Importance of Uterine Cell Death, Renewal, and Their Hormonal Regulation in Hamsters That Show Progesterone-Dependent Implantation

Abstract

This study was initiated to investigate the significance of uterine cell death and proliferation during the estrous cycle and early pregnancy and their correlation with sex steroids in hamsters where blastocyst implantation occurs in only progesterone-primed uteri. The results obtained in hamsters were also compared with mice where blastocyst implantation occurs in progesterone-primed uteri if estrogen is provided. Apoptotic cells in the uterus were detected by using terminal deoxynucleotide transferase-mediated deoxyuridine triphosphate nick end labeling (TUNEL) technique. Uterine cell proliferation was determined by 5-bromo-2′-deoxyuridine labeling followed by immunohistochemistry and methyl-tritiated [³H]thymidine labeling. Active caspase-3, an executor protein of cell death, expression was assayed by immunohistochemistry/immunofluorescence. Our results demonstrate that epithelial proliferation on the second day after mating marks the initiation of pregnancy-related uterine changes in both species despite their differences in hormonal requirements. Hamsters and mice showed subtle differences in uterine proliferative and apoptotic patterns during early pregnancy and in response to steroids. There existed almost a direct correlation between apoptosis and caspase-3 expression, suggesting uterine cell death mostly involves the caspase pathway. Consistent with these findings, we showed, for the first time, that execution of uterine epithelial cell apoptosis by caspase-3 is important for blastocyst implantation because a caspsase-3 inhibitor N-acetyl-DEVD-CHO when instilled inside the uterine lumen on d 3 of pregnancy inhibits implantation in hamsters and mice. The overall results indicate that uterine cell apoptosis and proliferation patterns are highly ordered cell-specific phenomena that play an important role in maintaining the sexual cycle and pregnancy-associated uterine changes.

The uterus undergoes cellular remodeling during each sexual cycle for a possible pregnancy. In the absence of pregnancy, uterine changes are reversible, permitting preparation for pregnancy in a subsequent cycle. In the event of mating and successful fertilization, however, the changes in the uterus take another route to support pregnancy. The uterine cellular changes during the cycle and pregnancy are regulated by the circulating levels of ovarian sex steroids progesterone (P₄) and estrogen (E). One of the basic events of uterine cellular change is the systemic cell turnover that consists of cell death by apoptosis and cell renewal by proliferation. Cell apoptosis and cell proliferation have been studied in the uterus of mice and rats (1–4). However, there has been no detailed study of uterine cell apoptosis and proliferation during the estrous cycle and early pregnancy and their possible relation to circulating sex steroid levels in hamsters. Unlike mice and rats that require both ovarian P₄ and E for uterine receptivity and implantation (5–7), hamsters need only ovarian P₄, but not ovarian E (8–12). Furthermore, rabbits, pigs, monkeys, and perhaps humans show ovarian P₄-dependent implantation like hamsters (13–17).

Previous studies in mice and rats have reported an inverse correlation between cell death and cell proliferation in the uterus (4). Although uterine epithelial proliferative indices are at their highest level on the proestrous day, epithelial apoptotic indices are their lowest level on this day (18). A reverse scenario was demonstrated between uterine cell proliferation and apoptosis on the estrous day in rats (19). The increased uterine epithelial apoptosis on the estrous day was attributed to a decline in serum E level (4). A number of studies in humans also showed that preovulatory ovarian E secretion stimulates proliferation of uterine glandular epithelial cells during the proliferative phase (20). This also remains relatively high in the secretory phase when predecidual cells also show increased DNA synthesis. A rapid increase in the apoptotic index was noted in both epithelium and stroma during the last days of the cycle (late secretory phase) with a maximum index on the second day of menstruation (21). However, the incidence of apoptotic stromal cells during late secretory phase is still controversial because others have shown few apoptotic stromal cells at any stage of the cycle (22, 23).

In regard to the direct actions of P₄ and E on uterus cell proliferation, animal studies have shown that whereas treatment of 17β-estradiol (E₂) alone to ovariectomized mice stimulated the uterine epithelial cell proliferation, treatment with P₄ alone induced stromal cell proliferation (1). However, treatment of P₄-primed animals with E₂ mainly maintained the proliferation in stromal cells but does not stimulate proliferation in epithelial cells (1, 24). In the mouse and hamster uteri, apoptotic epithelial cell death mainly occurs because of E withdrawal and can be blocked by E₂ treatment (25, 26).

After a successful mating, preimplantation uteri of the mouse first show luminal and glandular epithelial cell proliferation on the second day of pregnancy. On d 3 of pregnancy, however, epithelial cell proliferation ceases and stromal cell proliferation begins because of increased P₄ synthesis by corpora lutea. This stromal cell proliferation is further enhanced on d 4 in response to ovarian E secretion before initiation of implantation in mice (1, 27, 28). The combined effect of ovarian P₄ and E in stromal cell proliferation is a prerequisite uterine change for blastocyst implantation in this rodent. The apoptotic patterns in uterine cells during the preimplantation period have not been studied in detail.

Initiation of blastocyst implantation induces proliferation of stromal cells that are situated only immediate to the implanting blastocyst (1). These proliferating stromal cells fail to undergo cytokinesis and become polyploid, forming the primary decidual zone (PDZ) (29, 30). As the pregnancy continues, cells of the PDZ on d 6 stop proliferating, but stromal cells next to the PDZ undergo proliferation and polyploidy, forming the secondary decidual zone (SDZ) (1, 3, 29). In regard to apoptosis, the general consensus is that antimesometrial luminal epithelial cells surrounding the implanting embryo die during implantation (31). As gestation progresses, epithelial degeneration extends to the mesometrial side. Because stromal decidualization is more at the antimesometrial area compared with the mesometrial area, cell death starts first at the PDZ and then at the SDZ of the antimesometrial area, causing a stepwise regression of decidual tissues and creating space for the growing embryo. Besides steroid hormones, various local factors such as growth factors and cytokines have been shown to be crucial mediators of apoptosis. However, our knowledge of the molecular mechanisms underlying uterine apoptosis is still poor. There are two major pathways to activate proteases of the caspase family for apoptosis. The extrinsic pathway involves the death receptors and their ligands. Best studied are the Fas ligands, TNF-α, TGF-β, and their receptors. The intrinsic pathway induces oligomerization of the cytosolic apoptotic protease-activating factor-1 (Apaf-1) and apopto-some formation by cytochrome c release from the mitochondria. The caspase cascade is activated to execute the apoptotic cells in both pathways. Caspase-8 and caspase-9 serve as initiators and are situated at the top of the cascade. Caspase-3 and caspase-6 are effector execution caspases that degrade cells (3).

