| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Reproductive Molecular Research Group,5 Department of Pathology, University of Cambridge, Cambridge CB2 1QP, United Kingdom
Department of Obstetrics and Gynaecology,6 University of Cambridge Clinical School, The Rosie Hospital, Cambridge CB2 2SW, United Kingdom
ABSTRACT
The endometrium is a dynamic tissue that undergoes periodic growth, remodeling and breakdown under the influence of ovarian steroid hormones. To investigate the molecular mechanisms underlying these processes, we used a murine model to mimic the decidualization and regression observed in humans. Ovariectomized mice were treated sequentially with steroid hormones, and subsequently, to induce decidualization, oil was injected into the uterine lumen. The animals were then divided into progesterone-maintained and progesterone-withdrawal groups. In the latter group, a process similar to menstruation was induced. The uterine tissues were collected at several time-points after the induction of decidualization. Histological analysis demonstrated that decidualization and tissue degeneration were successfully induced with similar features to those observed during the human menstrual cycle. Immunohistochemical, morphometric, and microarray-based techniques were used to study the cellular and molecular changes. The volume fractions of leukocytes, macrophages, and neutrophils, but not endothelial cells, increased in decidualized uteri and decreased after major tissue degradation was completed. The microarray data show that the levels of many transcripts that encode immune-related factors changed during the time-course used for this model, and the transcript levels of many of these factors paralleled the changes observed in the volume fractions of the immune cells. The results of the present study suggest that this model is a useful alternative to the use of non-human primates. Our findings also show that immune cells are recruited into the menstruating endometrium, and that immune-related genes are regulated in the uterus throughout menstruation.
female reproductive tract, growth hormone, immunology, uterus
The endometrium is a dynamic tissue that undergoes periodic growth, remodeling, and breakdown, with these changes being orchestrated by the ovarian steroid hormones. The human menstrual cycle can be divided into three phases: the proliferative, secretory, and if embryo implantation does not occur, the menstrual phase. This last phase is characterized by a rapid decline in progesterone and estrogen levels, leading to numerous changes in the endometrium, including constriction of the coiled spiral arterioles, which cuts off the blood flow to the functional layer of the endometrium. These events lead to rapid tissue breakdown and the loss of the functional layer of the endometrium [1, 2]. Histological analyses show hemorrhage, fragmentation of the stroma, disruption of vessels, and infiltration of neutrophils. Tissue shedding is followed by regeneration, and re-epithelialization occurs by extension of the residual glandular epithelium over the denuded surface [2].
Menstruation represents a unique example of endocrine regulation of complex tissue remodeling that may lead in humans to menstrual disorders, such as menorrhagia (heavy menstrual bleeding) and dysmenorrhea (painful periods), which constitute a significant burden of disease affecting 10–30% and 43–90% of women of reproductive age, respectively [3]. Unraveling the molecular mechanisms at work in the menstruating endometrium remains a significant challenge, in part due to the lack of readily available model systems.
There are similarities in menstruation between human and nonhuman primates, making these animals suitable for investigations of the mechanisms that regulate menstruation. However, nonhuman primates are very expensive to maintain, and ethical considerations limit their use. Among nonprimates, the elephant shrew (Elephantus myuras jamesoni) and the bat (Glossophaga soricina) have been reported to menstruate [4]. However, neither of these species can be easily kept and handled. Therefore, utilizing common laboratory rodents to establish a reproducible model for menstruation that is cheaper and easily accessible is an attractive alternative. Furthermore, as these animals are amenable to genetic manipulation, a wider range of powerful techniques can be considered.
A model of endometrial breakdown in the mouse was developed during the 1980s [4]. A defined sequence of estrogen and progesterone is administered to ovariectomized mice, followed by the injection of a small drop of oil into the uterine lumen, which induces a uterine decidual response. Subsequently, progesterone support is withdrawn from the artificially decidualized endometrium, to mimic the fall in serum progesterone that occurs in women following luteal regression. Features associated with the human menstruating endometrium, including the influx of leukocytes and tissue degeneration, have been observed in this model.
In the present report, we describe a modified mouse model of decidualization and menstruation based on the model published by Finn and Pope but with more tightly defined tissue harvest time-points [4]. The cellular and molecular changes were investigated using immunohistochemistry, morphometry, and microarray-based methods. We found that the volume fractions of leukocytes, macrophages, and neutrophils, but not endothelial cells, increased in decidualized uteri and decreased after the completion of major tissue degradation. Microarray data showed that the levels of many transcripts that encode immune-related factors changed throughout the time-course used with this model. A group of immune-related genes, which included chemokine (C-X-C motif) ligands, secretory leukocyte protease inhibitor, and lipocalin 2, showed similar changes in their transcript levels throughout menstruation as the volume fraction of the immune cells changed. The present findings show that immune cells are recruited into the menstruating endometrium, and immune-related genes are regulated in the uterus throughout menstruation.
