Uterine-Associated Serine Protease Inhibitors Stimulate Deoxyribonucleic Acid Synthesis in Porcine Endometrial Glandular Epithelial Cells of Pregnancy 1

BIOLOGY OF REPRODUCTION 61, 380±387 (1999)

Lokenga Badinga, Frank J. Michel, and Rosalia C.M. Simmen2 Animal Molecular and Cell Biology Interdisciplinary Concentration, Department of Animal Science, University of Florida, Gainesville, Florida 32611-0910

ABSTRACT Consistent with this, uteri from mammalian species with distinct placentation types express common classes of pro- Protease inhibitors are major secretory components of the tease inhibitors (e.g., tissue inhibitors of metalloproteases, Downloaded from https://academic.oup.com/biolreprod/article/61/2/380/2734487 by guest on 24 September 2021 mammalian uterus that are thought to mediate pregnancy-as- TIMPs) as well as distinct ones (e.g., secretory leukocyte sociated events primarily by regulating the activity of proteolytic protease inhibitor, SLPI, and uterine plasmin/trypsin inhib- enzymes. In the present study, we examined the mitogenic potentials of two serine protease inhibitors, namely secretory leu- itor, UPTI) [4, 8, 9]. Since embryos from all species, re- kocyte protease inhibitor (SLPI) and uterine plasmin/trypsin in- gardless of placentation type, exhibit invasive properties hibitor (UPTI) in primary cultures of glandular epithelial (GE) when placed into ectopic sites [10], the limiting of blasto- cells isolated from early pregnant (Day 12) pig endometrium, cyst invasiveness, albeit to varying extents, is most likely using the [3H]thymidine incorporation assay. Purified porcine an important function of pregnancy-associated uterine anti- SLPI (pSPLI), porcine UPTI (pUPTI), or recombinant human SLPI proteases. (rhSLPI), all of which exhibited anti-trypsin activity, increased The recent demonstration that TIMPs 1 and 2 exhibit (p Ͻ 0.05) labeled thymidine incorporation into DNA of serum- growth-promoting activity in various mammalian cell types, deprived GE cells when tested at a range of 10–1000-ng/ml con- independent of their anti-proteolytic activity, points to a centrations. Polyclonal antibodies directed against either hSLPI potentially novel function(s) for anti-proteases in pregnancy or pSLPI abrogated the effect of SLPI. Co-addition of pSLPI and events [11±13]. The mechanism(s) underlying the growth pUPTI increased DNA synthesis in these cells to a level higher factor activity of TIMPs is presently unclear; however, re- (p Ͻ 0.05) than that observed with either protease inhibitor. The cent studies have indicated that this may occur through spe- glycosaminoglycan heparin, which has been previously shown ci®c cell surface receptors [13], similar to pathways utilized to increase the anti-protease activity of SLPI, exhibited a ten- by classical growth factors. The possibility that distinct cell to enhance SLPI and UPTI induction of cellular types of the uterine endometrium represent target sites for (0.08 ؍ dency (p DNA synthesis. Reverse transcription-polymerase chain reaction the growth-promoting action of TIMPs or any other pro- indicated that the messenger RNAs for both protease inhibitors tease inhibitors has not been previously explored. were present in the endometrium throughout pregnancy and, The pig uterus of pregnancy synthesizes several low- within this tissue, in GE cells to a greater extent (p Ͻ 0.05) than molecular-weight anti-proteases, which include SLPI [8], in stromal fibroblastic cells. Results demonstrate that, in addition to their well-documented anti-protease activities, SLPI and UPTI [9], and TIMPs 1 and 2 [14]. The apparent restricted UPTI may constitute autocrine growth promotants for the uter- uterine expression of SLPI and UPTI, but not of TIMPs, to ine epithelium. These data suggest a novel mechanism whereby species with noninvasive placentation, implicates their locally produced protease inhibitors may modulate periimplan- function in the maintenance of an intact utero-placental in- tation events and embryo-maternal communication. terface [15, 16]. Additionally, unlike that of TIMPs, ex- pressions of SLPI and of UPTI in the pig uterine endo- INTRODUCTION metrium are induced by the presence of periimplantation conceptuses, and more speci®cally for SLPI, by their se- The mammalian uterus of early pregnancy is a major site cretory products [14, 17]. This modulation, a form of em- of synthesis for different classes of protease inhibitors [1± bryo-maternal communication, probably represents a mech- 4]. This high level of production probably represents a ma- anism that potentiates the uterine functions of SLPI and jor defense mechanism through which it fends off proteases UPTI to positively affect early pregnancy events. The na- that originate from within itself, the blastocyst, the feto- ture of these uterine function(s), distinct from inhibition of placental unit, or in®ltrating neutrophils and mast cells that protease activity, has not been elucidated. accompany implantation events [5, 6]. Although the action The present study examined whether uterine-derived of proteases is essential for uterine tissue remodeling and SLPI and UPTI, similar to TIMPs 1 and 2, exhibit addi- the initiation of interaction with the implanting conceptus, tional functions distinct from their protease inhibitor activ- this activity must be carefully controlled. Thus, a biological ities and, if so, whether these effects occur in an autocrine rationale exists for the presence of a wide variety of pro- manner. Towards this end, the mitogenic potentials of pu- tease inhibitors with overlapping substrate speci®cities ri®ed preparations of these porcine uterine proteins, alone within the uterine micro-environment. Indeed, it has been or in combination, was tested in vitro, using primary cul- suggested that the extent of embryo-maternal interaction, tures of glandular epithelial cells isolated from endometri- and hence the nature of placentation, is dependent upon the um of early pregnancy. levels and types of proteases and protease inhibitors [7]. MATERIALS AND METHODS Accepted March 9, 1999. Received November 12, 1998. Materials 1This work was supported in part by USDA grants 94-37205-1164 and 96-35205-3745, and NIH HD21961 to R.C.M.S. This is Journal Series No. The polymerase chain reaction (PCR) optimizer kit and R-06687 from the Florida Agricultural Experiment Station. cDNA cycle kit were purchased from Invitrogen (San Di- 2Correspondence. FAX: 352 392 7652; e-mail: [email protected]fl.edu ego, CA). Taq DNA polymerase was obtained from Boeh- 380 UTERINE PROTEASE INHIBITORS AS EPITHELIAL MITOGENS 381 ringer Mannheim (Indianapolis, IN). [␣-32P]Deoxycytidine gel and visualized by ethidium bromide staining. In some triphosphate (SA 3000 Ci/nmol) and