BIOLOGY OF REPRODUCTION 50, 481-491 (1994)

Distribution of Transforming Growth Factor a Precursors in the Mouse Uterus duringthe Periimplantation Period and after Steroid Hormone Treatments'

B.C. PARIA,3 S.K. DAS,4 Y.M. HUET-HUDSON,S and S.K. DEY2'3'4

Departments of Obstetrics-Gynecology3 and Physiology,4 Ralph L. Smith Research CenterUniversity of Kansas Medical Center, Kansas City, Kansas 66160-7338

Department of Biology,5 University of North Carolina at Charlotte, Charlotte, North Carolina 28223

ABSTRACT

Temporal and cell-type specific distribution of transforming growth factor a (TGFa) precursor (proTGFa) was examined inthe mouse uterus during the periimplantation period, and after steroid hormone treatments of ovariectomized adult mice byimmunohistochemistry using antibodies that recognize the precursor forms of the growth factor. These studies were comple-mented by immunoblot analysis of proTGFa in separated uterine cell-type preparations. The specificity of the antibodies usedin these studies was confirmed by use of pancreas or lactating mammary glands from transgenic mice in which mutated proTGFa,lacking recognition sites for proteolytic cleavages, was targeted for expression under a tissue-specific enhancer/promoter. Analysisof histochemical studies revealed accumulation of immunoreactive proTGFa primarily in luminal and glandular epithelial cellson Day 1 of pregnancy or pseudopregnancy followed by little or no accumulation on Days 2 and 3. However, immunoreactiveproTGFa started to reappear in the luminal epithelium on the morning of Day 4 and became more prominent in the afternoon.In pregnant mice, immunostaining persisted in these cells at the implantation sites during the time of attachment reaction (2130h on Day 4), but disappeared by morning of Day 5. Immunostaining appeared to be situated at the apical border of the luminalepithelium. No positive immunostaining could be detected in the nonreceptive uterus on Day 5 or 6 of pseudopregnancy. Con-sistent with the immunohistochemistry results, Western blot analysis detected two species of precursor proteins (14.5 and 17kDa) in isolated luminal epithelial cell-enriched preparations on Day 4, but not on Day 5, of pseudopregnancy. The results suggestthat proTGFa accumulates in the luminal epithelium of the receptive uterus prior to implantation. The effects of ovarian steroidson uterine accumulation of proTGFa were examined in ovariectomized adult mice by immunohistochemistry and immunoblot-ting. Whereas an injection of estradiol-17P (E,) or progesterone (P4) had little or a modest effect on epithelial accumulation ofproTGFa, P4 priming for several days resulted in distinct accumulation of proTGFa in epithelial cells. The superimposition ofan E, treatment on P4 priming showed a biphasic response, with an initial gradual loss of immunostaining through 12 h followedby a return by 24 h of E2 treatment. The combined hormone treatment schedule employed here is similar to the situation ofinducing implantation with E in P4-primed delayed implanting mice. The results suggest a paracrine/"juxtacrine" role for thisgrowth factor in implantation.

INTRODUCTIONIn rodents, uterine receptivity for implantation occurs

only for a limited period during pregnancy or pseudo-pregnancy, or when the animal is treated with progester-one (P4) and estrogen appropriately [1]. In the pregnant orpseudopregnant mouse, the uterus attains the receptive stateonly on Day 4 (the day of implantation). However, by Day5 (as examined by embryo transfer in pseudopregnant mice),the uterus becomes refractory and fails to respond to thepresence of blastocysts for implantation [2]. Ovariectomyprior to implantation ovarian estrogen secretion on themorning of Day 4 results in blastocyst dormancy and de-layed implantation. This condition can be maintained bycontinued P4 treatment. However, a single injection of es-trogen can initiate implantation in P4-treated delayed im-planting mice [3, 4]. The mechanisms by which estrogen

Accepted October 12, 1993.Received August 5, 1993.'This work was supported, in part, by grants from NICHD (HD 12304 and HD

29968). A center grant (HD02528) provided access to Histology Care Facilities forthis study.

'Correspondence: S.K. Dey, Ph.D., Departments of Obstetrics-Gynecology & Phys-iology, University of Kansas Medical Center, MRRC 37/317, Kansas City, KS 66160-7338. FAX: (913) 588-5677.

transforms the P4-treated uterus into a receptive state forblastocyst implantation and the process by which receptiveuterus automatically proceeds to nonreceptivity are stillpoorly understood. The initiation of blastocyst implantationis preceded by uterine luminal closure, which results ininterdigitation of microvilli of trophectoderm and luminalepithelia (apposition) followed by a more intimate contact(adhesion or attachment reaction) [5, 6]. The attachment re-action coincides with increased localized uterine endo-metrial vascular permeability at sites of the blastocyst. Thisincreased vascular permeability is one of the earliest dis-cernible prerequisite events in the process of implantationand can be monitored by an intravenous injection of a bluedye solution (uterine blue reaction) [1].

Although the initial interaction during implantation in-volves "cross-talk" between the trophectodermal epithe-lium of the blastocyst and the luminal epithelium of theuterus, and although this interaction primarily depends onthe coordinated effects of P4 and estrogen, the mechanismsby which this process is regulated by steroids are not clearlyunderstood. The expression of several growth factors andtheir receptors in the periimplantation uterus and/or em-bryo suggests that these growth factors can influence em-bryonic and/or uterine development and functions in an

481

PARIA ET AL.