In the present study, cell-type-specific apoptosis and proliferation was investigated in the uterus of hamsters during the estrous cycle and early pregnancy and after steroid treatment in ovariectomized animals. Uterine apoptotic patterns were correlated with expression of active caspase-3. Our apoptotic and proliferation data in the hamster uterus were also compared with findings in the mouse uterus. We also show here the importance of caspace-3-mediated uterine cell apoptosis in initiation of implantation both in the hamster and mouse.

Materials and Methods

Animals

Adult virgin male and female golden hamsters (Mesocricetus auratus) and CD1 mice were purchased from Charles River Laboratories (Wilmington, MA). They were maintained in a 14-h light, 10-h dark cycle in the Laboratory Animal Facility of the Vanderbilt University (Nashville, TN) with unlimited access to water and food according to the institutional and National Institutes of Health guidelines on the care and use of laboratory animals. All experimental protocols were approved by the Vanderbilt University Committee of Use and Care of animals.

Uterine tissue collection during the estrous cycle of the hamster

The 4-d estrous cycle of hamsters was monitored by the presence of characteristic vaginal discharge on the morning of the estrous day, which was also designated d 1 of the estrous cycle (32). Hamsters with at least three consecutive regular 4-d cycles were used in this study. Hamsters were killed at the estrus, metestrus, diestrus, and proestrus at 0830–0900 h, 2 h after an injection (1 ml/100 g body weight) of 5-bromo-2′-deoxyuridine (BrdU) (catalog no. RPN201) purchased from Amersham Biosciences (Piscataway, NJ) to detect proliferative cells. Their uteri were quickly removed, cut into small pieces, and rapidly frozen in cold Super Friendly Freeze’it (Curtin Matheson Scientific, Houston, TX) and stored at −70 C for cell proliferation, cell apoptosis, and caspase-3 expression studies. Unless otherwise mentioned, all materials were purchased from Sigma (St. Louis, MO) or Fisher Scientific (Pittsburgh, PA).

Preparation of pregnant hamsters and mice and uterine tissue collection

Female hamsters that showed three consecutive 4-d estrous cycles were housed with fertile males overnight on the evening of proestrus. Finding of sperm in the vaginal smear the next morning (estrus) indicated the first day (d 1) of pregnancy (32). Three CD1 female mice were placed overnight with one fertile male for mating. Pregnancy was confirmed the following morning by checking for the presence of a copulatory plug in the vagina (d 1 = vaginal plug) (7). Animals were injected ip with either methyl-tritiated [³H]thymidine (25 μCi/0.1 ml; catalog no. NET027Z; specific activity, 70–90 Ci/mmol; PerkinElmer Life Sciences, Boston, MA) (1) or BrdU (1 ml/100 g body weight) 2 h before they were killed.

Hamsters and mice on d 1–3 of pregnancy were killed at 0830–0900 h. Their whole uteri were removed and cut into small pieces after confirmation of pregnancy by flushing and recovering embryos from oviducts (32). Although whole uteri were collected on the morning of d 4 (0900 h), implantation sites were collected on the morning (0900 h) of d 5 after an iv injection of Chicago Blue B dye solution [0.25 ml of 1% (wt/vol) dye in saline] in both species (7, 32). Implantation sites on these days were visualized by intermittent blue bands along the horns. On d 6–8, implantation sites were distinct and were identified visually without blue dye injection. Uterine tissues were immediately frozen in cold Super Friendly Freeze’it and stored at −70 C for cell proliferation, cell apoptosis, and caspase-3 expression studies.

To determine the effects of P₄ and E on uterine cell proliferation and apoptosis, mice and hamsters were ovariectomized regardless of their stage of estrous cycle and rested 15 d to eliminate circulating steroids. Control animals were injected with vehicle of steroids in sesame seed oil (0.1 ml/animal). Some animals were injected with a single injection of either P₄ (2 mg/0.1 ml per mouse; 2 mg/0.1 ml per hamster) or E₂ (100 ng/0.1 ml per mouse; 1000 ng/0.1 ml per hamster) or a combination of the same doses of P₄ plus E₂ and killed 24 h later for uterine tissue collection. Ovariectomized animals were also injected with P₄ for 2 or 3 d (once each day) with or without an injection of E₂ on the third day of P₄ treatment. The dose of P₄ was selected on the basis of its ability to maintain pregnancy in ovariectomized or hypophysectomized hamsters (8, 32). The dose of E₂ was selected based on its ability to stimulate heparin-binding epidermal growth factor-like growth factor gene expression in ovariectomized uteri of hamsters (32). All animals were killed 24 h after the last injection. Animals were injected with either [³H]thymidine or BrdU 2 h before they were killed.

Apoptosis detection

DNA fragmentation during apoptosis was detected by the terminal deoxynucleotide transferase (TdT)-mediated deoxyuridine triphosphate (dUTP) nick end labeling (TUNEL) technique using an apoptosis detection kit (Promega, Madison, WI; catalog no. G7130). TUNEL assay was performed according to the manufacturer’s instruction with minor modification. Briefly, fresh-frozen sections (12 μm) were fixed in 10% (vol/vol) formalin for 30 min, washed in water, dehydrated in ascending concentrations [30, 50, 70, 85, 95, and 100% (vol/vol)] of ethyl alcohol, washed in xylene and acetone, rehydrated in descending concentrations of ethyl alcohol, and finally washed in water and PBS before being incubated with proteinase K (20 μg/ml) for 5 min. The sections were then washed in PBS, refixed in 10% (vol/vol) formalin for 5 min, and divided into three groups. At the beginning of the experiment, the positive control sections were incubated with DNase (RQ1 RNase-free DNase from Promega; catalog no. M6101) for 5 min and rinsed in water. Next, all three groups (positive, treatment, and negative) were incubated with equilibration buffer for 20 min. Subsequently, 250 μl TUNEL reaction mixture containing biotinylated nucleotide and TdT was added to sections of the positive and treatment groups for 60 min at 37 C in the dark. For negative control, TdT was omitted from the reaction mixture. The reaction was terminated by transferring all the slides to the termination buffer for 15 min. The sections were rinsed in PBS, incubated in 0.3% (vol/vol) hydrogen peroxide (H₂O₂), washed in PBS, and incubated with streptavidin horseradish peroxidase solution for 60 min. After the sections were washed in PBS, aminoethylcarbazole (AEC) single solution (Zymed Laboratories Inc., San Francisco, CA; catalog no. 00-1111) was added to the slide and developed color to detect biotinylated nucleotide. The sections were counterstained with hematoxylin and photographed in an ECLIPSE 80i Nikon microscope connected to a computer with ACT-2U program for the DS-5M camera.

