The epithelium-specific Ets transcription factor ... - The FASEB Journal

The epithelium-specific Ets transcription factor ESX is associated with mammary gland development and involution R. NEVE,* C.-H. CHANG,† G. K. SCOTT,† A. WONG,* R. R. FRIIS,‡ N. E. HYNES,* AND C. C. BENZ*,†,1 *Friedrich Miescher Institute, Basel CH-4058, Switzerland; †Department of Medicine, University of California at San Francisco, California 94143, USA; and ‡Department of Clinical Research, University of Bern, CH-3004, Bern, Switzerland

To study mammary gland expression of the epithelium-restricted Ets factor, ESX, mouse cDNA and genomic sequences were cloned and a Ç350 bp proximal promoter region with ú80% mouse–human homology was identified that mediates ESX induction by serum, heregulin (HRG), or epidermal growth factor (EGF). ESX mRNA expression progressively increases during embryonic mouse development from day 7 and is detectable in virgin mammary glands; it shows little if any change during pregnancy, then declines to barely detectable levels after 3 days of lactation. Similarly, cultured HC11 cells from midpregnant mouse mammary epithelium show an increase in ESX expression upon reaching lactogenic competency (in the presence of EGF or HRG), with a decline to barely detectable levels upon exposure to lactogenic hormones that induce milk protein (b-casein) expression. In contrast, involuting mouse and rat mammary glands show maximal ESX expression. High ESX levels are also seen in the involuting ventral prostate gland of rats. These findings, including the persistence of up-regulated ESX in fully regressed mammary glands, suggest that ESX expression can be induced by soluble growth factors and is maximally up-regulated in those partially committed epithelial cells destined to survive both the apoptotic and remodeling phases of glandular involution.—Neve, R., Chang, C.-H., Scott, G. K., Wong, A., Friis, R. R., Hynes, N. E., Benz, C. C. The epithelium-specific Ets transcription factor ESX is associated with mammary gland development and involution. FASEB J. 12, 1541–1550 (1998) ABSTRACT

which is transcriptionally up-regulated in a subset of early breast tumors and breast cancer cell lines and is thought to transactivate the Ets-responsive mammary gland oncogene, erbB2 (4, 5). Subsequent to this report, four groups have published on the potential biological and developmental importance of this epithelium-specific Ets factor (variably named ESE-1, Elf-3, Jen, or ERT; now identified as ESX in HUGO/GDB:6837498) in nonmammary epithelial systems, where ESX is thought to transactivate such genes as the transforming growth factor b type II receptor (TGF-bRII), endo-A/keratin-8, and several markers of epidermal cell differentiation including transglutaminase 3, SPRR2A, and profilaggrin (6–9). Although its expression profile suggests that ESX is associated with the development of both simple and stratified epithelium (10), detailed studies have been performed only in the latter and have shown that ESX is unique among transcription factors generally, and Ets factors specifically, for its restricted expression in the most terminally differentiated of epidermal cells (6, 8). A limited in situ analysis of normal human mammary tissue demonstrated low but heterogeneous levels of ESX transcript expression restricted to the polarized simple epithelium of ductules and terminal ductal-lobular units (5). To evaluate ESX expression during all differentiation stages of mammary epithelium, mouse ESX cDNA and genomic sequences were cloned and compared to their human counterparts, then used to study postnatal rodent models of mammary gland development. The inductive influences controlling ESX expression were explored by transient transfection of an ESX promoter-reporter con-

Key Words: Ets family · ESX promoter · ESX expression and mammary gland differentiation 1

Ets TRANSCRIPTION FACTORS regulate stage- and tissuespecific gene programs in fetal development and are overexpressed or rearranged in a variety of vertebrate and human malignancies (reviewed in refs 1–3). We recently cloned and characterized a novel 42 kDa Ets factor, ESX (epithelial-restricted with serine box),2

Correspondence: Division of Hematology-Oncology, Box 1270, 3rd and Parnassus, University of California, San Francisco, CA 94143–1270, USA. E-mail: [email protected] 2 Abbreviations: EGF, epidermal growth factor; ESX, epithelium-specific Ets transcription factor (epithelial-restricted with serine box); HRG, heregulin; luc, luciferase; SFM, serum-free medium; TGF-bRII, transforming growth factor b type II receptor; UTR, untranslated region.

0892-6638/98/0012-1541/$02.25 Q FASEB

/ 383a no13 Mp 1541 Monday Oct 19 04:09 PM LP–FASEB no13

1541

struct into a breast cancer cell line (SKBr3) responsive to heregulin-b (HRG) and epidermal growth factor (EGF). RNA samples probed for ESX expression were derived from different stages of cultured HC11 mouse mammary epithelial cells, first made competent for lactogenesis and then hormonally induced to synthesize b-casein (11, 12). These results were compared to ESX Northern blots of virgin, pregnant, lactating, and involuting mouse mammary glands. Last, mouse and rat mammary glands collected during involution were also compared to the involuting ventral prostate gland of rats to demonstrate that maximal induction of ESX occurs during this stage of glandular regression, suggesting an association with epithelial apoptosis.

