In situ hybridization and immunofluorescence were used to study the expression of a lymphocyte poreforming protein (perforin) in the uterus of pregnant mice. Cells expressing perforin mRNA were detected as early as gestation day 5, whereas perforin protein was detected one or two days later. Although the number of cells expressing both perforin mRNA and perforin protein varied subsequently with time, they were consistently observed from the day of implantation until parturition. The highest levels of mRNA expression were observed sometime dining midgestation. The high abundance of this cytolytic protein in the metrial gland during pregnancy and the time course of its expression thus suggest that GMG perforin expression is tightly regulated, probably hormonally, by the uterus, and that GMG cell perforin plays an important role in the normal completion of pregnancy.
As a fetus inherits paternal transplantation antigens, it can be viewed as a semiallogeneic graft that survives fastened to the maternal uterus until birth, but it is not understood how the fetus is protected from the maternal immune response (Beer and Billingham, 1971; Pence et al. 1975; Smith and Powell, 1977; Clark and McDermott, 1981; Chaouat, 1986; Baines and Gendron, 1990; Redman, 1990). In addition, barriers must exist to prevent transmission of viral infections to the fetus, and these have yet to be properly identified.
The metrial gland is a transient uterine tissue that develops near the placenta in pregnant rodents and consists primarily of two cell types, a conspicuous granulated metrial gland (GMG) cell and a fibroblast-like stromal cell; both cell types are believed to be maternal in origin. The metrial gland starts to develop near the embryo a few days after implantation and is associated with the major blood vessels supplying the placenta. Since the metrial gland is thought to be involved in the immunology of pregnancy (Mitchell et al. 1981; Mitchell and Peel, 1987; Parr et al. 1987), it became of interest to determine whether cytotoxic mediators from the cellular branch of immunity might be present. Specifically, the presence of a pore-forming protein, called perforin or cytolysin (Henkart, 1985; Podack, 1985; Young, 1989; Tschopp and Nabholz, 1990), was measured during the course of pregnancy.
Perforin had previously been identified in the cytoplasmic granules of in vitro cell lines of cytotoxic T lymphocytes (CTL) and natural killer (NK) cells. The purified protein has potent lytic activity, and it can lyse a wide variety of targets by forming tubular lesions ultrastructurally resembling those produced by complement (Henkart, 1985; Podack, 1985; Young, 1989; Tschopp and Nabholz, 1990); by primary sequence analysis, it was also found that perforin is homologous to complement (Stanley and Luzio, 1988). In vivo, perforin appears to be produced by the effector lymphocytes primarily in situations where high local concentrations of interleukin-2 (IL-2) are generated (Ojcius and Young, 1990), such as in autoimmune disorders, transplant rejections, cancer, and acute viral infections.
The observations described here are the outgrowth of research in our laboratory on the in vivo function and distribution of perforin. Using the in situ hybridization technique, which allows for detection of perforin mRNA transcription at the single cell level during different stages of pregnancy, we here show that the metrial gland contains larger numbers of perforin-synthesizing cells than formerly observed in any of the pathological conditions studied until now (Young et al. 1989a,6, 1990; Müller et al. 1989). These measurements were made for perforin mRNA from day 5 of gestation (just after implantation) to day 19 (right before parturition).
Materials and methods
Pregnant Swiss mice were used (Charles River Breeding Laboratories, Wilmington, MA). The day at which a vaginal plug was found was designated as day 0 of gestation. The animals were deeply anesthesized with methoxyflurane and killed by rapid decapitation on the indicated day of gestation.
In situ hybridization technique
Sixty four uteri were taken at embryonic ages day 4 to day 19 from 32 pregnant mice. Cryostat sections (6 mn) were mounted onto polylysine-coated slides and fixed in 4 % paraformaldehyde, 0.1M phosphate-buffered saline (PBS) for 5 min, rinsed, and dehydrated in increasing concentrations of ethanol. Slides were then acetylated in 0.25% acetic anhydride containing 0.1M triethanolamine (pH 8.0,10 min), rinsed for 10 min with 0.2×SSC (SSC is 0.15 M NaCl, 0.015 M sodium citrate, pH 7.0), and dehydrated in increasing concentrations of ethanol. Sections were incubated for 3–4 h with prehybridization buffer, followed by a heat-denatured hybridization buffer (McCabe et al. 1986) containing 36S-labeled sense or antisense ribonucleotide probe. The riboprobes were transcribed from a 1600 base-pair (bp) fragment of the mouse perforin gene (Kwon et al. 1989) cloned into a transcription vector with bacteriophage T7 and T3 promoters arranged to produce riboprobes of both orientations. Hybridization continued for 2 days at 50°C, and sections were subsequently rinsed with 2×, l× and 0.5×SSC, treated with RNase A, rinsed with RNase-free buffer, rinsed again with SSC, and finally incubated in 0.1×SSC, 10mM dithiothreitol, at 45°C overnight. The sections were then rinsed for 1 min each in 300 mM ammonium acetateiethanol, 1:1, 3:7, 1:9 (v/v), and lastly in absolute ethanol. After drying, the sections for dark-field photomicroscopy were dipped in Kodak. NTB2 emulsion (38 °C), and stored dry in light-tight boxes at 4 °C for 4 days. Autoradiograms were developed, fixed, hematoxylin counter-stained, and examined with a Zeiss photomicroscope.