autocrine/paracrine manner (reviewed in [7-18]). An at-tractive hypothesis is that a growth factor is expressed inthe luminal epithelium to interact with its receptors ex-pressed on the embryonic trophectoderm cell surface inan early interaction that initiates the process of implanta-tion in a paracrine or "juxtacrine" fashion. Alternatively, thetrophectoderm may express the growth factor ligand andthe luminal epithelium, the receptor. Therefore, a detailedanalysis of the temporal and spatial expression of a specificgrowth factor and its receptor in the uterus and embryo isnecessary to better understand the role of ligand-receptorsignaling in implantation. The tight regulation of the epi-dermal growth factor receptor (EGF-R) gene in the blas-tocyst by maternal steroid hormonal status at the time ofimplantation suggests that the EGF-related growth factorsexpressed in the uterus could be important for interactionswith the EGF-R on the trophectoderm for implantation[10, 15,16]. EGF-related growth factors, such as EGF, hep-arin-binding EGF-like growth factor (HB-EGF), transform-ing growth factor ao (TGFa), and amphiregulin are derivedfrom large transmembrane precursors by proteolytic cleav-ages [19-25]. They are structurally related, and their effectsare apparently mediated via binding to the EGF-R [20-30].An early molecular event mediated by these ligands afterbinding to EGF-R is stimulation of subcellular protein ty-rosine phosphorylation, including rapid receptor auto-phosphorylation [31-34]. Numerous early and delayed re-sponses are initiated by this signaling, which eventually leadto cell proliferation and differentiation [35]. There is evi-dence that cells expressing precursor forms of these growthfactors on their cell surface can interact with the EGF-R onneighboring cells in a "juxtacrine" fashion and initiate sig-nal transduction. This has been shown particularly for TGFot/EGF-R interactions [36-38]. Thus, both membrane-an-chored precursor and mature forms of these growth factorsare likely to be operative in influencing biological func-tions. Although previous studies have documented theexpression of TGFt in the mouse uterus during the peri-implantation period [11], the forms of TGFot expressed werenot examined. The present investigation examined the tem-poral and cell-specific accumulation of TGFoe precursors(proTGFat) in the mouse uterus during the periimplanta-tion period and under P4 and/or estrogen stimulation.

MATERIALS AND METHODS

Animals and TreatmentsCD-1 mice (Charles River Laboratories, Raleigh, NC) were

housed in the animal care facility at the University of KansasMedical Center in accordance with NIH standards for thecare and use of experimental animals. Adult female mice(20-25 g, 48 days old) were mated with fertile or vasec-tomized males of the same strain to induce pregnancy orpseudopregnancy, respectively. The morning of finding avaginal plug was designated Day 1 of pregnancy or pseu-

dopregnancy. Mice were killed between 0830 and 0930 hon Days 1-3 of pregnancy or pseudopregnancy (preim-plantation period). On Day 4 (the day of implantation),pregnant mice were killed at 0830-0930 h, 1700 h, and 2130h, while pseudopregnant mice were killed at 0830 and 1700h. On Day 5 (after initiation of implantation), pregnant micewere killed at 0530 and 0900 h, whereas pseudopregnantmice were killed at 0900 h. Pseudopregnant mice were alsokilled at 0900 h on Day 6. In pregnant mice, increased lo-calized endometrial vascular permeability around the uterusat sites of the blastocyst, indicating that the implantationprocess had been initiated, was monitored by an i.v. injec-tion of a Chicago Blue B solution [4]. These blue bands inour CD-1 pregnant mice first became evident between 2130and 2200 h on Day 4. Therefore, implantation (blue bands)and interimplantation sites late on Day 4 or early on Day5 of pregnancy were separated and used for immunohis-tochemistry.

To examine the effects of estrogen and/or P4 on the ac-cumulation of proTGFa in the uterus, mice were ovariec-tomized and rested for 2 wk. Acute effects of ovarian ste-roids were determined by giving them single injections ofestradiol-1713 (E2, 250 ng/mouse) or P4 (2 mg/mouse). Toinduce the state of uterine receptivity for implantation [4],ovariectomized mice received injections of P4 for three days,and on the fourth day they received an injection of P4 orP4 plus E2. Ovariectomized mice receiving vehicle (sesameoil, 0.1 ml/mouse) for four days and killed 12 h later servedas controls. All steroids were dissolved in sesame oil andinjected s.c. Animals were killed at various times after ste-roid injections as indicated in the figure legends.

Production of Antipeptide Antibodies to TGFa PrecursorSite-specific antibodies directed toward peptides with

amino acid sequences within the predicted region of therat TGFao precursor molecule [39] were generated in rab-bits, with synthetic peptides used as immunogens. Theseantibodies were directed toward epitopes within the car-boxy-terminal region of the molecule. An antiserum againsta peptide YCRHEKPSALLKGRTACCHSETW, correspond-ing to amino acids 137-159 in this region, but with a ty-rosine residue added at the amino terminus, was generatedand kindly provided by David Lee, Lineberger Cancer Re-search Center, Chapel Hill, NC [39]. Antibodies to a secondpeptide YCRHEKPSALLKGRTA, corresponding to aminoacids 137-151, but also with a tyrosine residue added atthe amino terminus, were raised by us using two differentconjugation methods. Peptides were synthesized and pu-rified by Multiple Peptide System (San Diego, CA).

Peptide (137-151) was coupled with a carrier protein,keyhole limpet hemocyanin, by use of m-maleimidoben-zoyl-N-hydroxysuccinimide ester [40], or BSA, by use of glu-taraldehyde [41]. The peptide conjugates were mixed withFreund's complete adjuvant (1:1) and injected i.d. into NewZealand white rabbits. All rabbits were bled prior to im-

482

TGFa PRECURSOR IN THE MOUSE UTERUS

munization to obtain preimmune sera. Rabbits were boostedevery 2 wk with the same immunogens, and blood was col-lected 10 days later. The antibody titers were determinedby enzyme-linked immunoassay [40].