All procedures and care of the animals were performed following United Kingdom Home Office regulations. Adult female C57BL/6 mice were purchased (Harlan UK, Oxon, U.K.) and maintained in standard housing. Before surgery, animals were anesthetized by i.p. injection of ketamine and xylazine (110 mg/kg and 11.7 mg/kg respectively; GenusXpress, Bury St. Edmunds, U.K.). As an analgesic, 0.03 mg Temgesic (GenusXpress) was administered i.m. to each animal. After surgery, the anesthesia was reversed with 25 µg atipamezole (antisedan; GenusXpress) administered i.p. to each animal.
Animals were ovariectomized, allowed to recover for at least 7 days, and then treated sequentially with steroid hormones. The doses and the schedule were adopted from the model of Finn and Pope [4].
On Day 1 (D1) and D2, animals were injected s.c. with 100 ng of 17β-estradiol (E2) dissolved in 0.1 ml arachis oil (Sigma, Poole, Dorset, U.K.). No hormones were administered on D3, D4, and D5. On D6, D7, and D8, 10 ng of E2 and 500 ng of progesterone (Sigma), dissolved in 0.1 ml arachis oil, were administered s.c. to the animals. On D8, 4–6 h after the last hormone injection, laparotomy was performed under general anesthetic and 0.02 ml of arachis oil was injected into each horn of the uterus, to induce decidualization. The time of hormone injection on D8 was set as time-point 0 (T0). Experimental animals were divided into two groups to study the effect of estradiol/progesterone on the decidualized uterus. The progesterone-withdrawal group did not received any further hormone treatment after T0, while the progesterone-maintained group (designated as the P group) were given 10 ng of E2 and 500 ng of progesterone dissolved in 0.1 ml arachis oil s.c. daily, to prevent regression of the endometrium. Animals were killed after 36 h (T1, T1P), 48 h (T2, T2P), 60 h (T3, T3P), and 84 h (T4, T4P). Uteri were collected and either frozen in liquid nitrogen for RNA isolation or fixed for histological studies. Figure 1 summarizes the schedule for hormone treatment and tissue collection.
|
A portion of one uterine horn from each animal was fixed in Bouin solution at room temperature for 4 h, and then embedded in paraffin wax for sectioning and routine hematoxylin and eosin (H&E) staining.
One segment of one uterine horn from each animal was fixed in zinc fixative (BD Biosciences, Oxford, U.K.) for 8 h, embedded in paraffin wax, and 5-µm sections were cut. The following antibodies were used for immunohistochemistry: for endothelial cells, anti-mouse CD31 (clone MEC 13.3; BD Biosciences); for leukocytes, anti-mouse CD45 (clone 30-F11; BD Biosciences); for macrophages, anti-mouse F4/80 antigen (clone CI:A3–1; Serotec, Oxford, U.K.); and anti-mouse neutrophil (clone 7/4; Serotec). A mouse-absorbed biotin-conjugated goat polyclonal antibody against rat IgG (Serotec) was used to detect the anti-mouse neutrophil antibody, while all the other primary antibodies were detected by a biotin-conjugated goat polyclonal antibody against rat IgG (Zymed, San Francisco, CA). Streptavidin (Vector Laboratories, Peterborough, U.K.) and DAB (Sigma) were used for secondary antibody detection and final visualization, and the cell nuclei were counterstained with hematoxylin.
The Computer Assisted Stereology Toolbox (CAST) 2.0 system (Olympus, Ballerup, Denmark) was used to perform all measurements. The volume fractions of endothelial cells, leukocytes, macrophages, and neutrophils at different time-points in the uteri from both the progesterone-maintained and progesterone-withdrawal groups were measured. Two sections stained for neutrophils and at least three sections stained for the other markers from each mouse were randomly selected and the fields to be counted were selected using meander sampling to cover approximately 80% of each section. This is a function of CAST that allows random and nonoverlapping fields of view within a sample to be identified. A 144-point grid was used to count endothelial cells, leukocytes and neutrophils, and a 255-point grid was used to count macrophages. The volume fractions of the different cell types at different time-points were calculated using the following equation: volume fraction = N(stained)/N(total), where N(stained) is the number of points falling on the immunostained cells, and N(total) is the total number of points falling within the tissue (including the stained cells). The volume fractions of each cell type counted at each time-point were tested for normal distribution with the Kolmogorov-Smirnov test, and all of the datasets passed this normality test. However, results obtained from the Bartlett test suggest that the standard deviations among the groups were often different. Therefore, an unpaired t-test and a nonparametric ANOVA test (Kruskal-Wallis test) were used in the data analysis.
RNA Isolation and Purification for Microarray Studies
Total RNA from individual uteri was isolated by homogenizing one uterine horn in Trizol reagent (Invitrogen, Paisley, U.K.). The RNA was further purified using the Qiagen RNeasy Mini kit (Qiagen, West Sussex, U.K.). The quality of the purified RNA was examined using the Agilent 2100 Bioanalyzer (Agilent Technologies UK, Stockport, U.K.).
