[Abstract]   [Full Text]   [PDF] Physiological genomics in PG and beyond: July to September 2005

A rationale for using steroids in the treatment of severe cases of H5N1 avian influenza including high-resolution figures, can be found at: Updated information and services

Implantation is an interaction between blastocyst and uterus. The successful implantation of an embryo is dependent on cellular and molecular dialogue between the embryo and the uterus. There are many factors involved in embryo implantation, including endocrine, paracrine, autocrine, and juxtacrine modulators that span cell-cell and cellmatrix interactions (1). Although many specific factors have been identified during the implantation period, the molecular mechanism of embryo implantation is still unknown.
Gene expression profiling on the genomic scale has been routinely applied in current biological and medical research. There are a variety of platforms for measuring gene expression. Oligonucleotide chips, cDNA microarrays, and serial analysis of gene expression (SAGE) 2 are among the most widely used methods (2)(3)(4). Microarray analysis has been used to investigate temporal and spatial gene expression profiles in the mouse uterus during the implantation period (5,6). Recently, Yoon et al. (7) used laser capture microdissection to isolate the luminal epithelium at the implantation site and interimplantation site for microarray analysis. However, cDNA microarrays are restricted to known sequences and have some limitations in quantification (8).
SAGE has been proven to be a powerful and versatile technique suitable for constructing digital gene expression data bases (9 -11). SAGE provides a large scale and quantitative analysis of the transcriptome while giving the opportunity of identifying new transcripts (4). Short nucleotide tags (usually 10 bp) from the defined position in the transcripts were used for the identification of expressed genes in SAGE analysis. The ligation of the tags into long concatemers and their sequencing result in the qualitative and quantitative gene expression profile of a particular tissue (12). Essentially, the SAGE technique not only measures the expression level of a gene, but also quantifies a "tag," which represents a gene transcript. It is possible to combine results from different laboratories to construct a SAGE library (13). One of the advantages of SAGE is its ability to profile transcript expression regardless of any previous knowledge with regard to its character (14). To date, there is only one SAGE library of mouse uterus, which was constructed with normal adult uterus regardless of estrous cycle (15). In this study, SAGE was applied to examine the global gene expression profiling in mouse uterus between the implantation site and interimplantation site.

MATERIALS AND METHODS
Animals and Treatments-Mature mice (Kunming White outbred strain) were maintained in a controlled environment with a 14-h light/ 10-h dark cycle. All animal procedures were approved by the Institutional Animal Care and Use Committee of Northeast Agricultural University. Adult female mice were mated with fertile males or vasectomized males of the same strain to induce pregnancy or pseudopregnancy (day 1 is the day of vaginal plug). The implantation sites on day 5 of pregnancy were identified by intravenous injection of 0.1 ml of 1% Chicago blue (Sigma) in 0.85% sodium chloride. The implantation sites and interimplantation sites from 20 mice were collected and stored at Ϫ70°C ready for SAGE.
To induce delayed implantation, pregnant mice were ovariectomized under ether anesthesia at 0830 -0900 h on day 4 of pregnancy. Proges-terone (1 mg/mouse; Sigma) was injected to maintain delayed implantation from days 5-7. Estradiol-17␤ (25 ng/mouse; Sigma) was given to progesterone-primed delayed implantation mice to terminate delayed implantation. The mice were sacrificed to collect uteri 24 h after estrogen treatment. Delayed implantation was confirmed by flushing the blastocysts from one horn of the uterus.
Hormonal treatments were initiated 2 weeks after mature female mice were ovariectomized. The ovariectomized mice were treated with an injection of estradiol-17␤ (100 ng/mouse), progesterone (1 mg/mouse), or a combination of the same doses of progesterone and estradiol-17␤ for 24 h. All steroids were dissolved in sesame oil and injected subcutaneously. Controls received the vehicle only (0.1 ml/mouse).