Purification of Antibodies

Antipeptide antisera were purified by passage throughAffi-Gel 10 columns (1.5 ml) (Bio-Rad, Richmond, CA) thathad been cross-linked to the appropriate peptides accord-ing to the manufacturer's instructions. In brief, 3 ml of gelslurry was washed with ice-cold water, and the resin wasincubated with 1-2 mg of peptides for 4 h at 4°C in 4 mlof 0.1 M HEPES, pH 7.5. Blocking of the unreacted activeester sites was accomplished by adding 0.1 ml of 1 M eth-anolamine hydrochloride followed by incubation for 1 h atroom temperature. The resin was then transferred to a col-umn and washed sequentially with PBS, 0.1 M sodium bi-carbonate and 0.5 M sodium chloride (pH 7.8), 0.1 M gly-cine (pH 8.0), and PBS. Antiserum was then loaded into thecolumn, and the column was washed extensively with PBS.The column was first eluted with a mixture of 0.1 M sodiumbicarbonate and 0.5 M sodium chloride (pH 7.8). The spe-cifically bound IgG was eluted with 0.1 M glycine hydro-chloride (pH 2.5) into tubes containing 100 l1 of 3 M Tris-HCl (pH 8.8) [42]. Protein contents of these eluates weremeasured in a spectrophotometer at 260 nm. All IgG frac-tions were concentrated by means of an Amicon concen-trator, and the concentrate was dialyzed against PBS.

Immunohistochemistry

Immunohistochemistry followed the procedure de-scribed previously [9-12]. In brief, Bouin's-fixed uterinepieces were dehydrated in ascending grades of ethanol,cleared in xylene, and embedded in paraffin. Paraffin sec-tions (7 pIm) were mounted onto poly-L-lysine-coated glassslides, deparaffinized in xylene, and rehydrated in descend-ing grades of ethanol followed by two final incubations (10min each) in PBS. Sections were then incubated in blockingsolution (10% normal goat serum) for 10 min prior to in-cubation in primary antibodies at 4C for 24 h. Immuno-staining was performed by use of a Zymed-Histostain-SP Kitfor rabbit primary antibody (Zymed Laboratories, San Fran-cisco, CA). This kit used a biotinylated secondary antibody,a horseradish peroxidase-streptavidin conjugate, and a sub-strate chromogen mixture. After immunostaining, sectionswere lightly counterstained with hematoxylin, mounted, andexamined under a bright-field microscope. Red deposits in-dicated the sites of immunoreactive proteins. Control sec-tions were incubated in preneutralized antibodies with a10-fold molar excess of the peptide immunogen, and thesesections did not show any positive staining. Sections of for-malin-fixed pancreas obtained from transgenic mice ex-pressing a mutant proTGFat lacking the recognition sites forproteolytic processing were used as a positive control. This

mutant construct was specifically targeted to the pancreasunder the control of the elastase enhancer/promoter [43].This tissue was kindly provided by David Lee. The mutantproTGFat has been shown to be localized at the surface ofcultured cells and is not cleaved to release mature TGFot[36].

Preparation of Membrane Vesicles

On Days 4 and 5 of pseudopregnancy or after steroidtreatments of the ovariectomized mice, luminal epithelial(LE) cells were isolated by gently squeezing the uterine hornsfrom ovarian to cervical end with a pair of small curvedforceps. LE were recovered as single sheets of cells. Al-though these preparations were enriched with LE cells, somestromal cell contamination could not be avoided. This pro-cedure essentially followed the technique developed byBitton-Casimiri et al. [44]. The uteri from which LE wereremoved are defined as LE-free uteri (Ut-LE). This prepa-ration, however, contained glandular epithelia. LE fromovariectomized oil-treated mice could not be separated andwas not used for membrane preparation. Tissues were ho-mogenized in ice-cold STE buffer (10 mM Tris-HCI, 250 mMsucrose, 2 mM EGTA, pH 7.4) containing 10 pg/ml leu-peptin, 20 ,ug/ml phenylmethylsulfonyl fluoride, and 10 pzg/ml pepstatin in a Polytron P-10 at a dial setting of 6. Ho-mogenates were centrifuged at 900 x g for 10 min at 4°C.Supernatants were saved, and pellets were rehomogenizedand centrifuged. Supernatants were combined together andcentrifuged as above. Pellets were discarded, and super-natants were centrifuged at 10 000 x g for 30 min. Pelletswere then suspended in STE buffer and placed over 8 mlof Percoll solution (7 ml Percoll, 1 ml of 2 M sucrose, 80mM Tris-HCI, 8 mM EGTA, and 32 ml STE buffer) into Corextubes. Membranes were layered on the top of the Percollafter centrifugation at 10 000 x g for 20 min. Membraneswere separated by means of a Pasteur pipette and washedby centrifuging at 10 000 x g for 10 min with 3 vol of 10mM Tris-HCI buffer (pH 7.4) containing 0.15 mM sodiumchloride and 1 mM EGTA. Resulting pellets were then washedtwice with 10 mM Tris-HCl buffer containing 250 mM su-crose and protease inhibitors by centrifuging at 10 000 xg for 10 min. Finally, pellets were suspended in the samebuffer, divided into aliquots, and stored at -700C [45].Membrane vesicles obtained from mammary glands specif-ically targeted for expressing a mutant proTGFa as de-scribed above [43] under the control of whey acidic proteinin the lactating transgenic mice were also processed as above.These tissues were kindly provided by Eric Sandgren fromR.L. Brinster's laboratory, University of Pennsylvania, Phil-adelphia, PA.