BioTrans nylon mem- experiments, the products were blotted onto nylon membranes (0.2 ␮m) were purchased from ICN Radiochemicals branes and analyzed by Southern blotting, using the pUPTI (Irvine, CA), and the nick-translation kit was from Amer- cDNA probe. For analysis of PCR band intensities, pho- sham Corp (Arlington Heights, IL). TRIzol and culture me- tographs of ethidium bromide-stained gels or autoradio- dia were from GIBCO BRL (Gaithersburg, MD). Trypsin grams were scanned at high resolution, and the integrated (type XIII, L-1-tosylamino-2-phenylethylchloramethyl ke- density of the band was calculated using the Alpha Imager tone-treated), trypsin inhibitor (type I-S), N-benzoyl-DL-ar- 2000 Documentation & Analysis System (Alpha Innotech ginine-p-nitroanilide, and heparin were purchased from Sig- Corp., San Leandro, CA). The intensities of the SLPI and ma Chemical Co. (St. Louis, MO). Recombinant human UPTI signals were normalized to that of the ␤2-microglob- SLPI (rhSLPI) was obtained from R&D Systems Inc (Min- ulin internal control. The PCR-generated fragments were neapolis, MN). also puri®ed from primers using PCR wizard prep columns (Promega, Madison, WI) and ligated into the TA cloning Cell Culture and RNA Extraction vector pCR2.1 (Invitrogen) to isolate pUPTI and pSLPI Downloaded from https://academic.oup.com/biolreprod/article/61/2/380/2734487 by guest on 24 September 2021 clones for nucleotide sequence analyses. The cDNA clones Uterine GE and stromal (ST) cells were isolated from were sequenced in both directions by the Sanger dideoxy- Day 12 pregnant pig endometrium as previously described 6 nucleotide chain termination method [20]. Sequence infor- [18]. Cells were resuspended (0.5 ϫ 10 cells/ml) in RPMI mation was analyzed using the Sequence Analysis Software 1640 medium containing 10% heat-inactivated fetal bovine package from the Genetics Computer Group (Madison, serum and 0.25 U/ml insulin, and cultured at 37ЊCina95% WI). The entire nucleotide sequences of pSLPI (500 bp) air:5% CO2 environment. Cells remained undisturbed (i.e., and pUPTI (479 bp) PCR fragments (data not shown) were without medium change) for the ®rst three days after plat- 100% identical to those previously reported [8, 15]. ing. Thereafter, each culture was replenished with fresh RPMI 1640 containing serum every two days until cells reached 100% con¯uence. Total cellular RNA was isolated Northern Blot Analysis from con¯uent cells using TRIzol reagent according to the manufacturer's instructions. After chloroform extraction, Total cellular RNA from porcine endometrial tissue was RNA was recovered from the aqueous phase and precipi- extracted by the guanidinium thiocyanate-phenol-chloro- tated with isopropanol. The RNA pellet was dissolved in form method. Thirty micrograms of total RNA was sub- sterile water and quanti®ed by spectrophotometry. jected to Northern blot analysis as previously described [8]. Membranes were hybridized with 32P-labeled pSLPI or pUPTI cDNA fragments, according to previously described Reverse Transcription (RT)-Polymerase Chain Reaction procedures [8]. (PCR), Southern Blot, and Sequence Analyses of PCR Products Purification of Porcine SLPI and Porcine UPTI Five micrograms each of total RNA isolated from GE and ST cells were reverse-transcribed as described previ- Samples of allantoic ¯uid (AF) collected from Day 60 ously [19]. Each RT reaction mixture was extracted with pregnant pigs were clari®ed by low-speed centrifugation, phenol-chloroform and then precipitated with ethanol at and proteins were precipitated with 80% ammonium sulfate Ϫ80ЊC overnight. The cDNA pellet was recovered by cen- at 4ЊC for 1 h. The pellet was recovered by centrifugation trifugation, resuspended in 20 ␮l of sterile water, and frozen (50 000 ϫ g) for 20 min and redissolved in 20 mM Tris- at Ϫ20ЊC until analyzed. HCl buffer (pH 8.0) to a ®nal volume of 32 ml/500 ml of The PCR analysis was subjected to preliminary valida- starting volume of AF. Basic proteins were enriched by al- tion and optimization steps to ensure that the assay was lowing them to bind to carboxymethyl (CM) cellulose (Bio- linear relative to the amount of input RNA and that ampli- Rad, Richmond, CA) and were then eluted with 20 mM ®cation of the products was in the exponential phase. These Tris-HCl buffer (pH 8.0) over a 0±1 M linear NaCl gradi- steps included 1) determining the optimal Mg2ϩ concentra- ent. Fractions were evaluated for the presence of SLPI by tion and pH for each DNA template-primer combination, a speci®c RIA, following protocols previously described using the PCR optimizer kit; 2) varying the template con- [21] and by a trypsin inhibition assay (see below). The centration, once the optimal buffer was identi®ed; and 3) active fractions were pooled and then size-fractionated by adjusting the number of cycles under conditions of optimal gel ®ltration over a Sephadex G-50 column previously buffer and template amount. For both SLPI and UPTI, the equilibrated with Tris buffer (20 mM, pH 8.0), at a ¯ow determined optimal PCR conditions were 1) 2 min at 95ЊC, rate of 40 ml/h. The eluted fractions were assayed for SLPI 1.5 min at 55ЊC, and 2 min at 72ЊC; 35 cycles; 2) buffer immunoreactivity and anti-trypsin activity, respectively; G (60 mM Tris-HCl [pH 9.0], 15 mM ammonium sulfate, those exhibiting peak activities in both assays were exam- 2.5 mM MgCl2); and 3) 0.3 ␮g of each primer and 2 ␮lof ined for purity by SDS-PAGE and then stained with Coom- diluted (1:10) RT reaction. The sequences of gene-speci®c assie blue or silver dye. Highly puri®ed fractions were oligonucleotide primers were designed from the published pooled and subjected to NH2-terminal amino acid sequenc- cDNA sequences of porcine SLPI (pSLPI) and porcine ing (see below). UPTI (pUPTI), respectively [8, 15], and were as follows: During the puri®cation of pSLPI, fractions exhibiting 1) pSLPI, forward: 5Ј-CGTGGAAGGTGCTGAAAATG- anti-trypsin activity but lacking immunoreactivity towards 3Ј; reverse: 5Ј-CATGTCCTGGAAGCCTACAG-3Ј; and 2) anti-SLPI antiserum, were identi®ed from CM cellulose pUPTI, forward: 5Ј-AGCCCTCCTCCTCCTCCTGG-3Ј; chromatography. These fractions were pooled and subjected reverse: 5Ј-CACTCAGAATGGGATGTCGA-3Ј. The ex- to further puri®cation by size-fractionation over a Sephadex pected product sizes were 500 base pairs (bp) and 479 bp, G-50 column previously equilibrated with Tris buffer (20 respectively, for pSPLI and pUPTI. Aliquots of the PCR mM, pH 8.0). The fractions exhibiting highest anti-trypsin reactions (15 ␮l) were electrophoresed in a 1.5% agarose activity were evaluated for purity by SDS-PAGE. The ®nal 382 BADINGA ET AL. con®rmation of the puri®ed protein's identity as UPTI was determined by NH2-terminal amino acid analysis.