Construction of SAGE Libraries-SAGE libraries were constructed using 20 g of total RNA from implantation sites or interimplantation sites utilizing the I-SAGE kit according to the manufacturer's recommendation (Invitrogen). Briefly, poly(A) ϩ RNA was isolated using oligo(dT) magnetic beads and converted to cDNA. The resulting cDNA library was digested with NlaIII (anchoring enzyme), separated with streptavidin-coated magnetic beads, and divided into two populations following extensive washing. Each population was ligated with one of the two annealed linker pairs. After unligated linkers were removed by extensive washing, the tags besides the NlaIII restriction site (CATG) of each transcript were released from the magnetic beads by cleavage with BsmFI (tagging enzyme). Then the two tag populations were blunted and ligated to produce ditags. After the ditags were amplified by performing 27 PCR cycles, the 102-bp band of target was collected by gel purification and extensively digested by NlaIII to recover 26-bp ditags. The band containing the ditags was excised and self-ligated to produce long concatemers. The concatemers of 700 -1,200 bp were isolated by 8% polyacrylamide gel. After the resulting concatemers were ligated into the pZErO-1, the ligated products were transformed into One Shot TOP10 electrocomp Escherichia coli, seeded onto a low salt LB plate containing 50 g/ml zeocin, and cultured at 37°C overnight. White colonies were screened by PCR to select long inserts (Ͼ600 bp) for automated sequencing. SAGE tags of 10 bp were extracted, filtered, and tabulated using SAGE2000 software. The tag-to-gene mappings were performed by matching tag sequences to the most reliable SAGEmap list (available on the Internet at ftp://ftp.ncbi.nlm.nih.gov/ pub/sage/map/mm) that were updated using the mouse gene version released on August 26, 2005. To increase the percentage of the single match tags, 11-bp tags were extracted, filtered, and tabulated using SAGE2000 software. The linker-derived sequences and duplicated ditags were deleted from both libraries. A SAGEmap of 11-bp tags was built by shortening a long SAGE library's tag from 17-bp tags to 11-bp tags, as described previously according to the NCBI SAGEmap reliable data base released on August 26, 2005 (11, 16, 17). Gene identification encoded by the tag sequences was performed using the SAGEmap of 11 bp data base. p values were calculated by SAGE2000 software using the Monte Carlo simulation method and represented a crude estimate for the possibility of detecting a difference for given abundances and -fold differences between two libraries. The two SAGE libraries were then normalized so that tag number was expressed as tags per million for comparison purpose.
Rapid Amplification of cDNA Ends-For those tags that could not be identified by comparison with the NCBI data base, it was necessary to generate additional 3Ј sequence to identify the source mRNA species. 3Ј-Rapid amplification of cDNA ends was performed according to the manufacturer's protocol (Takara Biotechniques, Dalian, China). Based on the method described by Chen et al. (18), the specific 5Ј-sense primers were designed for TTGTTGCTACT and GTGACCACGG as the following: GGATCCCATGTTGTTGCTACT and GGATCCCATG-GTGACCACGG, respectively. Reverse transcription was performed using oligo(dT) 3Ј-site adaptor primer under the conditions of 30°C for 10 min, 42°C for 30 min, 95°C for 5 min, and 5°C for 5 min. Both specific 5Ј sense primer and 3Ј site adaptor primer 5Ј-CTGATCTA-GAGGTACCGGATCC were used for PCR amplification under the following conditions: 94°C for 30 s, 55°C for 30 s, 72°C for 1 min for 30 cycles. The amplified products were purified, ligated into pGEM-T vector (pGEM-T Vector System 1, Promega, Madison, WI), and transformed into competent E. coli DH5␣. The recombinant clone screened on a 5-bromo-4-chloro-3-indolyl-␤-D-galactopyranoside (X-gal) plate was sequenced and subjected to a BLAST search in GenBank TM for corresponding sequences.
RT-PCR Confirmation of SAGE-Uterine tissues at both implantation sites and interimplantation sites were used for RNA extraction with TRIZOL reagent (Invitrogen). RNA samples were treated with RQ1 DNase I (Promega) to remove the contamination of genomic DNA, extracted with phenol/chloroform, precipitated, dissolved in formamide at a final concentration of 1 g/l, and stored at Ϫ70°C.
Total RNAs were reverse-transcribed in a reaction volume of 20 l containing 22 units of reverse transcriptase and 2.5 M oligo(dT) using a Takara BcaBEST RNA PCR kit (Takara Biotechniques, Dalian, China). The intensity of amplified bands was scanned with the UVP laboratory imaging and analysis system (UVP, Inc., Upland, CA). Primers, annealing temperatures, and expected product sizes for seven randomly selected significantly up-regulated genes and seven significantly downregulated genes are listed in Table 1. The band densities for each gene were normalized to Gapdh expression, and ratios were determined. RT-PCR for each gene was repeated at least three times. RT-PCR products of each gene were verified by sequencing. Statistical analysis was performed using SPSS software.
