HOXA1-Stimulated Oncogenicity Is Mediated by Selective Upregulation of Components of the P44/42 MAP Kinase Pathway in Human Mammary Carcinoma Cells

Oncogene (2007) 26, 3998–4008 & 2007 Nature Publishing Group All rights reserved 0950-9232/07 $30.00 www.nature.com/onc ORIGINAL ARTICLE HOXA1-stimulated oncogenicity is mediated by selective upregulation of components of the p44/42 MAP kinase pathway in human mammary carcinoma cells

KM Mohankumar1,4,XQXu2,4, T Zhu3, N Kannan1, LD Miller2, ET Liu2, PD Gluckman1, S Sukumar3, BS Emerald1 andPE Lobie 1

1The Liggins Institute and National Research Centre for Growth and Development, University of Auckland, Auckland, New Zealand; 2Microarray and Expression Genomics, Genome Institute of Singapore, Singapore, Republic of Singapore and 3Breast Cancer Program, Sidney Kimmel Comprehensive Cancer Center at Johns Hopkins, Baltimore, MD, USA

Expression of homeobox A1 (HOXA1) results in onco- Introduction genic transformation of immortalized human mammary epithelial cells with aggressive tumor formation in vivo. Homeobox A1 (HOXA1) is a member of the home- However, the mechanisms by which HOXA1 mediates odomain containing transcription factor family, which oncogenic transformation is not well defined. To identify plays a pivotal role during the normal growth and molecules that could potentially be involved in HOXA1- differentiation of mammalian tissues. HOX genes mediated oncogenic transformation, microarray analysis exhibit distinct patterns of expression during develop- was utilized to characterize and compare the gene ment, andalterations in their expression have been expression pattern in response to forced expression or linked to various diseases, including development of depletion of HOXA1 in human mammary carcinoma cells. neoplasia (Chen andSukumar, 2003). In the human Gene expression profiling identified that genes involved in mammary gland, the expression of the HOXA1 gene is the p44/42 mitogen-activated protein (MAP) kinase very low or absent during normal growth and differen- activation pathway (GRB2, MAP kinase kinase (MEK1) tiation, but is detectable in a variety of human breast and SDFR1) orp44/42 MAP kinase-regulated genes cancer lesions (Chariot andCastronovo, 1996). We have (IER3, EPAS1, PCNA and catalase) are downstream previously observedthat forcedexpression of HOXA1 expression targets of HOXA1. Forced expression of HOXA1 in immortalizedhuman mammary epithelial cells MCF- increased GRB2 and MEK1 mRNA and protein expres- 10A, concomitantly enhances proliferation andcell sion and increased p44/42 MAP kinase phosphorylation, survival, thus forming the basis for abnormal mammary activity and Elk-1-mediated transcription. Use of a MEK1 acinar morphogenesis, oncogenic transformation and inhibitor demonstrated that increased p44/42 MAP kinase tumor formation in vivo. Use of a pharmacological activity is required for the HOXA1-mediated increase in inhibitor of B-cell leukemia/lymphoma-2 (Bcl-2) com- cell proliferation, survival, oncogenicity and oncogenic pletely prevented anchorage-independent growth seen in transformation. Thus, modulation of the p44/42 MAP MCF-10A cells overexpressing HOXA1 (Zhang et al., kinase pathway is one mechanism by which HOXA1 2003). However, the molecular mechanisms by which mediates oncogenic transformation of the human mam- HOXA1 induces oncogenic transformation and pro- mary epithelial cell. motes tumorigenicity of human mammary epithelial Oncogene (2007) 26, 3998–4008; doi:10.1038/sj.onc.1210180; cells remains poorly understood. publishedonline 8 January 2007 Signalling through pathways involving mitogen- activatedprotein kinases (MAP kinases) has been impli- Keywords: HOXA1; gene expression; p44/42 MAP kinase; catedin the pathogenesis of cancer. The p44/42 MAP MEK; oncogenicity kinase pathway is a well-characterizedintracellular signalling pathway that transduces signals from the cell surface to the nucleus to activate transcription. Activa- tion of p44/42 MAP kinase occurs by dual phosphorylation of threonine andtyrosine residues(Seger and Krebs, 1995). This activation is executedby an enzyme known as MAP kinase kinase. p44/42 MAP kinase has Correspondence: Professor PE Lobie, The Liggins Institute, University been demonstrated to be involved in various cellular of Auckland, 2-6 Park Avenue, Private Bag 92019, Auckland, New processes, including cell proliferation, differentiation, Zealand. apoptosis andoncogenic transformation. Constitutive E-mail: [email protected] 4Both these authors contributedequally to this work. activation of this signalling cascade has been noted in Received5 September 2006; revised26 October 2006; accepted27 October the oncogenic transformation of various cell lines, 2006; publishedonline 8 January 2007 andincreasedp44/42 MAP kinase activity has been Mechanism of HOXA1 oncogenicity KM Mohankumar et al 3999 implicatedin the metastatic potential of human cancer a (Mansour et al., 1994). The p44/42 MAP kinase activity level alone has been suggestedto have prognostic significance for the survival of breast cancer patients (Mueller et al., 2000). Additionally, an orally active inhibitor of MAP kinase kinase (MEK), PD184352, was observedto inhibit significantly the growth of colon carcinomas of both murine andhuman origin (Sebolt-

Leopold et al., 1999). MCF7-scrambled-siRNA MCF7-HOXA1 siRNA MCF7-VECTOR MCF7-HOXA1 To elucidate the mechanisms by which HOXA1 HOXA1 induces oncogenic transformation of the human mam- β mary epithelial cells, we evaluatedthe changes in gene -ACTIN expression in human breast carcinoma cells, MCF-7, in response to HOXA1. Oligonucleotide microarray b 10 MCF7-scrambled-siRNA analysis revealedthat HOXA1 upregulates multiple com- ∗∗∗∗ MCF7-HOXA1siRNA ponents of the p44/42 MAP kinase pathway, resulting in MCF7-VECTOR increasedp44/42 MAP kinase activity. This increased 8 p44/42 MAP kinase activity mediates HOXA1-stimu- MCF7-HOXA1 latedoncogenicity of human breast carcinoma cells andoncogenic transformation of immortalizedhuman 6 mammary epithelial cells.