Western Blot Analysis

Western blot analysis was performed as described pre-viously [40]. Membrane proteins (160 ILg/sample) were

483

PARIA ET AL.

separated by SDS-polyacrylamide (13%) gel electrophore-sis under a reducing condition and electrophoreticallytransferred to nitrocellulose membranes. Blocking of non-specific binding was accomplished by incubating the mem-branes with 20% powdered milk (Carnation) dissolved inTBST buffer (20 mM Tris-HCI and 0.15 M sodium chloride,pH 7.4, containing 0.05% Tween-20) for 1 h at room tem-perature. Membranes were then incubated in antibodies di-luted in 5% milk solution (1:100) for 24 h at 4°C with con-stant shaking. After incubation, membranes were washedtwice (10 min each) with TBST buffer and then incubatedfor 1 h at room temperature with the secondary antibody(goat anti-rabbit IgG; Sigma Chemical Co., St. Louis, MO)conjugated with alkaline phosphatase (1:1000 dilution in5% milk solution). Membranes were then washed twice inTBST buffer, and alkaline phosphatase reaction was initi-ated with the addition of 0.5 ml of nitro blue tetrazolium(20 mg/ml in 70% dimethylformamide; Eastman Kodak Co.,Rochester, NY) and 1 ml of BCIP (5-bromo-4-chloro-3-in-dolyl phosphate, 5 mg/ml in 50% dimethylformamide[Sigma]) in 100 ml of buffer containing 100 mM Tris-HCl,100 mM sodium chloride, and 5 mM magnesium chloride.The color was developed for 10-20 min in the dark fol-lowed by thorough rinsing with distilled water. Specificityof the reaction was determined by preneutralizing the an-tibody with a 10-fold molar excess of immunogenic pep-tide.

RESULTS

Immunohistochemisty

Antiserum to the synthetic peptide corresponding to 137-159 or 137-151 amino acids was used for immunohisto-chemistry, and similar results were obtained with both ofthe antisera. However, the representative microphoto-graphs used here show results of immunostaining detectedby the antiserum to the peptide corresponding to 137-159amino acids. The red deposits indicate the sites of positiveimmunostaining. Multiple uterine sections from several mice(N = 4-5) in each group were examined.

Accumulation of immunoreactive proTGFa was primar-ily noted in the luminal and glandular epithelia on Day 1of pregnancy (Fig. 1). This immunostaining was greatly re-duced when the primary antibodies were preneutralizedwith the respective peptide immunogens (Fig. 1). Immu-nostaining was rarely detected on Day 2 of pregnancy, andon Day 3 epithelial cells of only a few glands showed pos-itive immunostaining (Fig. 1). However, immunoreactiveproTGFao again began to accumulate in the LE on the morn-ing of Day 4 (Fig. 2), and the accumulation became moreprominent in the afternoon (prior to implantation). In ad-dition, the epithelial cells of a few glands opening to thelumen also showed positive immunostaining at this time.Immunostaining apparently accumulated at the apical bor-der of the epithelial cells. With the initiation of the first

attachment reaction at the implantation sites at 2130 h onDay 4, the positive staining was still evident in the LE (Fig.2). In contrast, with the progression of the implantationprocess by the morning of Day 5, LE staining disappeared,but the decidualizing stroma began to exhibit accumulationof proTGFa (Fig. 2). On the other hand, immunostainingin epithelial cells persisted at the interimplantation sites evenon Day 5 of pregnancy (data not shown). The loss ofproTGFct from luminal epithelial cells adjacent to the im-planting blastocyst may have resulted from the fact that thesecells are about to undergo apoptotic cell death [46]. How-ever, another gene, metallothionein-I, was induced in thesecells at this time [47]. Immunostaining patterns in pseu-dopregnant mice from Day 1 through the evening of Day4 were similar to those seen in pregnant mice (data notshown). However, immunostaining could not be detectedin any uterine cell-types on Days 5 or 6 of pseudopreg-nancy (data not shown), when the uterus was in the non-receptive phase. In ovariectomized mice treated with oil,no immunostaining was detected in the uterus (Fig. 3A).However, an accumulation of immunoreactive proTGFt wasnoted in the luminal and superficial glandular epithelial cellsin P4 -primed uteri (Fig. 3B). The intensity of the stainingshowed a gradual decline with time (2-12 h) after an in-jection of E2 in the P4-primed uterus (Fig. 3, C-E). How-ever, the immunostaining again became distinctly visible inthe epithelial cells 24 h after E2 injection into the P4-primedanimals (Fig. 3F). Immunostaining was not detected in uter-ine cells at any time (2-24 h) after an injection of E2 alone,although immunostaining of modest intensity was noted inepithelial cells 12 h after an injection of P4 (data not shown).Heterogeneity in the distribution of immunoreactivity wasoften noted within the epithelial cells (Figs. 1-3). Pancreasfrom transgenic mice expressing mutated proTGFot that lacksthe recognition sites for proteolytic processing showedpositive immunostaining for proTGFao (Fig. 2).

Western Blot Analysis

Immunoblotting was performed by use of the affinity-purified antibody raised against the BSA-conjugated peptidecorresponding to 137-151 amino acids. Each experiment wasrepeated twice. Membrane proteins isolated from LE or LE-free uteri on Day 4 or 5 of pseudopregnancy were electro-phoresed on SDS-polyacrylamide gels, transferred to nitro-cellulose membranes, and probed with this antipeptideantibody. Membrane vesicles prepared from lactating mam-mary glands of transgenic mice specifically targeted for ex-pressing a mutant proTGFa lacking recognition sites forproteolytic processing were used as a positive control. Twopolypeptide species of approximately 14.5 and 17 kDa weredetected in LE, but not in LE-free uterine (Ut-LE), mem-brane preparations on Day 4 of pseudopregnancy (Fig. 4A).Polypeptides of identical molecular sizes were also de-tected in the transgenic mammary gland samples (Fig. 4A).No such polypeptides were detected either in LE or LE-free

