BIOLOGY OF REPRODUCTION 58, 719-727 (1998)

Expression of neu Differentiation Factor during the Periimplantation Period in theMouse Uterus'

Jeff Reese, 2 Naoko Brown, S.K. Das, and S.K. Dey

Department of Pediatrics and Department of Molecular and Integrative Physiology, Ralph L. Smith Research Center,University of Kansas Medical Center, Kansas City, Kansas 66160-7338

ABSTRACT

Complex cellular interactions occur between the blastocystand the uterus during implantation. The expression of variouspolypeptide growth factors and their receptors in the uterusand/or blastocyst during the periimplantation period suggestthat growth factors participate in the implantation process. Neudifferentiation factor (NDF) is a member of the epidermalgrowth factor (EGF) family of growth factors and is representedby multiple conserved isoforms. The expression of several EGF-like ligands in the periimplantation uterus has been character-ized, including EGF, heparin binding-EGF, transforming growthfactor oa, amphiregulin, betacellulin, and epiregulin. We ana-lyzed the expression pattern of NDF in the periimplantationmouse uterus because of its mitogenic and differentiation-pro-moting effects. By using Northern analysis and isoform-specificpolymerase chain reaction, we found that multiple isoforms areexpressed in the periimplantation uterus. NDF displays a highlyrestricted temporal and spatial expression, with autoradiograph-ic signals localized to the uterine stroma immediately surround-ing the implanting blastocyst. NDF expression was absent inmice with delayed implantation but briefly reappeared with thesame restricted distribution after termination of the delay by aninjection of estrogen. Taken together, these results suggest thatan activated blastocyst is required for the expression of NDFand that multiple isoforms may be involved in the complex net-work of cell-signaling events between the implanting blastocystand the receptive uterus.

INTRODUCTION

The process of implantation is dependent on coordina-tion between the development of the preimplantation em-bryo and differentiation of the uterus to a receptive state[1]. Aspects of embryo development that contribute to thisprocess include progression of the embryo to the blastocyststage, its escape from the zona pellucida, and its attainmentof an activated state [2]. Establishment of a uterus that isreceptive to the blastocyst for implantation requires prolif-eration and/or differentiation of heterogeneous cell types.This is primarily regulated by ovarian estrogen and pro-gesterone (P4 ). In ovariectomized adult mice, replacementof estrogen alone results in the proliferation of uterine lu-minal and glandular epithelia, while a combination of es-trogen and P4 is required for development of the uterinestroma [3, 4]. During the periimplantation period, a similarpattern of cell type-specific regulation by ovarian steroidsoccurs. On Days 1 and 2, epithelial cell proliferation is

Accepted October 14, 1997.Received September 10, 1997.'Supported by NIH grants P20RR11825 U.R.), ES07814 (S.K. Das),

HD12304, HD29968 (S.K. Dey), and NICHD center grants in Reproduc-tive Biology (HD33994) and Mental Retardation and Developmental Dis-abilities (HD02528).

2Correspondence: Jeff Reese, Department of Pediatrics, University ofKansas Medical Center, Kansas City, KS 66160. FAX: (913) 588-6317;e-mail: jreese@kumc.edu

under the influence of preovulatory ovarian estrogen secre-tion. On Day 3, P4 from newly formed corpora lutea initi-ates stromal cell proliferation. This is further stimulated bypreimplantation ovarian estrogen on Day 4. In contrast, thecombined effects of estrogen and P4 initiate epithelial celldifferentiation with concomitant inhibition of proliferationon Day 4. With the initiation of implantation, stromal cellsaround the implantation site undergo extensive proliferationand differentiation into decidual cells [4].

The earliest sign of implantation in the rodent is an in-creased endometrial vascular permeability at the sites ofblastocyst apposition. This is visualized as discrete bluebands along the uterus after an intravenous injection of amacromolecular blue dye [1]. This localized vascular per-meability coincides with the initial attachment reaction be-tween the uterine luminal epithelium and blastocyst tro-phectoderm [5]. In the mouse, the attachment reaction oc-curs in the evening (2200-2300 h) of Day 4 and is precededby uterine luminal closure, which results in an intimate ap-position of the trophectoderm with the luminal epithelium[1, 5, 6]. The attachment reaction is followed by localizeddecidualization of the stroma and apoptosis of the luminalepithelium at the site of blastocyst implantation [7], withsubsequent invasion of the trophoblast into the uterine stro-ma [8]., However, ovariectomy on the morning of Day 4results in blastocyst dormancy and failure of the attachmentreaction. This state of delayed implantation can be main-tained by P4 supplementation and is terminated by an in-jection of estrogen, with resulting blastocyst activation andthe initiation of implantation [2, 9].

The expression of several polypeptide growth factorsand their receptors in the embryo (reviewed in [10, 11]) anduterus [12-17] suggests an intimate cell-cell signaling path-way between the trophectoderm of an active blastocyst andthe luminal epithelium of a receptive uterus. In particular,the epidermal growth factor (EGF) family of mitogens andtheir receptors are expressed in the uterus and embryo in atemporal and cell type-specific manner that coincides withthe critical window for implantation [2, 12-17]. The EGFfamily of ligands includes EGF, transforming growth factora, heparin-binding (HB)-EGF, amphiregulin, betacellulin,epiregulin, and the neuregulins. Members of the EGF re-ceptor family include erbB 1, the prototype EGF receptor(EGF-R), erbB2 (neu/heregulin [HER] 2), erbB3 (HER 3),and erbB4 (HER 4). ErbB2 was originally identified as adominant transforming gene in the peripheral nervous sys-tem of rat embryos [18]. An activity that stimulated tyro-sine phosphorylation of neu/erbB2 was isolated from ras-transformed rat fibroblasts and termed neu differentiationfactor (NDF) [19-21], although later data suggested thatNDF does not bind or activate neu directly [22-24]. NDFis a 44-kDa soluble glycoprotein that binds to erbB3 anderbB4 and can induce signal transduction via heterodimer-ization with other erbB subtypes [25-31]. It exists in mem-brane-bound and secreted forms, with multiple isoforms of

719

REESE ET AL.

either mesenchymal or neuronal origin (reviewed in[32, 33]). NDF belongs to a family of polypeptides thatincludes heregulin, glial growth factors (GGFs), and ace-tylcholine receptor-inducing activity (ARIA) [34-36].These proteins appear to be alternate splice products of asingle gene that maps to human chromosome 8 [37]. Col-lectively termed neuregulins [35], these proteins contain aconserved structural motif with an immunoglobulin do-main, a glycosylated spacer region, an EGF-like domain, atransmembrane domain, and a cytoplasmic tail [21, 32].NDF may act as either a differentiation factor or a mitogen,depending on its cellular origin, isoform predominance, orreceptor preference [28, 29, 32, 38, 39].

