activated transcription factor Stat 91 and which binds nuclear proteins in a growth growth hormone-induced complexes formed were dehormone-dependent fashion. We term this element, whichhas pendent on tyrosine phosphorylation but did not conthe sequence TT CT G AG AA, the SPI y-activated sequence tain the interferon-y-activated transcription factor Stat (GAS)-like element 1 (SPI-GLE I). 91. Competition studies with oligonucleotides similar to the SPI-GLE 1 reveal the sequence of a consensus eleMATERIALSANDMETHODS ment that specifically binds growth hormone-regulated nuclear proteins. Cell Culture and Preparation of Nuclear Extracts-Buffalo rat liver cells werecultured in Dulbecco's minimal essential medium (LifeTechnologies, Inc.), and Chinese hamster ovary cells were cultured in F-12 Growth hormone (GH)' exerts its cellular effectsas a result Medium (Life Technologies,Inc.) (both containing 10%fetal calf serum (Life Technologies, Inc.) and 50 unitdml penicillin, 50 unitdm1 strepof binding toa specific receptor (1,2). This receptor a member is tomycin (Life Technologies,Inc.)). Before the addition of 100 I"recomof the cytokine receptor superfamily, which includes the recepbinant human (h) GH (Pharmacia) or 25 pg/pl mouse interferon-y tors for prolactin, erythropoietin, tumor necrosis factora, leu- (Boehringer Mannheim), the cells were starved of fetal calf serum for kemia inhibitory factor, oncostatin M, ciliary neurotrophic fac- 12-20 h. After treatment, the cultures (1-2 x 107/extract)were chilled tor, interleukins (IL) 2-7, and granulocyte- and granulocyte- on ice and rinsed with ice-cold phosphate-buffered saline, and nuclear macrophage colony-stimulating factors(3,4).The receptors for extracts were prepared by a modification of the method of Dignam etal. (13). The cells were then scraped into 10 ml of buffer (10 mM Tris, pH interferon-a and interferon-y also show structural homologies 7.4, 10 mM NaCl, 6 mM MgCI,, 1m dithiothreitol, 0.4 mM phenylmethto the cytokine receptors (5, 6). The members of the cytokine ylsulfonyl fluoride and 0.1 mM Na,VO,) and Dounce homogenized. The receptor superfamilyare characterized by(i) their possessionof nuclear pellet, after centrifugation, was resuspended in 3 volumes of a single putative membrane spanning domain, (ii) definedse- lysis buffer (20% glycerol,20 m HEPES pH 7.9,420 mM NaCl, 1.5 mM quence homologies in their extracellular domains and the re- MgCI,, 0.2 mM EDTA, 0.2 mM phenylmethylsulfonylfluoride, 1 mM gion of their intracellular domains immediately juxtaposed to dithiothreitol, 0.1 mM Na,VO,) and incubated on ice for 30 min. The nuclear extract was finally collected by centrifugation. the membrane, andiii) the absenceof a consensus sequence for Gel Electrophoresis MobilityShift Assay-Gel electrophoreticmobilIn addition, the receptors for GH,eryth- ity shift assays (GEMSA) were performed accordingto standard protoprotein kinase activity. cols (14). Nuclear extracts were incubated with32P-labeleddouble* This work was funded by Grant 8556 from the Swedish Medical stranded SPI-GHRE (GATCTACGCTTCTACTAATCCATGWCTGResearch Council and Grant505/93/1134 fromthe Czech Grant Agency. AGAAATCATCCAGTCTGCCCATG)or SPI-GLE 1 (TGmCTGAThe costs of publication of this article were defrayed in part by the GAAATA) oligonucleotides (core sequence shown in boldface)in abuffer payment of page charges. This article must therefore be hereby marked containing 20% Ficoll,60 mM HEPES pH 7.9,20 mM Tris, pH 7.9,0.5 mM "advertisement" in accordance with 18 U.S.C. Section1734solely to EDTA, 5 mM dithiothreitol, and 5 pg of poly(d1-dC). Antibody-induced indicate this fact. supershift analyses with nucIear extracts were performed as described Contributed equally to this work. 8 Visiting from The Research Institute of Biopharmacy and Veteri- (15). GEMSA Competition Studies-For competition studies, we used unnary Drugs, Jilovh, near Prague, Czech Republic during the period in labeled double-stranded SPI-GHRE, SPI-GLE 1 (TGITCTGAGAAwhich this work was carried out. ATA) oligonucleotides, double-stranded oligonucleotides representing (1 To whom correspondences should be addressed. The abbreviations used are: GH, growth hormone; IL, interleukin; different parts of the SPI-GHRE (7-(TACTAATCCATGTTC), 8-(GATCSPI, serine protease inhibitor; bp, base pair; GHRE, growth hormone TACGCTTCTAC), 15-(GATCTACGCTTCTACTAATCCATGTTCTGA)), response element; GAS, y-activated sequence; GLE, GAS-like element; or double-stranded oligonucleotides representing various GAS-like hGH, human growth hormone; GEMSA, gel electrophoresis mobility DNA elements (see Fig. 2 for references)(human GBP, ATTACTCTW human FcyRI, TTTCCCAGAAA, human ICSBP, TTTCTCCGAAA; hushift assay; CHO, Chinese hamster ovary; BRL, buffalo rat liver.