Caspase-3 immunocytochemistry

Freshly prepared cryosections were rapidly fixed (10 min) in acetone at room temperature for 10 min. After rinsing in PBS, the blocking solution of 10% (vol/vol) goat serum was added to the section for 10 min at room temperature followed by the primary antibody (antirabbit active caspase-3 obtained from Promega) (1:40 dilution) in PBS at 4 C overnight. Subsequent immunostaining was performed using biotinylated goat antirabbit secondary antibody, streptavidin-horseradish peroxidase conjugate, and a substrate chromogen mixture (AEC single solution) from Zymed. Sections were lightly counterstained with hematoxylin. Brown deposits indicated the sites of immunoreactive proteins (33). Negative control studies were performed in which PBS was used instead of primary antibody.

Double immunofluorescence (colocalization of caspase-3 staining and TUNEL labeling)

A slide having cryosections from d-1 uterus of the hamster, and d-5 implantation sites of both the hamster and mouse was first fixed in acetone for caspase-3 staining. The slide was first incubated with anti-rabbit active caspase-3 overnight at 4 C and visualized with a tetramethyl rhodamine isothiocyanate-labeled goat antirabbit secondary antibody (Zymed Laboratories Inc.). The same slide was then processed for TUNEL assay using the DeadEnd Fluorometric TUNEL system (Promega), which measures the fragmented DNA of apoptotic cells by catalytically incorporating fluorescein-12-dUTP at 3′-OH DNA ends using TdT. The double fluorescence was visualized directly by fluorescence microscopy using the Nikon Eclipse TS100 with an X-cite 120 fluorescence illumination system.

Cell proliferation assays

BrdU incorporation

Proliferating uterine cells were assessed by BrdU-labeled cells using a kit from Calbiochem (catalog no. HCS30). Sections were soaked in 70% (vol/vol) ethyl alcohol for 1 h and then washed in PBS for 5 min. Endogenous peroxidase was inactivated by covering the section with 0.3% (vol/vol) fresh H₂O₂ in water for 10 min. Slides were then washed and incubated with denaturing solution for 90 min. After washing, sections were first treated with blocking solution for 30 min and then incubated at 4 C overnight with anti-BrdU mouse monoclonal antibody conjugated with biotin. After being rinsed in PBS, sections were incubated for 90 min with streptavidin conjugated with peroxidase. The sections were then rinsed in PBS, and peroxidase activity was revealed through the use of AEC single solution as chromogen. Control uterine sections obtained from the animal without BrdU injection were stained in a similar manner. All sections were lightly counterstained with hematoxylin and photographed.

[³H]Thymidine incorporation

To identify proliferating uterine cells, hamsters and mice were treated ip with [³H]thymidine (25 mCi/0.1 ml in saline) 2 h before they were killed. Small pieces of uterine tissues were frozen for cryosections. Cryosections mounted in poly-l-lysine-coated slides were fixed in 4% (wt/vol) paraformaldehyde solution in PBS. Sections were rinsed in water; air dried, and dipped in NTB-2 emulsion. The slides were developed 5–6 d after dipping in NTB-2 emulsion (Eastman Kodak, Rochester, NY) to detect silver grains in the proliferating nuclei (1). Sections were also counterstained with hematoxylin and photographed (dark-field).

Blastocyst implantation after treatment of caspase-3 inhibitor in pregnant hamsters

The ability of the caspase-3 inhibitor to inhibit or delay the initiation of implantation was determined in pregnant hamsters and mice. In the present study, an irreversible caspase-3 inhibitor N-acetyl-Asp-Glu-Val-Asp-CHO (Ac-DEVD-CHO) (EMD Bioscience Co., Madison, WI; catalog no. 235420) was chosen because caspase-3 appears to be one of the key mediators of apoptosis (34). A single intraluminal injection of Ac-DEVD-CHO (10 μg/5 μl sterile saline in hamsters or 5 μg/2 μl sterile saline in mice) was administered in one uterine horn at the ovarian end on the morning of d 3 (0900 h) when embryos were still in oviducts. The contralateral horn received an intraluminal injection of equal volume of vehicle saline (5 and 2 μl sterile saline in hamsters and mice, respectively). Hamsters and mice were killed on the morning of d 5 (0900 h) 15 min after blue dye injection. The number of implantation sites (blue bands) was visually recorded. The uterine horn not showing implantation sites was flushed with culture medium to recover blastocysts, and morphological appearances of these blastocysts were checked by direct visualization under a stereo zoom microscope.

Embryo culture

Female hamsters were superovulated by ip injection of 20 IU pregnant mare serum gonadotropin before 0900 h on the estrous day and were housed with males overnight on the evening of proestrus (32). To study the effects of Ac-DEVD-CHO on preimplantation embryo development, eight-cell embryos from superovulated hamsters and normal pregnant mice were flushed out from the uterus and oviduct, respectively, on d 3 (1300–1400 h). Hamster embryos were cultured in a group of 10–15 embryos per drop (50 μl) in hamster embryo culture medium-6 under silicon oil in an atmosphere of 10% CO₂/90% air at 37 C for 24 h (35) in the presence or absence of Ac-DEVD-CHO (5 and 10 μg/ml). Mouse embryos were cultured in a group of eight to 10 embryos per drop (25 μl) in Whitten’s medium under silicon oil in an atmosphere of 5% CO₂/95% air at 37 C for 48 h (36) in the presence or absence of Ac-DEVD-CHO (10 μg/ml). At the end of culture, the number of embryos that developed to blastocysts was recorded.

Statistical analysis

All data were subjected to χ² test followed by Fisher’s exact test using the SAS 9.1 program (SAS Institute Inc., Cary, NC) to determine statistical differences among groups. In all cases, P < 0.05 was considered to be statistically significant.