MATERIALS AND METHODS Comparison of murine and human ESX genomic and cDNA sequences A lFIXII 129SV mouse genomic library (Stratagene, San Diego, Calif.) was screened using a 5* cDNA probe from hESX (5; Genbank accession number U66894) to isolate a clone from which a 7.75 kb BamHI fragment was subcloned into pBluescript SK (Stratagene). Upon full sequencing, this genomic clone was found to contain 3.6 kb of sequence upstream from the ATG-start codon (beginning exon 2), about 2.9 kb upstream of the transcriptional start site. The deduced organization of 9 exons (8 coding) and 8 introns spanning 4.9 kb of genomic sequence was subsequently found to be similar to that reported by Tymms et al. (7). This mESX genomic sequence was compared to a previously isolated and fully sequenced 1.8 kb BglII-BglII human genomic clone containing 1.5 kb of hESX promoter sequence upstream of exon 1 and the 5* half of intron 1. A 1.1 kb Bluescript subclone encoding the entire mESX cDNA was derived from 129SV mouse ES cell total RNA by reverse transcriptase-polymerase chain reaction using specific primers extending 5* and 3* from the respective ATG-start and TAA-stop codons in the genomic sequence, and the entire cDNA subclone was sequenced. All sequencing was performed on an ABI Prism Automated DNA Sequencer (model 377) using 3*-dye labeled ddNTP terminators. Computer alignments of genomic and cDNA mESX and hESX sequences were performed, and a comparison of genomic and cDNA mESX sequences was used to determine exon and intron boundaries. Conserved murine and human promoter elements as well as putative amino acid domain homologies were identified from PIR protein, SWISS-PROT, and PROSITE databases by GCG computer search (Genetics Computer Group, Wisconsin Package 3.0, Madison, Wis.). Growth factors and tissue culture conditions Recombinant human EGF was commercially obtained (Sigma, St. Louis, Mo.). Recombinant human HRG isoforms were kindly provided (Amgen; b1 isoform 177–228) or commercially obtained (NeoMarkers; full-length b1 isoform), with no significant difference in activity detected between the truncated and full-length b1 isoforms. SKBr3, MCF-7, and MDA435 breast cancer cell lines (5, 13) and NIH3T3 mouse fibroblasts were all maintained in Dulbecco’s modified Eagle’s medium (Life Technologies, Inc., Paisley, U.K.) supplemented with 10% fetal calf serum. HC11 cells, derived from midpregnant BALB/c mouse mammary gland tissue, were 1542

Vol. 12

November 1998

maintained in culture using a growth medium consisting of RPMI-1640, 10% heat-inactivated fetal calf serum, 5 mg/ml bovine insulin, and either 2 nM HRG or 2 nM EGF (11, 12). HC11 cells were induced into lactogenic competency by culturing them in growth media and maintaining them at confluence for 3 days (12). These competent cultures were then induced to terminally differentiate and produce b-casein by incubation for 1–6 days in DIP induction medium (RPMI1640, 5 mg/ml ovine prolactin, 5 mg/ml insulin, and 1 mM dexamethasone). Northern blot analysis of cell and tissue RNA samples RNA samples included commercial blots of polyA-RNA from 7- to 17-day-old mouse embryos (Clontech, Palo Alto, Calif.) and total RNA extracted from HC11 cell cultures, excised mouse (BALB/c) and rat (Sprague Dawley) inguinal mammary glands (virgin, pregnant, lactating, and involuting), and excised rat ventral prostate glands (pre- or postcastration). Extractions of total RNA were performed on snap-frozen (liquid nitrogen) cell pellets or excised glands (14), using either the guanidinium isothiocyanate or Trizol (Gibco BRL, Paisley, U.K.) method. When indicated, polyA-enriched RNA from mammary or prostate tissue was prepared using oligo dT-cellulose (Boehringer-Mannheim, Mannheim, Germany). Either 10 mg of total RNA or 5 mg poly(A)-enriched RNA/sample was electrophoresed into 1% agarose gels and transferred onto either nylon (Zeta-probe, BioRad, Richmond, Calif.) or nitrocellulose filters, which were then UV cross-linked using a Stratalinker 1800 (Stratagene). After ethidium bromide or acridine orange staining to quantitate transfer of 18S and 28S RNA, filters were hybridized with a randomly primed and [32aP]dATP-labeled 300 bp cDNA fragment from the aminoterminal mESX coding region and given final washes at 657C in 0.21 SSC prior to autoradiography. ESX promoter activation in transient transfection assay Luciferase (luc) reporter constructs (in pGL2-Basic Vector; Promega, Madison, Wis.) containing either 0.4 kb (0349 bp to /61 bp) of mESX proximal promoter having ú80% homology to hESX (mESX-luc) or 1.1 kb (0349 bp to /704 bp) of proximal promoter with an additional 5* untranslated sequence up to the ATG initiation codon (mESXL-luc) were constructed by polymerase chain reaction amplification from the murine genomic clone. Transient transfection of mESX-luc reporter (1 mg DNA) in 6 ml of lipofectamine (Gibco BRL) in serum-free medium (SFM) was performed in replicate tissue culture wells containing 60% confluent (Ç11105) cells. After 5 h, the lipofectamine-containing media was replaced with SFM; 12 h later, cell cultures were induced with 10% serumcontaining media { growth factor (HRG or EGF) at the concentrations indicated. The transiently transfected cell cultures were then harvested at 0– 24 h after serum { growth factor induction, extracts were prepared, and luciferase activity was measured as recommended by the vendor (Promega).