NDF is predominantly expressed in central and periph-eral nervous systems of the developing rodent [35-37, 40].However, it is also expressed in adrenal, liver, kidney, ova-ry, testis, prostate, lung, heart, and dermal tissues[38, 40, 41]. Expression of NDF in the uterus has not beendescribed previously. We are interested in defining the rolesof NDF in the implantation process because of its functionas a ligand for the EGF-R family of receptor tyrosine ki-nases, and its known mitogenic and differentiation-promot-ing effects.

MATERIALS AND METHODS

Animals and Tissue Preparation

CD-1 (Charles River Laboratory, Raleigh, NC) andC57BI/6J (Jackson Labs, Bar Harbor, ME) mice werehoused in the animal care facility at the University of Kan-sas Medical Center according to NIH and institutionalguidelines for the care of laboratory animals. CD-1 micewere used in all experiments unless otherwise stated. Adultfemale mice (20-25 g; 48-60 days old) were mated withfertile males of the same strain. The morning when a vagi-nal plug was found was designated Day 1 of pregnancy.Mice were killed by cervical dislocation between 0830 and0930 h on Days 1-8 of pregnancy, and whole uteri werecollected. Pregnancy on Days 1-3 was confirmed by re-covery of embryos from oviducts, and on Day 4, from theuterine lumen. On Day 4.5 (2200-2300 h) and on Day 5(0830-0930 h), mice were anesthetized by i.p. injections of2.5% tribromoethanol (0.35 ml/mouse), and implantationsites were visualized by i.v. injections of 0.1 ml ChicagoBlue B dye solution (1% in saline). The animals were killed5 min later to identify the blue bands (implantation sites)along the uterus [1, 2].

To induce and maintain delayed implantation, mice wereovariectomized in the morning (0800-0900 h) of Day 4 ofpregnancy and were given daily injections of progesterone(P4 , 2 mg/mouse) from Days 5-7 [2, 9]. To terminate de-layed implantation and to induce blastocyst activation, theP4-primed delayed pregnant mice were given an injectionof estradiol-173 (E2; 25 ng/mouse) on the third day of thedelay (Day 7). These steroids were reconstituted in sesameoil and injected s.c. Mice were killed at 16-24 h after E2injection. Under these conditions, the first visually detect-able implantation sites after blue dye injection became vis-ible 16-24 h after an E2 injection.

Reverse Transcription (RT)-Polymerase Chain Reaction(PCR)

Reverse transcription was performed with an NDF-spe-cific antisense primer in the region of the cytoplasmic tail(5'-TCT CTG GCA TGC CTG AGG-3') [38] or with oli-

godeoxythymidine (oligo[dT]), using 1 g of total RNAfrom uterine tissues on Days 1-8 under conditions as pre-viously described [17]. Negative control reactions lackedRT, while adult brain RNA was used as a positive control.PCR was carried out with primers derived from the rat pro-NDFot2c sequence and amplified a region of NDF that isconserved in all isoforms. The sense primer corresponds tonucleotides 169-193 (5'-TGA AGA GCC AGG AGT CAGCTG CAG G-3') and the antisense primer to nucleotides481-501 (5'-GGC TCG AGA CTC TGA GGA CAC ATAGG-3') [37]. RT products (2 pl) were denatured at 95°Cfor 3 min and amplified for 35 cycles, under the followingconditions: 94°C, 30 sec; 66°C, 30 sec; 72°C, 60 sec; and72°C extension for 5 min, in 2 0-pl reaction volumes. PCRproducts were visualized on 2.5% agarose gels and blottedonto nylon membranes for subsequent Southern hybridiza-tion, using a 32p end-labeled internal oligonucleotide (5'-AGG AAA TGA CAG TGC CTC-3').

Cloning and Sequencing of the Mouse Uterine NDFPartial cDNA

The 333-basepair (bp) product of RT-PCR from Day 8uterine tissues was recovered and cloned into pCR-ScriptSK + cloning vector (Stratagene, La Jolla, CA). Several col-onies were analyzed by restriction digestion, and the nu-cleotide sequence of one clone was determined on bothstrands by the dideoxy nucleotide chain termination method[42] and the Sequenase version 2.0 kit (U.S. Biochemical,Cleveland, OH). This cDNA clone was also used to createprobes for Northern, Southern, and in situ hybridizations.

Hybridization Probes

For Northern and Southern hybridizations, 3 2 P-labeledprobes were generated through production of antisensecomplementary RNA (cRNA) from a 1.4-kilobase (kb)cDNA of mouse NDFa2c (a gift from G. Plowman, SUG-EN Inc., Redwood City, CA), or from the NDF cDNAcloned from RT-PCR amplification of Day 8 pregnantmouse uterus. A cDNA of human fibroblast [3-actin [12, 43]served as a template for synthesis of a 32 P-labeled cRNAprobe. Specific oligonucleotides were end-labeled with 3 2pfor Southern blot analysis of RT-PCR products (5'-AGGAAA TGA CAG TGC CTC-3'), and for isoform-specificPCR products (NDFa2c, 481-501-nt antisense PCR prim-er) [44]. For in situ hybridization, sense and antisense 35S-labeled cRNA probes were generated using the appropriatepolymerases with the 1.4-kb or 333-bp mouse NDF c-DNAs. Probes had specific activities of approximately 2 x109 dpm/g.