Results ARBITRARY UNITS 2 Microarray analysis identifies downstream determinants of HOXA1 ∗ To identify potential mechanisms by which HOXA1 may result in oncogenic transformation, we resortedto 0 the use of the described microarray platform. Two Figure 1 Demonstration of forcedexpression andfunctional complementary pairs of cell lines were utilizedfor these silencing of HOXA1 upon stable transfection with HOXA1 cDNA studies. The first was the well-characterized mammary andHOXA1 siRNA. MCF-7 cells were stably transfectedwith carcinoma cell line (MCF-7) with forcedexpression of scrambledsiRNA vector (MCF-7-scrambled-siRNA), HOXA1 pSilencer (MCF-7-HOXA1 siRNA), pCMV Tag2B vector (MCF-7- HOXA1 andas comparedwith the vector transfected vector) andthe vector containing HOXA1 cDNA (MCF-7- control (Zhang et al., 2003). Secondly, we generated a HOXA1). (a) The level of HOXA1 mRNA was determined by small interfering RNA (siRNA) construct of human RT–PCR under serum replete conditions (densitometric analysis HOXA1 to target endogenous HOXA1 mRNA for demonstrated a 2.2170.17-fold decrease with siRNA to deplete degradation (Zhu et al., 2005) andstably expressedthis HOXA1 anda 3.7 70.31-foldincrease with forcedexpression of HOXA1, in HOXA1 mRNA level comparedto the respective siRNA in MCF-7 cells. For control purposes, we controls) and( b) HOXA1 transcriptional activity by reporter utilizedthe same vector with a scrambledsiRNA target assay, as indicated. b-Actin was usedas loadingcontrol where sequence. The expression levels of these two pairedcell appropriate. *Po0.01. lines were verifiedby reverse transcriptase–polymerase chain reaction (RT–PCR) analysis. MCF-7-HOXA1 demonstrated higher levels of HOXA1 mRNA when output as only those genes with the appropriate comparedwith that of MCF-7-vector, andMCF-7- directional regulation (i.e., increased expression with HOXA1 siRNA demonstrated lower levels of HOXA1 forcedexpression of HOXA1 anddecreasedexpression mRNA when comparedwith MCF-7-scrambled-siRNA with HOXA1 depletion or vice versa) were to be finally (Figure 1a). This was further verifiedby the analysis of selected. Firstly, to identify genes regulated by forced HOXA1 transcriptional activity by use of the EphA2- expression of HOXA1 in mammary carcinoma cells, r42B enhancer (Zhang et al., 2003). MCF-7-HOXA1 RNA from MCF-7-HOXA1 cells was hybridized to that cells exhibitedincreasedHOXA1-mediatedtranscrip- of the MCF-7-vector cells. The expression level of each tional activity when comparedwith MCF-7-vector gene was representedby pseudocolorin matrix format, cells, andMCF-7-HOXA1 siRNA demonstrateda with redrepresenting increasing mRNA levels andgreen lower level of HOXA1-mediated transcriptional activity denoting decreasing mRNA levels, and color inten- when comparedwith the MCF-7-scrambled-siRNA sity representing the magnitude of the expression ratio. (Figure 1b). As listedin Supplementary dataTable 1, microarray We subsequently usedthe RNA extractedfrom the analysis identified 2450 genes that exhibited a 1.5-fold or two pairedcell lines to perform microarray analyses to greater change with forcedexpression of HOXA1. identify HOXA1-regulated genes. The aim of this Among these, 1130 genes were upregulatedand1320 approach was to improve specificity of the microarray were downregulated. We next examined the changes in

Oncogene Mechanism of HOXA1 oncogenicity KM Mohankumar et al 4000 gene expression when HOXA1 was depleted in MCF-7 classifiedby broadbiological function. They encode mammary carcinoma cells. RNA isolatedfrom MCF-7- proteins that are involvedin a broadrange of cellular HOXA1 siRNA cell line was hybridized with RNA functions andmetabolic pathways. The functions of isolatedfrom MCF-7-scrambled-siRNA control cells. these genes include regulation of apoptosis, biosyn- Microarray analysis identified 1089 genes that exhibited thesis, cell adhesion, cell cycle, development, DNA/RNA a 1.5-foldor greater change in response to depletion of HOXA1. Among these, 486 genes were upregulatedand 603 were downregulated (Supplementary data Table 1). a We subsequently comparedthe expression profiles from the two sets of pairedcell lines to determinewhich genes were coordinately regulated and would therefore represent potential HOXA1 downstream effector mole- MCF7-HOXA1 MCF7-HOXA1 siRNA cules. Among 1130 upregulatedgenes in MCF-7 cells with forcedexpression of HOXA1, 424 corresponding 706 genes were observedto exhibit decreasedexpression in MCF-7-HOXA1 siRNA comparedwith the MCF- 424 7-scrambled-siRNA cell line. Of the 1320 genes with decreased expression upon forced expression of 179 HOXA1, the expression of 260 corresponding genes increasedin HOXA1-depletedcells. Thus, a total of 684 1060 genes demonstrated concordant directional regulation between the two pairedcell sets. This accounts for 28% 260 of total differentially expressed genes in cells with forced 226 expression of HOXA1 and63% of that in the HOXA1- depleted cells. A hierarchical clustering algorithm based on Pearson correlation coefficients was appliedto group b MCF7-HOXA1 genes on the basis of expression pattern similarity over MCF7-HOXA1siRNA two sets of expression profiles. The expression level of 3 0.89 each gene was representedby pseudocolor in matrix 2 format (Figure 2a). Interestingly, the foldchange of 1 each gene in the intersecting regions of the Venn 0 diagram in Figure 2a exhibited relatively similar –1 Log2 ratio magnitudes of expression in the appropriate direction. –2 The correlation efficiencies between these two indepen- –3 dent microarray experiments were significantly high, 50 100 150 200 250 0.89 for upregulatedgenes (Figure 2b) and0.60 for NUMBER OF GENES c 3 downregulated genes (Figure 2c). 0.60 2

Functional classiﬁcation of the differential responsive 0 genes to HOXA1 –1 To assess the potential functionality of genes regulated Log2 ratio by HOXA1, the gene products were grouped into Gene –2 –3 Ontology (http://www.geneontology.org) categories 50 100 150 200 250 with the assistance of StanfordSOURCE database NUMBER OF GENES (Diehn et al., 2003). Of the 424 upregulatedgenes, 334 d APOPTOSIS hadknown biological roles andof 260 downregulated HOXAI DOWN REGULATED GENES genes, 140 hadknown biological roles. These genes were BIOSYNTHESIS HOXAI UP REGULATED GENES CELL ADHERSION CELL CYCLE DEVELOPMENT Figure 2 HOXA1 regulation of gene expression. (a) Diagram DNA/RNA PROCESSING

depicting the number of genes differentially expressed in each IMMUNE RESPONSE group andtheir overlaps. Redindicatesincreasing mRNA levels METABOLISM and green denotes decreasing mRNA levels. The degree of color SIGNAL TRANSDUCTION saturation reﬂects the magnitude of the ratio that compares the relative expression levels of HOXA1-regulatedgenes between STRUCTURE/CYTOSKELETON overexpression andsilencing of HOXA1. The graph depicting STRESS RESPONSE 424 genes upregulatedby HOXA1 and260 genes downregulatedby TRANSCRIPTION REGULATION HOXA1 shows similar magnitude of fold changes in HOXA1 TRANSPORT overexpression versus HOXA1 silencing in MCF-7 cells, as OTHERS indicated by the correlation coefﬁcient value of 0.89 (b) and0.60 UNKNOWN (c). (d) Genes were groupedinto Gene Ontology (http://www. 020406080100 120 geneontology.org) categories with the assistance of Stanford SOURCE database. NUMBER OF GENES