RESULTS We isolated a 7.8 kb mESX genomic clone containing 4.9 kb of sequence specifying 9 exons (exons 2–9, coding) and 8 introns, consistent with the recently described genomic structure of mESX (7). Alignment of this genomic sequence with that determined from the 1.1 kb cDNA clone allowed us to compare the primary

The FASEB Journal


NEVE ET AL.

structures of murine and human ESX as shown in Fig. 1. This comparison of the primary structures reveals 87% amino acid identity and maps the 7 exon boundaries within the encoded 371 amino acids of ESX. Comparing the 2.9 kb of mESX promoter-containing sequence with that of a formerly cloned 1.5 kb

hESX promoter-containing genomic fragment (5), and aligning both with reference to exon 1 and a previously determined hESX 5* untranslated region (UTR) sequence (6), showed °50% bp homology between the most upstream genomic sequences (01500 bp to 0350 bp). In contrast, the proximal

Figure 1. Comparison of exon-encoded mouse (m) and human (h) ESX amino acid sequences. The 371 amino acid sequences encoded by genomic exons 2–9 were determined by comparing cloned mouse and human cDNA sequences with the 7 exon boundaries mapped (arrows) after comparison between mouse and human genomic sequences, as described in Materials and Methods. Amino acid identities (vertical lines) and similarities (single and double dots) are as indicated. ESX AND MAMMARY GLAND DEVELOPMENT


1543

ESX promoter regions (0350 bp to /50 bp) showed 83% homology at the nucleotide level between mouse and human genes, which is demonstrated in Fig. 2. The notable features in this proximal promoter region include conservation of six different

consensus response elements (Ets, AP-2, SP1/GC box, USF, Oct, NF-kB), a CCAAT box at 075 bp, and a putative pyrimidine-rich initiator element (Inr) capable of specifying transcript initiation from the TATA-less murine promoter.

Figure 2. Nucleotide sequences and consensus response elements conserved between mouse and human ESX promoters. Comparison of aligned mouse and human genomic sequences revealed 83% nucleotide identity (vertical lines) between the 0.4 kb of upstream sequences shown above. Sequence numbering is relative to a putative transcriptional start site (/1) within a conserved pyrimidine-rich type Inr (box with arrow) located Ç75 bp downstream from a conserved CCAAT sequence (box); this putative site was also identified as the most 5*-terminal nucleotide from a hESX cDNA clone and agrees with a previously determined hESX transcriptional start site (6). The locations of conserved consensus response elements for Ets, AP-2, SP1/GC box, USF, Oct, and NF-kB are indicated by horizontal bars; of the pair of Ets elements, the 5* element (GGAA) appears displaced by three nucleotides while the 3* element (TTCC) is positionally conserved in both promoters. As described in Materials and Methods, this 0.4 kb of mESX promoter sequence was cloned into pGL2-Basic to produce the mESX-luc reporter construct used in Fig. 3. 1544

Vol. 12

November 1998

The FASEB Journal


NEVE ET AL.

To verify that this homologous region of the proximal promoter can confer growth factor-induced transcriptional up-regulation of mESX, as is known to occur in hESX-overexpressing breast cancer cells (5), the activities of two different mESX promoter-reporter constructs (0.4 kb mESX-luc and 1.1 kb mESXL-luc) were assessed by transient transfection into cultured cells expressing negligible (NIH3T3), low (MCF-7, MDA-435), or high (SKBr3) levels of endogenous ESX (5). Since no significant differences in promoter activity were observed between the 1.1 kb mESXL-luc and the 0.4 kb mESX-luc constructs, the smaller mESX-luc construct was used for all consequent experiments. The negligible and low ESX-expressing cell lines consistently showed minimal reporter activity unresponsive to culture stimulation by serum { growth factors (NIH3T3, MDA-453) or estradiol (MCF-7). In contrast, mouse NIH3T3 cells engineered to overexpress human ErbB receptor pairs (ErbB2/ErbB1ÅNE2/1 cells; ErbB2/ErbB3ÅNE2/3 cells; ErbB2/ErbB4ÅNE3/4 cells) and with intracellular signaling upon appropriate ErbB ligand stimulation (unpublished results) showed ligand-inducible increases in mESX promoter activity in the presence of serum. NE2/1 cells produced mESXluc reporter up-regulation in response to EGF, whereas NE2/3 and NE2/4 cells responded similarly to HRG (data not shown), demonstrating the functionality of this ectopic mESX promoter within mouse cells activated by human ErbB receptors. SKBr3 cells, which overexpress ErbB2 and also moderately express ErbB1 and ErbB3 receptors, were used to study mESX promoter induction since they were known to produce an immediate increase in endogenous ESX transcripts after culture exposure to HRG (5). As shown in Fig. 3, treatment of SKBr3 for various intervals (0–24 h) produced serum and growth factor (HRG, EGF)-inducible increases in mESX-luc reporter activity. Serum supplementation alone produced a three- to fourfold maximal induction of promoter activity, which peaked within 8 h of treatment and then declined to near serum-free basal promoter activity by 24 h (Fig. 3A). When these SKBr3 cells were treated with 1 nM HRG in addition to serum supplementation, a sevenfold peak induction over basal promoter activity was observed at 8 h and promoter activity was still elevated nearly fourfold over basal levels 24 h after treatment (Fig. 3A). HRG concentrations from 0.1 to 2 nM produced comparable enhancements in mESX promoter activity after 8 h (data not shown), ranging from two- to threefold over the peak activity produced by serum alone (Fig. 3). At this same time point, EGF concentrations of up to 4 nM (in serum-supplemented media) also enhanced mESX promoter activity to a similar (although slightly lesser) degree as HRG (Fig. 3B). In the absence of serum, neither HRG nor EGF produced any significant mESX promoter induction. Insulin (5 mg/ml{serum-supplementation) had no significant effect on mESX promoter activity in SKBr3 cells (Fig. 3B).