Northern Blot Hybridization

Total RNAs were extracted from whole uteri pooledfrom 25-30 mice on the indicated days of pregnancy by amodified guanidine thiocyanate procedure [12, 45]. Poly-adenylated (poly[A] +) RNAs were isolated from totalRNAs by oligo(dT)-cellulose column chromatography [44].Poly(A)+ RNA (2 ,ug) was denatured, separated by form-aldehyde-agarose gel electrophoresis, transferred to nylonmembranes, and cross-linked by UV irradiation (Spectro-linker, XL-1500; Spectronics Corp., Westbury, NY). Mem-branes were prehybridized, hybridized, and washed as de-scribed previously [12, 13]. After hybridization, the blotswere washed under stringent conditions, and the hybrids

720

NDF EXPRESSION IN THE MOUSE UTERUS

were detected by autoradiography. The membranes werethen stripped and rehybridized with -actin.

Isoform-Specific RT-PCR

Antisense PCR primers were designed that recognizeonly the a or 3 isoform of NDF, based on sequence vari-ation in the terminal portion of the EGF domain: NDFa:5'-TTG GGT TTG GAC TT CAT GG-3', and NDFR3: 5'-GCT GGC CAT TAC GTA GTT TTG-3'. A single up-stream sense primer (5'-CCA ACG AGT TCA TCA CTGGC-3'), which anneals in the conserved region for all iso-forms, was used for both a- and 13-specific reactions. Theabove described RT products (1 Il1) from uterine or brainsamples were amplified for 35 cycles with incorporation of[3 2 P]dCTP to label the PCR products, under the followingparameters: 94°C, 30 sec; 60°C 30 sec; 72°C 30 sec. Re-action products were visualized on a 6% polyacrylamidegel, since the predicted PCR products vary in length byonly 12 bp. A 1.0-kb ladder (Gibco BRL, Gaithersurg, MD)was end-labeled with 32p for use as a size marker.

In Situ Hybridization

In situ hybridization was performed as described previ-ously [12, 14, 17]. Uteri were cut into 4- to 6-mm piecesand flash-frozen in freon. Whole embryos from Day 15 ofgestation [37] or adult brain samples [37, 40] were alsorecovered as controls. Frozen sections (11 m) from severaltime points on Days 4 and 5 of gestation, and control sec-tions from whole embryo or brain specimens were mountedonto poly-L-lysine-coated slides, fixed in cold 4% parafor-maldehyde solution in PBS, acetylated, and hybridized at45°C for 4 h in 50% formamide hybridization buffer con-taining the 35S-labeled antisense cRNA probe. After hy-bridization and washing, the sections were incubated withribonuclease A (RNase A; 20 Rig/ml) at 37°C for 15 min.RNase A-resistant hybrids were detected by autoradiogra-phy using Kodak NTB-2 liquid emulsion (Eastman Kodak,Rochester, NY) after 3- to 4-wk exposure time. Parallelsections hybridized with the sense probe served as negativecontrols. Slides were poststained with hematoxylin and eo-sin.

RESULTS

NDF Expression in the Mouse Uterus by RT-PCR

To evaluate whether NDF is expressed in the periim-plantation mouse uterus, we first performed RT-PCR onuterine tissues from Day 1-8 pregnant CD-I mice. Com-plementary DNAs of the predicted size were present inuterine tissues from all days of pregnancy (data not shown).Amplification yielded a single 333-bp band from both brainand uterine tissues. The lack of amplification products fromRNA, RT-negative, and water controls supports NDF gene-specific expression in the mouse uterus. These results wereconfirmed by Southern hybridization of the PCR productswith 3 2P-labeled NDFa2c (Fig. 1).

Sequence of Mouse Uterine NDF Partial cDNA

To confirm that the amplified RT-PCR products werespecific to mouse tissues and to evaluate its nucleotide se-quence, the 333-bp RT-PCR product from Day 8 pregnantuterine tissues was recovered and cloned into a suitablevector. Sequence analysis of this clone revealed a uniqueA--G base pair variation at nucleotide position 283 (from

FIG. 1. Demonstration of NDF expression in the mouse periimplantationuterus. Using primers derived from the rat NDFa2c sequence, RT-PCRproducts from whole Day 8 uterine total RNA were electrophoresed, blot-ted, and hybridized with an internal oligonucleotide. A single 333-bpband was detected in brain (positive control) and uterine tissues. +,-,Presence or absence of RT in the RT-PCR reaction.

rat proNDFa2c) [38], which has been observed by otherinvestigators (G. Plowman, personal communication). Re-striction enzyme cleavage was not altered by this differ-ence, but a conserved amino acid change from lysine (rat)to arginine (mouse) at amino acid position 80 is predictedby this sequence variation.

Northern Analysis of NDF Expression in thePeriimplantation Uterus

To evaluate changes in the pattern of NDF expressionduring the periimplantation period, the levels of NDF mes-sage were analyzed by Northern hybridization on Days 1-8using 3 2 P-labeled NDF cRNA. Expression of multiple tran-scripts was detected in total and poly(A)+ RNA samples.Transcript sizes varied from 1.5 to 9 kb, but there wereno predominant transcripts on Days 4-5 around the time ofimplantation (Fig. 2). Adult CD-1 brain RNA served as apositive control and also revealed the expression of multi-ple bands. RNA loading was standardized by acridine or-ange gel staining and through hybridization with 3-actin.The observed pattern of expression was consistent over 5blots from different groups of Day 1-8 pregnant mouseuteri. There exist at least six pro-NDF isoforms and morethan 12 isoforms of processed or secreted NDF in rodents,which appear to arise from alternate splicing events [38].Tissue-specific expression of NDF has been documentedfrom a variety of organs in the rat and human. Reproductivetissues, such as the ovary, testis, and placenta, show mul-tiple NDF transcripts [21, 34], although no expression wasdetected in the human uterus [34]. The observed expressionof multiple transcripts on our Northern blots correspondswith the production of a single RT-PCR band, since theregion selected for PCR amplification is conserved amongall known NDF species.