Oncogene Mechanism of HOXA1 oncogenicity KM Mohankumar et al 4001 processing, metabolism, signal transduction, cellular a b structure, stress response, transcription regulation and cellular transport. Regulatory molecules involvedin signal transduction, metabolism and biosynthesis com- prisedthe bulk of the gene lists (Figure 2d).The genes with known or predictedfunctions are also presentedin Supplementary data Tables 2 and 3 with fold change siRNA MCF7-VECTOR MCF7-HOXA1siRNA MCF7-HOXA1 MCF7-scrambled-siRNA indicated. -

GRB2 siRNA MEK1 HOXA1 scrambled p44/42MAP kinase pathway candidate genes are VECTOR - - -HOXA1 upregulated by HOXA1 SDFR1 - MCF7- MCF7 MCF7 MCF7 To obtain a more detailed understanding of the IER3 MCF7 molecular mechanisms potentially utilizedby HOXA1 to stimulate oncogenicity, we focusedour analysis on EPAS1 GRB2 components of speciﬁc pathways utilizedfor signal PCNA MEK1 transduction. We initially observed from the HOXA1- Catalase β-ACTIN regulatedgene list that components of the p44/42 MAP β kinase pathway were modulated by HOXA1. Indeed, we -ACTIN observedincreasedexpression of several genes impor- c tant in p44/42 MAP kinase-mediated signalling, im- plicating this pathway as a potential expression target of HOXA1. Supplementary data Table 4 represents 29 MDA-MB-231 MDA-MB-435 MDA-MB-468 HS578T SkBr3 ZR75-1 21PT T47D 21MT 21PT HBL100 MCF10A genes that are involvedin the p44/42 MAP kinase MCF7 pathway andregulation of their expression by HOXA1. MEK1 To verify the ﬁdelity of data generated from the microarray analysis, we performedsemiquantitative HOXA1

RT–PCR with the different microarray identified genes β-ACTIN that activate, (SDFR1) regulate (GRB2 andMEK1) or are regulated(IER3, EPAS1, PCNA andcatalase) by Figure 3 HOXA1 regulates the expression of components of the the p44/42 MAP kinase pathway. As observedin p44/42 MAP kinase pathway. (a) MCF-7-scrambled-siRNA, Figure 3a, forcedexpression of HOXA1 concomitantly MCF-7-HOXA1 siRNA, MCF-7-vector andMCF-7-HOXA1 cells increasedthe mRNA levels of GRB2, MEK1, SDFR1, were culturedin complete mediaandthe mRNA level of GRB2, IER3, EPAS1, PCNA andcatalase. Similarly, with MEK1, SDFR1, EPAS1, PCNA andcatalase was determinedby semiquantitative RT–PCR. b-Actin was usedas control. Densito- siRNA-mediated HOXA1 depletion in MCF-7 cells, the metric quantification shows GRB2 (2.9870.12-folddecreasewith mRNA expression of each of these genes was decreased. siRNA to deplete HOXA1/2.3270.18-foldincrease with forced Thus, HOXA1 specifically regulates the expression of expression of HOXA1), MEK1 (3.2970.22-folddecreasewith genes directly involved in the p44/42 MAP kinase siRNA to deplete HOXA1/2.5470.13-foldincrease with forced activation pathway, specifically GRB2 andMEK1. expression of HOXA1) mRNA level in MCF-7-scrambled-siRNA, MCF-7-HOXA1 siRNA, MCF-7-vector andMCF-7-HOXA1 The increasedexpression of GRB2 andMEK1, in cells. (b) The protein levels of GRB2 andMEK1 were determined response to increasedexpression of HOXA1, was by Western blotting. b-Actin was usedas control. Densitometric further substantiatedby Western blot analysis. MCF-7- quantification shows GRB2 (2.8670.19-folddecreasewith siRNA HOXA1 cells exhibitedhigher expression of GRB2 and to deplete HOXA1/2.0370.11-foldincrease with forcedexpression of HOXA1), MEK1 (2.7870.33-folddecreasewith siRNA to MEK1 protein as comparedwith the MCF-7-vector cell deplete HOXA1/2.470.27-foldincrease with forcedexpression of line (Figure 3b). Similarly, expression of both GRB2 HOXA1) protein level in MCF-7-scrambled-siRNA, MCF-7- andMEK1 protein was decreasedin MCF-7 cells with HOXA1 siRNA, MCF-7-vector andMCF-7-HOXA1 cells. siRNA-mediated depletion of HOXA1 (Figure 3b). (c) The mRNA expression level of MEK1 andHOXA1 in different Thus, components of the p44/42 MAP kinase activation mammary epithelial andmammary carcinoma cell lines was estimated by semi-quantitative RT–PCR as indicated. b-Actin pathway are downstream expression targets of HOXA1. was usedas control. To determine whether HOXA1 expression would broadly correlate with expression of MEK1, we performedsimilar semiquantitative RT–PCR experiments on RNA extractedfrom a variety of mammary p44/42MAP kinase activity is increased by forced epithelial andmammary carcinoma cells. As illustrated expression of HOXA1 in Figure 3c, there was a tight correlation between As HOXA1 increasedexpression of components of HOXA1 andMEK1 mRNA expression in 9 of the 12 the p44/42 MAP kinase activation pathway, it could cell lines testedfor both genes, the exceptions being the be expectedthat forcedexpression of HOXA1 would T47D, MDA-MB-231 andMDA-MB-468 cell lines. also result in increasedactivity of p44/42 MAP kinase. This suggests that HOXA1 is one of the major The activation of p44/42 MAP kinase by various regulators of MEK1 expression in mammary carcinoma growth factors has previously been demonstrated to cells. require dual phosphorylation on threonine and tyrosine