Results

Apoptosis and proliferation in uterine cells during the estrous cycle in hamsters

Apoptosis and proliferation in uterine cells during the estrous cycle of hamsters is summarized in Fig. 1. Hamsters show regular 4-d (estrus, metestrus, diestrus, and proestrus) estrous cycles in our colony. The apoptosis in luminal and glandular epithelial cells was first detected on the estrous day and continued to the metestrous day with lower frequency. Only a few apoptotic cells were noticed in stroma and muscle layers on both the estrous and metestrous days (Fig. 1). The number of apoptotic cells was much less in the epithelial, stromal, and muscular layers on the diestrous and proestrous days (Fig. 1). Caspase-3 expression was also noticed both in the luminal and glandular epithelial cells on the estrous day, which correlated to the pattern of TUNEL staining. On the metestrous day, caspase-3 staining was observed in the luminal and glandular epithelial cells when epithelial cells were still undergoing apoptosis (Fig. 1). Immunoreactivity for caspase-3 was also noticed in a few subepithelial stromal cells. Caspase-3 staining appeared to be localized in the nucleus. The number of caspase-positive cells were only a few on the diestrous and proestrous days when there was almost no cellular apoptosis noticed (Fig 1). We next studied the cell proliferation pattern in the cyclic uterus of hamsters because apoptosis is followed by regeneration of new cells. We observed only a few proliferating cells in epithelial, stromal, and muscular compartments on the estrous day. There was a very small increase in the number of proliferating cells in epithelial and stromal layers on the metestrous day (Fig. 1). A noticeable increase in the number of stromal cell and glandular epithelial proliferations was noticed on the diestrous day, whereas no luminal epithelial cell proliferation was found (Fig. 1). On the proestrous day, however, we observed an increase in the number of proliferating luminal epithelial cells. Simultaneously, a reduction in the number of stromal and glandular proliferating cells was noticed on this day of the cycle. Cell proliferation in muscular layers remained unchanged on both the diestrous and proestrous days (Fig. 1).

Fig. 1.

Sections of hamster uteri from each day of the estrous cycle were immunostained to demonstrate cell-specific apoptosis and proliferation. Apoptosis were determined by TUNEL assay and active caspase staining. Cell proliferation was determined by BrdU incorporation. Arrow indicates positive immunostaining. Data represent two to three animals from each day of the cycle. cm, Circular myometrium; ge, glandular epithelium; le, luminal epithelium; s, stroma.

Separate effects of P₄ and E₂ in uterine cell apoptosis and proliferation

The effects of ovarian steroid hormones on uterine cell apoptosis and proliferation are shown in Figs. 2–4. TUNEL results revealed the presence of apoptotic cells more in the glandular epithelial cells and less in the luminal epithelial cells in the ovariectomized sesame oil-treated (vehicle for dissolving steroids) hamster (Fig. 2). However, there were more apoptotic cells in the luminal epithelium than the glandular epithelium in the mouse (Fig. 3). There were only a few apoptotic cells noticed both in the stroma and myometrium in both species. Although a single injection of P₄ or E₂ significantly attenuated the number of apoptotic cells in mice (Fig. 3), it has apparently no effect in hamsters (Fig. 2).

Fig. 2.

Sections of ovariectomized hamster uteri treated with vehicle, P₄ and E₂ were immunostained to demonstrate hormonal effects on cell-specific apoptosis and proliferation. Apoptosis were determined by TUNEL assay. Cell proliferation was determined by BrdU incorporation. Arrow head indicates positive immunostaining. Data represent at least two to three animals in each treatment group. cm, Circular myometrium; ge, glandular epithelium; le, luminal epithelium; lm, longitudinal myometrium; s, stroma.

Fig. 4.

Dark-field pictures of uterine sections showing tritiated thymidine incorporation in the nuclei of cells after treatment of sesame oil (vehicle), P₄, and/or E₂ to ovariectomized hamsters. Data represent two animals in each treatment group. cm, Circular myometrium; le, luminal epithelium; lm, longitudinal myometrium; s, stroma.

Fig. 3.

Sections of ovariectomized mice uteri treated with sesame oil (vehicle), P₄, and E₂ were immunostained to demonstrate hormonal effects on cell-specific apoptosis and proliferation. Apoptosis was determined by TUNEL assay. Cell proliferation was determined by BrdU incorporation. Arrowhead indicates positive immunostaining. Data represent at least two to three animals in each treatment group. cm, Circular myometrium; ge, glandular epithelium; le, luminal epithelium; lm, longitudinal myometrium; s, stroma.

In regard to the uterine cell proliferation in ovariectomized mice and hamsters, we noticed only a few scattered [³H]thymidine-positive (Fig. 4) or BrdU-positive (Figs. 2 and 3) endometrial cells in the uterus of sesame oil vehicle-treated animals. Although a single injection of P₄ did not cause any proliferation of uterine cells in the ovariectomized hamster at 24 h (Figs. 2 and 4), an injection of the same dose of P₄ to the ovariectomized mouse showed significant increase in stromal cell proliferation at 24 h (Fig. 3). Treatment with P₄ for 48 h (once a day for 2 d), however, showed a moderate increase in the number of proliferating uterine stromal cells in hamsters (Fig. 2) compared with oil-treated control and P₄ treatment for 24 h. A similar P₄ regimen in ovariectomized mice reduced the proliferating stromal cell numbers at 48 h compared with proliferating stromal cell numbers at 24 h after P₄ treatment (Fig. 3). A single injection of E₂ to ovariectomized hamsters stimulated proliferation in both the glandular and luminal epithelia but not in the stroma, as detected by the BrdU incorporation method (Fig. 2). A single injection of E₂ to ovariectomized mice was also effective in stimulating epithelial and glandular cell proliferation (Fig. 3). However, treatment of a single injection of E₂ together with P₄ or to a P₄-primed animal resulted in a significant increase in the number of [³H]thymidine-labeled stromal nuclei by 24 h, suggesting that the presence of E₂ enhances P₄ actions on stromal cell proliferation (Fig. 4). However, some epithelial DNA synthesis was also detected in these uteri (Fig. 4). These results suggest that uterine cell apoptosis and proliferation processes are regulated by sex steroids in a species-specific manner.