+

26208

Growth Hormone Regulationof Transcription

26209

man IFP-53, ATTCTCAGAAA, human IRF-I, TTTCCCCGAAA; mouse B .r .E .E .r.r IRF-1, "TCCCCGAAA;mouse LyGE, ATTCCTGTAAG mouse mig, E E E E E CTTACTATAAA). Assemhly of Reporter Constructs-Double-stranded oligonucleotides containing three tandem SPI-GLE 1 sequences (CTAGTGPCTGAGMTGAACGGTTCTGAGAAAGTACAGGTTCTGAGAAAT) were liB") gated, using standard techniques(141, into the XbaI siteof pBLCAT2, a reporter plasmid containing a thymidine kinase minimal promoter adjacent to the chloramphenicol acetyltransferasecDNA (16). Cell Tkansfection and Chloramphenicol Acetyl "Fansferase AssayThese were carried out a s described previously (17). Briefly, BRL-4 cells were grown in 30-mm cell culture dishes to 75% confluency and then washed twice with phosphate-buffered saline. Transfections were carriedoutinserum-free Dulbecco's minimalessentialmediumusing DOTAP (Boehringer Mannheim) and 1.25 pg of CsCl density gradient purified plasmid DNNwell. Cells were incubated with DOTAPDNA for 12 h, and then the medium was changed to serum free Dulbecco's minimal essential medium containing 100 nM hGH. After a further 12 h, cells were washed with phosphate-buffered saline and then scraped into 0.25 M Tris, pH 8.0. Following three roundsof freeze-thaw lysis and heat FIG.1. GH activation of SPI-GHRE binding activity in BRL-4 treatment at 65"C for 10 min, extracts were centifuged to removecell debris and then assayed for chloramphenicol acetyl transferase activity. cell nuclear extracts (panel A ) and CHO-4 cells(paneZB).BRL-4 20 pg of protein was incubated with 3 pl of ['Tlchloramphenicol (Am- cells grown to confluency (approximately 2x 10') in Dulbecco's modified ersham Corp., 50-62 mCi/mmol, 25 pCi/ml) and 25 pg of butyryl coen- Eagle medium with 10% fetal calf serum (Life Technologies, Inc.) and CHO-4 cells grown to confluency in Ham's F-12 medium (Life Technolozyme A (Boehringer Mannheim)for 3 h a t 37 "C. Butyrylated chloramgies, Inc.) with 10% fetal calf serum (Life Technologies, Inc.) were sephenicol was then extracted with xylene and detected using a Wallac rum-starved for 12 h and then treated with 100 nM hGH (Pharmacia)for scintillation counter. 0, 5, 10, 15, and 60 min. Nuclear extracts were prepared a s described and analyzed by GEMSA using a "'P-labeled 45-bp SPI-GHRE probe. RESULTS Two GH-dependent specific mobility shifts are observed, A and R.

Gel Electrophoresis Mobility Shift Assays with 45-bp SPIGHRE Probe-It has previously been reportedthat GH induces the bindingof proteins extractedfrom rat liver nucleito a 45-bp DNA element (SPI-GHRE) in the 5' flank of the serine protease inhibitor 2.1 gene (12). To facilitate investigation of the mechanism of GH regulation of gene transcription, we first wished to reproduce this effect using cultured cell lines. We, therefore, prepared nuclear extracts from Chinese hamster ovary (CHOK1) and buffalo rat liver (BRL-3A) cells that had been stably transfected with rat GH receptor cDNA and shown to respond to hGH (17). These cell lines will hereafter be referred to a s CHO-4 and BRL-4. The extracts weresubjected to GEMSA with a "P-labeled oligonucleotide probe consisting of a single copy of the SPI-GHRE. We observed a hGH-dependent formation of two complexes, A and B (Fig. l).The GH-dependent complexes first appeared a t 5 min, reached a maximum between 10 and 15 min, and were decreased in intensity by 60 min. That the addition of an excess (50-fold molar ratio) of unlabeled SPI-GHRE oligonucleotide, but not of an unrelated DNA (pSV2-Luc), resulted in a loss of electrophoretic mobility shift, indicates that theobserved DNA binding wasspecific for the probe used (data not shown). Similar results were observed in both cell lines. Definition of the GH-activated Complex-binding Domain of SPZ-GHRE-The 45-bp SPI-GHRE contains two regions that share homology to the GAS DNA element (Fig. 2). We termed these two regions SPI-GLE 1(-129 to -118 bp) andSPI-GLE 2 (-144 to -136 bp). In order to define minimal regions of the SPI-GHRE required for the response to GH, we used several double-stranded oligonucleotides representing discrete partsof the SPI-GHRE (including both SPI-GLE 1 and SPI-GLE 2) as competitors in GEMSA experiments with the45-bp SPI-GHRE probe (Fig. 3). We found that the oligonucleotides corresponding to SPI-GLE 1 were the most efficient inhibitors of complex formation (Fig. 3) abolishing both complexes A and B. Oligonucleotides covering SPI-GLE 2 abolished complex A but not complex B. Since SPI-GLE 1appeared to havea higher affnity forGH-activated proteins,furtherexperimentationwas focused on this element. Gel Electrophoresis Mobility Shift Assays with a SPZ-GLE 1 Probe-To determine whether SPI-GLE 1 alone was sufficient for formation of GH-induced DNA-binding complexes, we la-