A pooled common reference RNA containing equal proportions of RNAs extracted from the 32 individual mice was generated. We compared the abundance of transcripts within this reference RNA (labeled with Cy-5) to the abundance of transcripts within each of the 32 experimental samples (labeled with Cy-3) by hybridization to a mouse cDNA microarray. This microarray (MmcDNAv1) was printed onto glass slides in the Department of Pathology, University of Cambridge. It was composed of the NIA 7.4K mouse cDNA set, 400 hand-picked clones from the NIA 15K mouse cDNA set, and ~8500 hand-selected clones from eight cell-type-specific subtracted mouse cDNA libraries. The SMART-based template-switching protocol published by the Human Genome Mapping project center [5] was used to label fluorescently the RNAs. Briefly, double-stranded cDNA was synthesized from 1 µg of the individual RNAs using a template switching oligonucleotide, cDNA synthesis primer, and Powerscript reverse transcriptase (Invitrogen). The cDNA was amplified (14 cycles of 65°C annealing and 68°C extension) using the AmpliTaq DNA Polymerase II Kit (Applied Biosystems, Warrington, U.K.). The amplified samples were labeled with Cy3-dCTP or Cy5-dCTP using random hexamer primers and Klenow fragment (Invitrogen). Labeled cDNA was purified through AutoSeq G50 columns (Amersham Biosciences, Little Chalfont, U.K.). Purified sample and reference cDNAs were mixed and coprecipitated with human Cot-1 DNA at –20°C overnight. The cDNA pellet was then dissolved in 50°C prewarmed hybridization buffer (40% deionized formamide, 5x SSC, 5x Denhardt solution, 1 mM sodium pyrophosphate, 50 mM Tris (pH 7.4), 0.1% SDS, 1 µg human Cot-1 DNA, 2 µg poly(A)40–60, and 2 µg yeast tRNA). This mixture was incubated at 95°C for 5 minutes immediately before applying to the array slides, and then hybridized in a humidified chamber at 50°C for 16 h to 18 h. After hybridization, the array slides were washed sequentially with 2x SSC, 0.1x SSC plus 0.1% SDS, and 0.1x SSC, and the microarray images were scanned using the GenePix Personal 4100A Microarray Scanner and GenePix Pro 4.1 software (Axon Instruments, Union City, CA). The microarray images were then numerified using the GenePix Pro 3.0 software (Axon Instruments). Transcripts for which both the Cy-5 and Cy-3 signals were in the lowest 17th percentile of intensity on any of the 32 chips were removed from the analysis. Transcripts that the GenePix scanning software flagged as problematic on any of the 32 chips were also removed from the analysis. Within each chip, the Cy-5 and Cy-3 signals were normalized against one another using the Loess method, and for each gene, the ratio of the Cy-3 signal (individual experimental RNA) to the Cy-5 signal (pooled common control RNA) was calculated using the Genespring software (Silicon Genetics, Redwood City, CA). To allow comparisons of transcript abundance between the 32 mouse uteri, the Cy-3:Cy-5 ratios were log2-transformed and then normalized across all the chips using the Loess function of the ‘R’ programming environment (http://cran.r-project.org/). To select differentially expressed transcripts, one-way ANOVA analysis was performed to rank the transcripts according to the significance of transcript abundance variation over the time course.
The four RNAs within each of the experimental groups T1, T2, T3, and T4 were pooled together for Affymetrix microarray analysis using U74Av2 Chips (Affymetrix UK, High Wycombe, U.K.). The cRNAs were produced according to the manufacturers protocol (Affymetrix). Briefly, double-stranded cDNA was synthesized from 10 µg RNA using the SuperScript ds-cDNA Synthesis Kit (Invitrogen). After phenol extraction, biotin-labeled cRNA was synthesized using the Enzo BioArray High Yield RNA Transcript Labelling Kit (Cambridge Bioscience, Cambridge, U.K.). The cRNA was purified using the Qiagen RNeasy Mini Kit, eluted with RNase-free water, and hybridized to gene chips according to the Affymetrix protocols (MRC Geneservice, Cambridge, U.K.). The microarray images were scanned using the Affymetrix GeneChip Scanner 3000, and numerified using the Microarray Analysis Suite (MAS) 5.0 software (Affymetrix). The quality of the expression data from the chips was assessed using the Affymetrix MAS 5.0 and dChip software. The transcript abundance data from all the chips were then scaled in an intensity-dependent manner using the Loess function of the ‘R’ programming environment [6]. The gene array results collected from these Affymetrix experiments and the cDNA arrays were further analyzed using the GeneSpring software and Microsoft Excel.