Uterine cell apoptosis and proliferation during first 8 d of pregnancy in relation to preparation of the receptive uterus for implantation, initiation of implantation, and stromal decidualization and progression of implantation process for placentation

Apoptosis during d 1–4 of pregnancy in hamsters and mice

TUNEL and caspase-3 staining results during d 1–4 of early pregnancy are presented in Figs. 5–7. During the days of uterine preparation for implantation, TUNEL staining was primarily observed in uterine glandular and luminal epithelial cells on d 1 of pregnancy in hamsters (Fig. 5). Although TUNEL staining was not found in the negative control without the TdT, it was present in all cells in the positive control treated with DNase (data not shown). Caspase-3 staining almost parallels TUNEL assay on d 1 and showed its presence in the luminal and glandular epithelial cells (Fig. 5). The negative control without the primary caspase-3 antibody showed no staining on d 1 (data not shown). We next tried to colocalize TUNEL staining with expression of caspase-3 in a d-1 uterine section to demonstrate caspase-mediated cell death in the uterus. We observed that cells expressing caspase-3 correspond to the cells showing TUNEL staining (Fig. 6). These findings suggest that most of the caspase-positive cells were undergoing apoptosis on d 1. Compared with d 1 of pregnancy, the number of apoptotic cells in epithelial cells was reduced on d 2 of pregnancy but still remained high compared with d 3 and 4 of pregnancy (Fig. 5). Although the number of apoptotic cells was less in the stromal bed, apoptotic stromal cells were more on d 2 compared with d 1, 3, and 4. Caspase-3 staining was not observed in the epithelial cells on d 2, but some cells in the stroma and muscle layers showed caspase-3 staining. Only a few TUNEL- and caspase-positive cells were present in the endometrial compartment of the uterus on d 3 and 4 (Fig. 5). In the mouse, however, apoptosis was primarily observed in both the stromal and luminal epithelial layers on d 2 compared with other days (d 1, 3, and 4) of pregnancy (Fig. 7). The expression of caspase-3 was positively correlated with cell apoptosis showing maximum staining in luminal epithelial cells on d 2 of pregnancy (Fig. 7).

Fig. 5.

Sections of hamster uteri from d 1–4 of pregnancy were immunostained to demonstrate cell-specific apoptosis and proliferation. Apoptosis was determined by TUNEL assay and active caspase staining. Cell proliferation was determined by BrdU incorporation. Data represent at least two animals on each day of pregnancy. Arrowhead indicates positive immunostaining. cm, Circular myometrium; ge, glandular epithelium; le, luminal epithelium; lm, longitudinal myometrium; s, stroma.

Fig. 7.

Sections of mice uteri from d 1–4 of pregnancy were immunostained to demonstrate cell-specific apoptosis. Apoptosis was determined by TUNEL assay and active caspase staining. Arrowhead indicates positive immunostaining. Data represent at least two to three animals on each day of pregnancy. cm, Circular myometrium; ge, glandular epithelium; le, luminal epithelium; lm, longitudinal myometrium; s, stroma.

Fig. 6.

Colocalization of caspase-3 (red) and TUNEL (green) labeling as detected by double immunofluorescence in sections from d-1 uterus of the hamster. Note the colocalization of caspase-3 and TUNEL in the merger (arrowhead). Photographs were captured at ×200 magnification. le, Luminal epithelium; s, stroma.

Apoptosis during d 5–8 of pregnancy in hamsters and mice

Uterine cell apoptosis during d 5–8 of pregnancy in hamsters and mice is shown in Figs. 8–10. After the initiation of implantation on d 5, we observed a few focal apoptotic cells in the uterine epithelium surrounding the implanting blastocyst in hamsters (Fig. 8) as well as in mice (Fig. 9). We also observed apoptotic cells in the trophoblast cell layers in both species (Figs. 8 and 9). There was no indication of apoptosis in stroma as well as luminal epithelial cells away from the implantation chamber in hamsters (Fig. 8). However, we noted the presence of a few apoptotic cells in the subepithelial stromal layer surrounding the implanting embryo in the mouse (Fig. 9). Although a few caspase-3-positive luminal epithelial cells were noticed surrounding the implanting blastocysts in mice (Fig. 9), only one or two caspase-3-staining cells were found in this area in hamsters (Fig. 8). Double-immunofluorescence analyses were next performed to examine the colocalization of caspase-3 and TUNEL labeling in d-5 implantation sites of the hamster and mouse (Fig. 10). The results showed the localization of TUNEL labeling in the luminal epithelial cells surrounding the implanting blastocyst almost corresponded to that of caspase-3 protein staining in most of these cells.

Fig. 8.

Sections through implantation sites of hamsters from d 5–8 of pregnancy were immunostained to demonstrate cell-specific apoptosis and proliferation. Apoptosis was determined by TUNEL assay and active caspase staining. Cell proliferation was determined by BrdU incorporation. Arrowhead indicates positive immunostaining. AM, Antimesometrial side; bl, blastocyst; cm, circular myometrium; em, embryo; M, mesometrial side; PDZ, primary decidual zone; SDZ, secondary decidual zone.

Fig. 10.

Colocalization of caspase-3 (red) and TUNEL (green) labeling as detected by double immunofluorescence in sections from d-5 implantation sites of hamsters and mice. Note the colocalization of caspase-3 and TUNEL in the merger (arrowhead). Photographs were captured at ×200 magnification. bl, Blastocyst; le, luminal epithelium; s, stroma.

Fig. 9.

Sections through implantation sites of mice from d 5–8 of pregnancy were immunostained to demonstrate cell-specific apoptosis. Apoptosis was determined by TUNEL assay and active caspase staining. Arrowhead indicates positive immunostaining. AM, Antimesometrial side; bl, blastocyst; em, embryo; ge, glandular epithelium; M, mesometrial side, PDZ, primary decidual zone; SDZ, secondary decidual zone.

On d 6, TUNEL staining was observed in luminal epithelium which starts to degenerate surrounding the implanting embryo in both the species (Figs. 8 and 9). We also observed apoptotic cells in the PDZ. The apoptotic pattern was very similar on d 7 in both the species. Althoughapoptosis was noticed in decidual cells surrounding the embryo, luminal epithelial cells on the mesometrial side also showed apoptosis. On d 8, apoptotic cells were more prominent in the hamsters compared with mice surrounding the growing embryo (Figs. 8 and 9). TUNEL-positive uterine cells were mainly observed just surrounding the embryo. Apoptotic cells were also seen in the growing embryo. Caspase-3 expression in uterine cells was almost positively correlated with the apoptotic patterns on d 6–8 in mice. However, we observed fewer of caspase-3-positive cells compared with apoptotic cells in hamsters from d 5–8 of pregnancy (Fig. 8).