-130 -119 TTCTACTAAT CCAGTCTGC cT AC A C T T Rat SPI-GLEl.............GTT CT GAG A?A Rat SPI-GLE2.. CTT CT A CT M A T

...........

H u m a InF P - 5 3 . . . . . . . . . . . .ATT H u m a Fn c Y R I TTT H u m a InC S B P . . . . . . . . . . . . .T T T H u m a nG B P . . . . . . . . . . . . . . .ATT H u m a InR F - 1 TTT TTT Mouse I R F - 1 Mouse Ly6E . . . . . . . . . . . . . .ATT Mouse Mig . . . . . . . . . . . . . . .C T T GTT H u m a n C Fos B o v i n e p l a c t o g l o b u l i n . . . ATT Rat c a s e i n . . . . . . . . . . . . .C T T TTT H u m a nF C E R I I b Mouse C Y 1 . . . . . . . . . . . . . . . . ATT CTT Rata2-macroglobulin H u m a n Y - f i b r i n o g e n . . . . . . .T T T

.............

............. .............

..............

p

Palindrome..

............ .....

.............

CT CC C CT AC CC CC CC AC CC CG CT TT CAC CT CT

C AG AAA AG AAA C GG AAA T C T AAA C CG AAA C CG AAA T GT AAC T AT AAA C G T CAA G GG AAC T GG AAT A AG AAA AT GAA G GG AAT G G T AAG

(a)

(a) (a)

(a) (a) (a)

(a) (a) (b)

(b) (c)

(c) (d) (d)

TT CT N AG AA

FIG.2. Lineup between SPI-GAS-like element, interferon-y response elements (a,Ref. 24), a GAS-like element from the c-fos promoter, milk protein binding factor elements, aGAS-like element from the p casein promoter (h, Ref. 281, interleukin-4 nuclear-activated factor binding elements ( c , Refs. 25 and 26), and interleukin-6-activated acute phase response factor binding elements (d, Ref. 27).

beled SPI-GLE 1 with "P for use in GEMSA. We observed the formation of a single, specific, GH-dependent complex, thereby demonstrating thatSPI-GLE 1is indeed sufficient (Fig.4). The formation of this complex was critically dependent on the concentration of KC1 used in the bindingbuffer (data not shown). The induction of the SPI-GLE 1 DNA binding complex showed similar kinetics withrespect to GH treatment to that observed using the SPI-GHRE probe. A dose-response curve (Fig. 4) indicated SPI-GLE 1 binding over a range from 10 PM to 10 p~ hGH. An oligonucleotide probe corresponding to SPI-GLE 2 failed to yield a GH-dependent complex (data not shown). SPI-GLE 1 Is a Functionally Active GH-dependent Panscription Enhancer Element-Although protein binding to a particular DNA element in vitro suggests that thesemolecules have a role in transcriptional regulation in vivo, this needs to be confirmed with functional data. In orderto address this issue,we

26210

Hormone Regulation Growth

of Dunscription

FIG.3. Competition study with oligonucleotides corresponding to discrete regions of SPI-GHRE. Nuclear extracts from BRL-4 treated for 10 min with 100 nM human GH were prepared as described previously. A "2P-labeled SPIGHREwas used for GEMSA together with a 50-fold molar excess of the indicated competing oligonucleotides.

" "