In all, 280 transcripts with transcript levels above two and which changed more than two-fold between any pair of time-points were identified from the Affymetrix microarray analysis. The functions of these transcripts were further analyzed using GOStat [7] (http://gostat.wehi.edu.au/), which defines which GO categories are over- or under-represented in the gene list in comparison to the transcripts represented on the entire array.
In order to verify the results obtained from the microarray experiments, real-time PCR (TaqMan) verification was performed for the mRNAs of four genes: metallothionein 1 (Mt1), cold-inducible RNA binding protein (Cirbp), protein disulfide isomerase-associated 6 (Pdia6), and arginine-rich, and mutated in early stage tumors (Armet). Predesigned real-time PCR primers and FAM dye-labeled MGB probes for these four genes and the eukaryotic 18S rRNA endogenous control were purchased from Applied Biosystems. First-strand cDNA was synthesized using random hexamers (Amersham) and SuperScript III reverse transcriptase (Invitrogen). Standard curves were generated using cDNA from a normal mouse uterus. Each PCR experiment was carried out in triplicate on cDNA samples from all 32 animals. The cycle threshold (Ct) value of each sample was obtained and the equivalent dilution was calculated according to the standard curve, and then normalized to that of 18S rRNA. The relative cDNA levels were then compared with the normalized target signals from the microarray results.
Sequential hormone treatment was given to the ovariectomized mice as described in Figure 1. The uteri of the progesterone-maintained and progesterone-withdrawal groups at the same time-point were macroscopically indistinguishable. Microscopically, individual variations in uterine morphology and histology were noted, although similar changes could be observed among samples collected at the same time-point. H&E staining of uteri collected at different time-points is shown (Fig. 2, A–H). Both the progesterone-withdrawal and progesterone-maintained animals showed swollen uteri 36 h after T0 (i.e., T1 and T1P). Decidualized cells were detected in several samples, indicating successful induction of artificial decidualization (Fig. 2, C, C-hp, F, and G). The uteri were more enlarged in the T2, T2P, T3, and T3P animals. The enlarged endometrium was clearly visible by eye as a pink layer within the uterus. Upon histological examination, blood vessels were more apparent in the uteri from groups collected at 48 h and 60 h (T2, T2P, T3 and T3P). In most of the uteri harvested at 60 h (T3) from the progesterone-withdrawal group, the pink layer could be separated from the uterus by squeezing the uterine horn. At this stage, the tissue structure of the progesterone-withdrawal endometrium became fragile and the central parts of the endometrium were detached from the uterine tube in some of the tissues. Moreover, blood patches could be seen by eye in some of these T3 uteri. Tissue histology showed that many vessels were visible at T3. Hemorrhage-like blood masses were also observed in some of the animals at this time-point. However, extravascular erythrocytes were not seen within the sections collected for histology. In the progesterone-withdrawal group of mice, endometrial degeneration and shedding was complete before 84 h. There were few glands in the progesterone-withdrawal endometrium (T4; Fig. 2D), and the glands were small and compact compared to those in the progesterone-maintained group (T4P; Fig. 2H). The T4P endometrium also remained enlarged and loose.
|
Uterine tissues collected at different time-points were fixed in zinc fixative, and 5-µm-thick paraffin sections were used for immunohistochemical identification of endothelial cells, leukocytes, macrophages, and neutrophils. All four types of cells were visible at all time-points of the decidualization model (Fig. 2, I–X and Fig. 3, A–P).
|
The volume fraction, also known as the volume density and porosity, is the proportion of each unit volume of the reference space taken up by the object of interest. Our measurement was based on a method that gives an unbiased estimate of volume fraction, in that it randomly positions a point grid over a section and the volume fraction is calculated by dividing the number of points hitting the cell type of interest by the number of points hitting the whole section [8]. The CAST 2.0 system (Olympus) was used to select random and nonoverlapping fields of view within a sample and to apply a randomly positioned point grid to each field [9–11]. The fractions of the uterine volume occupied by these cells were determined (Fig. 4). The volume fractions of endothelial cells in the progesterone-withdrawal and progesterone-maintained groups did not differ at any time-point in the decidualization model (Kruskal-Wallis test, progesterone-withdrawal group P = 0.0548, progesterone-maintained group P = 0.1319). The volume fraction of leukocytes in the progesterone-withdrawal group did not change significantly between 36 h and 60 h after T0 (T1 to T3; Kruskal-Wallis test, P = 0.2957) but decreased significantly at 84 h after T0 (unpaired t-test, P < 0.05). In contrast, the volume fraction of leukocytes in the progesterone-maintained group increased significantly between 36 h and 60 h after T0 (T1P to T3P; Kruskal-Wallis test, P < 0.01) and decreased sharply at 84 h after T0 (unpaired t-test, P < 0.01). The volume fraction of macrophages changed in both hormone treatment groups and increased between 36–60 h after time 0 (Kruskal-Wallis test, progesterone-withdrawal group P < 0.05, progesterone-maintained group P < 0.01) and decreased at 84 h after T0 (unpaired t-test, P < 0.01). The volume fractions of neutrophils in both the progesterone-withdrawal group and progesterone-maintained group did not change within the first 60 h (Kruskal-Wallis test, progesterone-withdrawal group P = 0.8147, progesterone-maintained group P = 0.1202) but decreased at 84 h after T 0 (unpaired t-test, P < 0.01).
|
Identification of Genes with Altered Transcript Levels Throughout the Decidualization Model
Mouse cDNA microarray.