Northern blot analysis
Uteri were removed from pregnant mice at the indicated times after the beginning of gestation, dissected as described for preparation of tissues for in situ hybridization, and stored frozen for isolation and analysis of total RNA (Chirgwin et al. 1979). A 20 μg sample of RNA was either transferred directly to a GeneScreen Plus membrane (New England Nuclear) for dot analysis or fractionated on a 1 % agarose-formaldehyde gel before transferring to the membrane. In either case, the membrane was then probed with a 32P-labeled cDNA probe for perforin (Kwon et al. 1989). Prehybridization and hybridization were performed at 42°C in a solution containing IM NaCl, 50% formamide, 10% dextran sulfate, 1% SDS and 100 μg ml-1 of salmon sperm DNA. After hybridization, the blots were washed twice with 2×SSC/l% SDS at room temperature for 15min, followed by two washes with 0.5×SSC/l % SDS at 60°C.
Tissue preparation and immunostaining were performed by a technique as previously described (Zheng et al. 1988), with polyclonal rabbit antiserum against purified perforin (Young et al. 1986a) as the primary antibody. Briefly, the cryostat sections were fixed either with 4% paraformaldehyde in 0.1M PBS for 10 min or Zamboni’s fixative for 20 min. After rinsing, the sections were preincubated with 2% normal goat serum for 30 min, after which they were incubated in the primary antiserum for 48 h at 4°C. Finally, the sections were incubated in a 1:1000 dilution of sheep anti-rabbit IgG coupled to fluorescein isothiocynate (Aldrich Chemical Co., Milwaukee, WT) for Ih at room temperature.
The specificity of labelling was tested by two procedures. First, the tissue was incubated with rabbit pre-immune serum instead of primary antiserum against perforin; and second, the incubation in primary antiserum was omitted. No staining was found in either of these negative controls.
Specificity of in situ hybridization
Fig. 1A shows the histological structure of a whole uterus of a pregnant mouse at day 11 of gestation. At higher magnification, GMG cells are readily distinguished in the metrial gland and decidua basalis stained with the glycogen-specific dye PAS (data not shown; and Zheng et al. 1991), which reveals the presence of large numbers of cells containing glycoprotein-rich granules. By in situ hybridization with sense and anti-sense riboprobes for perforin mRNA, the specificity of the probe was next tested on two adjacent sections of the metrial gland. Ln comparison with the section hybridized with the antisense riboprobe (Fig. IB), a dark-field photomicrograph of the gland hybridized with the control, sense riboprobe did not yield any labeling (Fig. 1C). The cross-section (Fig. IB) shows that strongly labelled cells are present, indicating that the perforin message is highly abundant. Although perforin-positive cells were mostly visible in the metrial gland, scattered cells were also found to some extent in the decidua basalis. They were rarely seen to invade the layer of spongy trophoblasts, and none were found in the labyrinthine region of the placenta.
Time course of perforin gene expression in the uterus
As implantation normally takes place on day 4.5 of gestation, the search for perforin-expressing cells was begun on day 4. No perforin-expressing cells were detected in the uterus at day 4, whereas low levels of GMG cells expressing the perforin gene were already detected at day 5 (data not shown). These levels increased significantly by day 6 (Fig. 2A). Thereafter, there is a dramatic rise in the number of perforin-expressing cells as midgestation is approached (Fig-2 B-D). At day 11 (Fig. 2D), labeled cells tend to be clustered and to localize preferentially towards the mesometrial side of the metrial gland. By day 12 (Fig. 2E), the labeled cells become widely distributed throughout the metrial gland. The maximal number of perforin-positive GMG cells detected by in situ hybridization was observed sometime between days 9 and 12 (Fig. 2B-E), after which the signal began to diminish, reaching a minimum immediately before parturition, usually on day 20 (Fig. 2H-L). These experiments were performed for each day with two separate uteri, and qualitatively reproducible results were obtained each time. In addition, all of the in situ hybridization experiments described above were done in parallel with an adjacent section hybridized with the sense control, which was negative in all cases (data not shown).
Northern blot analysis of perforin expression
These results were complemented by Northern (RNA) blot (Fig. 3A) and dot blot (Fig. 3B) analysis. As noted with the in situ hybridization technique above, a faint signal is already evident on day 6 (Fig. 3). The intensity of the perforin band varies with time, confirming the in situ hybridization results.
The mRNA of a murine CTL line, CTLL-R8 (Kwon et al. 1989), was used as a positive control in Fig. 3A (lane 5), while mRNA extracted from the uterus before day 5 of gestation did not hybridize with the perforin probe (data not shown).