Site 2

Sie 1

S~,-GHRE GATCTACGCTTCTACTMTCCATGTTCTQAQMATCATCCAGTCTGCCCATG ATGCGAAQATQATTAGGTACMMTCTTTAGTAGGTCAGACTACCTAG

Oligo 7

I

Oligo 8 SPI-GLE1

TACTAATCCATG ATTAGGTACAAGA

I

GATCTACGCTTC ATGCGAAGATG TGTTCTGAGAAATC AAGACTCT'TTAGTA

oligo 15 GATCTACGCTTCTACTAATCCATGTT ATGCGAAGATGATTAGGTACAAGACT

placed three copies of the 9-bp SPI-GLE1upstream of a minimal latent transcription factors haspreviously been demonstrated to be mediated by tyrosine phosphorylation(18, 19, 20). In thymidinekinase promoter driving chloramphenicolacetyltransferase cDNA expression. This construct, when transiently order to determine the involvement of phosphorylation in GH transfected into BRL-4 cells, gave a 5-6-fold GH-dependent induction of nuclear protein binding to the SPI promoter, we enhancement of chloramphenicol acetyltransferase expression made nuclear extractsfrom CHO-4 cells co-incubated with 100 (Fig. 5 ) , which was similar to thedegree of enhancement seen nM hGH and 100 nM staurosporine for 15 min (Fig. 6). Such previously been demonwhen a construct containing the SPI-GHRE was used (17). A treatmentwithstaurosporinehas chloramphenicol acetyltransferase reporter construct (16)con- strated to prevent GH-induced phosphorylation of proteins in GH stimulation of lipid synthetaining the minimal thymidine promoter alone gave no such 3T3 adipocytes(21) and inhibit sis (22).No SPI-GLE 1binding activity was detected in nuclear GH-dependent effects. These results demonstrate that SPIGLE 1 is able to confer a functional response to GH and en- extracts prepared from cells treated with staurosporine. Thus it seems that activation of cellular kinases and cellular protein hance transcription of the SPI genes. The GH-activated SPI-GLE 1 Binding Protein Does Not Re- phosphorylation are required for transcription factor activaquire ProteinSynthesis for Activation-The time-dependent ac- tion. To investigate whether tyrosine phosphorylation of trantivation of nuclear protein-binding to SPI-GLE 1 is consistent scription factors isnecessary for their GH-activated binding to SPI-GLE 1, we incubated nuclear extracts from GH-treated with GH activating preexisting but latent transcription factors rather than increasing their expression. This is the case for CHO-4 cells with tyrosine-specific phosphatase 1B (UBI). This several other cytokine molecules (18, 19). We tested this hy- treatment specifically removed SPI-GLE 1 bindingactivity pothesis by inhibiting cellular protein synthesis with cyclohex- from nuclear extractsof GH treated cells (Fig. 7). Furthereviimide (50 pg/ml) for 15 min before the addition of hGH. These dence for the importanceof phosphorylation for protein binding conditions (dose and time) havepreviously been demonstrated to SPI-GLE 1 is provided by the reduction of intensity of the to completely inhibit cellularprotein synthesis (20). As ob- band corresponding to the GH-dependent complex seen when served in Fig. 6, cycloheximide had no effect on the ability of nuclear extracts are preincubated withantiphosphotyrosine an GH to induce the protein-binding to SPI-GLE 1 in nuclear monoclonal antibody but not with a control antibody (antichymotrypsin; data not shown). extracts of CHO-4 cells. The Complexes Between Nuclear Proteins and SPI-GLE 1 or Phosphorylation Is Required forActivation of the SPI-GLE1 Binding Protein-The hormone/cytokine-induced activation of IRFl Probes from Interferon-y-or GH-treated Cells Are Differ-

Hormone Regulation Growth

of

Dunscription

26211

B

A

.r

.-

o m

0 0

.G

€ E

C

E

-2

Relative binding activtty

c

c

0

0

” _

pM nM PM Concentrationof growh hormone

100

Relative binding activity

80

60 40

20 0 0 0.5 1

2

5

10 15 60

Time of growth hormone stimulation (min)

FIG.4. A, GH induction of nuclear proteins binding to SPI-GLE 1.CHO-4 cells grown to confluency (approximately 2 x 10’) in Ham’s F-12medium (Life Technologies, Inc.) with 10% fetal calf serum (LifeTechnologies, Inc.) were serum starvedfor 12 h and then treated with 100 nM human GH (Pharmacia)for 0,5,10,15, and60 min. Nuclear extracts were prepared a s described and analyzed by GEMSA using a‘”P-labeled SPI-GLE 1 probe. B, specificity of binding of GH-activated nuclear proteins to SPI-GLE 1. GEMSA was carried out as described using nuclear extracts from GH-treated CHO-4 cells with the addition of a 5- or 50-fold excess of unlabeled SPI-GLE 1oligonucleotides or unrelated DNA. C , CHO-4 cells were treated for 10 min with various concentrations of human GH. Nuclear extracts were analyzed by GEMSAusing a SPI-GLE1 probe, and the results were analyzed using a Fuji PhosphorImager. D,time course of activation of binding of nuclear proteins to SPI-GLE 1 by GH. CHO-4 cells were treated for various periods of time with 100nM human GH. Nuclear extracts were analyzedby GEMSA using a SPI-GLE1 probe, and the results were analyzed usink a Fuji PhosphorImager.