Amino Acid Sequencing Active fractions containing SLPI and UPTI, respectively, were separated on a one-dimensional SDS-PAGE gel, and proteins were electrophoretically transferred onto polyvi- nylidene ¯uoride (PVDF) membranes by the method of Towbin [22]. The membranes were rinsed in water, stained with 0.01% Coomassie Blue dye in 50% methanol, and destained in 50% methanol. The protein bands of desired molecular weights were sequenced using an Applied Bio- systems (Foster City, CA) model 470A Gas Phase Protein Downloaded from https://academic.oup.com/biolreprod/article/61/2/380/2734487 by guest on 24 September 2021 Sequencer with an on-line analytical HPLC system, at the Interdisciplinary Center for Biotechnology Research Pro- tein Core, University of Florida.

Trypsin Inhibition Assay The ability of pSLPI, pUPTI, and rhSLPI to inhibit the enzymatic activity of trypsin was examined according to previously described protocols [15]. Brie¯y, known concentrations of these proteins, as measured by the Bio-Rad assay (Bio-Rad Technical Bulletin) were incubated with 100 ␮l of trypsin (20 ␮g/ml) at room temperature for 15 min. The total volume of the reaction was adjusted to 500 ␮l by addition of 250 ␮l of 50 mM Hepes buffer (pH 7.0) and 50 ␮lof50mMN-benzoyl-DL-arginine-p-nitroanilide. After an additional 25-min incubation at 37ЊC, the reaction was terminated by the addition of 500 ␮l of soybean trypsin inhibitor (10 ␮g/ml), and the absorbance at 410 nm was measured. The inhibitory activity in each test sample was expressed as a percentage of the residual trypsin activity, with the activity of trypsin in the absence of added SLPI or UPTI considered to be 100%.

[3H]Thymidine Incorporation into Cellular DNA Mitogenic responses of uterine endometrial GE cells to FIG. 1. Gel electrophoretic analysis of pSLPI (A) and pUPTI (B). The SLPI and UPTI, alone or in combination, and in the pres- purified proteins from Day 60 pregnant pig allantoic fluid, were analyzed ence or absence of heparin were examined via the incor- by one-dimensional SDS-PAGE. poration of [3H]thymidine into DNA as previously described [23]. Con¯uent GE cells were washed twice in ϫ pig or dose ϫ pig were used to test the ®xed effects of Hanks' Balanced Salt Solution and preconditioned in se- treatment and dose, respectively. The mathematical model rum-free medium for 24 h. Medium was then changed, and for heparin effect included treatment (control vs. heparin), the cells were incubated with or without the indicated pro- protease inhibitor (control, SLPI, UPTI, SLPI ϩ UPTI), tease inhibitors for an additional 24 h. At 4 h before the treatment ϫ protease interaction, pig, treatment ϫ pig in- end of the incubation period, cells were pulse-labeled with 3 teraction, protease inhibitor ϫ pig interaction, and treatment 2 ␮Ci of [ H]thymidine. Cells were rinsed with PBS (pH ϫ protease inhibitor ϫ pig interaction. Effects of pig and 7.4), DNA was precipitated with 5% trichloroacetic acid, interactions involving pig were analyzed as random effects and incorporated radioactivity was measured by liquid scin- and used as error terms for appropriate upstream ®xed ef- tillation counting. fects. Signi®cant effects (p Ͻ 0.05) of dose or treatments were separated by preplanned orthogonal contrasts. Results Statistical Analysis represent means Ϯ SEM of two to three independent ex- Differences in mitogenic responses due to protease in- periments, with each experiment representing 3 separate hibitor and/or heparin treatments and numerical data from cultures of endometrial GE cells isolated from an individual densitometric analysis of PCR or hybridization bands were gilt at Day 12 of pregnancy. examined by least-squares ANOVA using the General Lin- ear Models procedure of the Statistical Analysis System RESULTS [24]. The effects of protease inhibitors were analyzed using a mathematical model that included treatment (control, Purification of Porcine SLPI and UPTI SLPI, UPTI, SLPI ϩ UPTI), pig, and treatment ϫ pig in- Porcine SLPI and UPTI were puri®ed from Day 60 preg- teraction. The dose effects were tested using a statistical nant pig AF by successive gel ®ltration over CM cellulose model that included dose, pig, and pig ϫ dose interaction. and Sephadex G-50 columns, respectively. In each case, In both mathematical models, the pig effect was analyzed eluted fractions containing the desired proteins were iden- as a random effect, and the interactions between treatment ti®ed by evaluating their reactivity towards anti-SLPI an- UTERINE PROTEASE INHIBITORS AS EPITHELIAL MITOGENS 383