Relationship between perforin transcription and translation
The temporal relationship between perforin transcription and perforin translation is shown in the microphotographs in Fig. 4. A strong signal detectable by the in situ hybridization technique with riboprobe (Fig. 4A) was already seen early on day 6, whereas the presence of perforin protein as seen by immunofluorescence (Fig. 4B) is relatively sparse at the same time, suggesting that either there is a lag time between the expression of perforin mRNA and perforin protein or the amounts of perforin protein are too low to be detectable by immunocytochemistry. By visual inspection, it is seen that a maximal signal from in situ hybridization is reached by day 11, while the maximal signal from immunohistology does not appear until day 13. Fig. 4C-D shows the mRNA and protein expression on day 11. On day 19, the last day of gestation (Fig. 4E-F), there is a marked diminution in the expression of perforin mRNA, while the number of cells expressing perforin protein, although decreasing, is still large. For unknown reasons, the perforin-expressing cells also begin to aggregate shortly before term (e.g. Fig. 4F).
Perforin is a pore-forming protein known to be present in the granules of cytolytic lymphocytes. The purified monomer has been shown to bind to a wide array of cells, where it traverses the plasma membrane and polymerizes into nonselective, heterogeneous pores as large as 20 nm in diameter (Henkart, 1985; Podack, 1985; Young, 1989; Tschopp and Nabholz, 1990). According to the granule exocytosis model of cell-mediated killing, interaction between the effector and target cells leads to effector cell degranulation, which causes perforin to be released into the extracellular space between the effector and target cells. Pore formation on the target subsequently results in small ions and other solutes diffusing down their electrochemical gradient into the target, so that the cell swells and eventually bursts.
As most of the evidence for this model had been derived with CTL and NK cell lines maintained in long-term cultures in IL2-containing medium, it became crucial to determine the distribution of perforin in primary lymphocytes actively engaged in cell killing. Perforin has not been detected in resting cytolytic lymphocytes from spleen, liver or peripheral blood of healthy animals. In vivo, its presence has rather been restricted to pathological conditions where high local concentrations of IL-2 are generated, such as lymphocytic choriomeningitis virus-infected tissues (Müller et al. 1989; Young et al. 1989a), pancreata from non-obese diabetic mice (Young et al. 19895), and herpesvirus infections (Young et al. 1990).
The results presented in this work show the specific expression of perforin mRNA in the metrial gland, as judged from the absence of labelled cells with the control, sense riboprobe. Surprisingly, the density of perforincontaining cells in the uterus is higher than in any pathological condition studied to date (Young et al. 1989a,b, 1990; Müller et al. 1989). Perforin message is detectable as early as day 5, and the protein is expressed at highest levels during midgestation. Significantly, implantation usually takes place on day 4.5, which coincides with the onset of both decidualization and perforin synthesis in GMG cells. These observations suggest that perforin expression in GMG cells is most likely under the control of the same hormones that induce decidualization.
The in vivo target of this cytolytic protein has yet to be determined. Two possibilities come readily to mind: that GMG cell perforin may be required (1) to prevent vertical transmission of viral infections to the fetus or (2) to lyse aberrant trophoblasts that may otherwise invade the maternal side. Either of these possibilities would be in agreement with the view that GMG cells are NK or NK-like cells (Peel et al. 1983; Peel and Stewart, 1984; Mitchell and Peel, 1984; Mukhtar et al. 1989; Parr et al. 1990). This view has recently been reinforced through a study of the subcellular location of perforin in GMG cells. By transmission electron microscopy (Zheng et al. 1991), it was observed that GMG cells contain an abundance of granules, which often have a characteristic vesicular cap over a fine granular matrix, as previously found for NK cells (Young et al. 19866; Groscurth et al. 1987). In addition, the subcellular distribution of perforin in GMG cells was studied by immunoelectron microscopy using anti-perforin antibodies and protein A-gold as a second ligand (Zheng et al. 1991). It was found that the gold particles are distributed mainly in the granular homogeneous core, and that all of the granules contain perforin. The pattern of immunogold labelling of perforin in GMG cells is thus similar to that observed in NK cells (unpublished observations), and would be consistent with perforin being released from GMG cells through degranulation.
Lastly, as abundant levels of GMG perforin are produced in a normal uterus during pregnancy, it is likely that the perforin is usually required for the successful completion of pregnancy. At the present time, however, one cannot exclude the possibility that, under pathological conditions, GMG cells may also be lytic to other cells besides trophoblasts or virally infected cells. Accordingly, we are currently attempting to establish what role GMG cells may play in the pathology of pregnancy. For these studies, the most reliable and sensitive GMG cell marker currently available is the expression of the perforin gene.
We are grateful to Drs Zanvil A. Cohn and Ralph M. Steinman for constant support and encouragement. This work was supported by grants from the Juvenile Diabetes Foundation, the American Diabetes Association, the American Cancer Society, the Lalor Foundation, and the National Institutes of Health.