ent-Interferon-y stimulation of cells induces the phosphoryl- The elementsused included thosefrom the GBP, FcyR1, ICSBP, ation and subsequently the binding of Stat 91(STAT la) to the IFP-53 IRF-1, LyGE, and Mig genes. These experiments define GAS DNA element (18). It possible is that GH could also induce a palindromic recognition sequence for DNA-binding of GHStat 91 binding to SPI-GLE 1. In order to compare the GH- activated nuclear proteins. activated SPI-GLE 1 binding complex with interferon-y-acti-b vated Stat91, nuclear extractswere prepared from BRL-4 cells T T C C C A GAA TGT 6 treated with recombinantmouse interferon-y or hGH and subjected to GEMSA with a SPI-GLE 1 or IRFl probe (24). The SEQUENCE 1 GH-activated transcription complex migrated more slowly under gel electrophoresis than the interferon-y-activated Stat 91 This sequence is similar but not identical tothe GAS consensus (Fig. 7) with both SPI-GLE 1 and IRF 1 probes. To determine sequence. whether Stat 91 is present in both the GH- and interferon-ystimulated complexes, we performed supershift analyses with TTC ACNNNAA four different anti-Stat 91 antisera produced against four separate GST fusion peptides corresponding to amino acids 2-66, SEQUEXCE 2 515-607,609-716, and 715-750 (26). We were unable to supershift the GH-stimulated complex with any of these antisera, The principal difference is the greaterconservation of the palalthough they were able torecognize a interferon-y-stimulated indrome in the SPI-GLE l . This requirement for a palindomic complex with either the SPI-GLE or IRFl probes (Fig. 7; data sequence is supportedby the observation that SPI-GLE 2 does not shown). shown for the antisera raised against amino acids 2-66). These not bind nuclear proteins in GEMSAanalyses (data results clearly show that the GH-induced DNA-binding com- In addition the GAS’S absolute requirement for a cytosine a t plex is specific to GH and antigenically distinct from that gen- position four is relaxed to a pyrimidine in the SPI-GLE 1. erated by interferon-y. DISCUSSION DNA Binding Specifity for GH-activated Nuclear ProteinsGrowth hormone is a member of a family of cytokines and Some degree of DNA-binding specificity must existif cytokine activation of the Stat 91-like transcription factor family is to growth factors that exert their cellular effects a s a result of have nonoverlapping transcriptional effects. We have investi- binding tospecific membrane-bound receptors (3,4,5,6). These gated this using oligonucleotide competitors with sequences cytokine receptors, in addition to having a number of common structural features, mayalso share related signal transduction representingvariousmembers of the GAS family inthe GEMSA analyses of nuclear extracts from GH-treated CHO-4 pathways. The pathway for interferon-y may be considered to Following ligand binding, interferon-y cells (See Fig. 8 for GEMSA analyses andFig. 2 for sequences). represent the paradigm.

Regulation Hormone Growth

26212

of

Danscription .=0

1

m

a

xw

A C

.-

I

J

a

.2

- Growth hormone

+ Growth hormone

FIG.5. GH-dependentenhancement of TK promoter activity by SPI-GLE 1. Double-stranded oligonucleotides containing three tandem SPI-GLE 1 sequences (CTAGTG'JTCTGAGAAATGAACGGTTCTGAG.@GTACAGGTTCTGAGAAAT) were ligated into the XhaI site of pRLCAT2, a reporter plasmid containing a thymidine kinase minimal promoter adjacentto the chloramphenicol acetyltransferasecDNA. This construct, pSPI-GLE 1-TK-CAT, was used in transient transfection experiments using BRL-4 cells a s described (17). The results of transient transfection experiments using the empty pBLCAT2 (TK-CAT) construct and pBLCAT2 with six tandem repeats of the 45-bp SPI-GHRE cloned into its BamH I site (SPI-GHRE-TK-CAT) (17) are shown for comparison. Each column represents the averageof three independent of percentage of chloramexperiments. Results are expressed in terms phenicol acetyltransferase induction relative to transfected but nonGH-treated cells.

causes the activation of members of the Janus kinase family. These in turn activate, by phosphorylation, the transcription factor Stat 91. Phosphorylated Stat 91 is then translocated to the nucleus where it binds to specific y-activated DNA sequences in interferon-y-activated gene promoters, thereby enhancing transcription (24). In order togain insight into the signal transduction mechanism of growth hormone, we have used GH receptor cDNAtransfected cells and gel electrophoresis mobility shift competitionassaysto define aminimal GH-responsive element, termed SPI-GLE 1, present in the promoters of the growth hormone-regulated serine protease inhibitor 2.1 and 2.2 genes. SPI 2.1,2.2, and 2.3 constitute a family of closely related serine protease inhibitors. The expression levels of SPI 2.1 and 2.2, unlike those for SPI 2.3, are tightly controlled by GH in rat liver. These differences in regulation have been attributed to differences in promoter sequences (23). The SPI 2.1 and 2.2 promoters are virtually identical, while the 2.3 promoter, although sharingsome homology with the 2.112.2 consensus, also containsuniquestretches of sequence. Two independent growth hormone responsive regions have been described in the SPI 2.112.2 promoters; one proximal to the transcription start site between positions -41 and +8 and another, thepreviously mentioned SPI-GHRE, more distal between positions -175 and -114 (23). SPI-GLE 1 lies within the distal region, and since i t is functionally active when placed upstream of a minimal thymidine kinase promoter-driving chloramphenicol acetyltransferase, cDNA expression is likely to be the major contributor to its GH responsiveness. The mechanism of GH regulation of SPI expression via the proximal promoter region is unknown, al-