Epithelial-specific ESX mRNA expression has been shown for various mouse tissues after fetal day 17, but not during earlier embryonic development or adult mouse mammary gland differentiation (7). As shown in the Fig. 4 Northern blot, mouse embryos exhibit progressive induction of a 2.2 kb ESX transcript after fetal day 7, with 17-day transcript levels about 10-fold higher than those of 11-day-old embryos, which is consistent with the earliest onset of epithelial differentiation and progressive fetal growth of epithelial organs and tissues. Before day 17, embryos show no detectable evidence of the alternatively spliced larger ESX transcript (3.8–4.1 kb) noted in later stage fetal and adult organs and malignant tissues (5–8). Postembryonic mammary gland expression of ESX was evaluated in three separate experiments where RNA was isolated from mouse glands taken at various stages of differentiation including that of virgin, pregnant, lactating, and involuting mammary glands. In general, a basal level of ESX expression was seen in virgin and first-pregnancy glands, which declined to undetectable levels after 2–3 days of lactation and increased to maximal levels after weaning and involution. Figure 5 shows a representative stage-specific Northern blot profile of ESX expression. RNA samples from 8- to 12-day involuting glands revealed persistently high ESX expression. These later time points are beyond the active phases of mammary gland involution and after most of the molecular and histologic correlates of apoptosis and tissue remodeling have already peaked (14– 16). A fully regressed mouse mammary gland resected 8 wk after weaning also showed maximal ESX expression comparable to peak transcript levels observed within the first 12 days of involution (Fig. 5; lane R vs. involuting day 10 lane). A panel of rat mammary gland RNA samples was also probed and confirmed this stage-specific profile of ESX expression (Fig. 6). Rat mammary ESX expression was basal during pregnancy, undetectable during lactation, and showed reinduction to maximal levels within 3 days of weaning and involution. Given that normal prostate expresses ESX (5) and regressing prostatic tissue shows morphological and biochemical features similar to involuting mammary tissue (14), we looked for changes in ESX expression during castration-induced involution of the adult rat ventral prostate. As with the rodent mammary tissue, rat prostate expression of ESX appears highest during glandular involution (Fig. 6). Detailed studies in stratified epithelium have shown that ESX expression is restricted to the most terminally differentiated epidermal keratinocytes (6, 8). Since ESX transcripts decline to undetectable levels during lactation, when the mammary gland is composed of fully differentiated secretory epithelium, we tried to simulate this in vivo observation using cultured HC11 cells, which can be induced into lactogenic competency on exposure to growth factors and then hormonally

ESX AND MAMMARY GLAND DEVELOPMENT


1545

Figure 3. Serum and growth factor (HRG, EGF) induced ESX promoter activity in SKBr3 cells. A) SKBr3 cells, known to express high endogenous levels of ESX mRNA, were transiently transfected in serum-free replicate cultures with a mESX-luc reporter (containing the 0.4 kb promoter sequence shown in Fig. 2) and subsequently treated with serum-containing media { HRG (1 nM) for the indicated intervals (0–24 h), as described in Materials and Methods. Mean ({SEM) values of reporter (luciferase) activity normalized for culture protein content are expressed as fold induction over that measured in transfected and untreated control cultures (left in serum-free media for 0 or 24 h). Maximum induction of promoter activity by serum { HRG is observed in the 8 h treatment conditions. B) mESX-luc transfected SKBr3 cells were treated in serum-containing media with either HRG (0.25 nM), EGF (4 nM) or Insulin (5 mg/ml) for the optimal 8 h interval; mean ({SEM) reporter activity was measured and expressed as fold induction over that of transfected and untreated control cultures (left in serum-free media for 8 h).

stimulated to differentiate and produce the milk protein, b-casein (11, 12). Proliferating HC11 cells express basal levels of ESX (Fig. 7; lane G) until they reach confluence and a state of lactogenic competence (2–3 days after culture confluence), when ESX expression increases dramatically (Fig. 7; lanes C vs. C2 and C3). Upon growth factor (HRG or EGF) withdrawal and administration of lactogenic hormones (DIP induction medium), these competent and terminally differenti1546

Vol. 12

November 1998

ating cells express increasing amounts of b-casein while ESX transcript levels fall concurrently to basal levels (Fig. 7; lanes C3 vs. D1, D3, and D6).