The cDNA microarray is composed of ~16 300 probes generated from individual cDNA clones. Each probe detects the RNA transcripts from one gene, although some transcripts are detected by more than one probe. RNAs extracted from the uteri of all 32 animals were fluorescently labeled and hybridized to the cDNA microarrays. The transcript abundance detected by each probe within each of the 32 RNA samples was compared to the transcript abundance within a pooled common reference RNA. Transcripts for which high quality data were not obtained for all 32 animals (as indicated by the gene array scanning software), as well as transcripts for which the expression levels were too low for reliable analysis (as indicated by an expression level lower than the 17th percentile) were removed from the analysis. To select transcripts that might be regulated in association with this model of decidualization, particularly during the progesterone-withdrawal phase, ANOVA was performed on the normalized data from the progesterone-withdrawal group time series. Signals from the 494 probes with P 
0.05 were then further filtered based on the concordance of regulation between replicates, and 13 probes, which detected transcripts from nine annotated genes that showed tight concordance between replicates, were selected for further analysis (Table 1). The cDNA microarray data discussed in this publication have been deposited in the EBI ArrayExpress Database (http://www.ebi.ac.uk/arrayexpress; accession no. E-MEXP-932).
|
Affymetrix microarray analysis. To investigate a broader range of transcripts than was investigated using the cDNA arrays, we pooled the RNAs from four progesterone-withdrawal animals at the same time-point, and four hybridizations were performed. The progesterone-maintained group of animals was not included in the Affymetrix microarray study. The Affymetrix microarray data discussed in this publication have been deposited in the NCBI Gene Expression Omnibus (http://www.ncbi.nlm.nih.gov/geo/; GEO Series accession no. GSE6359).
Data generated from the Affymetrix microarrays were analyzed using the GeneSpring software. Two groups of transcripts were identified as being of particular interest. The first group comprised transcripts with levels that were: 1) identified as Present or Marginal at one or more of the four time-points in the time-course, and 2) changed more than 10-fold between any time-point. The 37 transcripts in this group (Transcript group A; Table 2) were grouped according to biological function. The complete gene lists with descriptions are provided in Supplemental Table 1A (available online at www.biolreprod.org).
|
The second group comprised transcripts with generally higher levels but showing more modest changes in transcript level between time-points. All the transcripts present in the Affymetrix Murine Genome U74Av2 gene chips were filtered using their normalized expression levels. There were 4734 transcripts with normalized expression levels above 2 (as indicated by an expression level above the 62nd percentile) at any of the four time-points, which covered about one third of all the transcripts, and 280 of these transcripts showed more than two-fold changes in level between any two time-points. These 280 transcripts (Transcript group B; Table 2 ) were grouped according to biological function. Supplemental Table 1B shows the complete gene lists with descriptions (available online at www.biolreprod.org).
Expression Profile Analysis of the Menstruation Model
The 280 transcripts with transcript levels above 2 and which showed more than two-fold changes in level between any pair of time-points were clustered (k means) according to expression pattern similarities across all four time-points using standard correlation. Several k mean classifications were performed to determine the optimal number of clusters, which resulted in the highest observed variability and lowest redundancy between similar clusters. Nine clusters were obtained from the 280 transcripts (Fig. 5).
|
Real-Time PCR Analysis of Selected Regulated Genes Identified by Microarray Analysis
Real-time PCR (TaqMan) analysis was carried out on four genes, to verify the findings of the microarray analysis. These genes were selected from nine transcripts (Table 1), which were determined to have changed transcript levels after the NIA 7.7 cDNA array analysis. Three of these genes, Armet, Pdia6, and Mt1, were chosen because they were also in the list of 280 genes (Table 2 and Supplemental Table 1B) that showed higher transcript levels and changed more than two-fold between any two time-points in the Affymetrix Expression Array. The Cirbp gene was chosen arbitrarily from the rest of the list. Figure 6 shows the relative cDNA levels of these four genes detected by real-time PCR compared to the normalized target signals detected by microarray experiments. As shown in Figure 6, the real-time PCR results for all four genes confirm the results obtained from the microarray experiments.
|
We developed a model of decidualization and menstruation, as described by Finn and Pope [4]. In the original experiment, earliest decidualization occurred in approximately 50% of the experimental mice 36 h after the hormone treatment on D8 (T1). However the decidual response was not apparent in the whole uterus but was observed as swollen areas in only parts of the uterus. All of our animals showed enlarged uterine horns on D8, and decidualization was uniform in a generally swollen uterus. Histologically defined decidualized cells appeared in all the animals at T1, indicating successful decidualization. Although there were individual differences in uterine morphology and histology within the different treatment groups, histological changes between groups could be clearly identified. The individual differences in uterine histology were reflected in the volume fraction data for different cell types. The volume fractions of the same cell type within each group were quite scattered (Fig. 4), although three to four animals were used in each experimental group and at least two sections from each animal were counted. It is possible that increasing the number of animals used in each group would give more stringent data and reveal statistically significant differences.