Cell proliferation during d 1–4 of pregnancy in hamsters

Cell proliferation in the hamster uterus was studied by applying two well known methods: BrdU and [³H]thymidine incorporations. Only a few proliferating luminal epithelial cells were noted in d-1 uterine section by the [³H]thymidine incorporation method (Fig. 11), but not by BrdU incorporation (Fig. 5). Proliferating cells as detected by the BrdU incorporation method were equally distributed in all compartments of the uterus as observed in the first day of the estrous cycle, the estrous day. Uterine cell proliferation was dramatically increased on d 2 of pregnancy. Proliferation was seen mainly in the luminal and glandular epithelial cells (Figs. 5 and 11). The negative control section obtained from the animal without the treatment of BrdU showed no staining on d 2 (data not shown). We also saw some increase in the number of proliferating cells in stromal and muscular compartments by [³H]thymidine incorporation (Fig. 11) but not by the BrdU incorporation method (Fig. 5). Proliferation of epithelial cell completely ceased after d 2 of pregnancy. In contrast, we noticed a significant increase in the number of proliferating cells in the stromal compartment on d 3 and 4 of pregnancy (Figs. 5 and 11). Proliferating cells were also observed in muscle layers from d 4.

Fig. 11.

Dark-field pictures of uterine sections were showing [³H]thymidine incorporation in the nuclei of cells during d 1–6 of pregnancy in hamsters. AM, Antimesometrial side; bl, blastocyst, cm, circular myometrium; em, embryo; ge, glandular epithelium; lm, longitudinal muscle; M, mesometrial side; PDZ, primary decidual zone; SDZ, secondary decidual zone.

Cell proliferation during d 5–8 of pregnancy in hamsters

With the initiation of blastocyst implantation on d 5, an additional increase in stromal cell proliferation was observed at the implantation site surrounding the implanting blastocyst. Increased cell proliferation was also noticed in muscle layers. Implanting blastocysts also showed cellular proliferation (Figs. 8 and 11). However, with the gradual progression of the implantation process and transformation of stromal cells to decidual cells from d 5 to d 8, gradual inhibition of stromal cell proliferation was noticed surrounding the implanting embryo (Figs. 8 and 11). However, stromal cells further away from the implanting embryo still showed proliferation. A survey of serial sections showed the presence of some proliferating cells in the embryo as well in muscle layers (Figs. 8 and 11).

A close similarity was observed in uterine cell proliferation patterns during d 1–6 of pregnancy using two well established methods. Only a subtle difference in uterine cell proliferation pattern between these two methods was noticed on d 3 of pregnancy. Incorporation of [³H]thymidine was noticed more in stromal cells just beneath the luminal epithelium, compared with equal distribution of BrdU incorporation in the stromal area. We observed more proliferating cells by the [³H]thymidine incorporation method compared with the BrdU incorporation method, suggesting that the former method is more sensitive than the latter one (Figs. 5, 8, and 11).

Caspase-3 inhibitor Ac-DEVD-CHO blocked blastocyst implantation in hamsters and mice

Because the activation of caspase-3 plays an important role in executing apoptosis, we next determined whether its inhibitor, Ac-DEVD-CHO, inhibits implantation by blocking luminal epithelial cell apoptosis. Compared with the vehicle treatment, uterine horns from five pregnant hamsters and six pregnant mice that received administration of Ac-DEVD-CHO showed complete absence of implantation sites on d 5 as determined by the blue-dye method (Table 1 and Fig. 12). A total of six and 26 blastocysts were recovered upon flushing drug-treated uterine horns of hamsters and mice, respectively (Table 1). Because hamster blastocysts do not undergo delay in implantation, these embryos mostly degenerate in the event of interruption of the implantation process. Hence, blastocyst recovery from drug-treated hamster uterine horns was low, and only six apparently normal looking zona-free blastocysts were recovered from three horns. Twenty-six blastocysts that were recovered from six drug-treated uterine horns in mice had no sign of degeneration. Twenty of those blastocysts were zona free and the rest were zona encased. We observed a normal number of implantation sites in vehicle-treated uterine horns of both species. This drug is not reported to be toxic in mice (34). This drug is also not toxic to preimplantation embryos because normal blastocyst formation occurs when eight-cell embryos of hamsters and mice were cultured in vitro in the presence of this compound at doses of 5 and 10 μg/ml (Table 2). Under our culture conditions, about 87% of hamster and 94% of mouse eight-cell embryos developed into blastocysts in the absence of Ac-DEVD-CHO (vehicle control). In the presence of Ac-DEVD-CHO, we observed no significant change in embryo development from eight-cell stage to the blastocyst stage (Table 2), suggesting no inhibitory or toxic effects of Ac-DEVD-CHO on embryo development. The morphological appearance of those blastocysts that formed in the presence of the drug was indistinguishable from blastocysts that developed in the absence of the drug. These results demonstrate that intraluminal administration of Ac-DEVD-CHO inhibits implantation in hamsters and mice by inhibiting uterine epithelial cell apoptosis at the implantation site.

TABLE 1.

Effect of caspase-3 inhibitor Ac-DEVD-CHO on blastocyst implantation in hamsters and mice

Treatment	No. of pregnant animals	No. of uterine horns	Total no. of implantation sites(mean ± SEM)
Hamster	5
Vehicle-treated horn		5	30 (6.00 ± 0.63)
Inhibitor-treated horn		5	1.0
Mouse	6
Vehicle-treated horn		6	33 (5.45 ± 0.10)
Inhibitor-treated horn		6	00

Drug (hamster: 10 μg/5 μl saline per horn; mice: 10 μg/2 μl saline per horn) and vehicle (hamster: 5 μl saline per horn; mice: 2 μl saline per horn) were administered intraluminally in separate uterine horn of the same animal on d 3 of pregnancy at 0900 h. Implantation sites were determined by the blue dye method on d 5 at 0900 h. A total of six blastocysts were recovered from uterine horns that did not show implantation sites on d 5 of pregnancy in the presence of Ac-DEVD-CHO. Because hamsters do not exhibit delay in implantation, blastocysts that do not implant in normal time usually degenerate. A total of 26 blastocysts were recovered from uterine horns that did not show implantation sites on d 5 in the presence of Ac-DEVD-CHO in mice.

Fig. 12.