In Situ Hybridization-Total RNAs from the mouse uterus on day 8 of pregnancy were reverse-transcribed and amplified with the corresponding forward and reverse primers ( Table 1). The amplified fragment of each gene was recovered from the agarose gel and cloned into pGEM-T plasmid. The cloned fragment was further verified by sequencing. These plasmids were amplified with the primers for T7 and SP6 to prepare the templates for labeling sense and antisense probes. Digoxigenin-labeled antisense or sense cRNA probes were transcribed in vitro using a digoxigenin RNA labeling kit (Roche Applied Science).
Uteri were cut into 4 -6-mm pieces and flash-frozen in liquid nitrogen. Frozen sections (10 m) were mounted on 3-aminopropyltriethoxysilane (Sigma)-coated slides and fixed in 4% paraformaldehyde solution in PBS. In situ hybridization was performed as previously described (19). All of the sections were counterstained with 1% methyl green. The positive signal was visualized as a dark brown color.

Construction and Analysis of SAGE Libraries-
The SAGE libraries of 10-bp tags constructed from mouse uteri at the interimplantation site and implantation site were deposited in the GEO repository under accession number GSE2892 (GSM63231 for the interimplantation site and GSM63232 for the implantation site). The total numbers of tags sequenced were 51,306 tags for the interimplantation library and 53,214 tags for the implantation library, corresponding to 16,964 and 16,733 unique tags, respectively. All of our 10-bp SAGE library data were shown in Table 2. The percentages of the tags with single match were 45.92% for the interimplantation site and 45.40% for the implantation site.
In order to increase the percentage of the tags with single match, we also used the same sequence files of SAGE libraries to construct two SAGE libraries of 11-bp tags. The total numbers of tags in 11-bp libraries were 48,121 for the interimplantation site and 50,227 for the implantation site. A comparison between 10-and 11-bp tag SAGE libraries was listed in Table 2. In this way, 390 tags with multiple matches of 10-bp libraries were single matched in 11-bp libraries.
All of the following analysis was based on our 11-bp SAGE libraries. In order to avoid the possibility of sequencing errors and to perform a valuable comparison between two libraries, the tags with a combination of 1:0, 0:1, or 1:1 tags between two libraries were discarded. The remaining tags were designated as meaningful unique tags (i.e. 5,531 tags for the interimplantation site and 5,743 tags for the implantation site). In the 11-bp libraries, the percentages of the tags with a single match were 50.75% for the interimplantation site and 50.50% for the implantation site.
Based on the total tag number in both libraries, the 30 most abundant tags were listed in Table 3, of which 11 tags were single matched and 19 tags were multiple matched. There were three tags that expressed over 1% of total tags. Tpt1 (tumor protein, translationally controlled 1) was significantly expressed at the interimplantation site (16,729 versus 13,379 tags), whereas Cybb (cytochrome b 245 , ␤ polypeptide) was significantly expressed at the implantation site (9,788 versus 12,663). Tag GTGGCTCACAA was significantly expressed at the implantation site (6,567 versus 15,948) but was multiply matched.
There were 1,039 tags specifically expressed at the interimplantation site and 1,252 tags specifically expressed at the implantation site. Based on p value, there were 195 tags significantly up-regulated at the interimplantation site and 261 tags significantly up-regulated at the implan-  tation site, of which 100 genes were single matched at the interimplantation site and 127 genes were single matched at the implantation site, respectively (supplemental Tables 4 and 5). Functional Category of Significantly Changed Genes-Functional classification of the significantly changed tags was assigned according to Gene Ontology Consortium criteria and shown in Fig. 1. Many genes were involved in basic cellular activity, including catalytic activity, protein metabolism, cell growth and/or maintenance, structural molecule activity, protein binding, hydrolase activity and so on. However, among many functional groups, there were different numbers or groups of genes between the interimplantation site and implantation site. Com-pared with the implantation site, the genes responsible for immune response, cytoskeleton organization, structural component of ribosome, and structural molecule activity were significantly up-regulated at the interimplantation site. Furthermore, there were more genes responsible for apoptosis, cell cycle, signal transducer activity, cell communication, catalytic activity, transporter activity, and so on at the implantation site compared with interimplantation site.