Isoform-Specific NDF Expression

Isoform-specific RT-PCR amplification, which identifiesalpha or beta subtypes of NDF, was performed with anti-sense PCR primers designed to anneal within the highlyvariable 18-21-amino acid region of the EGF moiety. Con-trol (brain) tissues yielded bands of the predicted size forboth NDFa and NDFP, with some predominance of theNDF{3 isoform, although these PCR methods are only semi-quantitative. Uterine tissues produced cDNAs of equal in-tensity from both NDFa and NDF3 reactions, with a cor-responding lack of amplification in all RT-negative and wa-

721

REESE ET AL.

FIG. 3. Isoform-specific NDF mRNA expression. RT-PCR was performedwith primers that discriminate between c(- and 1-NDF isoforms, based onsequence variation within the EGF domain. Complementary DNAs werelabeled by random incorporation of 32p and revealed the expression ofboth isoforms in brain (positive control) and uterus, with a correspondinglack of expression in controls without RT. +,-, Presence or absence ofRT. M, 32P-labeled 1-kb ladder.

FIG. 2. Northern hybridization of NDF mRNA in the periimplantationmouse uterus. Approximately 2 Ig of poly(A) + RNA from uteri of Days1-8 of pregnancy and adult brain (Br) tissues were electrophoresed indenaturing gels, blotted, and hybridized with the 333-bp mouse NDFcDNA. Arrows indicate transcripts of approximately 1.5, 1.8, 2.2, 3.0,4.4, and 9 kb. Membranes were rehybridized with a 3-actin probe tostandardize RNA loading [15, 21-23].

ter control reactions (Fig. 3). These results indicate thatuterine tissues express both a and 3 subtypes of NDE

In Situ Hybridization

The temporal and cell-specific localization of NDF ex-pression was evaluated by in situ hybridization (Fig. 4).Although Northern analysis did not indicate that an NDFmessage was up-regulated at the time of implantation, animplantation-specific expression of NDF, like that of HB-EGF [12], amphiregulin [14], and betacellulin and epire-gulin [13], was detected by in situ hybridization. No cell-specific autoradiographic signals were detected in the uteruson the mornings of Days 1-4 of pregnancy, or until 1600h on the afternoon of Day 4 (Fig. 4B). However, on theevening of Day 4 (2200-2300 h) at the time of the attach-ment reaction and on the morning of Day 5 of pregnancy,autoradiographic signals were detected in a limited regionof stroma beneath the uterine luminal epithelium, restrictedto the area immediately surrounding the implanting blas-tocyst (Fig. 4, D and F). Thereafter, the NDF signal wasno longer detected, except at very low levels within thedeveloping embryo (Day 8). Asymmetry of NDF mRNAdistribution was noted on several sections, as has been ob-served for betacellulin [13]. This asymmetry may representpreferential expression of these ligands at the sites of initialattachment or could be due to selection of sections fromdifferent planes during sectioning. No distinct expressionwas detected over background in the interimplantationregions. The specificity of the hybridization was confirmedby the lack of signals in tissue sections hybridized with anNDF sense probe. The control brain and whole embryosections revealed NDF mRNA expression in discrete cellpopulations, corresponding to prior observations [37, 40].

Results were identical among multiple implantation sitesfrom the 10 CD-1 mice examined. We also evaluated sitesfrom three C57BI/6J mice by in situ hybridization andfound an identical pattern of restricted cellular expression.Although signal intensity was highest on the evening ofDay 4 (2200-2300), C57B1/6J mice had no detectable sig-nal by the morning of Day 5, suggesting that more temporalrestriction may occur in other mouse species, or that vari-ations in the timing of implantation may exist, which alterthe anticipated expression pattern of tightly restrictedgenes.

Regulation of NDF Expression by Blastocysts

Ovariectomy and supplementation with ovarian steroidswas used to examine whether the limited expression ofNDF is dependent on the presence of an active blastocyst.In situ hybridization was performed on uterine sections ob-tained from mice with P4-treated delayed implantation orafter the initiation of blastocyst activation and implantationby an E2 injection (Fig. 5). There were no hybridizationsignals in the uterus adjacent to the dormant blastocysts thathad been in close apposition to the luminal epithelium dur-ing P4 supplementation and delayed implantation (Fig. 5B).However, the injection of E2 resulted in temporal and cell-specific expression of the NDF signal in a pattern similarto expression observed during normal implantation (Fig.5D). The intensity of the NDF signal was strongest at 16h after E2 injection, with some diminution in expression by24 h after injection. The expression pattern was again high-ly localized to the subepithelial uterine stroma in the im-mediate vicinity of the implantation chamber. Interimplan-tation sites lacked specific NDF signal. Taken together,these data suggest that the presence of a blastocyst is re-quired for uterine NDF expression, and that estrogen acti-vation is necessary for induction of NDF expression in theuterine stroma.

DISCUSSION

The present study examined the expression of NDF, amember of the EGF family of ligands, during the periim-plantation period. This is the first demonstration of the ex-pression of the NDF gene in the mouse uterus. We foundthat NDF mRNA displays a restricted spatial expressionpattern similar to that of other members of the EGF familyof ligands [12, 14] and the most limited temporal expres-

722

NDF EXPRESSION IN THE MOUSE UTERUS

FIG. 4. In situ hybridization of NDF mRNA in the mouse uterus at the time of implantation. Frozen sections (11 m) were obtained at various timepoints on Day 4 and the morning of Day 5. Sections through implanting blastocysts (identified using blue dye injection) on Day 4 night and Day 5morning were hybridized with a 35S-labeled NDF cRNA probe. A, C, and E show brightfield images, B, D, and F, corresponding darkfield images;x100 (reproduced at 94%). A, B) Day 4, 1600 h; C, D) Day 4, 2300 h; E, F) Day 5, 0900 h. NDF signal is confined to the uterine stroma immediatelyadjacent to the luminal epithelium in the region surrounding the blastocyst. bl, Blastocyst; le, luminal epithelium; ge, glandular epithelium; st, stroma.