Oncogene Mechanism of HOXA1 oncogenicity KM Mohankumar et al 4002 (Thr202 andTyr204) (Seger andKrebs, 1995). We, dual phosphorylated p44/42 MAP kinase was higher therefore, examinedfor activatedp44/42 MAP kinase in with forcedexpression of HOXA1 comparedwith that MCF-7 cells with forcedexpression of HOXA1, andin of vector-transfectedcontrol cells (Figure 4a). Consis- cells with siRNA-mediated depletion of HOXA1, using tent with this, the level of dual phosphorylated p44/42 an antibody that speciﬁcally recognizes dual phosphory- MAP kinase was lower in MCF-7 cells with siRNA- latedp44/42 MAP kinase (phospho-ERK1/2). We mediated depletion of HOXA1 compared with that observedthat neither forcedexpression nor siRNA- of scrambled-siRNA vector-transfected control cells mediated depletion of HOXA1 inﬂuenced total p44/42 (Figure 4a). To verify this observation, we also MAP kinase protein expression. However, the level of examinedthe effect of HOXA1 on actual p44/42 MAP

a HOXA1 MCF7-scrambled-siRNA siRNA MCF7-HOXA1 MCF7-VECTOR MCF7-

phospho p44/42 MAP kinase

p44/42 MAP kinase activity

p44/42 MAP kinase

bc7 7 MCF7-scrambled-siRNA ∗∗ MCF7-HOXA1 siRNA ∗∗ 6 6 MCF7-VECTOR MCF7-HOXA1 5 5

4 4

3 3

2 2 ARBITRARY UNITS ARBITRARY UNITS ∗ 1 ∗ 1

0 0

de50 14 ∗∗ ∗∗ 12 40 10 30 8

20 6

4 ARBITRARY UNITS ARBITRARY UNITS 10 2 ∗ ∗ 0 0 Figure 4 Activation of p44/42 MAP kinase and p44/42 MAP kinase-dependent transcription by HOXA1 in human mammary carcinoma cells. (a) Western blot analysis with anti-phospho-p44/42 MAP kinase and in vitro kinase assay to demonstrate the activity of p44/42 MAP kinase in MCF-7-scrambled-siRNA, MCF-7-HOXA1 siRNA, MCF-7-vector and MCF-7-HOXA1 cells. Cells were grown to 80% conﬂuence andwere processedfor p44/42 MAP kinase assay andWestern blot as describedin‘Materials andmethods’. All four different stable cell lines were grown to 60–80% conﬂuence and were transiently transfected with the Elk promoter-luciferase construct in serum-free (b) andserum conditions( c). Cells were processed for luciferase assay as described in ‘Materials and methods’. The same procedure was followed for Egr-3 luciferase construct in serum free (d) andserum conditions( e). Results represent mean of triplicate determinations; bars, s.d. *Po0.001.

Oncogene Mechanism of HOXA1 oncogenicity KM Mohankumar et al 4003 kinase activity by assessing the in vitro activity of p44/42 mammary carcinoma cell number (Zhang et al., 2003) MAP kinase activity in MCF-7-HOXA1 andMCF-7- is mediated by p44/42 MAP kinase. As shown in HOXA1 siRNA cells comparedwith their respective Figure 5a, the HOXA1-stimulatedincrease in mammary controls. Concordant with the level of p44/42 MAP carcinoma cell number was largely preventedby kinase phosphorylation, we observedthat increased PD098059. Increase in total mammary carcinoma cell expression of HOXA1 significantly increasedp44/42 number, stimulatedby HOXA1, is consequent on MAP kinase activity, whereas decreased HOXA1 increases in both cell proliferation andcell survival expression resultedin decreasedp44/42 MAP kinase (Zhang et al., 2003), albeit the proliferative effect activity (Figure 4a). HOXA1, therefore, regulates p44/ stimulatedby HOXA1 is minimal. The p44/42 MAP 42 MAP kinase activity in human mammary carcinoma kinase dependence of HOXA1-stimulated proliferation cells. was, therefore, determined by incorporation of 50- bromo-20-deoxyuridine (BrdU) during entry to S phase. PD098059 preventedthe HOXA1-stimulatedincrease in HOXA1 activates Elk-1-mediated transcription BrdU incorporation in mammary carcinoma cells as Increasedphosphorylation andactivity of p44/42 MAP shown in Figure 5b. Thus, HOXA1-stimulatedincreases kinase will result in transcriptional activation of down- in human mammary carcinoma cell proliferation is p44/ stream effector molecules. One mechanism by which 42 MAP kinase dependent. p44/42 MAP kinase regulates transcriptional activation We next examined the dependence of HOXA1- is by use of Elk-1-mediated transcription (Gille et al., stimulatedcell survival on p44/42 MAP kinase activity. 1995). Elk-1-mediated luciferase activity was increased As observedin Figure 5c, forcedexpression of HOXA1- almost sixfoldby forcedexpression of HOXA1 and protectedmammary carcinoma cells from apoptosis due decreased to 63% by HOXA1 depletion when compared to serum deprivation and exposure of cells to PD098059 with that of the respective vector control in serum-free abolishedthis protective effect of HOXA1. We have medium (Figure 4b). Stimulation with serum similarly previously demonstrated that forced expression of resultedin an almost sixfoldincrease of Elk-1-mediated HOXA1 in mammary carcinoma cells resultedin transcription by forcedexpression of HOXA1 compared decreased apoptotic cell death mediated by a specific with control, andElk-1-mediatedtranscription was increase in Bcl-2 (Zhang et al., 2003). Thus, HOXA1 decreased to 70% by depletion of endogenous HOXA1 may utilize p44/42 MAP kinase to mediate upregulation expression comparedwith control (Figure 4c). of Bcl-2 requiredfor cell survival. We examinedthe We have further validated the effect of altered effect of inhibition of p44/42 MAP kinase on the HOXA1 expression on Elk-1 transcriptional activation HOXA1-stimulatedincrease in Bcl-2 mRNA. Concor- by use of the Egr-3 luciferase reporter construct to dant with previous observation, MCF-7-HOXA1 cells examine p44/42 MAP kinase-mediated transcriptional exhibitedhigher Bcl-2 mRNA in comparison with activation. The Egr-3 luciferase reporter construct MCF-7-vector cells in serum-free media (Figure 5d). contains a luciferase coding sequence, downstream of As observedin Figure 5d,the HOXA1-stimulated an Egr-1 promoter fragment, comprising 624 bp 50 of the upregulation of Bcl-2 mRNA was significantly reduced initiation codon. This 624 bp Egr-1 promoter fragment by treatment with PD098059. Thus, the HOXA1- possesses five SRF/Ets response elements, that are stimulatedincreases in Bcl-2 expression andcell survival responsive to p44/42 MAP kinase stimulation through is p44/42 MAP kinase dependent. phosphorylation of Elk-1 andSap1a (Clarkson et al., 1999), We observeda 45-foldincrease in Elk-1- and Sap1a-mediated transcription with forced expression of HOXA1-mediated oncogenicity is p44/42MAP HOXA1, and a decrease to 67% in HOXA1-depleted kinase-dependent cells comparedwith their respective vector controls in Anchorage-independent growth is one characteristic serum-free condition (Figure 4d). In the presence of feature of oncogenically transformedcells. This process serum, Elk-1- andSap1a-mediatedtranscription was can be simulated in vitro by measurement of the cell’s increased12-foldupon forcedexpression of HOXA1, ability to grow independently of substrate adhesion and and it decreased to 16% by siRNA-mediated depletion to form colonies on a soft agar (Shin et al., 1975). of HOXA1 (Figure 4e). HOXA1, therefore, regulates Forcedexpression of HOXA1 in mammary carci- p44/42 MAP kinase-mediated transcription in human noma cells significantly increasedthe number of colo- mammary carcinoma cells. nies formedon soft agar comparedwith control (Figure 6a) in accordwith the previous demonstration HOXA1-stimulated cellular proliferation and survival is that HOXA1 increases oncogenicity (Zhang et al., p44/42MAP kinase-mediated 2003). The ability of HOXA1 to stimulate the formation Given the pivotal role of p44/42 MAP kinase activity in of colonies on soft agar was significantly reduced by oncogenic transformation (Mansour et al., 1994), it is treatment with PD098059 (Figure 6a). We verifiedthis plausible that HOXA1 increasedp44/42 MAP kinase further by use of suspension culture, another measure activity mediates HOXA1-stimulated oncogenicity of for anchorage-independent growth (Figure 6b). Thus, human mammary carcinoma cells. To address this, the HOXA1-stimulatedincrease in oncogenicity of we utilizedthe MEK1 inhibitor, PD098059, to human mammary carcinoma cells is p44/42 MAP assess whether the HOXA1-stimulatedincrease in total kinase-dependent.