RT-PCR Confirmation-In order to validate SAGE data, RT-PCR was performed to examine differential expression of 14 randomly selected tags with single match, 7 for significantly up-regulated ones, and 7 for down-regulated ones. Each amplified product was verified by sequenc- ing. The means and S.D. values of the corresponding genes were shown in Fig. 2. For all of the genes examined, the trend of relative expression of their transcripts was comparable between SAGE and RT-PCR. However, the expression ratio between the implantation site and interimplantation site was not matched well between SAGE and RT-PCR. The -fold changes detected by SAGE were greater than RT-PCR. For example, 1810014L12Rik was significantly up-regulated by 10.4 s (21 versus 219 tags) at the implantation site under SAGE analysis. However, the expression ratio between implantation and interimplantation sites was only 1.28 s by RT-PCR.

Confirmation of SAGE Data by in Situ
Hybridization-To further confirm SAGE data, in situ hybridization was performed to examine differential expression of three significantly up-regulated genes in mouse uterus at the implantation site and interimplantation site on day 5 of pregnancy and pseudopregnancy.
1810014L12Rik was not detected in mouse uterus at the interimplantation site (Fig. 3A) but was strongly expressed in subluminal stromal cells at the implantation site (Fig. 3B). When we checked mouse uterus on day 5 of pseudopregnancy, there was no detectable signal (Fig. 3C). Psmb5 was weakly expressed in the luminal epithelium at interimplantation site (Fig. 3D) but was strongly expressed in the subluminal stromal cells and weakly in luminal and glandular epithelia at the implantation site (Fig. 3E). In the uterus on day 5 of pseudopregnancy, there was a low level of Psmb5 expression in the luminal epithelium (Fig. 3F).
Ddx39 was significantly up-regulated at the implantation site (42 versus 279 tags, p ϭ 0.002) in our SAGE libraries. Except for verifying Ddx39 expression on day 5 of pregnancy, we also examined Ddx39 expression from days 1-8 of pregnancy, days 1-5 of pseudopregnancy, under delayed implantation, and under steroid hormonal treatments in the ovariectomized mice.
During early pregnancy, there was a low level of Ddx39 expression in the luminal epithelium on days 1 and 2 (Fig. 4A). Ddx39 expression was moderately observed in the subluminal stromal cells on day 3 (Fig. 4B) but weakly seen in the subluminal stromal cells on day 4 (Fig. 4C). On day 5, Ddx39 signals were strongly shown in the subluminal stromal cells at the implantation site (Fig. 4D) but not seen at the interimplantation site (Fig. 4E). From days 6 to 8, Ddx39 was mainly expressed in the secondary decidua and implanted embryos (Figs. 4, G and H). During pseudopregnancy, the expression pattern of Ddx39 was similar to early pregnancy from days 1-4 (data not shown). On day 5 of pseudopregnancy, there was no detectable Ddx39 signal in the uterus (Fig. 4F).
Under delayed implantation, there was no detectable Ddx39 expression in the uterus (Fig. 5A). After delayed implantation was terminated  by estrogen treatment and embryo implantation occurred, Ddx39 was strongly expressed in the subluminal stromal cells (Fig. 5B), similar to that on day 5 of pregnancy. In the ovariectomized mice, Ddx39 expression was undetected in the uterus (Fig. 5C). Progesterone had no effect on Ddx39 expression (Fig. 5D). After the ovariectomized mice were treated with estrogen, Ddx39 expression was strongly stimulated in the luminal epithelium (Fig. 5E). However, there was only a basal level of Ddx39 expression in the subluminal stromal cells after the ovariectomized mice were treated with a combination of estrogen and progesterone (Fig. 5F).
Except for the tags with single match, we also chose three tags with multiple match for further analysis, because 47% of the unique tags in our SAGE were equivocal for assigning genes to SAGE tags. Tag GAGT-GGATTCT was significantly up-regulated at the implantation site (1,121 versus 1,852 tags, p ϭ 0.002) but multiply matched to Cd63 (Cd63 antigen) and Hsd17b4 (hydroxysteroid (17-␤)-dehydrogenase 4). Because Cd63 had the highest score for this tag, Cd63 was then chosen for in situ hybridization. In the uterus on day 5 of pregnancy, Cd63 expression was not detected at the interimplantation site (Fig. 6A), whereas was strongly observed in the subluminal stromal cells at the implantation site (Fig. 6B).