The immunohistochemical data show the presence of endothelial cells, leukocytes, macrophages, and neutrophils in both the progesterone-withdrawal and progesterone-maintained decidualized mouse uteri at all time-points.
Although endometrial angiogenesis has been suggested to be one of the key features of the menstrual cycle [12–14], the precise details remain unclear [15]. As shown in Figure 2, most of the vessels in the decidualized uteri were small and could not be reliably identified without immunohistochemical staining. The volume fraction determination showed there was no significant difference in the fractions of endothelial cells either at different time-points or between the progesterone-withdrawal and progesterone-maintained groups. However, the distribution of blood vessel was very uneven throughout the whole uterine tissue, which suggests that this issue is more complex. An assumption made when using the CAST system is that features of interest are evenly distributed within the counting area. Unfortunately, lack of material precluded a more detailed quantitative examination of the regional distributions of endometrial vessels.
Since capillary sprouts were not observed in the endometrium, nonsprouting mechanisms, such as elongation and intussusception angiogenesis, have been suggested as the main mechanisms of human endometrial angiogenesis [15, 16]. However, since the counting grids were carefully selected so that each blood vessel was only counted once in each section, our data do not rule out the possibility of vessel elongation and enlargement during decidual tissue development.
An influx of macrophages and neutrophils occurs during the secretory phase of the human menstrual cycle [17–19]. The numbers of these cells are relatively low in the proliferative phase while tissue regeneration takes place. In the mouse, progesterone antagonizes the estrogen-induced influx of neutrophils and macrophages [20]. However, this occurs in a model with no decidual stimulus. Our progesterone-maintained groups showed a pattern that was more similar to the human secretory phase in which some leukocyte populations gradually increased between T1P and T3P. However, the fractions of neutrophils and macrophages decreased at T4P, in contrast to the situation in the nonpregnant human cycle. Neutrophil numbers increase in rat endometrium around early pregnancy and in decidualized mouse endometrium [21]. However, since leukocytes and macrophages are mostly clustered around the implantation site [18, 19], the decreases in the volume fractions of these cells at T4 may be due to the lack of embryo implantation.
The group of Salamonsen have developed a similar model of menstruation induction in the murine uterus, in which they used progesterone implants instead of daily s.c. injections of progesterone [22]. In their study, progesterone was withdrawn 49 h after the induction of decidualization, by which time the endometrium was already well-decidualized, and all the tissues are harvested within 48 h of progesterone withdrawal (49–97 h after the induction of decidualization). In the present study, animals received their last progesterone injection 4 h prior to the induction of decidualization in the progesterone-withdrawal group and therefore, the in vivo hormone level would start to decrease soon after decidualization. This may account for some of the differences in the biological responses that we observed. However, the time-points selected for the present study include those close to the time when decidualization was induced (32–80 h after induction), which enabled us to study the process of decidualization in the progesterone-maintained group. In the previous study, the number of leukocytes in the uterine tissue was determined by random selection of counting fields throughout the designated region of interest. No change in leukocyte number was found within 24 h of progesterone withdrawal. In contrast, our present data show that the volume fractions of leukocytes in the progesterone-withdrawal group were maintained at similar levels between 36 h and 60 h but decreased significantly at 84 h after T0 (unpaired t-test, P < 0.05).
The method of Brasted does not take into account changes in tissue edema, weight, and size when counting leukocytes and therefore, potential changes could be masked [22]. The last time-point for leukocyte counting examined by these authors was 24 h after progesterone withdrawal and their histological data suggest that this particular time-point is similar to T3 in our present study, as the uterus was still in the process of degenerating at this time-point. This may explain why no decrease in the leukocyte population was observed in their study.
The data generated from the two gene array analyses using two microarray platforms were used to identify transcripts that changed in abundance throughout the time-course of progesterone withdrawal. Affymetrix microarrays were used to investigate a broader range of transcripts. However, due to the expense of Affymetrix gene chips, only four Affymetrix microarrays were performed, each using a pooled sample of all of the animals in a single experimental group. The progesterone-maintained animals were not included in the Affymetrix microarray study, as we focused on the process of menstruation. The individual variation in the pooled samples and the absence of replicate arrays for each time-point render a high likelihood of false discovery. Therefore, the data generated in the Affymetrix analyses should not be overinterpreted, and additional validation, involving real-time PCR or Northern blotting, should be performed before any gene is selected for further functional studies.