723

REESE ET AL.

FIG. 5. In situ hybridization of NDF mRNA in the delayed implanting mouse uterus. A and C show brightfield micrographs, B and D, correspondingdarkfield images; x100. A, B) A representative uterine section of P4-treated delayed implanting uterus containing a dormant blastocyst. C, D) A repre-sentative uterine section after termination of the delayed implantation by E2. NDF mRNA accumulation is detected in stromal cells adjacent to anactivated blastocyst (D). bl, Blastocyst; le, luminal epithelium; ge, glandular epithelium.

sion of any of these factors, although epiregulin and beta-cellulin are also ligands of the EGF family with a restricteduterine expression pattern [13]. The expression of NDF inembryo-uterine interactions was investigated because of itsaction as an EGF-like ligand and the dual nature of itseffects on cell growth. Depending on its context, NDF hasbeen observed to function either as a cellular mitogen oras a stimulus to differentiation with corresponding growtharrest (reviewed in [32]). The mitogenic properties of theNDF-related glial growth factors (GGFs) in bovine neuraltissues allowed its identification as a member of the neu-regulin family [35]. Similarly, growth-promoting effectswere noted for human breast cancer cells exposed to dif-ferent isoforms of recombinant NDF [34]. In contrast, NDF

stimulation of particular rat breast cell carcinoma cell linesexposed to NDF resulted in differentiation and increasedproduction of breast milk components with concomitant in-hibition of cell growth [20]. Stimulatory monoclonal anti-bodies directed to the human neu/HER2/erbB2 receptoralso induced differentiation of human mammary tumorcells to a milk-producing and growth-arrested state [46]. Inaddition, NDF may have an effect other than differentiationor proliferation on epithelial cells. For example, treatmentof experimentally induced dermal wounds in the rabbit earwith recombinant human NDFot2 increased epithelial cellmigration and skin thickness [41]. Therefore, NDF expres-sion in the periimplantation mouse uterus may be relatedto the complex series of events leading to growth and pro-

724

NDF EXPRESSION IN THE MOUSE UTERUS

liferation of the uterine luminal epithelium and underlyingstroma.

A variety of pro-NDF isoforms are expressed in differentrat tissues [38]. The results of our RT-PCR and demonstra-tion of a unique mouse partial cDNA by sequence analysisconfirmed the presence of a mouse-specific NDF message.The detection of multiple NDF transcripts by Northern hy-bridization is consistent with previous reports. Althoughmost rodent and human tissues express 2 or 3 major tran-scripts (1.7-6.8 kb) [21, 34], detection of additional tran-scripts has been reported with longer exposure [21] or ex-amination of specific tissues [34]. Since our probes werederived from the region of NDF that is identical among allknown isoforms, it was expected that many transcripts andboth the secreted and precursor forms could be identified.Insignificant alteration in uterine NDF mRNA levels onDays 1-8 of pregnancy as detected by Northern hybridiza-tion is similar to that observed for HB-EGF and betacellulin[12, 13]. Time-dependent changes in cell type-specific ex-pression may be obscured by the dilution of RNAs derivedfrom heterogeneous uterine cell types. Thus, these resultsdo not reflect the importance of the temporal and cell-spe-cific expression of these ligands around the time of im-plantation as examined by in situ hybridization. Recogni-tion of other neuregulins may also contribute to the obser-vation of multiple RNA transcripts in Northern analysis.Two recently identified members of the neuregulin family,neuregulin-2 and neuregulin-3, also display multiple tran-scripts on Northern hybridization, although sequence ho-mology to the neuregulin-1 family (NDF, GGFs, ARIA,heregulins) is only 19-45% [47-49].

Within the EGF family, NDF is unique in this redun-dancy and its representation by multiple isoforms. In par-ticular, the areas of highest sequence variation are in theEGF domain and in the adjacent region connecting the EGFand transmembrane domains [38]. Designation as a or 13subtypes is based on variation in the EGF sequence in theC terminal (18-21 amino acids) of this domain [34]. NDFaisoforms bind less avidly to target cells, appear to serve aslow-affinity ligands, and are present primarily in mesen-chymal tissues, while the 3 isoforms and the GGFs andARIA, which share the beta configuration, are found pre-dominantly in neural tissues [35, 36, 38, 40, 50]. Meyer andBirchmeier showed the coordinate expression of both sub-types in mouse embryonic intestine using an RNase pro-tection assay [50]. We demonstrated the expression of botha and 3 isoforms in Day 5 pregnant mouse uterine tissuesby isoform-specific RT-PCR, with the expected predomi-nance of 3 isoforms in the brain samples. Although othernon-neuronal tissues express NDF3 [38, 40, 50], coexpres-sion of both subtypes in the periimplantation uterus sug-gests that the tissue-specific alternate splicing events, whichgenerate o and 13 isoforms, occur in a very small populationof stromal cells in a tightly restricted temporal and spatialmanner. The significance of this coexpression is not clear,but it may be important since a and 3 subtypes differ intheir mitogenic potential [41, 51, 52] and their ability toinduce various combinations of erbB receptor heterodimers[39].

Previously, erbB3 and erbB4 were identified as low- andhigh-affinity receptors, respectively, for most NDF isoforms[23, 24, 53], with erbB3 possessing an impaired kinase ac-tivity [54]. However, the low-affinity binding of erbB3 isimproved through coexpresssion with erbB2 [27], and theerbB3/erbB2 heterodimer is the predominant form of NDFreceptor in epithelial cells [55]. A hierarchy of receptor