FIG. 2. Induction of [3H]thymidine incorporation into cellular DNA by SLPI. Prima- ry cultures of serum-deprived GE cells were incubated with increasing concentrations of pSLPI (A) or rhSLPI (B) for 20 h and then pulse-labeled with [3H]thymidine for 4 h. Results represent the mean Ϯ SEM of three independent experiments, with each experiment representing 3 separate cultures of GE cells isolated from an individual gilt on Day 12 of pregnancy. The speciﬁcity of SLPI effects was examined after preincubation of pSLPI (100 ng/ml) and rhSLPI (100 ng/ml) with anti-pSLPI (30 ␮g/ well; C) and anti-hSLPI (15 ␮g/well; D) an-

tisera, respectively, for 2 h before treat- Downloaded from https://academic.oup.com/biolreprod/article/61/2/380/2734487 by guest on 24 September 2021 ment implementation. Means without a common superscript differ (p Ͻ 0.05).

tiserum by RIA [21] and by their ability to inhibit trypsin. Mitogenic Effects of SLPI and UPTI in Uterine GE Cells Fractions corresponding to peak activity were further examined for homogeneity by one-dimensional SDS-PAGE. To evaluate the mitogenic potentials of SLPI and UPTI The puri®ed pSLPI protein used in all in vitro assays con- towards uterine epithelial cells, primary cultures of GE cells at 100% con¯uence were incubated in serum-free me- tained two bands migrating slightly below the Mr 20 000 and above the M 14 200 standards (Fig. 1A). NH -terminal dium for 24 h and then treated with increasing concentra- r 2 tions (10±1000 ng/ml) of pSLPI, rhSLPI, and pUPTI for amino acid sequence analyses of these two proteins re- 3 vealed identical sequences within the ®rst 17 amino acid 20 h. The cells were then pulse-labeled with [ H]thymidine residues (band 1: ALKGGACPPRKIVQCLRYEK; band 2: for the last 4 h of the incubation period. All three protease ALKGGACPPRKIVQCLRYE). These sequences corre- inhibitors increased (p Ͻ 0.05) DNA synthesis in GE cells over control (untreated) cells at the different concentrations sponded exactly to those predicted from the nucleotide se- 3 quence of cloned porcine uterine SLPI cDNA [8]. The pu- tested (Figs. 2 and 3). SLPI stimulation of [ H]thymidine ri®ed pUPTI protein migrated as a doublet on a one-di- incorporation into DNA of GE cells was completely abro- mensional SDS-PAGE gel (Fig. 1B), with estimated mo- gated upon preincubation of pSLPI and rhSLPI with anti- lecular weights for the top and bottom bands of 17 000 pSLPI (IgG fraction; 30 ␮g/well) and anti-hSLPI (IgG frac- (intact) and 14 000 (truncated), respectively. The NH2-terminal amino acid sequences of these two proteins (top: VRAGPPAFCREPPYTGP; bottom: VRAGPPAFCREP- PYTG) corresponded exactly to those predicted from the nucleotide sequence of cloned porcine UPTI cDNA [15]. The pUPTI protein of 17 kDa was puri®ed from the 14- kDa protein by electroelution from acrylamide gel slices using the Bio-Rad electroeluter (Bio-Rad) according to the manufacturer's instructions and was used in all assays described in the present study.

Trypsin Inhibition by SLPI and UPTI The endogenous anti-trypsin activity of pSLPI, pUPTI, and rhSLPI was examined as previously described [15]. All three proteins inhibited trypsin activity at protein amounts within the nanogram-to-microgram range. Porcine SLPI FIG. 3. Induction of [3H]thymidine incorporation into cellular DNA by and UPTI had 10% and 20% trypsin inhibitory activities, pUPTI. Primary cultures of serum-deprived GE cells were incubated with respectively, at 650 ng, and these increased to 40% and increasing concentrations of pUPTI for 20 h and then pulse-labeled with [3H]thymidine for 4 h. Results represent the mean Ϯ SEM of three inde- 70%, respectively, at 6.5 ␮g. Recombinant hSLPI, at 200 pendent experiments, where each experiment corresponds to 3 separate and 400 ng, had 20% and 40% trypsin inhibitory activities, cultures of GE cells isolated from an individual gilt on Day 12 of preg- respectively. nancy. Means without a common superscript differ (p Ͻ 0.05). 384 BADINGA ET AL.