Tag TGAAATAAACT was also significantly up-regulated at the implantation site (892 versus 2,549 tags, p ϭ 0) and was multiply matched to Npm1 (nucleophosmin 1), Hpse (heparanase) and one transcriber locus (Unigene Mm. 376925). We chose Npm1 for in situ hybridization because of its highest score. There was a basal level of Npm1 expression in the subluminal stromal cells at the interimplantation site (Fig. 6C), whereas Npm1 at the implantation site was highly expressed in the subluminal stromal cells (Fig. 6D).
Additionally, tag GAAAAAAAAAA was significantly up-regulated at the implantation site (685 versus 1,613 tags, p ϭ 6.6 ϫ 10 Ϫ6 ) and was reliably matched to at least 31 genes. Nevertheless, the scores for Fads3 (3,003,007) and Tagln2 (3,002,008) were much higher than other genes (Յ1,000,501). Both Fads3 and Tagln2 were further analyzed by in situ hybridization. In mouse uterus on day 5 of pregnancy, Fads3 mRNA expression was strongly detected in the subluminal stroma at the implantation site (Fig. 6F), but not seen at interimplantation site (Fig.  6E). Similarly, Tagln2 was also highly expressed at the implantation site (Fig. 6H), but not at the interimplantation site (Fig. 6G).

DISCUSSION
Comparison with Other SAGE Data-In our 10-bp SAGE libraries, 51,306 tags for the interimplantation site and 53,214 tags for the implantation site were sequenced. Our data were very similar to two mouse ovary libraries, in which 51,528 tags and 53,696 tags were sequenced, respectively (20). In another study on porcine embryos, 42,389 tags from day 11 porcine embryos and 42,391 tags from day 12 embryos were sequenced to construct the SAGE library (21).
In our 11-bp tag libraries, there were 48,121 tags for the interimplantation site and 50,227 tags for the implantation site. To date, there was only one available SAGE library on normal mouse uterus regardless of estrous status (15). That library was also constructed with an 11-bp tag and contained 44,484 tags, 14,543 of which were unique tags (15). Our two libraries were slightly bigger than theirs. In their library, there were 75 tags highly expressed in mouse uterus compared with muscle (15), of which 34 tags were found in our SAGE. Similarly, in the four 11-bp libraries constructed with mouse adipose tissue, there were 48,107 tags sequenced in each library (22). By studying gene expression profiles in normal and cancer cells, Zhang et al. (23) estimated the transcriptome of a given cell population to be composed of 14,000 -20,000 different species of transcripts. In our SAGE libraries, the numbers of unique tags or transcripts were 16,677 for the interimplantation site and 16,696 for The percentages of tags with single match were 45.92% for the interimplantation site and 45.40% for the implantation site in our 10-bp libraries. After 11-bp tag libraries were constructed from the same sequenced files, the percentages of tags with single match increased to 50.75% for the interimplantation site and 50.50% for the implantation site. Through this strategy, 390 tags, about 5% of the tags with multiple matches in 10-bp libraries, were single matched in our 11-bp libraries and could be used for further analysis. In two libraries constructed with embryonic day 8.5 yolk sac tissues from Tie2 wild-type and the Tie2null littermates, 62% of the tags were uniquely mapped to single Uni-Gene clusters, whereas 38% of the tags were ambiguously mapped to multiple clusters (14). In two SAGE libraries of mouse testis, 54.3% tags were single matched, 38.3% multiple matched and 7.4% with no match based on NCBI SAGEmap reliable mapping (24). Although longer SAGE tags can provide more sequence information for more accurate gene identification, the analysis throughput is compromised (25). Saha et al. (26) showed that Ͼ75% of 17-bp tags occurred only once in the then known human gene sequence. Additionally, SAGE also has limited sensitivity related to the number of sequenced tags. Even with 100,000 tags sequenced in one library, a gene that is expressed on average less than one copy/cell cannot be accurately quantified or detected (13).
It has been shown that prostaglandin I synthase, spermidine synthase, leukemia inhibitory factor receptor, Il11ra1 (interleukin-11 receptor, ␣ chain 1), and calbindin-28K were important for embryo implantation (5,(27)(28)(29)(30). Although these genes were found in our SAGE data, they were multiply matched. Because there were more than one gene matched, it was hard to know the expression level of these genes.