DISCUSSION The ú30 known metazoan members of the Ets family of transcription factors are recognized for their roles

The FASEB Journal


NEVE ET AL.

Figure 4. Progressive induction of ESX expression in developing mouse embryos. A commercially obtained membrane containing polyA-RNA samples (5 mg/lane) from whole mouse embryos of the indicated age was probed with a 32Plabeled cDNA fragment specific for the amino-terminal mESX coding region. A single 2.2 kb ESX transcript band was detectable in the 7-day embryo sample and increased progressively in intensity in the older embryo samples. Alternative ESX transcripts of larger (3.8–4.1 kb) or smaller (õ2 kb) size, noted in some adult organs and malignant cell lines (5–7), were not detectable on these embryo blots.

in embryonic development and tissue maturation, where they direct stage-specific and tissue-restricted programs of gene expression targeted by a highly conserved Ç85 amino acid Ets DNA binding domain (1–3). As a new member of this family, the 371 amino acid ESX transactivator possesses a typical Ets DNA binding domain located in its carboxyl-terminal region (5–8). Outside this conserved DNA binding domain, ESX contains several other structural motifs not found in other Ets proteins, a situation thought to result from ú500 million years of evolutionary recombination and exon shuffling (5–8). The present

Figure 5. ESX expression at various stages of mouse mammary gland development and differentiation. Representative Northern blot of total RNA (Ç10 mg/lane) isolated from adult virgin (V), first-pregnant, lactating, involuting, and fully regressed (8 wk postinvolution, R) mouse mammary glands. Numbers over the lanes signify days since onset of pregnancy, lactation, or involution (weaning). Only the 2.2 kb ESX transcript band was detected in all samples, with lowest levels present in ú3 day lactating samples and highest ESX transcript levels present in ú3 day involuting and fully regressed samples (upper panel). The ethidium bromide-stained 18S ribosomal RNA band from the same blot (lower panel) is shown as a control for total RNA lane loading and membrane transfer.

Figure 6. Increased ESX expression during involution of rat mammary and ventral prostate glands. Northern blot detection of 2.2 kb ESX transcripts in polyA-RNA samples (5 mg/ lane) isolated from pregnant (P), lactating (L), and involuting (I: days 2, 3, 4, and 6 postweaning) rat mammary glands, as well as from normal adult rat prostate (N) and involuting (I: days 2, 4, and 6 postcastration) glands. These rat polyA-RNA samples demonstrate variable amounts of larger (3.8–4.1 kb) and smaller (õ2 kb) size ESX transcripts, as noted in some adult human tissues and malignant cell lines (5–7). The acridine orange staining of residual 18S RNA present on the

findings support a domain-based modular structure for ESX, as shown by the overall high degree of amino acid sequence homology (87% identity) between mouse and human ESX and by the fact that all

Figure 7. ESX expression in cultured HC11 mammary epithelial cells induced into lactogenic competency and terminal differentiation. As described in Materials and Methods, HC11 cells were induced into lactogenic competency by culturing them in growth media and maintaining them at confluence for 3 days; these competent cultures were then hormonally induced into terminal differentiation and production of bcasein by incubation for 6 days in dexamethasone, insulin, and prolactin (DIP). Northern blotting detected only the 2.2 kb ESX transcript band (upper panel) in total RNA (Ç10 mg/ lane) isolated from cultures of subconfluent growing (G), near confluent (C), 2-day confluent (C2), and 3-day confluent (C3) cells, as well as competent cells hormonally induced with DIP for 1 (D1), 3 (D3), and 6 (D6) days. b-Casein transcript levels (lower panel) were detectable only in the hormonally induced cells, with increasing milk protein expression inversely correlated with ESX transcript levels. The ethidium bromide-stained 18S ribosomal RNA band from the same blot (middle panel) is shown as a control for total RNA lane loading and membrane transfer.



1547

putative domains within ESX are encoded by only one or two exons (Fig. 1). In addition to the exon 8and 9-encoded Ets DNA binding domain, these other structural modules include the exon 3-encoded Pointed (A-region) domain, the exon 5- and 6-encoded amphipathic helix and serine-rich box, and the exon 7-encoded bipartite nuclear targeting sequence (5). Similar patterns of epithelial-specific ESX mRNA expression have been noted in human and mouse tissues (7), suggesting common mechanisms of transcriptional control and promoter regulation in both murine and human genes. Results from this study indicate that mouse and human ESX promoters share a high degree of homology (83% nucleotide identity) over a relatively short region extending Ç0.4 kb upstream from the putative transcriptional start site (/1) and just beyond a conserved pair of Ets binding sites adjacent to an AP-2 consensus response element (Fig. 2). The mESX promoter lacks the TATA box sequence present in hESX (041 bp). However, both promoters have a typical CCAAT box located Ç75 bp upstream of a conserved pyrimidine-rich type Inr, making it likely that both mESX and hESX function as TATA-less promoters. No significant differences in promoter activity were observed between the 1.1 kb mESXL-luc and the 0.4 kb mESX-luc constructs, suggesting that the 0.7 kb of 5* UTR between the Inr and ATG initiation codon (beginning exon 2) does not contain strong promoter regulatory elements. In addition to conserved Ets and AP-2 response elements, both murine and human ESX proximal promoters share consensus elements for SP1/GC, USF, Oct, and NF-kB. Any combination of these response elements could account for the development- and tissue-specific profile of ESX expression common to both mouse and human tissues (7). These same response elements likely contribute to the differential up-regulation of ESX promoter activity observed between high (SKBr3) and low (MCF-7, MDA-435) ESX-expressing cell lines and in SKBr3 and ErbB receptor overexpressing NIH3T3 cells (NE2/1, NE2/3, NE2/4) upon exposure to serum and growth factors (HRG, EGF). HRG, in particular, appears to synergistically enhance ESX proximal promoter activity two- to threefold over the primary threefold stimulatory effect of serumsupplementation alone, consistent with our previous report of HRG-induced up-regulation of ESX mRNA in cultured SKBr3 cells (5). Additional studies are under way to determine the mechanisms and response elements mediating serum, HRG, and EGF induction of ESX promoter activity in these cells. The dramatic changes in ESX mRNA levels observed during normal mammary epithelial differentiation in vitro and in vivo may also be mediated by these same growth factor-responsive promoter elements. 1548