When the patterns of change in transcript level were compared, there was considerable concordance between these two platforms. For example, there was agreement for all but one (Cirbp) of the transcripts identified as being significantly regulated in the cDNA array study for which data was available from the Affymetyrix study. However, in the case of Cirbp, there was an apparent lack of agreement between the two array platforms, which may explain why the change in transcript level detected by the Affymetrix microarray is not significant (less than two-fold) even though the profile is similar to that detected by the cDNA microarray.
The design of Affymetrix expression arrays allows us to compare the relative transcript levels within a single sample on the same chip. There are 12 488 probe sets on the Affymetrix Murine Genome U74 Av2 gene chips. The signal levels of the transcripts detected in the Affymetrix expression array were ranked, to define the most abundant transcripts in the mouse uteri in the model of menstruation. Within the 200 probes with the highest signals, the most common were for genes that encode ribosomal proteins (44 transcripts), cytoskeletal proteins (30 transcripts), and proteins that are involved in metabolism and biosynthesis (28 transcripts). The 200 probes with the highest signals over the four time-points are shown in Supplemental Table 2 (available online at www.biolreprod.org). The 200 probes with the highest signal in the uteri collected from different time-points were mostly overlapping, with small variations; 240 probes covered all 200 probes with the highest signal at any of the four time-points, and 164 (68.3%) of these probes were found to be most abundant at all four time-points. These 240 probes with the highest signals throughout the model of menstruation were compared with 200 probes with the highest signals in human endometrium at the proliferative phase and secretory phase, excluding the internal control and genes that encode ribosomal proteins [23]. Overall, 46 out of 187 (24.6%) of the probes with the highest signals in the human endometrium were similar to the probes with the highest signals in the menstruating mouse uteri.
The 40 most-abundant shared transcripts were identified in the heart, lung, kidney, stomach, small intestine, liver, bone marrow, mammary gland, ovary, placenta, umbilical cord, and uterus tissues (microarray data downloaded from NCBI Geo DataSets website, accession no. GDS182; http://www.ncbi.nlm.nih.gov/geo/gds/gds_browse.cgi?gds=182). These 40 transcripts were also amongst the 240 probes with the highest signals in the menstruating mouse uterus, which indicates that there are physiological processes that occur in several mouse organs and also in the menstruating mouse uterus. In contrast, 39 transcripts were found to be abundant in the menstruating endometrium but not in other normal mouse tissues. These included Hif1a, Lamp1, Fkbp4, Hsp90b1 (formerly known as Tra1), Cd63, Ctnnb1, and Lsp1. HIF1 protein is known to be present in secretory-phase and menstrual-phase human endometrium and its RNA is present in human endometrium throughout the entire human menstrual cycle and in the mouse uterus during early pregnancy [24, 25]. The transcript that encodes CD63 (a transmembrane-4 superfamily protein) is present at a higher level in human decidual NK cells than in peripheral NK cells [26]. LSP1 (lymphocyte-specific protein 1) is a cytoskeleton-associated protein in leukocytes. It is expressed in the endothelium and is essential for neutrophil emigration [27].
Molecules that are involved in a wide variety of biological processes have been found in the endometrium, for example, inflammatory factors [28], matrix metalloproteinases [29, 30], growth factors [31, 32], angiogenesis-stimulating factors [15, 33], and molecules that mediate cell-cell interactions [31, 34].
Our microarray data show that the levels of many transcripts involved in different biological processes change over the time-course of the model of decidualization and menstruation. The 280 transcripts identified from the Affymetrix analysis as having expression levels greater than 2 and which exhibited more than two-fold changes in level between any pair of time-points, were clustered (k means). Several k mean classifications were performed to determine the optimal number of clusters, which resulted in the highest observed variability among clusters and the highest similarity within each cluster.
It is interesting to note that some related biological components tend to show similar expression profiles. For example, the transcript levels of immunoglobulins, which were mostly clustered into cluster 3, gradually increased across the time-points and were highest at T4. Many genes that are involved in immune responses (e.g., those for chemokine (C-C motif) ligand 9 (Ccl9), haptoglobin, and adipsin) show similar patterns of transcript levels to those of the immunoglobulins across the progesterone-withdrawal time-points. Other genes, e.g., those for secretory leukocyte protease inhibitor (Slpi), lipocalin 2 (Lcn2, also known as 24p3), and several C-X-C motif chemokines (Cxcl1, Cxcl5 and Cxcl14), show related patterns, with maximum transcript levels at T3 and decreased transcript levels at T4. The changes in transcript levels of these immune-related genes following progesterone withdrawal support the notion that menstruation involves an inflammatory-like response and offer a mechanistic insight into the regulation of this process [35, 36]. Some of these immune-related transcripts are also suggested to be involved in embryo implantation. Haptoglobin and lipocalin 2 are reported to be present in the endometrium and are suggested to be involved in endometrial receptivity [37–39].