Heparin Effects on SLPI- and UPTI-Induced [3H]Thymidine Incorporation The glycosaminoglycan heparin had been previously demonstrated to bind hSLPI, resulting in the increased rate of inhibition of its substrate elastase [25]. To determine whether interaction with heparin similarly in¯uences the mitogenic potentials of SLPI and UPTI, the effect of added heparin (0.5 ␮g/ml) on SLPI- and UPTI-induced [3H]thymidine incorporation into GE cellular DNA was examined. All three protease inhibitors increased (p Ͻ 0.05) DNA synthesis in GE cells above control values, and the effects of added pSLPI, hSLPI, and pUPTI on these cells

were comparable (p Ͼ 0.05) (Fig. 5). However, heparin had Downloaded from https://academic.oup.com/biolreprod/article/61/2/380/2734487 by guest on 24 September 2021 no effect (p Ͼ 0.05) on the level of labeled thymidine incorporated by DNA of these cells, and exhibited only a FIG. 4. Effect of pSLPI and pUPTI on uterine GE cell DNA synthesis. tendency (p ϭ 0.08) to enhance the individual effects of Primary cultures of uterine GE cells were incubated in serum-free media all three protease inhibitors. for 24 h and then treated with pSLPI (100 ng/ml), pUPTI (100 ng/ml), or their combination (100 ng/ml each) for 20 h, before pulse-labeling with SLPI and UPTI Transcript Levels in Porcine Endometrium [3H]thymidine for 4 h. Results represent the mean Ϯ SEM of three independent experiments, where each experiment corresponds to 3 separate during Pregnancy cultures of GE cells isolated from an individual gilt on Day 12 of preg- Steady-state pSLPI and pUPTI mRNA levels in endo- nancy. Means without a common superscript differ (p Ͻ 0.05). metrium as a function of stage of pregnancy and of cell type were evaluated by Northern analysis and semiquantitative RT-PCR. Northern blot analysis showed the presence tion; 15 ␮g/well) antisera, respectively, for 2 h before treat- of SLPI and UPTI transcripts in Day 12 and Day 60 preg- ment (Fig. 2, C and D). Neither antiserum alone had any nant pig endometrium, with the level of each transcript effect on labeled thymidine incorporation by these cells greater at Day 60 than at Day 12 (Figs. 6A and 7A). Con- (data not shown). The lack of available anti-pUPTI anti- sistent with this, the levels of SLPI mRNA increased (p Ͻ serum precluded the conduct of similar experiments with 0.05) between Days 12 and 60 and remained high through Day 90 of gestation (Fig. 6B), as measured by RT-PCR. UPTI. SLPI transcript levels were consistently present and higher To examine further whether SLPI and UPTI increase (p Ͻ 0.05) in endometrial GE, compared to ST, cells (Fig. uterine epithelial cell DNA synthesis via common signaling 6C). The levels of UPTI transcript were readily detected at pathways, con¯uent GE cells were incubated with pSLPI Day 12, peaked by Day 60, and were diminished by Day (100 ng/ml) and pUPTI (100 ng/ml) alone or in combina- 90 of pregnancy (Fig. 7B). Similar to SLPI, a tendency (p tion in serum-free media. Simultaneous addition of pSLPI ϭ 0.07) for higher levels of UPTI transcript in GE vs. ST and pUPTI to these cells resulted in stimulation of cells was observed in the endometrium of early pregnancy 3 [ H]thymidine incorporation into DNA to a level higher (p (Fig. 7C). In all cases, the levels of ␤2-microglobulin tran- Ͻ 0.05) than that obtained with either anti-protease alone script [26], the internal control in the PCR reactions, did (Fig. 4). not change [27].

DISCUSSION The recent demonstration of a growth-promoting activity for two of the four members of the TIMP family of protease inhibitors, namely TIMP-1 and TIMP-2, whose normal function in cellular processes is the control of metallopro- tease activity, has introduced a new level of complexity to the biological functions of anti-proteases. In this report, we demonstrate for the ®rst time that two uterine-associated serine protease inhibitors of pregnancy, namely SLPI and UPTI, exhibit potential growth factor activity as measured by their induction of cellular DNA synthesis, and that a likely target for these inhibitors is the glandular epithelium of the uterus. Since these anti-proteases have distinct substrate speci®cities from TIMPs but are coordinately expressed with TIMPs in the early pregnant pig endometrium FIG. 5. Effect of heparin on SLPI and UPTI-induced GE cell DNA syn- [14], results suggest that induction of uterine epithelial cell thesis. Primary cultures of serum-deprived GE cells were incubated with mitogenesis during early pregnancy may be a common or without rhSLPI (100 ng/ml), pSLPI (100 ng/ml), or pUPTI (100 ng/ml) function for protease inhibitors involving mechanisms that in the absence or presence of heparin (0.5 ␮g/ml) for 20 h and then pulse- are unrelated to their anti-proteolytic effects. labeled with [3H]thymidine for 4 h. Responses to each treatment combi- SLPI (also known as antileukoprotease), although widely nation, as determined by [3H]thymidine incorporation into cellular DNA, were determined in triplicate per experiment, and results represent the present in human mucosal secretions [28, 29], was de- mean Ϯ SEM of two independent experiments. Each experiment repre- scribed in the pig initially as a pregnancy-associated uterine sents cells isolated from an individual gilt on Day 12 of pregnancy. protease inhibitor [8]. Subsequent studies indicated that its UTERINE PROTEASE INHIBITORS AS EPITHELIAL MITOGENS 385