FIG.6. A, effects of cycloheximide and staurosporine on GH-dependent DNA binding of nuclear proteinsfrom CHO-4 cells using a SPI-GLE 1 probe. CHO-4 cells grown as described were pretreated with 50 pghl cycloheximide or 100 mM staurosporine for 15 min. Treatment was continued for another 10 min with the addition of 100 nhf human GH. Nuclear extracts were prepared and subjected to GEMSA with a R2Plabeled SPI-GLE 1 probe. B , effects of a tyrosine-specific phosphatase on GH-dependent SPI-GLE1 binding. CHO-4 cells grown as described, and nuclear extracts were prepared (13).Prior to GEMSA with a "Plabeled SPI-GLE I probe, nuclear extracts were treated with tyrosinespecific phosphatase 1B (UBI).

though theidentification of several C/EBP binding sites raises the possibility that this transcriptionfactor family may mediate GH responses in some way (23) andmay also explain why SPI mRNA levels are almost undetectable in transformed cell lines (17). DNA elements similar to SPI-GLE 1 are also found in promoters activated by interferon-y (24), interleukin-4 (25, 26) interleukin-6 (271, and prolactin (28) (all hormones binding to members of the cytokine receptor family). The SPI-GHRE contains a second GAS-like element upstream of SPI-GLE 1(Fig. 9). This element showed no binding activity for nuclear proteins in GEMSA analyses. However the fact that it is disrupted by a 42-bp insertion in the GH refractory SPI 2.3 promoter indicates that itmay have a role in vivo. This is interestingin light of the recentdiscovery that both the prolactin-regulated ?./ casein (29, 30) and the interferon-y-regulated mig genes (31) contain tandem repeated GAS-like DNA elements thatmediate transcription regulation. The two GAS-like elements in the /.? casein promoter that bind the mammarygland factorare separated by a binding site for the transcription factor W1 (32). A sequence similar to theYY1 binding consensus, CCATnT, also separates the GAS elements in the SPI-GHRE. This fact, together with the observation that BRL cells transiently cotransfected with a rabbit prolactin receptor expression plasmid and pSPI-CAT and treated with lactogenic hormone express chloramphenicol acetyltransferase activity,* indicates that prolactin and GH may share similar mechanisms of transcription enhancement. We are currently investigating the possibility

T.J. J. Wood, unpublished observations.

Growth Hormone Regulation of ?Fanscription SPI-GLE 1 probe

26213

IRF 1 probe

FIG.7. GEMSA comparison of interferon-activated Stat 9lSTAT and GH-activated GAS-binding factor. Nuclear extracts were prepared from BRL-4 cells treated with either 100 nM human GH or 25 pg/ml mouse interferon-y and analyzed by GEMSA. In the indicated lanes, extracts were incubated with 1 1.11 of anti-Stat 91 sera.

FIG.8. Effect of competition with GAS-like oligonucleotides on GH-dependent binding of nuclear proteins to SPI-GLE 1 DNA. Nuclear extracts were prepared from GH (100 nsl for 10 min) treated CHO-4 cells as described. 50-fold excesses of various GAS-like oligonucleotides (24) were used a s competitors. See Fig. 2 for key.