Vol. 12

November 1998

Mammary epithelium requires not only membrane-activated ErbB receptor family members for normal ductal development (17), but also the ErbB receptor ligands EGF and HRG, which are potent in vivo stimulators of mammary epithelial proliferation and differentiation (18–20). The in vivo situation can be simulated in vitro using HC11 cell cultures, in which both HRG and EGF are mitogenic and either one can be used to promote HC11 lactogenic competency, a state of commitment essential for subsequent hormonal induction of terminal differentiation and milk expression (11, 12). The mechanisms associated with lactogenic competency are incompletely understood, but are partially mediated by responses to increased cell–cell interactions and to a reorganized extracellular matrix (21, 22). Our present study demonstrates that in HC11 cells, growth factor-promoted lactogenic competency is associated with a dramatic up-regulation in ESX expression (Fig. 7). The changes in ESX expression associated with in vitro induction of HC11 terminal differentiation mimicked some but not all the features of ESX transcript profiles observed during in vivo mammary gland development. Pregnancy represents a developmental stage in which epithelial cell proliferation and increasing commitment to terminal differentiation occur. Unlike the ESX up-regulation observed when proliferating HC11 cells become lactogenically competent, glands from sexually mature virgin and first-pregnancy mice showed no significant variation in their level of ESX expression. However, with in vivo terminal differentiation of mouse and rat mammary epithelium into milk-producing lobuloalveolar units, there was a marked decline in ESX expression (Figs. 5 and 6) consistent with the fall in ESX transcript levels observed with hormonal induction of b-casein expression in competent HC11 cells (Fig. 7). This dramatic decline in ESX expression upon terminal differentiation of mammary epithelial cells in vitro and in vivo is in unique contrast to stratified epithelial systems where ESX expression is up-regulated and restricted to the most terminally differentiated forms of epidermal keratinocytes (6, 8). ESX may now be added to a small but growing list of epithelial genes known to be repressed during lactogenesis and then dramatically up-regulated with weaning and initiation of mammary gland involution (14–16, 22–24). Increasing ESX transcript levels are evident in the involuting mammary glands of both mouse and rat beginning as early as 1–2 days after weaning. In the rat gland, this induction reaches peak levels within 4 days, whereas in the mouse gland expression is maximal by 8 days and remains high for at least 8 wk, a point when apoptosis and remodeling are completed and the gland is fully regressed. The persistence of high ESX transcript levels in fully regressed mammary glands suggests that the involu-

The FASEB Journal


NEVE ET AL.

tional induction of ESX is occurring in newly committed epithelial cells that are destined to survive both apoptotic and remodeling phases of involution. Future in situ analysis will address the possibility that ESX up-regulation occurs in a subpopulation of partially committed and pluripotential ductal epithelium poised to regenerate a fully differentiated milk-producing gland with the next cycle of pregnancy and lactation. Molecular markers that potentially distinguish virgin mammary epithelium from partially or terminally differentiated ductal-lobular elements are of biological and medical interest, as they might ultimately serve to identify women whose breast tissue is more or less vulnerable to malignant transformation (25, 26). Like the mammary gland, prostatic tissue is subject to involutional changes and its epithelium regresses in a reversible manner after surgical castration or pharmacologically induced androgen ablation. Although the regressing ventral prostate shows morphological and biochemical features of epithelial apoptosis analogous to those of involuting mammary gland, unlike the latter it shows little evidence of tissue remodeling with slight induction of ECM proteinases of the matrix metalloproteinase and serine protease families (14). In contrast, the transition from lactating to involuting mammary gland is well characterized by two distinct phases of apoptosis: an early proteinase-independent phase and a prominent proteinase-dependent later stage (24). In the initial phase (days 1–3 after weaning), the gland’s alveoli and supporting mesenchyme remain largely intact, but chromatin cleavage and DNA laddering become detectable along with induction of the same apoptosis-associated genes upregulated during prostatic involution (14–16, 22– 24, 27). During the second stage (days 3–10 after weaning), massive apoptotic cell loss (Ç50% of the gland’s cellularity and ¢95% of all alveolar epithelium) results in collapse and dissolution of all milkproducing glands, necessitating a much more extensive protease-mediated ECM remodeling process than that required by the involuting prostate. Despite these differences between involuting breast and prostate glands, ESX transcript levels increased in both in a similar manner (Fig. 6). Maximal upregulation of prostatic ESX occurred within 2–4 days of hormonal ablation, concurrent with increases in other apoptosis-associated prostatic transcripts (e.g., sulfated glycoprotein-2, tissue transglutaminase, p53, DDC-4, TGF-b1, TGF-bRII) previously demonstrated in these same RNA samples (14) or by other groups (27). Although a number of ECM proteases are known to be transcriptionally regulated by Ets factors (28, 29), the early and comparable extent of ESX up-regulation observed during involution of prostate and mammary glands and the persistence of up-regulated