FIG. 6. SLPI mRNA levels in porcine endometrium during pregnancy. A) Northern blot analysis of SLPI transcript in Day 12 and Day 60 pregnant pig endometrium. Total cellular RNA (30 ␮g) from each sample was analyzed. B) Stage of pregnancy-dependent expression of SLPI in porcine endometrium. Total cellular RNA extracted from Days 12, 60, and 90 pregnant pig endometrium (n ϭ 3 individual animals for each pregnancy day) was subjected to

semiquantitative RT-PCR analysis using pSLPI gene-speciﬁc primers. Expected product size is 500 bp. C) SLPI gene Downloaded from https://academic.oup.com/biolreprod/article/61/2/380/2734487 by guest on 24 September 2021 expression in endometrial ST and GE cells. Total cellular RNA isolated from ST and GE cells of early pregnancy (Day 12 post-estrus) was subjected to semiquantitative RT-PCR using pSLPI gene-speciﬁc primers. Each lane represents cellular RNA isolated from GE cells prepared from endometrium of an individual gilt on Day 12 of pregnancy. Px, Pregnancy. high level of uterine endometrial expression during preg- Interestingly, another uterine-derived serine anti-protease nancy appears to be limited to mammalian species with in the pig, namely uterine plasmin/trypsin inhibitor [9, 15], noninvasive placentation (e.g., pig, cow, horse). Species ex- induced [3H]thymidine incorporation into DNA of GE cells, hibiting more intimate contact of the maternal and tropho- similar to that of SLPI (this study). Indeed, several features blast tissues (e.g., rat, mouse) were found to show unde- of the mitogenic activities of SLPI and UPTI are worth tectable endometrial levels of this mRNA and protein [16]. noting. First, their respective activities are achieved at con- The latter observation implies that uterine SLPI may largely centrations (i.e., ng/ml) at which classical growth factors underlie the maintenance of an intact feto-maternal inter- have been shown to be effective ([31] and references there- face in species with epitheliochorial placenta [21], thus pri- in). Second, their near-maximal mitogenic activities are ob- marily acting as an anti-protease within the uterine envi- served at concentrations at which their anti-trypsin activi- ronment. However, the ®nding reported here suggests that ties are less than optimal. Third, their sum activities are SLPI may have an additional function during early preg- higher than that individually. Finally, their activities appear nancy that involves its modulation of uterine glandular ep- to be modulated, to a limited extent, by heparin. Taken ithelial cell growth. This novel activity of SLPI is speci®c, together, these data indicate that the mitogenic potentials of since it can be abrogated by the co-addition of speci®c anti- SLPI antiserum and is demonstrated by SLPI isolated from SLPI and UPTI probably involve mechanisms distinct from two mammalian species (human and pig). Additionally, the their anti-protease functions. TIMPs have been shown to pig and human SLPI proteins were active within the same modulate target cell mitogenesis via cell surface receptors concentration range (10±1000 ng/ml), eliminating any pos- [13]; however, our preliminary data indicate lack of binding sibility that minor contaminants in the pig SLPI protein of labeled pSLPI or hSLPI to monolayer cultures of GE might underlie the observed activity. The latter observation cells or membrane-enriched fractions prepared from these is particularly interesting since human and porcine SLPI cells (data not shown). This suggests that cell surface re- exhibit only a 68% similarity in primary structures [8]; this ceptors probably do not mediate the mitogenic activity of suggests that the conserved core regions within the two SLPI. Potential mediators might include components of the proteins, which include cysteine- and proline-rich residues extracellular matrix such as integrins, which have been as well as a two-domain structure [8, 30], are suf®cient for shown to directly in¯uence cellular differentiation, survival, these proteins' mitogenic potentials. and function [32]. In this regard, SLPI has been localized