484

TGFa PRECURSOR IN THE MOUSE UTERUS

FIG. 1. Immunohistochemical localization of proTGFa( in the mouse uterus on Days 1-3 of pregnancy. Uteri werecollected on the indicated days of pregnancy, and Bouin's-fixed paraffin sections (7 m) were mounted onto poly-L-lysine coated slides. After deparaffinization and hydration, sections were incubated with primary antibody at 1:2000dilution in PBS for 20 h at 4°C. The well-characterized rabbit polyclonal antiserum, raised against a synthetic pep-tide corresponding to 137-159 amino acid sequences of the precursor region of rat proTGFa, provided by David Lee,was used 39]. Immunostaining employed the avidin-biotin-peoxidase complex system, in which red deposits indi-cate positive immunostaining. Representative microphotographs of uterine sections on Days 1-3 (D,-D3) are shownat x200. DIN exhibits a representative uterine section on Day 1 of pregnancy in which a marked decrease in immu-nostaining was noted after incubation with the primary antibody preneutralized with an excess of the antigenic peptide(x200). LE, luminal epithelium, GE, glandular epithelium; S, stroma; CM, circular muscle; LM, longitudinal muscle.

uterine (Ut-LE) membrane preparations on Day 5 of pseu- tion of P4 or E2 after 3 days of P4 priming also exhibiteddopregnancy (Fig. 4A). The reactivity of the antibody was similar species of polypeptides when immunoblotted withspecific, since neutralization of this antibody with an excess this antibody (Fig. 5). Again, the molecular sizes of theseof the peptide immunogen failed to recognize these spe- species were identical to those detected in mammary glandcies of polypeptides (Fig. 4B). LE membrane preparations membrane preparations of transgenic mice with targetedobtained from ovariectomized mice treated with an injec- expression of the mutant proTGFao (Fig. 5).

485

PARIA ET AL.486

TGFot PRECURSOR IN THE MOUSE UTERUS

DISCUSSION

The major immunohistochemical finding of the presentinvestigation is the temporal and cell type-specific distri-bution of proTGFa in the mouse uterus during the periim-plantation period. Of particular interest is the apical expres-sion of proTGFao in the luminal epithelium on the afternoonof Day 4 of pregnancy that continued through the eveningat the time of attachment reaction. This observation raisesthe possibility of a "juxtacrine" interaction between the lu-minal epithelial proTGFa and trophectodermal EGF-R[11, 13-16] in the initiation of implantation. Furthermore,the observation of the absence of proTGFa in the luminalepithelium at the implantation sites by the early morningof Day 5, but its continued presence in the epithelium atthe interimplantation sites, is consistent with an interactionbetween the embryo and the luminal epithelium. An iden-tical distribution pattern of uterine proTGFa in pregnantand pseudopregnant mice on the morning or afternoon ofDay 4 suggests that the luminal epithelial accumulation ofproTGFa is not dependent upon the presence of an em-bryo. However, its absence in the uterus on Day 5 or 6 ofpseudopregnancy indicates that proTGFat either does notaccumulate or is not expressed during the refractory state.Expression of TGFao mRNA by Northern blot and in situhybridization, as well as distribution of TGFa protein byimmunohistochemistry using an antibody directed to themature region of the molecule, have been examined pre-viously in the periimplantation rodent uterus [11, 48]. In mice,colocalization of the mRNA with the immunoreactive pro-tein was noted in the luminal and glandular epithelial cellson Day 1, followed by a decrease on Day 2, an increase onDay 3, and a further increase on Day 4 [11]. In addition,stromal cells on Days 3 and 4 also expressed TGFa mRNAand protein. Mature TGFa (50 amino acids) is processedfrom a large precursor molecule (159 amino acids) by pro-teolytic action on alanine- and valine-rich sequences ap-parently by an elastase-type protease [21]. Analysis of theresults of the immunohistochemistry using antibodies di-rected to the mature [11] or precursor region suggests thatthe processing of TGFa during early pregnancy is regu-lated. The accumulation of proTGFa primarily in epithelialcells, but not in stromal cells, during the preimplantation

FIG. 2. Immunohistochemical localization of proTGFa in the mouseuterus on Days 4 and 5 of pregnancy. Immunolocalization was performedas described in the legend of Figure 1. Microphotographs of sections ofuteri obtained on Day 4 at 0930 h and 1700 h are shown at x100 (left) andx200 (right). While a representative microphotograph of sections ob-

tained from uteri on Day 4 of pregnancy at 2130 h is shown at x200, thoseobtained from uteri on Day 5 at 0530 h and 0900 h are shown at x100.Positive immunostaining (red) is shown in acinar cells (AC) in a repre-sentative pancreatic section (Pan) obtained from a transgenic mouseshowing targeted expression of mutant proTGFcx lacking recognition sitesfor proteolytic cleavages (x100). IL, islets of Langerhans; LE, luminal epi-thelium; GE, glandular epithelium; S, stroma; E, embryo; DS, decidualiz-ing stroma.

period suggests that the elastase-type proteases could bemore abundant in the stroma.

In the present investigation, the detection of approxi-mately 14.5- and 17-kDa proteins in luminal epithelial cell-enriched membrane preparations obtained from Day 4pseudopregnant mice by immunoblotting using antibodiesdirected to the precursor region of TGFa sequence is con-sistent with earlier observations in other cell types [39].Furthermore, the detection of identical species in the lac-tating mammary glands of transgenic mice expressing mu-tant proTGFea lacking recognition sites for proteolytic cleav-ages establishes that the antibodies used in the present studyspecifically recognized TGFa precursor molecules. The de-tection of these species in the luminal epithelium of pseu-dopregnant mice on Day 4, but not on Day 5, is consistentwith the immunohistochemical results described above.