Confirmation of SAGE Data with RT-PCR and in Situ
Hybridization-When RT-PCR was used to verify the SAGE data of 14 randomly chosen genes, the trend of increasing or decreasing between the interimplantation site and implantation site was comparable between the two approaches. However, the ratio of the tag number between the interimplantation site and implantation site was not completely matched with our RT-PCR data and was usually greater than the -fold change from RT-PCR. Then several genes were randomly chosen to be verified by in situ hybridization. 1810014L12Rik, Ddx39, and Psmb5 were significantly up-regulated at the implantation site based on our SAGE data. By in situ hybridization, these genes were strongly expressed in the subluminal stromal cells surrounding the implanting blastocyst, but were not detected or weakly seen at the interimplantation site. Lepourcelet et al. (31) reported that 38 of 40 transcripts tested independently by RT-PCR showed expression changes consistent with the SAGE data and eight of nine independent genes verified by Northern analysis. Additionally, in our SAGE data, the tag for Gapdh gene was 706 for the interimplantation site and 558 for the implantation site (p ϭ 0.21). Gapdh was shown to be expressed at a constant level in mouse uterus (32).
In SAGEmap, tag-to-gene associations are classified by a reliability score, and the tag-to-gene associations with the top two reliability scores are considered reliable (12). Cd63 and Npm1 were both multiple matched genes but had the highest score for their corresponding tags. By in situ hybridization, both Cd63 and Npm1 were strongly expressed in the subluminal stromal cells at the implantation site compared with interimplantation. Among the multiple matched tags, GAAAA-AAAAAA was reliably matched to at least 31 genes. The top two genes (Fads3 and Tagln2) with the highest scores were chosen to be confirmed by in situ hybridization. Both Fads3 and Tagln2 were highly expressed in the subluminal stroma at the implantation site compared with inter-  In the ovariectomized mice, Ddx39 expression was undetected in the uterus (C). Progesterone had no effect on Ddx39 expression (D). After the ovariectomized mice were treated with estrogen, Ddx39 expression was strongly detected in the luminal epithelium (E). However, there was only a basal level of Ddx39 expression in the subluminal stromal cells after the ovariectomized mice were treated with a combination of estrogen and progesterone (F). Bar, 60 m.
implantation. The strong expression of Cd63, Npm1, Fads3, and Tagln2 at the implantation site suggested that some of multiple matched genes, especially for the genes with high score, were expected to be very important candidate molecules for embryo implantation. Lepourcelet et al. (31) combined single matched genes and multiple matched genes with the highest score for their subsequent analysis. It is necessary to do further analysis for the multiple matched tags to search for the implantation-related molecules.
Genes Known to be Associated with Implantation-Because the whole uterine fragments at the implantation site and interimplantation site on day 5 of pregnancy were used for microarray analysis by Reese et al. (5), our SAGE data should be comparable with their data. In our SAGE libraries, follistatin was significantly up-regulated at the implantation site (187 versus 418 tags). Follistatin was also detected in mouse uterus by microarray and up-regulated by 1.77-fold at the implantation site (5). In the rat uterus, follistatin mRNA was strongly detected at the implantation site and in the decidua (33). Ly6c was significantly up-regulated at the implantation site in our SAGE data (104 versus 378 tags). In a previous study, Ly6c was down-regulated in luminal epithelial cells during days 1-5 of pregnancy, but up-regulated in decidualized stromal cells around the implanted embryo at the time of implantation (34). In our study, Hsp105 was significantly up-regulated at the implantation site (83 versus 378 tags). By microarray, Hsp105 was up-regulated 2.07fold at the implantation site (5). Igfbp4 was significantly up-regulated at the implantation site in our SAGE data (104 versus 478 tags). In mouse uterus, Igfbp4 mRNA was strongly detected in mouse uterus at the implantation site and then declined after the establishment of embryo implantation (35,36). In our SAGE data, Igfbp3 was significantly downregulated at the implantation site (457 versus 199 tags). By microarray analysis, Igfbp3 was down-regulated 2.43-fold at the implantation site (5).
In another microarray study on mouse implantation, the luminal epithelium isolated from implantation site and interimplantation site by laser capture microdissection was used for microarray analysis (7). Follistatin-like 1 was increased by 1.37 s at the implantation site (7) (270 versus 518 tags in our data). Col1a1 (procollagen, type I, alpha 1) was increased by 1.97 s at the implantation site (7) and increased by 2.31 s (1354 versus 582 tags) in our SAGE. P4hb (prolyl 4-hydroxylase, ␤-polypeptide) increased by 1.23 s (7), from 291 to 518 tags in our SAGE.