FIG. 7. UPTI mRNA levels in porcine endometrium during pregnancy. A) Northern blot analysis of UPTI transcript in Day 12 and Day 60 pregnant pig endometrium. Total cellular RNA (30 ␮g) from each sample was analyzed. B) Stage of pregnancy-dependent expression of UPTI in porcine endometrium. Total cellular RNA extracted from Days 12, 60, and 90 pregnant pig endometrium (n ϭ 2 individual animals per pregnancy day) was subjected to semiquantitative RT-PCR analysis using pUPTI gene speciﬁc primers. Expected product size is 490 bp. C) UPTI gene expression in endometrial ST and GE cells. Total cellular RNA isolated from ST and GE cells of early pregnancy (Day 12 post-estrus) was subjected to semiquantitative RT-PCR using pSLPI gene-speciﬁc primers. PCR products were electrophoresed in an agarose gel, blotted onto a nylon membrane, and hybridized with 32P-labeled porcine UPTI cDNA fragment. Each lane represents cellular RNA isolated from GE cells prepared from endometrium of an individual gilt on Day 12 of pregnancy. Px, Pregnancy. 386 BADINGA ET AL. to the extracellular matrix via its association with elastin 2. Simmen RCM, Simmen FA. Regulation of uterine and conceptus se- ®bers in the human lung [33]. cretory activity in the pig. J Reprod Fertil Suppl 1990; 40:279±292. The functional domain(s) involved in the biological ac- 3. Lala PK, Graham CH. Mechanisms of trophoblast invasiveness and their control: the role of proteases and protease inhibitor. Cancer Me- tivity of SLPI can only be speculated on at the present time. tastasis Rev 1990; 9:369±379. The observation that alkylated TIMPs had no anti-metal- 4. Salamonsen LA, Nagase H, Wooley DE. Matrix metalloproteinases loprotease activity, while maintaining signi®cant cell-pro- and their tissue inhibitors at the ovine trophoblast-uterine interface. J liferating activity [13], indicate that distinct domains are Reprod Fertil Suppl 1995; 49:29±37. responsible for the mitogenic and anti-protease activities of 5. Strickland S, Reich E, Sherman MJ. Plasminogen activator in early these protease inhibitors. The anti-trypsin activity of SLPI embryogenesis. Enzyme production by trophoblast and parietal en- doderm. Cell 1976; 9:231±240. has been localized to its carboxy-terminal domain [34, 35], 6. Bischof RJ, Brandon MR, Lee C-S. Cellular immune responses in the while the positively charged amino acid residues that bind pig uterus during pregnancy. J Reprod Immunol 1995; 29:161±178. heparin are dispersed in clusters throughout the molecule 7. Roberts RM, Bazer FW. The functions of uterine secretions. J Reprod [36]. Since heparin appears to enhance SLPI-induction of Fertil 1988; 82:875±892. DNA synthesis, albeit to a much lesser extent that its in- 8. Farmer SJ, Fliss AE, Simmen RCM. Complementary DNA cloning Downloaded from https://academic.oup.com/biolreprod/article/61/2/380/2734487 by guest on 24 September 2021 hibitory activity [24, 37], the speci®c domains involved in and regulation of expression of the messenger RNA encoding a preg- these distinct functions of SLPI cannot be assigned with nancy-associated porcine uterine protein related to human antileukoproteinase. Mol Endocrinol 1990; 4:1095±1104. certainty. 9. Fazleabas AT, Bazer FW, Roberts RM. Puri®cation and properties of Finally, the biological rationale for the mitogenic activity a progesterone-induced plasmin/trypsin inhibitor from the uterine se- of SLPI and UPTI towards uterine GE cells remains un- cretions of pigs and its immunocytochemical localization in the preg- clear. The three major cell types of the uterine endometrium nant uterus. J Biol Chem 1982; 257:6886±6897. (GE, LE, and ST) synthesize a diverse array of classical 10. Samuel CA. The development of pig trophoblast in ectopic sites. J growth factors, which together would appear to be suf®- Reprod Fertil 1971; 27:494±495. 11. Bertaux B, Hornebeck W, Eisen AZ, Dubertret L. Growth stimulation cient to provide an environment optimal for growth [38]. of human keratinocytes by tissue inhibitor of metalloproteases. J In- Our previous ®nding that a conceptus secretory product, vest Dermatol 1991; 97:679±685. transforming growth factor ␣, induces endometrial expres- 12. Hayakawa T, Yamashita K, Tanzawa K, Uchijima E, Iwaka K. sion of the pSLPI gene [14], whose product is now dem- Growth-promoting activity of tissue inhibitor of metalloproteinases-1 onstrated to exhibit endogenous mitogenic activity, sug- (TIMP-1) for a wide range of cells. A possible new growth factor in gests that a period critical for periimplantation may require serum. FEBS Lett 1992; 298:29±32. a rapid but sustained local ampli®cation of growth. At later 13. Hayakawa T, Yamashita K, Ohuchi E, Shinagawa A. Cell growth- promoting activity of tissue inhibitor of metalloproteinase-2 (TIMP- pregnancy stages (e.g., mid-pregnancy), when SLPI and 2). J Cell Sci 1994; 107:2373±2379. UTPI synthesis are maximal (this study; [8, 15]) and, as in 14. Reed KL, Blaeser LL, Dantzer V, Green ML, Simmen RCM. Control the case for SLPI, the protein is transported to the fetal of secretory leukocyte protease inhibitor gene expression in the por- circulation [21], the possibility that these serine protease cine periimplantation endometrium: a case of maternal-embryo com- inhibitors function less as mitogenic agents and more as munication. Biol Reprod 1998; 58:448±457. anti-proteases is likely. Alternatively, SLPI may have ad- 15. Stalling-Mann ML, Burke MG, Trout WE, Roberts RM. Puri®cation, ditional functions distinct from the above. A novel role for characterization, and cDNA cloning of a kunitz-type proteinase inhibitor secreted by the porcine uterus. J Biol Chem 1994; 269:24090± SLPI in the inhibition of matrix metalloproteinase produc- 24094. tion by monocytes [39] suggests its anti-in¯ammatory ac- 16. Badinga L, Michel FJ, Simmen RCM. Pregnancy-associated endo- tion, independent of its serine protease substrates. metrial expression of antileukoproteinase gene is correlated with ep- In conclusion, our results provide evidence for a novel itheliochorial placentation. Mol Reprod Dev 1994; 38:357±363. autocrine function for SLPI and UPTI in early pregnancy 17. Fazleabas AT, Geisert RD, Bazer FW, Roberts RM. Relationship be- events. These data, taken in light of similar demonstrated tween release of plasminogen activator and estrogen by blastocysts activities for members of another protease inhibitor family, and secretion of plasmin inhibitor by uterine endometrium in the pregnant pig. Biol Reprod 1983; 29:225±238. namely TIMPs 1 and 2, suggest that control of cell growth 18. Reed KL, Badinga L, Davis DL, Chung TE, Simmen RCM. Porcine may constitute a general function for protease inhibitors in endometrial glandular epithelial cells in vitro: transcriptional activities mammals. The elucidation of the mechanisms that underlie of the pregnancy-associated genes encoding antileukoproteinase and this activity distinct from their respective anti-protease uteroferrin. Biol Reprod 1996; 55:469±477. functions may provide a clear understanding of the signal- 19. Green ML, Simmen RCM, Simmen FA. Developmental regulation of ing mechanisms and novel strategies for ensuring success- steroidogenic enzyme gene expression in the periimplantation porcine ful implantation. conceptus: a paracrine role for insulin-like growth factor-I. Endocri- nology 1995; 136:3961±3970. 20. Sanger F, Nicklen S, Coulson AR. DNA sequencing with chain-ter- ACKNOWLEDGMENTS minating inhibitors. Proc Natl Acad Sci USA 1977; 74:5463±5467. The authors thank Dr. Frank A. Simmen for discussion and critical 21. Simmen RCM, Michel FJ, Fliss AE, Smith LC, Fliss MFV. Ontogeny, review of this manuscript, members of Drs. Rosalia C.M. and Frank A. immunocytochemical localization, and biochemical properties of the Simmens' laboratories for help with the collection of tissues and breeding pregnancy-associated uterine elastase/cathepsin-G protease inhibitor, of animals, and Ge Zhao for formatting of ®gures and densitometric anal- antileukoproteinase (ALP): monospeci®c antibodies to a synthetic ysis. We extend our appreciation to Dr. Jeff Vallet (MARC, Clay Center, peptide recognize native ALP. Endocrinology 1992; 130:1957±1965. NE) for kindly providing Day 60 pregnant pig allantoic ¯uid, Dr. Pieter 22. Towbin H, Staehelin T, Gordon J. Electrophoretic transfer of proteins S. Hiemstra (Leiden University Medical Center, The Netherlands) for gift from polyacrylamide gels to nitrocellulose sheets: procedures and ap- of rabbit anti-human SLPI, and Dr. William Trout (University of Missouri, plications. Proc Natl Acad Sci USA 1979; 76:4350±4354. Columbia, MO) for advice on the anti-trypsin assay. 23. Ko Y, Simmen RCM, Lee CY, Simmen FA. Reversion to the non- transformed phenotype of tsSV40-transformed uterine endometrial ep- REFERENCES ithelial cells is accompanied by decreased mitogenic responsiveness to IGFs and by enhanced secretion of at least ®ve different IGF bind- 1. Fazleabas AT, Geisert RD, Bazer FW, Roberts RM. Relationship be- ing proteins. Endocrine 1994; 2:495±504. tween release of plasminogen activator and estrogen by blastocysts 24. SAS. SAS/Stat User's Guide, Release 6.04 ed. Cary, NC: Statistical and secretion of plasmin inhibitor by uterine endometrium in the preg- Analysis System Institute, Inc.; 1988. nant pig. Biol Reprod 1983; 29:225±238. 25. Ying Q-L, Kemme M, Simon SR. Functions of the N-terminal domain UTERINE PROTEASE INHIBITORS AS EPITHELIAL MITOGENS 387