Estrogen has been shown to induce the expression ofTGFa in the immature mouse uterus, predominantly in ep-ithelial cells, in a dose- and time-dependent manner; P4treatment was not effective in this response [49]. The tem-poral and cell-type-specific accumulation of uterine proTGFaduring the periimplantation period in pregnant mice sug-gested that the processing of TGFa was regulated by ovar-ian steroids. Indeed, analysis of our immunohistochemicalstudies suggests that P4 and/or E2 treatments influenced theaccumulation of proTGFa in the adult ovariectomized mouseuterus. While an injection of E2 had little effect on uterineaccumulation of proTGFoa, an injection of P4 showed a modestaccumulation in epithelial cells. In contrast, accumulationof proTGFa in epithelial cells was prominent after P4 prim-ing for several days. However, the superimposition of anE2 injection on P4 priming appeared to cause gradual lossof immunostaining in epithelial cells through 12 h, withimmunostaining again becoming prominent in these cellsat 24 h of E2 treatment. It is possible that E2 rapidly, buttransiently, stimulated the proteolytic apparatus that partic-ipated in the processing of proTGFa. Since P4 priming ofthe uterus is essential for estrogen to initiate the processof implantation, the initial uterine loss of proTGFa after E2

treatment of the P4-primed mice suggests that secretion ofthe mature form of the growth factor acts on the embryoin a paracrine fashion. Indeed, estrogen treatment has beenshown to cause secretion of high levels of mature TGFainto uterine luminal fluids of immature mice [49]. On theother hand, proTGFoa shown to be accumulated again inthe luminal epithelium by 24 h after E2 treatment of the P4-primed uterus may interact with blastocyst EGF-R in a "jux-tacrine" manner to initiate the process of implantation. Itis to be noted, however, that there were no significantchanges in the levels of two species of protein (14.5 and17 kDa) detected in luminal epithelial cell-enriched mem-brane preparations obtained from ovariectomized mice af-ter steroid hormone treatments. This could be due to ran-dom contamination of the preparations with stromal cells.However, the caveat of these findings is that neither the

487

PARIA ET AL.

FIG. 3. Immunohistochemical localization of uterine proTGFoc in steroid-treated ovariectomized adult mice. Immunolocalization was performed asdescribed in the legend of Figure 1. Mice were ovariectomized and rested for at least 2 wk. One group of mice received daily oil injections for 4 days andwere killed 12 h later. These mice served as controls (A). For hormone treatments, mice were given daily injections of P4 for 3 days, and on the fourthday, they received either an injection of P4 or P4 + E2. The P4-primed mice were killed 12 h later (B), while P4 + E2 mice were killed at 2 (C), 6 (D), 12(E), or 24 (F) h after the final injection. Microphotographs are shown at x100.

488

TGFat PRECURSOR IN THE MOUSE UTERUS

FIG. 5. Western blot analysis of proTGFa in uterine membrane prep-aration from ovariectomized mice after steroid treatment. Ovariectomizedmice were treated with P4 for 3 days. On the fourth day they receivedeither an injection of P4 or P4 + E2. P4-treated mice were killed 12 h later(lane 3), while P4 + E2-treated mice were killed at 12 (lane 2) and 24 h(lane 1) after the last injection. Luminal epithelia were isolated as de-scribed in Materials and Methods. Membrane preparations from lactatingmammary glands (MG) of transgenic mice as described in the legend ofFigure 4 and those from LE of Day 4 pseudopregnant mice (lane 4) wereused as positive controls. The two species of protein (14.5 and 17 kDa)detected are indicated by arrows.

FIG. 4. Western blot analysis of proTGFa in mouse uterine mem-brane preparations. A) Immunoblotting was performed with membranepreparations isolated from luminal epithelia (LE) and LE-free uteri (Ut-LE)on Days 4 and 5 of pseudopregnancy. A membrane preparation frommammary glands (MG) from lactating transgenic mice expressing mutantproTGFa lacking recognition sites for proteolytic cleavages was used as apositive control. Immunoblotting was performed with a rabbit polyclonalantiserum raised to a synthetic peptide corresponding to 137-151 aminoacids of the precursor region of rat proTGFa. The two species of protein(14.5 and 17 kDa) detected are indicated by arrows. B) Specificity of theantibody was confirmed by immunoblotting the same membrane prepa-rations with primary antibody after neutralization with an excess of theimmunogenic peptide. Std; molecular weight markers.

immunohistochemistry nor the immunoblotting experi-ments used in the present investigation were quantitative,and thus, the differential effects of ovarian steroids on theuterine status of proTGFoa should be interpreted with res-ervation.

Overall, the results suggest that proTGFoa shows pref-erential accumulation in the luminal epithelial cells in thereceptive uterus prior to implantation, and could be usedas a marker for determining uterine receptivity for implan-tation. Although an interaction between luminal epithelialproTGFt and trophectodermal EGF-R for implantation seemspossible, the recent findings of apparently normal embryodevelopment and implantation in TGFao "knockout" miceraises concern regarding the physiological relevance of TGFotin reproduction and development [50, 51]. One simple ex-planation could be that this growth factor is not an impor-tant participant in embryo development and implantation.Alternatively, TGFoa may be operative in these processes,but in "knockout" mice its deficiency was perhaps com-pensated by other EGF-related growth factors. Indeed, ourrecent findings document temporal and cell-specificexpression of HB-EGF in the mouse uterus during the peri-implantation period (Das et al., unpublished results). EGFwas also shown to be expressed in uterine epithelial cells,but only on Day 1 of pregnancy [52, 53]. This finding raisesdoubts about EGF's role in implantation. Whether amphi-regulin is expressed in the uterus and/or embryo is a sub-ject of future investigation. In any case, TGFa does not ap-pear to be an essential molecule for embryonic developmentand implantation, although it still could play an importantrole under normal conditions.

489

490 PARIA ET AL.

ACKNOWLEDGMENTS

Thanks are due to Brenda Boese for technical assistance and to Ansley Spencerfor assistance in manuscript preparation.

REFERENCES

1. Psychoyos A Endocrine control of egg implantation. In: Greep RO, Astwood EG,Geiger SR (eds.), Handbook of Physiology. Washington, DC: American Physio-logical Society; 1973: 187-215.