Oncogene Mechanism of HOXA1 oncogenicity KM Mohankumar et al 4004 a 1000 ∗ MCF7-VECTOR b )

3 60 MCF7-HOXA1 10 800 ∗ ؋

600 40 ∗∗ ∗∗

400 20 200 BrdU INCORPORATION TOTAL CELL NUMBER (

0 PROPORTION OF CELLS WITH 0 VEHICLE + + – – VEHICLE + + – – PD098059 – – + + PD098059 – – + +

cd20

15 ∗∗ VECTOR VECTOR HOXA1 VECTOR VECTOR HOXA1 VECTOR VECTOR - - - 10 - MCF7-HOXA1 MCF7 MCF7-HOXA1 MCF7 MCF7- MCF7- MCF7 MCF7

5 Bcl-2

% APOPTOTIC NUCLEI ∗ β-ACTIN + + – – 0 VEHICLE PD098059 – – + + VEHICLE + + – – PD098059 – – + + Figure 5 p44/42 MAP kinase mediates HOXA1-stimulated increase in total cell number, cell proliferation and survival. MCF-7- vector andMCF-7-HOXA1 cells were culturedin complete media 720 mM PD098059. (a) Total cell number, (b) cell cycle progression (BrdUrd incorporation), (c) apoptotic cell death and (d) the level of Bcl-2 mRNA was determined in both cell lines as described in ‘Materials and methods’. Results represent mean of triplicate determinations; bars, s.d. *Po0.001.

Forcedexpression of HOXA1 has also been demon- When culturedin a reconstitutedbasement membrane stratedto result in de novo oncogenic transformation of (Matrigel), immortalizedhuman mammary epithelial immortalizedhuman mammary epithelial cells (Zhang cells undergo a series of morphogenetic processes, et al., 2003). We, therefore, stably transfectedimmorta- resulting in the formation of acinus-like structures lizedhuman mammary epithelial cells (MCF-10A) with containing a single layer of polarizedcells surrounding pCMV Tag2B vector (MCF-10A-vector) or the vector a hollow lumen. Formation of the luminal space in this containing HOXA1 cDNA (MCF-10A-HOXA1). The model follows the development of apico-basal polarity, expression levels of these stable cell lines were veriﬁedby andinvolves apoptosis andan autophagy-like process in RT–PCR analysis andHOXA1-mediatedtranscrip- the centrally localizedcells (Debnath et al., 2003). In tional activity. The MCF-10A-HOXA1 stable cell line contrast, oncogenically transformedmammary epithe- exhibitedhigher levels of HOXA1 mRNA andHOXA1- lial cells form large, non-polarized, undifferentiated mediated transcriptional activity when compared with colonies without lumina when grown in matrigel (Zhu MCF-10A-vector (Figure 6c andd). et al., 2005). We usedthis ex vivo model to examine To determine whether HOXA1-stimulated oncogenic potential disruptive effects of HOXA1 on the architec- transformation per se was dependent on p44/42 MAP ture of acinar structures formedby immortalizedhuman kinase activity, we examinedthe ability of PD098059 to mammary epithelial cells. Three-dimensional (3D) aci- prevent soft agar colony formation in MCF-10A cells nar structures were generatedby plating MCF-10A cells consequent on forcedexpression of HOXA1. As as single cells in matrigel. Confocal laser scanning expected, forced expression of HOXA1 in MCF-10A microscopic analyses of acini, labelledwith basal surface cells produced colonies in soft agar, whereas the control marker (a6 integrin) andapical polarity marker vector-transfectedcells exhibitedonly a small number of (GM130), indicated that the acini formed consisted of spontaneously formedcolonies (Figure 6e), as described polarizedmammary epithelial cells anda hollow lumen previously. As observedin Figure 6e, use of PD098059 (Figure 7a). Forcedexpression of HOXA1 in immorta- completely preventedHOXA1-stimulatedsoft agar colo- lizedhuman mammary epithelial cells resultedin the ny formation by the MCF-10A-immortalizedhuman generation of large disorganized multi-acinar structures mammary epithelial cell line. Thus, p44/42 MAP kinase with ﬁlledlumina, andalso disruptedcellular polariza- mediates HOXA1-induced oncogenic transformation of tion (Figure 7b). Fifty-eight percent of the formed human mammary epithelial cells. acinar structure in MCF-10A-HOXA1 exhibit large