of secretory leukoprotease inhibitor. Biochemistry 1994; 33:5445± 33. Kramps JA, Te Boekhorst AHT, Fransen JAM, Ginsel LA, Dijkman 5450. JH. Antileukoprotease is associated with elastin ®bers in the extracel- 26. Reddy NR, Wilkie BN, Mallard BA. Construction of an internal con- lular matrix of the human lung. Am Rev Respir Dis 1989; 140:471± trol to quantitate multiple porcine cytokine mRNAs by RT-PCR. Bio- 476. techniques 1996; 21:868±870. 34. Eisenberg SP, Hale KK, Heimdal P, Thompson RC. Location of the 27. Simmen RCM, Chung TE, Imataka H, Michel FJ, Badinga L, Simmen protease-inhibitory region of secretory leukocyte protease inhibitor. J FA. Trans-activation functions of the Sp-related nuclear factor, basic Biol Chem 1990; 265:7976±7981. transcription element binding protein, and progesterone receptor in 35. Kramps JA, van Twisk C, Appelhans H, Meckelein B, Nikiforov T, endometrial epithelial cells. Endocrinology 1999; (in press). Dijkman JH. Proteinase inhibitory activities of antileukoprotease are 28. Wallner O, Fritz H. Characterization of an acid stable proteinase in- represented by its second COOH-terminal domain. Biochim Biophys hibitor in human cervical mucus. Hoppe-Seyler's Z Physiol Chem Acta 1990; 1038:178±185. 1974; 355:709±715. 36. Masuda K, Kamimura K, Watanabe K, Suga T, Kanesaki M, Takeuchi 29. Ohlsson K, Tegner H, Akesson U. Isolation and partial characteriza- A, Imaizumi A, Suzuki Y. Pharmacological activity of the C-terminal and N-terminal domains of secretory leukoprotease inhibitor in vitro. tion of a low mol wt acid stable protease inhibitor from human bron- Br J Pharmacol 1995; 115:883±888. chial secretion. Hoppe-Seyler's Z Physiol Chem 1977; 358:583±589.

37. Fath MA, Wu X, Hileman RE, Linhardt RJ, Kashem MA, Nelson Downloaded from https://academic.oup.com/biolreprod/article/61/2/380/2734487 by guest on 24 September 2021 30. Stetler G, Brewer MT, Thompson RC. Isolation and sequence of a RM, Wright CD, Abraham WM. Interaction of secretory leukocyte human gene encoding a potent inhibitor of leucocyte proteases. Nu- protease inhibitor with heparin inhibits proteases involved in asthma. cleic Acids Res 1986; 14:7883±7896. J Biol Chem 1998; 273:13563±13569. 31. Green ML, Chung TE, Reed KL, Modric T, Badinga L, Yang J, Sim- 38. Simmen FA, Simmen RCM. Peptide growth factors and proto-onco- men FA, Simmen RCM. Paracrine inducers of uterine endometrial genes in mammalian conceptus development. Biol Reprod 1991; 44: spermidine/spermine N1-acetyltransferase gene expression during ear- 1±5. ly pregnancy in the pig. Biol Reprod 1998; 59:1251±1258. 39. Zhang Y, DeWitt DL, McNeely TB, Wahl SM, Wahl LM. Secretory 32. Boudreau N, Sympson CJ, Werb Z, Bissell MJ. Suppression of ICE leukocyte protease inhibitor suppresses the production of monocyte and apoptosis in mammary epithelial cells by extracellular matrix. prostaglandin H synthase-2, prostaglandin E2, and matrix metallopro- Science 1995; 267:891±893. teases. J Clin Invest 1997; 99:894±900.