ESX in fully regressed mammary glands suggest that this Ets family transactivator may be regulating other genes in addition to proteases in cells destined to survive both the apoptotic and remodeling phases of glandular involution. Transglutaminase 3 and TGFbRII are two of the few genes identified to date as being transcriptionally up-regulated by ESX (6, 8, 9). The former is closely related to tissue transglutaminase, which, along with TGF-bRII, is up-regulated in concert with ESX during involution. Thus, our findings should not only stimulate the search for ESXregulated genes associated with involution and apoptosis, but also provide greater incentive to identify ECM and growth factor-sensitive response elements within the ESX promoter that account for its transcriptional up-regulation during prostate and mammary gland involution. We thank Zena Werb and Jane Wiesen (UCSF) for their advice and assistance. This work was supported in part by NIHsponsored grants (P01-CA44768, R01-CA36773) as well as the Hazel P. Munroe and Janet Landfear memorial funds. C.C.B., as Visiting Scientist and Professor at the Friedrich Miescher Institute and University of Basel, thanks N. Hynes, M. Burger, R. Herrmann, C. Rochlitz, and E. Rosenbaum for their assistance and support this year. R.N. and C.C.B. were recipients of Krebsliga beider Basel research awards.

REFERENCES 1.

2. 3. 4.

5.

6.

7.

8.

9.



Wasylyk, B., and Nordheim, A. (1997) Ets transcription factors: partners in the integration of signal responses. In Transcription Factors in Eukaryotes (Papavassiliou, A. G., ed) pp. 253–286, Springer-Verlag GmbH & Co. KG., Heidelberg, Germany Hromas, R., and Klemsz, M. (1994) The ETS oncogene family in development, proliferation and neoplasia. Int. J. Hematol. 59, 257–265 Macleod, K., Leprince, D., and Stehelin, D. (1992) The ets gene family. Trends Biochem. Sci. 17, 251–256 Scott, G. K., Daniel, J. C., Xiong, X., Maki, R. A., Kabat, D., and Benz, C. C. (1994) Binding of an Ets-related protein within the DNase-I hypersensitivity site of the HER2/neu promoter in human breast cancer cells. J. Biol. Chem. 269, 19848–19858 Chang, C.-H., Scott, G. K., Kuo, W.-L., Xiong, X., Suzdaltseva, Y., Park, J. W., Sayre, P., Erny, K., Collins, C., Gray, J. W., and Benz, C. C.. (1997) ESX: a structurally unique Ets overexpressed early during human breast tumorigenesis. Oncogene 14, 1617–1622 Oettgen, P., Alani, R. M., Barcinski, M. A., Brown, L., Akbarali, Y., Boltax, J., Kunsch, C., Munger, K., and Libermann, T. A. (1997) Isolation and characterization of a novel epithelium-specific transcription factor, ESE-1, a member of the ets family. Mol. Cell. Biol. 17, 4419–4433 Tymms, M. J., Ng, A. Y. N., Thomas, R. S., Schutte, B. C., Zhou, J., Eyre, H. J., Sutherland, G. R., Seth, A., Rosenberg, M., Papas, T., Debouck, C., and Kola, I. (1997) A novel epithelial-expressed ETS gene, ELF3: human and murine cDNA sequences, murine genomic organization, human mapping to 1q32.2 and expression in tissues and cancer. Oncogene 15, 2449–2462 Andreoli, J. M., Jang, S.-I., Chung, E., Coticchia, C. M., Steinert, P. M., and Markova, N. G. (1997) The expression of a novel, epithelium-specific ets transcription factor is restricted to the most differentiated layers in the epidermis. Nucleic Acids Res. 25, 4287–4295 Choi, S.-G., Youngsuk, Y., Kim, Y.-S., Kato, M., Chang, J., Chung, H.-W., Hahm, K.-B., Yang, H.-K., Rhee, H. H., Bang, Y.-J., and Kim, S.-J. (1998) A novel ets-related transcription factor, ERT/ ESX/ESE-1, regulates expression of the transforming growth factor-b type II receptor. J. Biol. Chem. 273, 110–117

1549

10. 11.

12.

13.

14.

15. 16.

17.

18. 19.