cross-talk has been proposed with a preference of NDF forerbB3/erbB2 [29, 31]. Although erbB2 is an orphan recep-tor, it functions as a common subunit for all other receptorsubunits [28, 30] and is preferred for heterodimerizationwhen NDF is bound to erbB3 [29]. Alpha and 13 isoformsof NDF are equivalent in their ability to activate erbB3when coexpressed with erbB2, but only NDF3 can induceheterodimers of erbB3 and erbB1 [31]. In the mouse uterus,the full-length transcript of erbB1 is absent in the luminalepithelium but present in the uterine stroma and myome-trium [15], and a truncated form of erbB1 is present in theluminal epithelium as well as stroma and myometrium [16].ErbB3 and erbB4 serve as the primary NDF receptors[28, 30], but the uterine expression pattern of erbB3 is stillunknown. Uterine erbB4 expression is limited to the lu-minal epithelium and stroma in the immediate vicinity ofthe implantation chamber (unpublished results), suggestingthat erbB and erbB4 interaction is possible in the regionin which NDF is localized. Uterine expression of erbB2occurs primarily in the epithelial cells on Days 1-4, but itbecomes localized to the luminal epithelium and decidual-izing stromal cells by Day 5 and is not limited to the im-plantation site alone [17]. In that study [17], recombinantNDFot and NDF3 activated erbB2 in the Day 5 pregnantuterus but to a lesser extent than did other EGF-family li-gands. In addition, NDFa showed diminished transphos-phorylation relative to NDF13, as noted by others [38, 39].Therefore, the presence of a and 3 NDF subtypes in com-bination with erbB2 and erbB4 expression in a distinct re-gion surrounding the blastocyst is consistent with an emerg-ing model of lateral signaling or transmodulation [31, 56-58] that is mediated by receptor heterodimerization of erbBreceptors and by ligand induction of heterodimer formationin the initial steps of blastocyst-uterine cell signaling.

The absence of NDF expression in the uterine stromalcells at the sites of blastocyst apposition during P4-treateddelayed implantation, but its induction with the onset ofblastocyst activation and termination of the delay by anestrogen injection suggest that the expression of NDF inthe uterus requires the presence of active blastocysts. Blas-tocyst activation is also required for the uterine expressionof HB-EGF, betacellulin, and epiregulin [12, 13], suggest-ing a network of tightly regulated communication steps be-tween the uterus and embryo. The specific contribution ofNDF to the implantation process is difficult to assess. Micewith targeted disruption of NDF have an embryonic lethalphenotype, with severe cardiac malformations and abnor-mal formation of the central and peripheral nervous system[59]. The cell types affected by this mutation are alsoknown to express erbB3 or erbB4. Interestingly, mice thatlack erbB2 or erbB4 are also embryonic lethal and are ob-served to have similar phenotypes [60, 61]. The contribu-tion of neuregulin-2, which binds to erbB3erbB4 hetero-dimers [47,48], and neuregulin-3, which binds to erbB4[49], to NDF activity requires further investigation. Anal-ysis of the functional contribution of NDF to implantationwill remain incomplete until the expression and interactionof erbB3 and erbB4 receptors in the uterus and blastocystare fully characterized. However, the short "window" ofexpression of NDF around the active blastocyst during theinitial phase of implantation and the coexpression of erbBreceptors in this vicinity supports a role for this growthfactor in implantation.

REFERENCES1. Psychoyos A. Endocrine control of egg implantation. In: Greep RO,

Astwood EG, Geiger SR (eds.), Handbook of Physiology. WashingtonDC: American Physiological Society; 1973: 187-215.

725

REESE ET AL.

2. Paria BC, Huet-Hudson YM, Dey SK. Blastocyst's state of activitydetermines the "window" of implantation in the mouse receptive uter-us. Proc Natl Acad Sci USA 1993; 90:10159-10162.

3. Finn C, Martin L. The role of the oestrogen secreted before oestrusin the preparation of the uterus for implantation in the mouse. J En-docrinol 1970; 47:431-438.

4. Huet-Hudson YM, Andrews GK, Dey SK. Cell type-specific localiza-tion of c-myc protein in the mouse uterus: modulation by steroid hor-mones and analysis of the periimplantation period. Endocrinology1989; 125:1683-1690.

5. Enders AC, Schlafke S. A morphological analysis of the early im-plantation stages in the rat. Am J Anat 1967; 120:185-226.

6. Enders AC. Anatomical aspects of implantation. J Reprod Fertil 1976;25:1-15.

7. Parr EL, Tung HN, Parr MB. Apoptosis as the mode of uterine epi-thelial cell death during embryo implantation in mice and rats. BiolReprod 1987; 36:211-225.

8. Schlafke S, Enders AC. Cellular basis of interaction between tropho-blast and uterus at implantation. Biol Reprod 1975; 12:41-65.

9. Yoshinaga K, Adams CE. Delayed implantation in the spayed, pro-gesterone treated adult mouse. J Reprod Fertil 1966; 12:593-595.

10. Adamson ED. Activities of growth factors in preimplantation embry-os. J Cell Biochem 1993; 53:280-287.

11. Paria BC, Dey SK. Preimplantation embryo development in vitro:Cooperative interactions among embryos and role of growth factors.Proc Natl Acad Sci USA 1990; 87:4756-4760.

12. Das SK, Wang X-N, Paria BC, Damm D, Abraham JA, Klagsbrun M,Andrews GK, Dey SK. Heparin-binding EGF-like growth factor geneis induced in the mouse uterus temporally by the blastocyst solely atthe site of its apposition: a possible ligand for interaction with blas-tocyst EGF-receptor in implantation. Development 1994; 120:1071-1083.

13. Das SK, Das N, Wang J, Lim H, Schryver B, Plowman GD, Dey SK.Expression of betacellulin and epiregulin genes in the mouse uterustemporally by the blastocyst solely at the site of its apposition is co-incident with the "window" of implantation. Dev Biol 1997; 190:178-190.

14. Das SK, Chakraborty I, Paria BC, Wang X-N, Plowman G, Dey SK.Amphiregulin is an implantation-specific and progesterone-regulatedgene in the mouse uterus. Mol Endocrinol 1995; 9:691-705.

15. Das SK, Tsukamura H, Paria BC, Andrews GK, Dey SK. Differentialexpression of epidermal growth factor receptor (EGF-R) gene andregulation of EGF-R bioactivity by progesterone and estrogen in theadult mouse uterus. Endocrinology 1994; 134:971-981.