Although leukemia inhibitory factor, cyclooxygenase 2, and microsomal prostaglandin E synthase 1 were important for mouse implantation (19,37,38), these genes were not detected in our SAGE data. The reason for missing these genes in our SAGE libraries might come from their limited expression in the specific location of the mouse uterus. Leukemia inhibitory factor, cyclooxygenase 2, and microsomal prostaglandin E synthase 1 were only expressed in a thin layer of subluminal stromal cells at the implantation site (19,39). Cyclooxygenase 2 was only expressed in the luminal epithelium and subluminal stromal cells immediately surrounding the blastocyst at the implantation site (39). However, in the uterus, there were 5-10% luminal epithelium, 30 -35% stroma, and 60 -65% myometrium (40). This problem could be solved through a combination of SAGE and laser capture microdissection. Cho-Vega et al. (41) successfully analyzed gene expression profiles of pure populations of prostate cancer cells obtained from fresh-frozen prostatectomy specimens and small initial quantities of RNA by combining laser capture microdissection and SAGE. Another reason for missing these genes is that SAGE will miss all of those transcripts that do not have a NlaIII anchoring enzyme site. It is previously estimated that this error is ϳ0.7% of transcripts (9). Furthermore, sequence features such as high GC content can affect SAGE analysis (42).
Functional Relevance of Significantly Changed Genes-During embryo implantation, the loss of the uterine epithelium surrounding the blastocyst is important in bringing the trophoblast into close association with the endometrial stroma in several laboratory rodents. Uterine epithelial cells surrounding mouse and rat embryos during implantation undergo apoptotic cell death, leading to their phagocytosis by trophoblast cells (43,44). Many apoptosis-related genes were expressed in the mouse uterus during embryo implantation, including Tnf␣, tumor necrosis factor receptor 1, Fas ligand, Bax, Bcl2, caspase-9, and caspase-3 (44). Although these genes were not significantly changed in our SAGE data, seven apoptosis-related genes were significantly up-regulated at the implantation site compared with the interimplantation site, including Nr4a1 (nuclear receptor subfamily 4, group A, member 1), Ptdsr (phosphatidylserine receptor), Cfdp1 (craniofacial development protein 1), Ripk3 (receptor-interacting serine-threonine kinase 3), Sulf1 (sulfatase 1), Birc1b (baculoviral IAP repeat-containing 1b) and Bcap31 (B-cell receptor-associated protein 31). The functional relevance between these genes and other apoptosis-related genes during embryo implantation should be further investigated.
There is no available information on the tissue distribution, regulation, and function of 1810014L12Rik, Psmb5, and Tagln2. Because both 1810014L12Rik and Tagln2 were highly expressed at the implantation site, their regulation and function should be further examined.
CD63 antigen, a recently identified member of the transmembrane 4 superfamily, is a lysosomal membrane glycoprotein (gp53) expressed in activated platelets. The expression, regulation, and function of Cd63 in mouse uterus are still unknown. The strong expression of Cd63 in mouse uterus at the implantation site suggests a possible key role during FIGURE 6. In situ hybridization of Cd63, Npm1, Fads3, and Tagln2 in mouse uterus. In the uterus on day 5 of pregnancy, the specific signals corresponding to each gene were not detected at the interimplantation site (A, C, E, and G), whereas they were strongly observed in the subluminal stromal cells at the implantation site (B, D, F, and H). Bar, 60 m. embryo implantation. In human endometrium, CD63 mRNA levels were significantly decreased (p Ͻ 0.05) during the secretory phase compared with levels during the proliferative phase. CD63 was down-regulated by progesterone in the secretory phase (45). Platelets from women with preeclampsia and normotensive pregnant women expressed higher basal CD63 levels than platelets from nonpregnant women (46). CD63 expression might be useful to identify a subgroup of patients with a high risk for development of preeclampsia (47).
NPM is a nucleolar protein directly implicated in cancer pathogenesis in humans. Npm1 Ϫ/Ϫ and Npm1(hy/hy) mouse mutants have aberrant organogenesis and die between embryonic day 11.5 and 16.5 due to severe anemia resulting from defects in primitive hematopoiesis. Npm1 inactivation leads to unrestricted centrosome duplication and genomic instability (48). The differential expression of Npm1 in mouse uterus at the implantation site was first reported in this study. The expression and function of Npm1 in mouse uterus during early pregnancy are still unknown.