antisera ina gel electrophoresis mobility shift assaydespite the that, as is the case for prolactin, GH enhances transcriptionby fact thatwe observed a supershift with interferon-y-stimulated cells, and (iii)GH but notinterferon-y is ableto drive chloramovercoming W1-dependent transcription repression. The time frame of appearance and the resistance to inhibi- phenicol acetyltransferase expression from the SPI-GLE L 2 tion by cycloheximide treatment of nuclear proteins binding to Since Stat 91 and thegrowth hormone-activated transcription the SPIpromoter are consistent with GH activating preexisting factor described here share rather similarDNA binding specibut latent transcription factors rather than increasing their ficities, it seems likely that they are related, belonging to a expression. Our results support the theory that this activationfamily of cytokine-activated Stat 91-like transcription factors. is the caused by tyrosine phosphorylation of transcription fac- This hypothesis is consistent with results demonstrating that tors. GH also induces tyrosinephosphorylation of microtubule- the IL-6-activated transcription factor APRF shares sequence associated protein kinase (33) and GH the receptor itself, and it similarities with Stat 91 (431, and a recent report describing is reasonable to speculate that of these all phosphorylations are the existence of a family of Stat proteins with distinct but mediated by the physical association of Janus kinase 2 to the overlapping hormonal specificites (44). As yet, there is very receptor (7). It is interesting that Stat 91activation by inter- little evidence to indicate how relatedtranscription factors feron-y requires a phosphorylated tyrosine residue located in binding to related DNA elements could maintain hormonal the carboxyl terminus (34)of the receptor. This phosphorylated specificity. It is possible that the small differences between tyrosine residue is alsoinvolved in the physical association of i n vitro may be much larger in vivo, DNA binding affinities seen Stat 91 with the interferon-y receptor. a contention supported by the GH-activated DNA-binding comThe Janus kinases are required for the activation of the plex's sensitivity to salt concentration in a gel mobility shift interferon-y-activated, GAS-binding transcription factor Stat 91. This activation, which results from phosphorylation of a assay. Additional factors notseen in vitro but capable of dictatsingle tyrosine, leadsto DNA binding via the formation of Stat ing specificity of transcription enhancement i n vivo may be 91 homodimers (11, 35-38). Since Janus kinase 2 associates present incytokine-activated transcription complexes. Alternawith the growth hormone receptor (71, one may postulate that tively the activation of different Stat molecules may show difGH also induces the phosphorylation of Stat 91. Indeed, during ferent kinetics as supported by the observation that IL-6 actithe preparationof this manuscript several groups reported thevates APRF more rapidly and for a longer period of time than activation by GH of a transcription factor sharing antigenic Stat 91(27). In conclusion, we have identified a y-activated sequence-like determinants with Stat 91 and the binding of a Stat 91-like growth hormone-regulated serine protease protein to the GAS-like SIE in the c-fos promoter (39-42). We DNA element in the inhibitor 2.1 and 2.2 genes that is capable of mediating GH have similarly tried to demonstrate the involvement of Stat 91 or a Stat 91-like molecule in theGH regulation of the SPIgene effects even in nondifferentiated cell lines by specifically intervia SPI-GLE 1. We have, however, obtained and describe here acting with one or more unique GH-activated tyrosine-phosseveral lines of evidence suggestive of a unique GH-induced phorylated transcription factors. Thiselement does not act factor mediating signal transductionvia the SPI-GLE 1.(i) The alone in the GH-dependent regulation of SPI expression, but since it is sufficient to confer GH responsiveness on a heterelectrophoretic mobility of the GH-induced factor binding to ologous promoter, can be regarded a s a paradigmatic growth SPI-GLE 1 is retarded in comparison to interferon-y-stimulated binding to the same element,suggestive of a higher mo- hormone responsive DNA elements and will, therefore, be of lecular weight. (ii) TheGH-induced factor binding to SPI-GLE value for the continued study of GHs mechanism of gene 1 is not supershifted by several different polyclonal Stat 91 regulation.