An intraluminal application of caspase-3 inhibitor AC-DEVD-CHO inhibits implantation both in the hamster and mouse. Pregnant hamsters and mice received a single intraluminal injection of AC-DEVD-CHO (10 μg/5 μl sterile saline in hamsters or 5 μg/2 μl sterile saline in mice) in one uterine horn and vehicle in the contralateral horn on d 3 of pregnancy when embryos were still inside the oviducts. Animals were examined for implantation sites at 0900 h on d 5 by the blue dye method. Arrowheads indicate the location of implantation sites in vehicle-treated uterine horns.

TABLE 2.

Development of blastocysts after culture of eight-cell embryos in the absence or presence of caspase-3 inhibitor Ac-DEVD-CHO

Ac-DEVD-CHO (μg/ml)	No. of eight-cell embryos cultured (no. of observations)	Duration of culture	No. of embryos developed to blastocysts	% Blastocysts developed
Hamster		24 h
0	63 (5)		55	87
5	37 (3)		31	84
10	75 (6)		64	85
Mouse		48 h
0	46 (5)		43	94
10	48 (5)		44	92

Eight-cell embryos were cultured in a group of 10–15 embryos per drop for hamsters and 8–10 embryos for mice. Ac-DEVD-CHO was first dissolved in water and later diluted in respective cultured medium for hamster and mouse embryos. No significant differences (P < 0.05) were observed between embryos cultured in the absence or presence of the caspase-3 inhibitor.

Discussion

In this study, we aimed to establish a correlation between apoptosis and proliferation in uterine cell turnover during the estrous cycle and early pregnancy in hamsters that show P₄-dependent implantation. The salient result of this study is the evidence of significant epithelial cell proliferation on the second day (d 2) after mating, marking entry of the uterus into the state of pregnancy in hamsters. This is also true in mice considering previously published results that showed significant increase in luminal epithelial cell proliferation on d 2 of pregnancy in this species (1). In general, an inverse relationship between apoptosis and cell proliferation patterns was noted except on d 2 of pregnancy in mice where both the proliferation (1) and apoptosis processes were happening simultaneously in epithelial cell layers. The frequency of apoptosis in the stromal bed was not similar to the frequency of proliferation during pregnancy, suggesting that a higher frequency of cell proliferation may well be related to rapid growth of the uterus during pregnancy. In respect to steroid hormonal regulations of the apoptosis and proliferation patterns in the uterus, our results clearly showed that E₂ was effective in stimulating epithelial cell proliferation in both species. However, although P₄ elicited stromal cell proliferation within 24 h in mice, it was not as effective in hamsters and took 48 h to obtain a moderate increase in stromal cell proliferation. In addition, although both P₄ and E₂ were ineffective in inhibiting apoptosis observed in ovariectomized hamsters, they were effective in inhibiting uterine cell apoptosis in ovariectomized mice. Our results also revealed the colocalization of caspase-3 in most of the apoptotic cells during early pregnancy. Thus, our observation that caspase-3 inhibitor Ac-DEVD-CHO adversely affects the implantation process shows, for the first time, that caspase-3-mediated uterine epithelial cell apoptosis may play an important role in initiating the implantation process in both the hamster and mouse.

Previously reported data demonstrated that uterine epithelial cells undergo proliferation on the proestrous day of the estrous cycle of mice and rats (18). This pattern is partly synchronous and corresponds to the period when serum E levels are increasing (37). Consistent with these data, we observed luminal epithelial cell proliferation on the morning of the proestrous day in hamsters. However, this could be attributed to a gradual increase in E levels that peak at 1400 h and remain elevated for several hours on the diestrous day (38). Apparently, the second and large preovulatory E surge that occurs on the estrous day between 0900 and 1500 h has no effect on further uterine epithelial cell proliferation in hamsters because there was no significant epithelial cell proliferation at the estrous day. Proliferation of epithelial cells in response to E₂ was demonstrated in mice and rats (1, 39). We also observed proliferation of epithelial cells in response to E₂ in ovariectomized hamsters and mice, suggesting that proliferation of epithelial cells in response to E₂ is a common phenomenon in the uterus. This was contradictory to the previous belief that the hamster uterus is not quite as sensitive to E₂ compared with rats (40). The rapid reduction in uterine epithelial cell proliferation was accompanied by the gradual increase in apoptosis on the estrous and metestrous days. These epithelial changes were noted when serum E levels were at their lowest limit after a peak on the proestrous day (39). This is consistent with the data published previously that E withdrawal causes degenerative changes in the epithelial cells (25, 26). The hamster shows stromal cell proliferation on the diestrous day when the first peak of P₄ occurs (41, 42). These results are in agreement with previously reported data that showed uterine stromal cell proliferation in response to P₄ alone or when P₄ and E₂ were given together, P₄ significantly suppressed the epithelial cell proliferation but increased the proliferation of stromal cells in mice (1, 24). We also observed a similar effect of E₂ and P₄ in the ovariectomized hamster uterus. It was reported previously that hamster uterus is more sensitive to P₄ than E₂ (40). However, our results did not quite support this belief because there was no noticeable increase in stromal cell proliferation by giving a single injection of P₄ at the dose of 2 mg to ovariectomized hamsters. The increase in stromal cell proliferation in response to E₂ plus P₄ suggests that, whereas P₄ suppresses the E₂ actions on epithelial cell proliferation, E₂ also potentiates P₄ actions on the stroma. However, the reason for the diminished and delayed effectiveness of P₄ in the hamster uterus in terms of stromal cell proliferation was not certain from our studies and will require further investigation.