2. Paria BC, Huet-Hudson YM, Dey SK. Blastocyst's state of activity determines thewindow of implantation in the mouse receptive uterus. Proc Natl Acad Sci, USA1993; 90:10159-10162.

3. Yoshinaga K, Adams CE. Delayed implantation in the spayed, progesterone-treatedadult mouse. J Reprod Fertil 1966; 12:593-595.

4. Huet YM, Dey SK. Role of early and late oestrogenic effects on implantation inthe mouse. J Reprod Fertil 1987; 81:453-458.

5. Kirby DRS, Cowell TP. Trophoblast-host interactions. In: Fleischmeyer R, Bil-lingham RE (eds.), Epithelial-Mesenchymal Interactions. Baltimore, MD: Williamsand Wilkins Co.; 1968: 64-77.

6. Enders AC, Schlafke S. Cytological aspects of trophoblast-uterine interactions inearly implantation. Am J Anat 1969; 125:1-25.

7. Murphy LJ, Murphy LC, Friesen HG. Estrogen induces insulin-like growth factor-I expression in the rat uterus. Mol Endocrinol 1987; 1:445-450.

8. Lingham RB, Stancel GM, Loose-Mitchell DS. Estrogen induction of the epidermalgrowth factor receptor mRNA Mol Endocrinol 1988; 2:230-235.

9. Tamada H, McMaster T, Flanders KC, Andrews GK, Dey SK Cell type-specificexpression of transforming growth factor-{1 in the mouse uterus during theperiimplantation period. Mol Endocrinol 1990; 4:965-972.

10. Kapur S, Tamada H, Dey SK, Andrews GK Expression of insulin-like growth fac-tor-I and its receptor in the peri-implantation mouse uterus, and cell-specificregulation of insulin-like growth factor-I gene expression by estradiol and pro-gesterone. Biol Reprod 1992; 46:208-219.

11. Tamada H, Das SK, Andrews GK, Dey SK. Cell-type-specific expression of trans-forming growth factor-a in the mouse uterus during the peri-implantation pe-riod. Biol Reprod 1991; 45:365-372.

12. Das SK, Flanders KC, Andrews GK, Dey SK. Expression of transforming growthfactor-y isoform (2 and 133) in the mouse uterus: analysis of the preimplantationperiod and effects of ovarian steroids. Endocrinology 1992; 130:3459-3466.

13. Paria BC, Dey SK Preimplantation embryo development in vitro: cooperativeinteractions among embryos and role of growth factors. Proc Natl Acad Sci USA1990; 87:4756-4760.

14. Rappolee DA, Brenner CA, Schultz R, Mark D, Werb Z. Developmental expressionof PDGF, TGF-a, and TGF-P genes in preimplantation mouse embryos. Science1988; 242:1823-1825.

15. Wiley LM, Jie-Xin W, Harari I, Adamson ED. Epidermal growth factor receptormRNA and protein increase after the four-cell preimplantation stage in murinedevelopment. Dev Biol 1992; 149:247-260.

16. Paria BC, Das SK, Andrews GK Dey SK. Expression of the epidermal growthfactor receptor gene is regulated in mouse blastocysts during delayed implan-tation. Proc Nad Acad Sci USA 1993; 90:55-59.

17. Heyner S, Smith RN, Schultz GA Temporally regulated expression of insulin andinsulin-like growth factors and their receptors in early mammalian development.BioEssays 1989; 11:171-176.

18. Paria BC, Jones KL, Flanders KC, Dey SK. Localization and binding of transform-ing growth factor-y isoforms in mouse preimplantation embryos and in delayedand activated blastocysts. Dev Biol 1992; 151:91-104.

19. Rail LB, Scott J, Bell GI, Crawford RJ, PenschowJD, Niall HD, Coghlan JP. MousepreproEGF synthesis by the kidney and other tissues. Nature 1985; 313:228-231.

20. Marquardt H, Hunkapillar MW, Hood LE, Twardzik DR, DeLarco JE, StevensonJR, Todaro GJ. Transforming growth factors produced by retrovirus transformedrodent fibroblasts and human melanoma cells. Amino acid sequence homologywith epidermal growth factor. Proc Natl Acad Sci USA 1983; 80:4684-4688.

21. Derynck R, Roberts AB, Winkler ME, Chen EY, Goeddel DV. Human transforminggrowth factor-a: precursor structure and expression in E. coli. Cell 1984; 38:287-297.

22. Higashiyama S, Abraham JA, Miller J, Fiddes J, Klagsbrun M. A heparin-bindinggrowth factor secreted by macrophage-like cells that is related to EGF. Science251:936-939.

23. Higashiyama S, Lau K, Besner GE, Abraham JA, Klagsbrun M. Structure of theheparin-binding EGF-like growth factor. Multiple forms, primary structure andglycosylation of the mature protein. J Biol Chem 1992; 267:6205-6212.

24. Shoyab M, Plowman GD, McDonald VL, Bradley JG, Todaro GJ. Structure andfunction of human amphiregulin: a member of the epidermal growth factor fam-ily. Science 1989; 243:1074-1076.

25. Plowman GD, Green JM, McDonald VL, Neubauer MG, Disteche CM, Todaro GJ,Shoyab M. The amphiregulin gene encodes a novel epidermal growth factor-related protein with tumor inhibitory activity. Mol Cell Biol 1990; 10:1969-1981.

26. Marquardt H, Hunkapillar MW, Hood LE, Todaro GJ. Rat transforming growthfactor I: structure and relation to epidermal growth factor. Science 1984; 223:1079-1082.

27. Brown JP, Twardzik DR, Marquardt H, Todaro GJ. Vaccinia virus encodes a4hbapoly-peptide homologues to epidermal growth factor and transforming growthfactor. Nature 1985; 313:491-492.