26214

Growth Hormone Regulation of Dunscription

22. Moller, C., Emtner, M., h e r , P. & Norstedt, G. (1994)Mol. Cell. Endocrinol. 99, 111-117 1. Matthews, L. S., Enberg, B. & Norstedt, G . (1989)J. Biol. Chem. 264, 9905- 23. Paquereau, L., Vilarem, M. J., Rossi, V., Rouayrenc, J. E & Le Cam, A. (1992) 9910 Eur: J. Biochern. 209,1053-1061 2. Leung, D.W., Spenser, S. A,, Cachianes, G., Hammonds, R. G., Collins, C., 24. Pellegrini, S. & Schindler, C. (1993)fiends Biochem. Sci. 18,338-342 Henzel, W. J., Barnard, R., Waters, M. J. & Wood, W. I. (1987)Nature 330, 25. Kotanides, H. & Reich, N. C. (1993)Science 262, 1265-1267 537-543 26. Schindler, C., Kashleva, H., Pernis, A,, Pine, R. & Rothman, P. (1994)EMBO 3. Cosman, D., Lyman, S. D., Idzerda, R. L., Beckman, M. P., Park, L. S., GoodJ. 13,1350-1356 win, R. G. & March, C. J. (1990)Dends Biochem. Sci. 16,265-269 27. Wegenka, U. M., Buschmann, J., Liitticken, C., Heinrich, P. C. & Horn, F. 4. Bazan, J. F. (1990)Roc.Natl. Acad. Sci. U. S. A. 87,6934-6938 (1993)Mol. Cell. Biol. 19, 276-288 5. Thoreau, E., Petridou, B., Kelly, P. A., Djiane, J. & Mornon, J. P. (1991)FEBS 28. Watson, C. J., Gordon,K. E., Robertson,M. & Clark, A. J. (1991)Nucleic Acids Lett. 282, 26-31 Res. 19,6603-6610 6. Aguet, M. (1990)J. Interferon Res. 10, 551-558 29. Meier, V. S. & Groner, B. (1994)Mol. Cell. Eiol. 14, 128-137 7. Argetsinger, L. S., Campbell, G. S., Yang, X.,Witthuhn, B.A., Silvennoinen, O., 30. Raught, B., Khursheed, B., Kazansky, A. & Rosen, J. (1994)Mol. Cell. Biol. 14, Ihle, J. N. & Carter-Su, C. (1993)Cell 74, 237-244 8. Witthuhn, B. A., Quelle, F. W., Silvennoinen, O., Yi, T., Tang, B., Miura, 0. & 1752-1763 Ihle, J . N. (1993)Cell 74, 227-236 31. Wong, P., Severns, C. W., Guyer, N. B. & Wright, T.M. (1994)Mol. Cell. Biol. 9. Silvennoinen, O., Witthuhn, B. A., Quelle, F. W., Cleveland, J. L., Yi, T., & Ihle, 14,914-922 J. N. (1993)Proc. Natl. Acad. Sci. U. S. A. 90, 8429-8433 32. Natesan, S. & Gilman, M.Z. (1994)Genes & Dev. 7, 2497-2509 10. Stahl, N., Boulton, T.G., Farruggell, T., Ip, N. Y., Davis, S., Witthuhn, B. A., 33. Moller, C., Hansson, A., Enberg, B., Lobie, P. E., Norstedt, G. (1993)J. Biol. Quelle, F. W., Silvennoinen, O., Barbieri, G., Pellegrini, S., Ihle, J. N. & Chem. 267,23403-23408 Yancopoulos, G. D. (1994)Science 263,92-95 34. Greenlund,A. C., Farrar,M. A., Viviano, B. L. & Schreiber, R. D. (1994)EMBO. 11. Watling, D., Guschin, D., Muller,M., Silvennoinen, O., Witthuhn, B. A,, Quelle, J. 13, 1591-1600 F. W., Rogers, N. C., Schindler, C., Stark, G. R., Ihle, J. N. & Kerr, I. M. 35. Decker, T., Lew, D. J., Mirkovitch, J. & Darnell, J. E., Jr. (1991)EMBO J. 10, (1993)Nature 366, 166170 927-932 12. Yoon, J.-B., Berry, S. A,, Seelia, S. & lbwle, H. C. (1990)J. Biol. Chem. 265, 36. Igarashi, K., David, M., Finbloom, D. S. & Larner, A. C. (1993)Mol. Cell. Biol. 19947-19954 13, 1634-1640 13. Dignam, J. D., Lebovitz, R. M. & Roeder, R. G. (1983)Nucleic Acids Res. 11, 37. Shuai, R,Ziemiecki, A., Wilks, A. F., Harpur, A. G., Sadowski, H. B., Gilman, 1475-1489 M. 2. & Darnell, J. E. (1993)Nature 366, 580-585 14. Ausubel, F. M., Brent, R., Kingston, R. E., Moore, D. D., Seidman, J. G., Smith, 38. Shuai, K., Horvath, C. M., Tsai Huang, L. H., Qureshi, S. A., Cowburn, D. & J. A. & Struhl, K. 11990)Current Protocol in Molecular Biology, John Wiley Darnell, J. E. Jr. (1992)Cell 76,821328 & Sons, Inc., New York 39. Wang, X., Xu, B., Souza, S. C. & Kopchick, J. J. (1994)Proc. Natl. Acad. Sci. 15. Silvennoinen, O., Schindler, C., Schlessinger, J. & Levy, D. E. (1993)Science U. S. A. 91, 1391-1395 261, 17361738 40. Meyer, D.J., Campbell, G. S., Cochran, B. H.,Argetsinger, L. S., Larner, A. C., 16. Luckow, B. & Schiitz, G. (1987)Nucleic Acids Res. 15, 5490 Finbloom, D. S., Carter-Su, C. & Schwartz, J. (1994)J . Biol. Chem. 269, 17. Enberg, B., Hulthen, A., Moller, C., Norstedt, G. & Francis, S. (1994)J. Mol. 47014704 Endocrinol. 12, 3946 41. Finbloom, D. S., Petricoin, E. F., Hackett, R. H., David, M., Feldman, G. M., 18. Larner,A. C., David, M., Feldman, G. M., Igarashi, K., Hacket, R. H., Webb, D. Igarashi, K.-I., Fibach, E., Weber, M. J., Thorner, M. O., Silva, C. M. & S., Swietzer, S. M., Petricoin, E. F. & Finbloom, D. S. (1993)Science 261, Larner, A. C. (1994)Mol. Cell. €201. 14, 2113-2118 1730-1733 42. Gronowski, A.M. & Rotwein, P. (1994)J. Biol. Chem. 269,7874-7878 19. Ruff-Jamison, S., Chen, K. & Cohen, S. (1993)Science 261, 1733-1736 20. Shuai, K., Schindler, C., Prezioso, V. R. & Darnell, J. E. (1993)Science 258, 43. Wegenka, U. M., Liitticken, C., Buschmann, J.,Yuan,J., Lottspiech,F., MiillerEsterl, W., Schindler, C., Roebb, E., Heinrich, P. C. & Horn, F. (1994)Mol. 1808-1812 Cell. Eiol. 14, 3186-3196 21. Campbell, G. S., Christian, L. J . & Catrer-Su, C. (1993)J. Biol. Chem. 268, 44. Zhong, Z., Wen, Z. & Darnell, J. E. (1994)Science 264, 95-98 7427-7434 REFERENCES