16. Tong BJ, Das SK, Threadgill D, Magnuson T, Dey SK. Differentialexpression of the full-length and truncated forms of the epidermalgrowth factor receptors in the preimplantation mouse uterus and blas-tocyst. Endocrinology 1996; 137:1492-1496.

17. Lim H, Dey SK, Das SK. Differential expression of the erbB2 genein the periimplantation mouse uterus: potential mediator of signalingby epidermal growth factor-like growth factors. Endocrinology 1997;138:1328-1337.

18. Shih C, Padhy LC, Murray M, Weinberg RA. Transforming genes ofcarcinoma and neuroblastomas introduced into mouse fibroblast. Na-ture 1981; 290:261.

19. Yarden Y, Weinberg RA. Experimental approaches to hypotheticalhormones: detection of a candidate ligand of the neu proto-oncogene.Proc Natl Acad Sci USA 1989; 86:3179-3188.

20. Peles E, Bacus SS, Koski RA, Lu HS, Wen D, Ogden SG, Levy RB,Yarden Y. Isolation of the neu/HER-2 stimulatory ligand: a 44 kdglycoprotein that induces differentiation of mammary tumor cells. Cell1992; 69:205-216.

21. Wen D, Peles E, Cupples R, Suggs SV, Bacus SS, Luo Y, Trail G, HuS, Silbiger SM, Levy RB, Koski RA, Lu HS, Yarden Y. Neu differ-entiation factor: a transmembrane glycoprotein containing an EGF do-main and an immunoglobulin homology unit. Cell 1992; 69:559-572.

22. Peles E, Ben-Levy R, Tzahar E, Liu N, Wen D, Yarden Y. Cell-typespecific interaction of neu differentiation factor (NDF/heregulin) withneu/HER-2 suggests complex ligand-receptor relationships. EMBO J1993; 12:961-971.

23. Plowman DG, Green JM, Culouscou J-M, Carlton GW, Rothwell VM,Buckley S. Heregulin induces tyrosine phosphorylation of HER4/pl80er bB4. Nature 1993; 366:473-475.

24. Tzahar E, Levkowitz G, Karunagaran D, Yi L, Peles E, Lavi S, ChangD, Liu J, Yayon A, Wen D, Yarden Y. ErbB3 and erbB4 function asthe respective low and high affinity receptors of all neu differentiationfactor/heregulin isoforms. J Biol Chem 1994; 269:25226-25233.

25. Kraus MH, Fedi P, Starks V, Muraro R, Aaronson S. Demonstrationof ligand-dependent signaling by the erbB-3 tyrosine kinase and itsconstitutive activation in human breast tumor cells. Proc Natl AcadSci USA 1993; 90:2900-2904.

26. Plowman DG, Culouscou J-M, Withney GS, Green JM, Carlton GW,Foy L, Neubauer MG, Shoyab M. Ligand-specific activation of HER4/pl80erbB4, a fourth member of the epidermal growth factor receptorfamily. Proc Natl Acad Sci USA 1993; 90:1746-1750.

27. Sliwkowski MX, Schaefer G, Akita RW, Lofgren JA, Fitzpatrick VD,Nuijens A, Fendly BM, Cerione RA, Vandlen RL, Carraway III KL.Coexpression of erbB2 and erbB3 proteins reconstitutes a high affinityreceptor for heregulin. J Biol Chem 1994; 269:14661-14665.

28. Riese II DJ, Raaij TM, Plowman GD, Andrews GC, Stern DE Thecellular response to neuregulins is governed by complex interactionsof the erbB receptor family. Mol Cell Biol 1995; 15:5770-5776.

29. Tzahar E, Waterman H, Chen X, Levkowitz G, Karunagaran D, LaviS, Ratzkin BJ, Yarden Y. A hierarchical network of interreceptor in-teractions determines signal transduction by neu differentiation factor/neuregulin and epidermal growth factor. Mol Cell Biol 1996; 16:5276-5287.

30. Karunagaran D, Tzahar E, Beerli RR, Chen X, Graus-Porta D, RatzkinBJ, Seger R, Hynes NE, Yarden Y. ErbB2 is a common auxiliarysubunit of NDF and EGF receptors: implications for breast cancer.EMBO J 1996; 15:254-264.

31. Pinkas-Kramarski R, Soussan L, Waterman H, Levkowitz G, Alroy I,Klapper L, Lavi S, Seger R, Ratzkin BJ, Sela M, Yarden Y. Diversi-fication of neu differentiation factor and epidermal growth factor sig-naling by combinatorial receptor interactions. EMBO J 1996; 15:2252-2467.

32. Peles E, Yarden Y. Neu and its ligands: from an oncogene to neuralfactors. Bioessays 1993; 15:815-824.

33. Ben-Baruch N, Yarden Y. Neu differentiation factors: a family of al-ternatively spliced neuronal and mesenchymal factors. Proc Soc ExpBiol Med 1994; 206:221-227.

34. Holmes WE, Sliwkowski MX, Akita RW, Henzel WJ, Lee J, Park JW,Yansura D, Abadi N, Raab H, Lewis GD, Shepard HM, Kuang W-J,Wood WI, Goeddel DV, Vandlen RL. Identification of heregulin, aspecific activator of pl85erbB2. Science 1992; 256:1205-1210.

35. Marchionni MA, Goodearl ADJ, Chen MS, Bermingham-McDonoghO, Kirk C, Hendricks M, Danehy F, Misumi D, Sudhalter J, KobayashiK, Wroblewski D, Lynch C, Baldassare M, Hiles I, Davis JB, HsuanJJ, Totty NE, Otsu M, McBurney RN, Waterfield MD, Stroobant P,Gwynne D. Glial growth factors are alternatively spliced erbB2 li-gands expressed in the nervous system. Nature 1993; 362:312-318.

36. Falls DL, Rosen KM, Corfas G, Lane WS, Fischbach GD. ARIA, aprotein that stimulates acetylcholine receptor synthesis, is a memberof the neu ligand family. Cell 1993; 72:801-815.