28. Pike LJ, Marquardt H, Todaro GJ, Gallis B, Casnellie JE, Bornstein P, Krebs EG.Transforming growth factor and epidermal growth factor stimulate the phos-phorylation of a synthetic, tyrosine-containing peptide in a similar manner. J BiolChem 1982; 257:14628-14631.

29. DeLarco JE, Todaro, GJ. Sarcoma growth factor (SGF): specific binding to epi-dermal growth factor (EGF) membrane receptors. J Cell Physiol 1980; 102:267-277.

30. Margolis BL, Lax I, Kris R, Dombalagian M, Honegger AM, Howk R Givol D,Ullrich A, Schlessinger J. All autophosphorylation sites of epidermal growth fac-tor receptor and HER2/neu are located in their carboxyl-terminal tails. J BiolChem 1989; 264:10667-10671.

31. Cohen S, Carpenter G, King LJr. Epidermal growth factor-receptor-protein kinaseinteractions. Copurification of receptor and epidermal growth factor-enhancedphosphorylation activity. J Biol Chem 1980; 255:4834-4842.

32. Downward J, Parker P, Waterfield MD. Autophosphorylation sites on the epider-mal growth factor receptor. Nature 1984; 311:483-485.

33. Hunter T, Cooper JA Epidermal growth factor induces rapid tyrosine phos-phorylation of proteins in A431 human tumor cells. Cell 1981; 24:741-752.

34. Carpenter G, King L Jr, Cohen S. Rapid enhancement of protein phosphorylationin A431 cell membrane preparations by epidermal growth factor. J Biol Chem1979; 254:4884-4891.

35. Schlessinger J. Allosteric regulation of the epidermal growth factor receptor ki-nase. J Cell Biol 1986; 103:2067-2072.

36. Wong ST, Winchell LF, McCune BK, Earp HS, Teixido J, Massague 3, Herman B,Lee DC. The TGF-a( precursor expressed on the cell surface binds to the EGFreceptor on adjacent cells, leading to signal transduction. Cell 1989; 56:495-506.

37. Massague J. Transforming growth factor-a. A model for membrane-anchored growthfactors. J Biol Chem 1990; 265:21393-21396.

38. Derynck R. Transforming growth factor-a. Mol Reprod Dev 1990; 27:3-9.39. Gentry LE, Twardzik DR, Lim GJ, Ranchalis JE, Lee DC. Expression and charac-

terization of transforming growth factor-a precursor protein in transfected mam-malian cells. Mol Cell Biol 1987; 7:1585-1591.

40. Deb S, Hashizume K, Boone K, Southard JN, Talamantes F, Rawitch A, Soares MJ.Antipeptide antibodies reveal structural and functional characteristics of rat pla-cental lactogen-II. Mol Cell Endocrinol 1989; 63:45-56.

41. Avrameas S, Ternyck T. The cross-linking of proteins with glutaraldehyde and itsuse for the preparation of immunoadsorbents. Immunocytochemistty 1969; 6:53-66.

42. Flanders KC, Roberts AB, Ling N, Fleurdelys BE, Sporn MB. Antibodies to peptidedeterminants in transforming growth factor-j and their applications. Biochem-istry 1988; 27:739-746.

43. Sandgren EP, Luetteke NC, Palmiter RD, Brinster RL, Lee DC. Overexpression ofTGF-a in transgenic mice: induction of epithelial hyperplasia, pancreatic meta-plasia, and carcinoma of the breast. Cell 1990; 61:1121-1135.

44. Bitton-Casimiri V, Rath NC, Psychoyos A A simple method for separation andculture of rat uterine epithelial cells. J Endocrinol 1977; 73:537-538.

45. Nakamura T, Tomomura A, Noda C, Shimoji M, Ichihara A Acquisition of a -andrenergic response by adult rat hepatocytes during primary culture. J BiolChem 1983; 258:9283-9289.

46. Parr EL, Parr MB. Epithelial cell death during rodent embryo implantation. In:Yoshinaga K (ed.), Blastocyst Implantation. Serono Symposia. Boston, MA: AdamsPublishing Group Ltd.; 1989: 105-115.

47. McMaster MT, Dey SK, Andrews GK Association of monocytes and neutrophilswith early events of blastocyst implantation in the mouse. J Reprod Fertil 1993;(in press).

48. Han VKM, Hunter III S, Pratt RM, Zendagui JG, Lee DC. Expression of rat trans-forming growth factor-a mRNA during development occurs predominantly inmaternal decidua. Mol Cell Biol 1989; 7:2335-2343.

49. Nelson KG, Takahashi T, Lee DC, Luetteke NC, Bossert NL, Ross K, Eitzman BE,McLachlan JA Transforming growth factor-a is a potential mediator of estrogenaction in the mouse uterus. Endocrinology 1993; 131:1657-1664.

TGFot PRECURSOR IN THE MOUSE UTERUS

50. Luetteke NC, Qiu TH, Peiffer RL, Oliver P, Smithies O, Lee DC. TGF-a deficiencyresults in hair follicle and eye abnormalities in targeted and waved-i mice. Cell1993; 73:263-278.

51. Bruce Mann G, Fowler KJ, Gabriel A, Nice EC, Linsay Williams R, Dunn AR. Micewith a null mutation of the TGF-a gene have abnormal skin architecture, wavyhair, and curly whiskers and often develop corneal inflammation. Cell 1993; 73:249-261.

52. Huet-Hudson YM, Chakraborty C, De SK, Suzuki Y, Andrews GK, Dey SK. Estro-gen regulates synthesis of EGF in mouse uterine epithelial cells. Mol Endocrinol1990; 4:510-523.

53. Huet-Hudson YM, Andrews GK, Dey SK. Epidermal growth factor and pregnancyin the mouse, In: Heyner S, Wiley LM (eds.), Early Embryo Development andParacrine Relationships, UCLA Symposium. New York: Wiley-Liss; 1990; 117:125-136.

491