Oncogene Mechanism of HOXA1 oncogenicity KM Mohankumar et al 4005

) 700 abMCF7-VECTOR 3

10 ∗ 3 MCF7-HOXA1 600 ؋ ∗ 500 ∗∗ 2 ∗∗ 400

300

1 200 OF COLONIES 100 RELATIVE NUMBER

0 TOTAL CELL NUMBER ( 0 + + – – VEHICLE + + – – VEHICLE – – + + PD098059 – – + + PD098059

cd12 e 10 ∗ ∗ MCF10A-VECTOR MCF10A-HOXA1 10 8

8 6 6 MCF10A-VECTOR MCF10A-HOXA1 4 HOXA1 4

β-ACTIN ARBITRARY UNITS ∗∗ 2 2

0 RELATIVE NUMBER OF COLONIES 0 VEHICLE + + – – PD098059 – – + +

Figure 6 Forced expression of HOXA1 enhances anchorage-independent growth in a p44/42 MAP kinase pathway-dependent manner. (a) Soft agar colony formation and( b) anchorage-independent growth in suspension culture of MCF-7-vector and MCF-7- HOXA1 in bacteriological dishes were performed as described under ‘Materials and methods’. MCF-10A cells were stably transfected with pCMV Tag2B vector (MCF-10A-vector) andthe vector containing HOXA1 cDNA (MCF-10A-HOXA1). ( c) The level of HOXA1 mRNA was determined by RT–PCR under serum conditions and (d) HOXA1 transcriptional activity by reporter assay, as indicated. (e) Soft agar colony formation by MCF-10A-vector andMCF-10A-HOXA1 cells with or with out 10 mM PD098059. The results are presented as the mean of triplicate determinations; bars, s.d. *Po0.001.

disorganized multiacinar structure with filled lumina cells andidentifiedcomponents of the p44/42 MAP (Figure 7c). After treatment with MEK1 inhibitor, only kinase pathway as one major effector of HOXA1- 14% of MCF-10A-HOXA1 cells formedlarge multi- mediated oncogenicity and oncogenic transformation. acinar structures andthe majority of acini possesseda The genome-wide analyses performed here demon- lumen (Figure 7dande). HOXA1, therefore, specifically stratedthat HOXA1 upregulatedthe expression of utilizes p44/42 MAP kinase for execution of its genes that are downstream of the p44/42 MAP kinase oncogenic program. pathway. Expression targets of HOXA1 include PCNA, SDFR1, IER3 andEPAS1 genes that are involvedin cell cycle regulation, angiogenesis andproliferation. One of the upregulatedgenes, SDFR1, a chemokine receptor, Discussion has been shown to activate p44/42 MAP kinase pathway during astrocyte proliferation (Bajetto et al., 2001). Dysregulatedexpression of many HOX genes has been Another upregulatedgene, IER3 (IEX-1), is an im- shown to be involvedin the oncogenic process (Chen mediate upstream regulator that potentiates p44/42 andSukumar, 2003). Owing to the simplicity of core MAP kinase activation in response to growth factors binding motifs for HOX genes (such as ATTA) that are andis also involvedin the regulation of cell death and present in many potential HOX target genes promoters, oncogenesis (Garcia et al., 2002). In MCF-12A mam- delineation of the mechanism of action of HOX genes in mary epithelial cells, constitutive activation of p44/42 neoplasia is complex. We hadpreviously demonstrated MAP kinase pathway also results in transcriptional that forcedexpression of HOXA1 in immortalized upregulation of IER3 (Grill et al., 2004). Endothelial human mammary epithelial cells resultedin an increase PAS-domain protein 1 (EPAS1) is phosphorylated by in cell number, primarily by the promotion of cell p44/42 MAP kinase pathway, andis also involvedin survival, mediated by the transcriptional upregulation angiogenesis of human renal carcinoma (Conrad et al., of Bcl-2 (Zhang et al., 2003). Here, we have used 1999). Proliferating cell nuclear antigen (PCNA) has a oligonucleotide microarray analysis to identify expres- pivotal role in the regulation of cell growth, acts as sion targets of HOXA1 in human mammary carcinoma a prognostic factor of metastasis in breast cancer andis

Oncogene Mechanism of HOXA1 oncogenicity KM Mohankumar et al 4006 a b

c e

MCF10A -VECTOR 70 MCF10A-HOXA1

d 40 30

20 ∗ 10

0 % SPHEROIDS WITH FILLED LUMEN VEHICLE + + – – PD098059 – – + + Figure 7 MEK inhibitor prevents HOXA1-mediated disruption of acinar morphology and luminal filling. MCF-10A-vector and MCF-10A- HOXA1 cells were culturedin Matrigel for 16 days 710 mM PD098059. Shown are representative confocal cross-sections of MCF-10A-vector (a) andMCF-10A-HOXA1 ( b). 3D structures were immunostainedwith GM130 (red)and a6 Integrin (green). Representative phase contrast images of MCF-10A-vector (c) andMCF-10A-HOXA1 cells 710 mM PD098059 (d). The percentage of acinar structure with filledlumen was quantified( e). The results are presented as the mean of triplicate determinations; bars, s.d. *Po0.001.

a substrate of p44/42 MAP kinase (Yu et al., 2005). The response to particular growth factors (Gale et al., 1993). level of PCNA correlates directly with the rates of A recent report demonstrated successful use of a MEK cellular proliferation andDNA synthesis (Prosperi, enzyme inhibitor CI-1040 as a chemotherapeutic agent 1997). Previously, we have also reportedthat p44/42 to treat non-small-cell lung, breast, colon andpancreatic MAP kinase-mediated upregulation of catalase results cancer. PD0325901, a secondgeneration MEK inhibi- in inhibition of oxidative stress-induced apoptosis in tor, has recently enteredclinical development with better human mammary carcinoma cells. Thus, p44/42 MAP therapeutic potential for MEK inhibition in cancer kinase-mediated upregulation of catalase by HOXA1 (Rinehart et al., 2004). Therefore, the identiﬁcation of may offer similar protection from oxidative stress (Zhu HOXA1 regulation of components of the p44/42 MAP et al., 2005b). kinase pathway further substantiates the oncogenic The p44/42 MAP kinase pathway is a major pathway potential of HOXA1. In immortalizedhuman mammary involvedin cell proliferation andsurvival. Activation of epithelial cells, forcedexpression of HOXA1 resultedin the MAP kinase pathway has been demonstrated to be a a dramatic increase in anchorage-independent prolifera- frequent event in tumorigenesis (Mansour et al., 1994). tion, primarily by the promotion of cell survival- Analysis of human breast carcinomas demonstrates that mediated by the transcriptional upregulation of Bcl-2 25–50% of these tumors express elevatedlevels of (Zhang et al., 2003). We have demonstrated that phosphorylatedp44/42 MAP kinase (Salh et al., 1999). HOXA1-stimulatedupregulation of Bcl-2 is mediated Previous studies have also reported that activation of by p44/42 MAP kinase pathway. Previous studies have the p44/42 MAP kinase pathway intermediate MEK1, also demonstrated that Bcl-2 upregulation occurs regulatedby HOXA1 herein, is sufﬁcient to transform through the p44/42 MAP kinase pathway, resulting in several cell types. Although GRB2 does not have increasedcell survival (Wang et al., 2005). Distinct from transforming potential when overexpressed(Cheng transcriptional upregulation of Bcl-2, p44/42 MAP et al., 1998), early studies demonstrated that GRB2 kinase pathway also phosphorylates Bcl-2 at Ser-70. overexpression couldamplify activation of Ras or Increases in Bcl-2 phosphorylation results in cell p44/42 mitogen-activatedprotein kinase (MAPK) in survival by stabilization of Bcl-2:Bax interaction