37. Orr-Urtreger A, Trakhtenbrot L, Ben-Levy R, Wen D, Rechavi G,Lonai P, Yarden Y. Neural expression and chromosomal mapping ofneu differentiation factor to 8p12-p21 . Proc Natl Acad Sci USA 1993;90:1867-1871.

38. Wen D, Suggs SV, Karunagaran D, Liu N, Cupples RL, Luo Y, Jans-sen AM, Ben-Baruch N, Trollinger DB, Jacobsen VL, Meng SY, LuHS, Hu S, Chang D, Yang W, Yanigahara D, Koski RA, Yarden Y.Structural and functional aspects of the multiplicity of neu differen-tiation factors. Mol Cell Biol 1994; 14:1909-1919.

39. Pinkas-Kramarski R, Shelly M, Glathe S, Ratzkin BJ, Yarden Y. Neudifferentiation factor/neuregulin isoforms activate distinct receptorcombinations. J Biol Chem 1996; 271:19029-19032.

40. Corfas G, Rosen KM, Aratake H, Krauss R, Fischbach GD. Differ-ential expression of ARIA isoforms in the rat brain. Neuron 1995; 14:103-115.

41. Danilenko DM, Ring BD, Lu JZ, Tarpley JE, Chang D, Liu N, WenD, Pierce GE Neu differentiation factor upregulates epidermal migra-tion and integrin expression in excisional wounds. J Clin Invest 1995;95:842-851.

42. Sanger F, Nicklen S, Coulson AR. DNA sequencing with chain-ter-minating inhibitors. Proc Natl Acad Sci USA 1977; 74:5463-5467.

43. Ponte Ng S-Y, Engel J, Gunning P, Kedes L. Evolutionary conser-vation in the untranslated regions of actin RNAs: DNA sequence ofa human beta-actin cDNA. Nucleic Acids Res 1984; 12:1687-1696.

44. Sambrook J, Fritsch EF, Maniatis T. Molecular Cloning: A LaboratoryManual. Cold Spring Harbor, NY; Cold Spring Harbor Press; 1989.

45. Han JH, Stratowa C, Rutter WJ. Isolation of full-length putative ratlysophospholipase cDNA using improved methods for mRNA isola-tion and cDNA cloning. Biochemistry 1987; 26:1617-1625.

46. Bacus SS, Stancobski I, Huberman E, Chin D, Hurwitz E, Mills GB,

726

NDF EXPRESSION IN THE MOUSE UTERUS

Ullrich A, Sela M, Yarden, Y. Tumor-inhibitory monoclonal antibodiesto the HER-2/neu receptor induce differentiation of human breast can-cer cells. Cancer Res 1992; 52:2580-2589.

47. Chang H, Riese DJ, Gilbert W, Stem DF, McMahan UJ. Ligands forerbB-family receptors encoded by a neuregulin-like gene. Nature1997; 387:509-512.

48. Carraway III KL, Weber JL, Unger MJ, Ledesma J, Yu N, GassmanM, Lai C. Neuregulin-2, a new ligand of erbB3/erbB4-receptor tyro-sine kinases. Nature 1997; 387:512-516.

49. Zhang D, Sliwkowski MX, Mark M, Frantz G, Akita R, Sun Y, HillanK, Crowley C, Brush J, Godowski PJ. Neuregulin-3 (NRG3): a novelneural tissue-enriched protein that binds and activates erbB4. ProcNatl Acad Sci USA 1997; 94:9562-9567.

50. Meyer D, Birchmeier C. Distinct isoforms of neuregulin are expressedin mesenchymal and neuronal cells during mouse development. ProcNatl Acad Sci USA 1994; 91:1064-1068.

51. Marikovsky M, Lavi S, Pinkas-Kramarski R, Karunagaran D, Liu N,Wen D, Yarden Y. ErbB-3 mediates differential mitogenic effects ofNDF/heregulin isoforms on mouse keratinocytes. Oncogene 1995; 10:1403-1411.

52. Ram TG, Kokeny KE, Dilts CA, Ethier SP Mitogenic activity of neudifferentiation factor/heregulin mimics that of epidermal growth factorand insulin-like growth factor-I in human mammary epithelial cells. JCell Physiol 1995; 163:589-596.

53. Carraway KL I, Sliwkowski MX, Akita R, Platko JV, Guy PM,Nuijens A, Diamonte AJ, Vandlen RL, Cantley LC, Cerione RA. The

erbB3 gene product is a receptor for heregulin. J Biol Chem 1994;269:14303-14306.

54. Guy PM, Platko JV, Cantley LC, Cerione RA, Carraway HI KL. Insectcell-expressed pl80 er bB3 possesses an impaired tyrosine kinase activ-ity. Proc Natl Acad Sci USA 1994; 91:8132-8136.

55. Chen X, Levkowitz G, Tzahar E, Karunagaran D, Lavi S, Ben-BaruchN, Leitner O, Ratzkin BJ, Bacus SS, Yarden Y. An immunologicalapproach reveals biological differences between the two NDF/here-gulin receptors, erbB-3 and erbB-4. J Biol Chem 1996; 271:7620-7629.

56. Heldin CH. Dimerization of cell surface receptors in signal transduc-tion. Cell 1995; 80:213-223.

57. Carraway III KL, Cantley LC. A neu acquaintance for erbB3 anderbB4: a role for receptor heterodimerization in growth signaling. Cell1994; 78:5-8.

58. Earp HS, Dawson TL, Li X, Yu H. Heterodimerization and functionalinteraction between EGF receptor family members: a new signalingparadigm with implications for breast cancer research. Breast CancerRes Treat 1995;35:115-132.

59. Meyer D, Birchmeier C. Multiple essential functions of neuregulin indevelopment. Nature 1995; 378:386-390.

60. Gassman M, Casagranda F, Orioli D, Simon H, Lai C, Klein R, LemkeG. Aberrant neural and cardiac development in mice lacking theerbB4 neuregulin receptor. Nature 1995; 378:390-394.

61. Lee KF, Simon H, Chen H, Bates B, Hung MC, Hauser C. Require-ment for neuregulin receptor erbB2 in neural and cardiac develop-ment. Nature 1995; 378:394-398.

727