Fatty acid desaturases are nonheme iron-containing enzymes that introduce a double bond between defined carbons of fatty acyl chains. Fads3 had a high degree of homology with Fads1 (⌬5 desaturase) and Fads2 (⌬6 desaturase), but the function of Fads3 is unknown (49). Although both Fads1 and Fads2 are responsible for arachidonic acid synthesis from linoleic acid (C18:2 n-6), Fads1 is the limiting step for the synthesis of arachidonic acid (50). Holman et al. (51) reported a deficiency of arachidonic acid in tissues of streptozotocin-induced diabetic rats and suggested alterations in Fads1 activity (52). Arachidonic acid is the substrate for the biosynthesis of eicosanoids of the prostaglandin, leukotriene, and thromboxane families. Among them, prostaglandin I 2 and prostaglandin E 2 are essential for mouse embryo implantation (19,27).
Regulation and Potential Function of Ddx39-Both Ddx39 and Bat1 (HLA-B-associated transcript 1) belong to the DEAD box family and share 89% homology at the amino acid sequence level but are encoded by two different genes (53,54). Using a yeast two-hybrid screen, both Bat1 and Ddx39 were identified as the proteins that interact with Hcc-1 (hemofiltrate CC chemokine-1) (55). Ddx39 was undetectable in the nontumor tissue but was observed in the tumor tissue. Hcc-1 and Bat1 were expressed with a higher transcript level in the nontumor tissue than in tumor cells. Bat1 is inhibited when Ddx39 expression increases, and this leads to an increase in cell proliferation, which may subsequently contribute to oncogenesis (55). Expression of Hcc-1 in cells led to growth inhibition. The differential expression of Bat1 and Ddx39 may suggest that both proteins interact with Hcc-1 at different stages of cellular events, with the Hcc-1-Bat1 complex possibly acting as a negative regulator to inhibit cell proliferation (55). The expression and function of Ddx39, Bat1, and Hcc1 in mouse uterus are still unknown. In human endometrium, HCC-1 was strongly produced by epithelial glands and highly decidualized tissues (56). Compared with proliferative phase, HCC-1 was up-regulated in the midsecretory phase. Additionally, HCC-1 in human endometrium was also stimulated by long acting progestin-only contraceptives (57). In our SAGE data, Hcc1 was not detected, and Bat1 expressed at a high level at the implantation site (0 versus 80 tags, p ϭ 0.064). The high level of Ddx39 in the subluminal stromal cells at the implantation site and in decidual cells may be beneficial to the proliferation of stromal cells during decidualization. Ddx39 expression at the implantation site was specifically up-regulated by implanting blastocyst, since Ddx39 was not expressed at interimplantation site on day 5 of pregnancy, on day 5 of pseudopregnancy, and under delayed implantation.
Additionally, Ddx39 expression was up-regulated by estrogen in the ovariectomized mice. Similarly, Ddx39 was also expressed on days 1 and 2 of pregnancy when estrogen was still at a high level during this period (58). Through the analysis of the promoter region, we found an estrogen response element (ERE) in the promoter of Ddx39 gene, suggesting a direct effect of estrogen on Ddx39. In the DEAD family, p68(Ddx5) and p72(Ddx17) can act as a transcriptional co-activator, specific for the activation function 1 (AF-1) domain of estrogen receptor ␣ (ER␣) (59,60). p68 can be recruited to the promoter of the ER␣ target gene pS2 and may be directly involved in transcriptional regulation (61,62). ERE is also present in the promoter region of Ddx23 (63). Because both estrogen and estrogen receptor ␣ are essential for mouse embryo implantation (64 -66), Ddx39 may play a key role during embryo implantation in mice.
In conclusion, the differential gene expression in mouse uterus at the implantation site was first profiled by SAGE analysis in this study. From our SAGE data, there were 127 genes significantly up-regulated and 100 genes significantly down-regulated at the implantation site, of which most genes were first reported to be highly expressed at the implantation site. The strong expression of 1810014L12Rik, Ddx39, Psmb5, Cd63, Npm1, Fads3, and Tagln2 in mouse uterus at the implantation site was verified by in situ hybridization. Our data will shed light on further study of mouse embryo implantation.