Oncogene Mechanism of HOXA1 oncogenicity KM Mohankumar et al 4007 (Ruvolo et al., 2001). Activation of p44/42 MAP kinase showing a X1.5-folddecreasein cells with HOXA1 over- couldpotentially target other signalling molecules, expression, but a X1.5-foldincrease in HOXA1 depletedcells. including signal transducer and activator of transcription (STAT), notably STAT3 andSTAT5, which RT–PCR and primers cooperate to confer the complete oncogenic potency of Semiquantitative RT–PCRs were performedfor the following HOXA1 (manuscript in preparation; KM Mohankumar genes as described (Mertani et al., 2001). The sequences of the andPE Lobie, unpublishedobservations). Activated primers usedare listedas Supplementary data5. STAT3 andSTAT5 pathways have also been reported to increase Bcl-2 expression levels (Lord et al., 2000; Western blot analysis Cells were treatedas describedabove andWestern blot Hattori et al., 2001). Thus, increasedHOXA1-mediated analysis was performedas described(Mertani et al., 2001), Bcl-2 upregulation could be mediated directly through using the following antibodies: anti-b-actin, anti-GRB2, anti- the activation of p44/42 MAP kinase andindirectly via ERK-2 antibodies were obtained from Santa Cruz Biotech- p44/42 MAP kinase regulation of STAT3/5-mediated nology (Santa Cruz, CA, USA); anti-MEK1 antibody and transcription. phospho p44/42 MAP kinase were from Cell Signaling In conclusion, we have demonstrated that HOXA1 in Technology (Beverly, MA, USA). human mammary carcinoma cells results in increased p44/42 MAP kinase activity by virtue of increasing p44/42 MAP kinase assay GRB2 andMEK1 expression. This increasedp44/42 Cells were grown to 80% confluence in complete media for MAP kinase activity is requiredfor HOXA1-stimulated 24 h, andp44/42 MAP kinase assay was performedaccording cell proliferation, survival andoncogenicity. to the protocol as described (Zhu et al., 2005b). Luciferase reporter assay Cells were culturedto 40–60% confluence in six-well plates. Materials and methods Transient transfection with EphA2-r42B, Egr-3 andElk-1 luciferase reporter constructs were performedby using SAINT Cell culture MIX according to the manufacturer’s instructions. Reporter MCF-7, MCF-10A, HBL100, T47D, ZR75-1, HS578T, SkBr3, assay was performed according to the protocol described 21MT, 21PT, MDA-MB-231, MDA-MB-435 andMDA-MB- previously (Zhu et al., 2005b). 468 cell lines were culturedusing ATCC recommended conditions. Cell number and oncogenicity assays MCF-7 andMCF-10A cell lines were stably transfectedwith 5 Â 104 of MCF-7-vector andMCF-7-HOXA1 cells were either HOXA1 expression plasmid(pCMV Tag2B-HOXA1) or seeded into six-well plates in monolayers in complete media. the empty vector (pCMV Tag2B). Stable cell lines MCF-7- Cell number was determined as previously described 720 mM vector, MCF-7-HOXA1, MCF-10A-vector andMCF-10A- MEK1 inhibitor, PD098059 (Zhu et al., 2005a). HOXA1 were established using methods as described (Zhang 0.2 Â 106 of MCF-7-vector andMCF-7-HOXA1 cells were et al., 2003). seeded in six-well plates and allowed to grow in complete The design of siRNA vector targeting HOXA1 has been medium. Cells were treated 720 mM PD098059 in fresh described by Zhu et al. (2005a). The MCF-7 cell lines stably complete media for an additional 24 h. BrdU incorporation transfectedwith the HOXA1 siRNA construct (pSilencer- and detection was performed as previously described 720 mM HOXA1) or the empty vector (scrambledsiRNA), were named PD098059 (Zhu et al., 2005a). MCF-7-HOXA1 siRNA andMCF-7-scrambled-siRNA, Apoptotic cell death was determined as described previously respectively. 720 mM PD098059 (Mertani et al., 2001). Anchorage-independent growth assays, including suspen- Microarray analysis sion culture, were performedas previously described 720 mM Total RNA was isolatedfrom MCF-7-scrambled-siRNA, PD098059 (Zhang et al., 2003). MCF-7-HOXA1 siRNA, MCF-7-vector, MCF-7-HOXA1, MCF-10A-vector andMCF-10A-HOXA1 cells using TRIZOL Morphogenesis assays reagent. The 3D culture of MCF-10A cells in basement membrane was Microarray analyses were performedas describedpreviously performedas describedpreviously(Zhu et al., 2005a). For (Xu et al., 2005). HOXA1-upregulatedgenes were identifiedas studying MEK1 inhibitor, morphogenesis assays were per- those showing a X1.5-foldincrease in cells with HOXA1 formedwith 10 mM PD098059. Immunostaining andimage overexpression, but X1.5-fold decrease in HOXA1-depleted acquisition of acinar structures was performedas described cells, whereas downregulated genes were identified as genes (Zhu et al., 2005a).

References

Bajetto A, Barbero S, Bonavia R, Piccioli P, Pirani P, Florio T Chen H, Sukumar S. (2003). Role of homeobox et al. (2001). Stromal cell-derived factor-1alpha induces genes in normal mammary glanddevelopmentand astrocyte proliferation through the activation of extracel- breast tumorigenesis. J Mamm Gland Biol Neoplasia 8: lular signal-regulatedkinases 1/2 pathway. J Neurochem 77: 159–175. 1226–1236. Cheng AM, Saxton TM, Sakai R, Kulkarni S, Mbamalu G, Chariot A, Castronovo V. (1996). Detection of HOXA1 Vogel W et al. (1998). Mammalian Grb2 regulates multiple expression in human breast cancer. Biochem Biophys Res steps in embryonic development and malignant transforma- Commun 222: 292–297. tion. Cell 95: 793–803.