Numerous studies have shown that many immune-related cells (macrophages, T cells, and neutrophils) are present in the endometrium during menstruation [19, 40–43]. The immunohistochemical data described in the present study indicate that these cells are present and that their numbers are regulated in the mouse model. Several transcripts associated with the migration, regulation or function (CXC chemokines, Il1b and Slpi) of these cell types show similar expression patterns, which parallel the changes seen in cell numbers. Human SLPI and CXCL1 have previously been described as being more abundant in the secretory phase of the human endometrium than in the proliferative phase, and their expression is under the regulation of interleukin 1β [44–46], which may explain the similar changes in transcript levels observed in this model.
The expression patterns of genes involved in cell growth and maintenance suggest that these genes can be divided into two functional groups: those that decrease over the four time-points (cluster 2) and those that increase over the time-course but especially at T4 (clusters 4 and 5). These complementary patterns suggest that processes associated with both degeneration and endometrial regeneration occur in the menstruating uterus. Interestingly, many genes involved in protein folding and proteolysis gradually increase in transcript level at T4 (cluster 3). For example, MMP12 is known as one of the matrix metalloproteinases that is involved in matrix degradation [47, 48], and its transcript level was upregulated over ten-fold at T4. The upregulated patterns of these genes imply that the uterine tissues are not only regenerating but are undergoing degeneration or remodeling at 84 h after T0 [49, 50].
Additional analysis using Gene Ontology has indicated that transcripts with related biological functions are over-represented in the 280 transcripts identified in the Affymetrix analysis which had expression levels greater than 2 and which exhibited more than two-fold changes in levels between any pair of time-points (Table 3); these include transcripts that are involved in immune responses and functions specifically associated with immunity (for example, chemokine activity).
|
The transcript levels of genes that encode metallothioneins (Mt1, Mt2), glutathione peroxidases (Gpx2, Gpx3), and glutathione S-transferases (Gstm2, Gstm5) decreased over the time-course of the model of menstruation. These transcripts are all present in the endometrium and are suggested to be hormone-regulated [23, 51–54]. Therefore, the decreases in the transcript levels of these genes over the time-course can be explained as a result of hormone withdrawal in the model. Our present data show that these genes are regulated in this mouse model of menstruation as in humans, which suggests that similar regulatory mechanisms operate in the human and mouse endometrium.
In conclusion, a murine model of decidualization and menstruation was used that provides a cheaper and more convenient alternative to nonhuman primate models. This model can be used for further studies of menstruation in which genetic manipulation may be useful. Microarray and immunohistochemical analyses were performed on the same tissue samples. The derived data indicate that genes related to immune function are regulated in these tissues. These genes/proteins have the potential to mediate the local effects of the steroid hormones that co-ordinate the tissue remodeling that occurs during decidualization and menstruation. The array analysis also identified for the first time in the remodeling endometrium regulated transcripts that encode factors involved in a wide range of biological processes. These data give some insight into hitherto unrecognized processes that occur in the remodeling endometrium, and may offer new approaches to understanding endometrial biology and pathology.
ACKNOWLEDGMENTS
We thank the Centre of Microarray Resources at the Department of Pathology, University of Cambridge for assistance with the microarray experiments. We also thank Mr. Barry Potter for assistance with the histology.
FOOTNOTES
3Current address: Faculty of Medicine, Level 2, Faculty Building, South Kensington Campus, Imperial College. London, SW7 2AZ, United Kingdom. 
4Current address: Department of Molecular Medicine & Pathology, School of Medical Sciences, University of Auckland, Private Bag 92019, Auckland, New Zealand.
1Supported by the UK Medical Research Council (MRC) Programme grant G9623012 to D.S.C.J., S.K.S., and C.G.P. The cDNA microarray data have been deposited in the EBI ArrayExpress Database (http://www.ebi.ac.uk/arrayexpress/) and are accessible through accession number E-MEXP-932. The Affymetrix microarray data have been deposited in the NCBI Gene Expression Omnibus (http://www.ncbi.nlm.nih.gov/geo/) and are accessible through GEO Series accession number GSE6359.
Correspondence: 2Ching-wen Cheng, Department of Pathology, University of Cambridge, Tennis Court Road, Cambridge, CB2 1QP, United Kingdom FAX: 44 1223 333346; e-mail: cwc28{at}cam.ac.uk
Received: 10 October 2006.
First decision: 8 November 2006.
Accepted: 15 January 2007.
REFERENCES
This article has been cited by other articles:
|
|
 |
|
  M. L. Hull, C. R. Escareno, J. M. Godsland, J. R. Doig, C. M. Johnson, S. C. Phillips, S. K. Smith, S. Tavare, C. G. Print, and D. S. Charnock-Jones Endometrial-Peritoneal Interactions during Endometriotic Lesion Establishment Am. J. Pathol., September 1, 2008; 173(3): 700 - 715. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||