1550

Davies, J. A., and Garrod, D. R. (1997) Molecular aspects of the epithelial phenotype. BioEssays 19, 699–704 Ball, R. K., Friis, R. R., Schoenenberger, C. A., Doppler, W., and Groner, B. (1988) Prolactin regulation of b-casein gene expression and a cytosolic 120-kd protein in a cloned mouse mammary epithelial cell line. EMBO J. 7, 2089–2095 Marte, B. M., Jeschke, M., Graus-Porta, D., Taverna, D., Hofer, P., Groner, B., Yarden, Y., and Hynes, N. E. (1995) Neu differentiation factor/heregulin modulates growth and differentiation of HC11 mammary epithelial cells. Mol. Endocrinol. 9, 14– 23 Daly, J. M., Jannot, C. B., Beerli, R. R., Graus-Porta, D., Maurer, F. G., and Hynes, N. E. (1997) Neu differentiation factor induces ErbB2 down-regulation and apoptosis of ErbB2-overexpressing breast tumor cells. Cancer Res. 57, 3804–3811 Bielke, W., Ke, G., Feng, Z., Buhrer, S., Saurer, S., and Friis, R. R. (1997) Apoptosis in the rat mammary gland and ventral prostate: detection of cell death-associated genes using a coincident expression cloning approach. Cell Death Differ. 4, 114– 124 Strange, R., Li, F., Saurer, S., Burkhardt, A., and Friis, R. R. (1992) Apoptotic cell death and tissue remodelling during mouse mammary gland involution. Development 115, 49–58 Wilde, C. J., Quarrie, L. H., Tonner, E., and Flint, D. J. (1997) Mammary apoptosis: physiological regulation and molecular determinants. In Biological Signalling and the Mammary Gland (Wilde, C. J., Peaker, M., and Taylor, E., eds) pp 103–114, Hannah Research Institute, Ayr, Scotland Xie, W., Paterson, A. J., Chin, E., Nabell, L. M., and Kudlow, J. E. (1997) Targeted expression of a dominant negative epidermal growth factor receptor in the mammary gland of transgenic mice inhibits pubertal mammary gland development. Mol. Endocrinol. 11, 1766–1781 Vonderhaar, B. K. (1987) Local effects of EGF, a-TGF, and EGFlike growth factors on lobuloalveolar development of the mouse mammary gland in vivo. J. Cell. Physiol. 132, 581–584 Coleman, S., Silberstein, G. B., and Daniel, C. W. (1988) Ductal morphogenesis in the mouse mammary gland: evidence supporting a role for epidermal growth factor. Dev. Biol. 127, 304– 315

Vol. 12

November 1998

20.

21.

22.

23.

24.

25. 26. 27.

28. 29.

Jones, F. E., Jerry, D. J., Guarino, B. C., Adrews, G. C., and Stern, D. F. (1996) Heregulin induces in vivo proliferation and differentiation of mammary epithelium into secretory lobuloalveoli. Cell Growth Differ. 7, 1031–1038 Chammas, R., Taverna, D., Cella, N., Santos, C., and Hynes, N. E. (1994) Laminin and tenascin assembly and expression regulate HC11 mouse mammary cell differentiation. J. Cell Sci. 107, 1031–1040 Lochter, A., and Bissell, M. J. (1997) Mammary gland biology and the wisdom of extracellular matrix. In Biological Signalling and the Mammary Gland (Wilde, C. J., Peaker, M., and Taylor, E., eds) pp 77–92, Hannah Research Institute, Ayr, Scotland Marti, A., Feng, Z., Jehn, B., Djonov, V., Chicaiza, G., Altermatt, H. J., and Jaggi, R. (1995) Expression and activity of cell cycle regulators during proliferation and programmed cell death in the mammary gland. Cell Death Differ. 2, 277–283 Lund, L. R., Romer, J., Thomasset, N., Solberg, H., Pyke, C., Bissell, M. J ., Dano, K., and Werb, Z. (1996) Two distinct phases of apoptosis in mammary gland involution: proteinase-independent and -dependent pathways. Development 122, 181–193 Chepko, G., and Smith, G. H. (1997) Three division-competent structurally-distinct cell populations contribute to murine mammary epithelial renewal. Tissue Cell 29, 239–253 Russo, J., and Russo, I. H. (1997) Role of differentiation in the pathogenesis and prevention of breast cancer. Endocrine-Related Cancer 4, 7–21 Nishi, N., Oya, H., Matsumoto, K., Nakamura, T., Miyanaka, H., and Wada, F. (1996) Changes in gene expression of growth factors and their receptors during castration-induced involution and androgen-induced regrowth of rat prostates. Prostate 28, 139–152 Higashino, F., Yoshida, K., Noumi, T., Seiki, M., and Fujinaga, K. (1995) Ets-related protein EIA-F can activate three different matrix metalloproteinase gene promoters. Oncogene 10, 1461–1463 D’Orazio, D., Besser, D., Marksitzer, R., Kunz, C., Hume, D. A., Kiefer, B., and Nagamine, Y. (1997) Cooperation of two PEA3/ AP1 sites in uPA gene induction by TPA and FGF-2. Gene 201, 179–187

The FASEB Journal


Received or publication April 30, 1998. Revised for publication June 1, 1998.

NEVE ET AL.