Oncogene Mechanism of HOXA1 oncogenicity KM Mohankumar et al 4008 Clarkson RW, Shang CA, Levitt LK, HowardT, Waters MJ. survival of primary breast cancer patients. Int J Cancer 89: (1999). Ternary complex factors Elk-1 andSap-1a mediate 384–388. growth hormone-induced transcription of egr-1 (early Prosperi E. (1997). Multiple roles of the proliferating cell growth response factor-1) in 3T3-F442A preadipocytes. nuclear antigen: DNA replication, repair andcell cycle Mol Endocrinol 13: 619–631. control. Prog Cell Cycle Res 3: 193–210. ConradPW, Freeman TL, Beitner-Johnson D, Millhorn DE. Rinehart J, Adjei AA, Lorusso PM, Waterhouse D, Hecht JR, (1999). EPAS1 trans-activation during hypoxia requires Natale RB et al. (2004). Multicenter phase II study of the p42/p44 MAPK. J Biol Chem 274: 33709–33713. oral MEK inhibitor, CI-1040, in patients with advanced Debnath J, Muthuswamy SK, Brugge JS. (2003). Morpho- non-small-cell lung, breast, colon, andpancreatic cancer. genesis andoncogenesis of MCF-10A mammary epithelial J Clin Oncol 22: 4456–4462. acini grown in three-dimensional basement membrane Ruvolo PP, Deng X, May WS. (2001). Phosphorylation of cultures. Methods 30: 256–268. Bcl2 andregulation of apoptosis. Leukemia 15: 515–522. Diehn M, Sherlock G, Binkley G, Jin H, Matese JC, Salh B, Marotta A, Matthewson C, Ahluwalia M, Flint J, Hernandez-Boussard T et al. (2003). SOURCE: a unified Owen D et al. (1999). Investigation of the Mek-MAP kinase- genomic resource of functional annotations, ontologies, and Rsk pathway in human breast cancer. Anticancer Res 19: gene expression data. Nucleic Acids Res 31: 219–223. 731–740. Gale NW, Kaplan S, Lowenstein EJ, Schlessinger J, Sebolt-LeopoldJS, DudleyDT, Herrera R, Van Becelaere K, Bar-Sagi D. (1993). Grb2 mediates the EGF-dependent WilandA, Gowan RC et al. (1999). Blockade of the MAP activation of guanine nucleotide exchange on Ras. Nature kinase pathway suppresses growth of colon tumors in vivo. 363: 88–92. Nat Med 5: 810–816. Garcia J, Ye Y, Arranz V, Letourneux C, Pezeron G, Seger R, Krebs EG. (1995). The MAPK signaling cascade. Porteu F. (2002). IEX-1: a new ERK substrate involvedin FASEB J 9: 726–735. both ERK survival activity andERK activation. EMBO J Shin SI, Freedman VH, Risser R, Pollack R. (1975). 21: 5151–5163. Tumorigenicity of virus-transformedcells in nudemice is Gille H, Kortenjann M, Thomae O, Moomaw C, Slaughter C, correlated specifically with anchorage independent growth Cobb MH et al. (1995). ERK phosphorylation potentiates in vitro. Proc Natl Acad Sci USA 72: 4435–4439. Elk-1-mediated ternary complex formation and transactiva- Wang CX, Song JH, Song DK, Yong VW, Shuaib A, Hao C. tion. EMBO J 14: 951–962. (2005). Cyclin-dependent kinase-5 prevents neuronal apop- Grill C, Gheyas F, Dayananth P, Jin W, Ding W, Qiu P et al. tosis through ERK-mediated upregulation of Bcl-2. Cell (2004). Analysis of the ERK1,2 transcriptome in mammary Death Differ 13: 1203–1212. epithelial cells. Biochem J 381: 635–644. Xu XQ, EmeraldBS, Goh EL, Kannan N, Miller LD, Hattori R, Maulik N, Otani H, Zhu L, Cordis G, Gluckman PD et al. (2005). Gene expression profiling to Engelman RM et al. (2001). Role of STAT3 in ischemic identify oncogenic determinants of autocrine human growth preconditioning. J Mol Cell Cardiol 33: 1929–1936. hormone in human mammary carcinoma. J Biol Chem 280: LordJD, McIntosh BC, Greenberg PD, Nelson BH. (2000). 23987–24003. The IL-2 receptor promotes lymphocyte proliferation and Yu FQ, Han CS, Yang W, Jin X, Hu ZY, Liu YX. (2005). induction of the c-myc, bcl-2, and bcl-x genes through Role of ERK1/2 in FSH induced PCNA expression the trans-activation domain of Stat5. J Immunol 164: andsteroidogenesisin granulosa cells. Front Biosci 10: 2533–2541. 896–904. Mansour SJ, Matten WT, Hermann AS, Candia JM, Rong S, Zhang X, Zhu T, Chen Y, Mertani HC, Lee KO, Lobie PE. Fukasawa K et al. (1994). Transformation of mammalian (2003). Human growth hormone-regulatedHOXA1 is a cells by constitutively active MAP kinase kinase. Science human mammary epithelial oncogene. J Biol Chem 278: 265: 966–970. 7580–7590. Mertani HC, Zhu T, Goh EL, Lee KO, Morel G, Lobie PE. Zhu T, Starling-EmeraldB, Zhang X, Lee KO, Gluckman PD, (2001). Autocrine human growth hormone (hGH) regula- Mertani HC et al. (2005a). Oncogenic transformation of tion of human mammary carcinoma cell gene expression. human mammary epithelial cells by autocrine human Identification of CHOP as a mediator of hGH-stimulated growth hormone. Cancer Res 65: 317–324. human mammary carcinoma cell survival. J Biol Chem 276: Zhu Z, Mukhina S, Zhu T, Mertani HC, Lee KO, Lobie PE. 21464–21475. (2005b). p44/42 MAP kinase-dependent regulation of Mueller H, Flury N, Eppenberger-Castori S, Kueng W, catalase by autocrine human growth hormone protects DavidF, Eppenberger U. (2000). Potential prognostic value human mammary carcinoma cells from oxidative stress- of mitogen-activatedprotein kinase activity for disease-free induced apoptosis. Oncogene 24: 3774–3785.

Supplementary Information accompanies the paper on the Oncogene website (http://www.nature.com/onc).

Oncogene