Uterine cell proliferation during early pregnancy in mice has been previously studied and showed epithelial, but not stromal, cell proliferation on the second day of pregnancy and stromal, but not epithelial, cell proliferation on the third and fourth days of pregnancy (1). We observed a similar pattern of uterine cell proliferation on d 2–4 of pregnancy in hamsters, suggesting that uterine cellular proliferating patterns in achieving uterine receptivity are similar in mice and hamsters. On the basis of these observations, it can be suggested that both in the mouse and hamster, mating on the day of ovulation leads to a significant increase in luminal epithelial cell proliferation on the following day (d 2 of pregnancy), marking the end of cycle-related and the beginning of pregnancy-related uterine changes. Another interesting finding with respect to cell apoptosis between mice and hamsters was that although epithelial apoptosis and caspase-3 activity were highest on d 1 of pregnancy in hamsters, there was almost no apoptosis in d-1 uteri of mice. On d 2 of pregnancy, however, the epithelial apoptosis and caspase-3 activity was at the maximum level in mice when epithelial proliferation is also reported to be occurring (1). This pattern is slightly different in hamsters. Fewer epithelial cells were undergoing apoptosis on d 2 in hamsters compared with mice. Maximum uterine cell apoptosis on d 1 of pregnancy in hamsters is well correlated with the low level of circulating E on this day (42, 43). Because circulating E levels gradually start increasing from d 2 of pregnancy in hamsters (42, 44), epithelial cells clearly receive an E stimulus to divide on d 2 in hamsters. If the fall of proestrous E surge is responsible for the induction of apoptosis in epithelial cells on d 2 of pregnancy, then the cause of epithelial cell proliferation on this day in mice remains a matter of debate because there was no significant change in circulating E level on this day (27). However, a previous study suggested that d-2 uterine epithelial cell proliferation in mice could also be a result of proestrus E-stimulated epithelial c-myc expression on d 1 of pregnancy (1). The occurrence of proliferation in the stroma and absence of apoptosis on d 3 and 4 of pregnancy in hamsters is probably the result of the gradual increase in circulating E and P₄ levels during early pregnancy in this species (42, 43). On the other hand, in mice, stromal cell proliferation on d 3 and 4 could be solely due to the increase in P₄ levels during early pregnancy (44). However, it has been postulated that the brief increase in E levels on d 4 of pregnancy in mice (28) may accelerate the P₄-induced stromal proliferation.

The immediate consequence of blastocyst implantation is the stimulation of stromal cell proliferation and death of luminal epithelium surrounding the implanting blastocyst (2). We observed that both of these events were occurring at the implantation sites of both mice and hamsters. The first epithelial cell death perhaps starts in a few cells at the antimesometrial side and then extends to all epithelial cells surrounding the embryo. From d 6–8, epithelial cell death also occurred in the mesometrial side of the luminal epithelium. Although the luminal epithelium starts apoptotic changes on d 5, stromal cells surrounding the implanting blastocyst simultaneously undergo massive proliferation. As gestation progresses, stromal cells immediately surrounding the implanting blastocysts stopped proliferation and showed sign of apoptosis, whereas stromal cells next to it still showed proliferation. Electron microscopic studies showed that deteriorating decidual cells were removed by autolytic activity and that their elimination was facilitated by trophoblastic phagocytosis (45). This elimination of decidual cells surrounding the implanting blastocyst is absolutely required to remodel the uterus to accommodate the growing embryo. Although we saw stromal cell death and proliferation in different regions of the decidua, we saw more proliferating stromal cells than dying stromal cells, suggesting net growth of the uterus after implantation. Ovarian hormones are the key regulatory elements for uterine apoptosis and proliferation during the cycle and preimplantation period, but what induces uterine epithelial cell apoptosis and stromal cell proliferation at the implantation site is unknown. Because the blastocyst initiates uterine processes for implantation, embryo-derived molecules certainly play key roles. One of those embryonic molecules in hamsters is possibly E (32, 46). Although E-induced apoptosis has not been demonstrated in the uterus, E-induced spatial and regional cell apoptosis has been demonstrated in pituitary and prostate glands (47, 48). Furthermore, antiestrogen inhibits implantation in hamsters (49). Thus, depending on the circumstances, it is possible that E produced by the embryo may induce epithelial cell apoptosis and stromal cell proliferation at the implantation site. However, these processes at implantation sites may also require the involvement of cooperative actions of numerous local paracrine, autocrine, or juxtacrine factors to execute the uterine remodeling process to accommodate the growing embryo.

Previous studies have proposed the involvement of caspase-3, a downstream executioner enzyme common to many paradigms of programmed cell death in mediating apoptosis of both germ and somatic cells (50). Because caspase-3 is an earlier event than DNA fragmentation and the apoptotic process lasts only a few hours (51), it is tempting to speculate that caspase-3 remains active in cells during DNA fragmentation. In our studies, although we saw colocalization of caspase-3 and TUNEL labeling in the same cells on d 1, we observed lower numbers of caspase-3-positive cells at the implantation sites as compared with TUNEL assay in hamsters. This is not surprising, because both techniques might yield different results in certain cell types and even in the same cell type on different times and days of pregnancy, depending on the stimuli used or due to different pathways in the apoptotic process in which caspase-3 may or may not be involved. Using caspase-3 knockout mice, it has been demonstrated that although caspase-3 is required for granulosa cell apoptosis during follicular atresia, it is dispensable for germ cell apoptosis in females, suggesting coexistence of more than one cell death mechanism (52). Luminal epithelial cell loss at the implantation site via apoptosis is increasingly recognized as a key component of implantation for trophoblast cell invasion and contact with the stromal/decidual cells. We studied the importance of uterine epithelial cell apoptosis in implantation by inhibiting epithelial cell apoptosis using a caspase-3 inhibitor. Our results showed that Ac-DEVD-CHO completely blocked implantation in the treated uterine horn as compared with the vehicle-treated contralateral uterine horn in both hamsters and mice. Thus, it appears that caspase-3-mediated uterine epithelial cell apoptosis plays an important role during the time of establishment of implantation in hamsters.

In summary, our comparative studies between hamsters and mice show that apoptosis is present in all cell types during the estrous cycle and periimplantation uterus, but the incidence of apoptosis was lower in the stroma and myometrium than the epithelium. There was a significant correlation of uterine cell apoptosis and proliferation with circulating steroid levels in hamsters. However, the reduced effectiveness of progesterone in ovariectomized hamsters needs further investigation. After initiation of implantation, epithelial cells only exhibited death, whereas stromal cells showed compartment-specific death and proliferation in response to growing embryo. Both in the mouse and hamster, stromal cells around the blastocyst first showed proliferation and then death a day later. This process starts at the subepithelial stroma and gradually moves toward the periphery. Substantial similarities have been noticed between the mouse and hamster in the pattern of uterine cell apoptosis and proliferation during early pregnancy. Further studies will confirm whether a change in the ratio between proapoptotic and antiapoptotic regulators control uterine cell death and proliferative mechanisms. Our findings of almost parallel expression of caspase-3 with TUNEL labeling and inhibition of implantation by caspase-3 inhibitor indicate, for the first time, that caspase-3-mediated cell apoptosis may play an important role in initiating the implantation process in hamsters and mice.