Analysis of Uterine Gene Expression in Interleukin-15 Knockout Mice Reveals Uterine Natural Killer Cells Do Not Play a Major Role in Decidualization and Associated Angiogenesis

Abstract

During decidualization, uterine natural killer cells are the most abundant immune cell types found in the uterus. Although it is well known that they play key roles in spiral arteriole modification and the maintenance of decidual integrity seen after mid-pregnancy, their roles in the differentiation of decidual cells and accompanying angiogenesis during the process of decidualization is less well-characterized. To address this, we used whole-genome Illumina BeadChip analysis to compare the gene expression profiles in implantation segments of the uterus during decidualization on Day 7.5 of pregnancy between wild-type and uterine natural killer cell-deficient (interleukin-15-knockout) mice. We found almost 300 differentially-expressed genes and verified the differential expression of approximately 60 using quantitative RT-PCR. Notably there was a lack of differential expression of genes involved in decidualization and angiogenesis and this was also verified by quantitative RT-PCR. Similar endothelial cell densities and proliferation indices were also found in the endometrium between the implantation site tissues of wild-type and knockout mice undergoing decidualization. Overall, the results of this study reveal that uterine natural killer cells likely do not play a major role in decidualization and accompanying angiogenesis during implantation. In addition, the study identifies a large number of genes whose expression in implantation-site uterine tissue during decidualization depends on interleukin-15 expression in mice.

Introduction

Decidualization, the differentiation of the endometrium into the decidua, is a key process during implantation of the conceptus in rodents and humans and has been covered by a number of reviews (Abrahamsohn & Zorn 1993, Dunn et al. 2003, Gellersen et al. 2007, Herington et al. 2009, Ramathal et al. 2010). Briefly, in mice, the decidual tissue provides the nutritive environment for approximately 5 days in which the embryo and placenta develop. After mid-pregnancy the functional placenta is formed and takes over providing the nutrients to the fetus. Several changes occur in the uterus during decidualization but the hallmark is the rapid proliferation and then trans-differentiation of fibroblast-like endometrial stromal cells into the epithelial-like decidual cells, usually called decidual cell differentiation. In the mouse this is dependent on the actions of several hormones including progesterone and BMP2 (Lee et al. 2007) and is accompanied by an increase in the expression of decidual markers such as liver/bone/kidney alkaline phosphatase (Akpl) (Herington et al. 2009) and prolactin family 8 subfamily a member 2 (Prl8a2) (Bany & Cross 2006) and a massive increase in uterine weight. To support this tissue growth, another important process during decidualization of the endometrium is angiogenesis or neo-angiogenesis (Torry et al. 2007, Laws et al. 2008, Demir et al. 2010). Critical to this is the endometrial expression of several genes such as vascular endothelial growth factor A (Vegfa)(Chakraborty et al. 1995, Halder et al. 2000), prostaglandin endoperoxide synthase 2 (Ptgs2) plus angiopoietin 2 (Angpt2) (Matsumoto et al. 2002), and gap junction protein alpha 1 (Gja1) (Laws et al. 2008).

The most abundant immune cell types in the uterus during decidualization are uterine natural killer (uNK) cells. Current dogma for mice is that there are two sources of uNK cells in uterus during implantation (Zhang et al. 2009, Zhang et al. 2011). One population of uNK cells are dolichos biflorus agglutinin (DBA) lectin-positive (DBA⁺) and have granules that stain positive using periodic acid Schiff staining (PAS⁺). The DBA⁺PAS⁺ uNK cells are believed to be derived from circulating lymphocyte progenitor cells, which upon entering the uterus as immature non-granulated uNK cells begin expressing DBA lectin and undergo maturation into large granulated NK cells. The other source of uNK cells in the uterus comes from the resident uNK cells which are DBA lectin-negative (DBA⁻) but are PAS⁺. A great deal of work has been conducted on the functions of uNK cells in mice and has involved the use of several genetic models, including interleukin-15 knockout (Il15^−/−) mice, where uNK cells are absent from the uterus during decidualization (Croy et al. 2003b). All of these studies conclusively showed that uNK cells play a key role in maintaining decidual integrity and the characteristic modification of the spiral arteries which is clearly seen only after mid-pregnancy. However, it should be noted that the uNK cell deficient Il15^−/− mice are fertile, have normal gestation times and comparable litter sizes to wild-type mice (Barber & Pollard 2003). However, although little is really known about the role of uNK cells in decidualization and accompanying angiogenesis, it has been speculated that they play a role in decidualization. For example, it has been speculated uNK cells play a role in uterine angiogenesis because in humans and mice they express Vegfc and Vegfa, respectively (Wang et al. 2000, Lash et al. 2006). However, a clear role for uNK cells in decidual cell differentiation and angiogenesis during decidualization is currently not established.

The present study was conducted to more closely examine the potential aberrant expression of genes involved in decidual cell differentiation and angiogenesis in the mouse uterus during decidualization in 1l15^−/− mice. The results support the hypothesis that uNK cells in mice do not play a major role in decidual cell differentiation as well as angiogenesis during decidualization. However, the microarray analysis reveals the identity of almost of 300 genes whose expression is altered in the implantation areas of the uterus of Il15^−/− mice, some of which were investigated further.

Results

Microarray Analysis

To determine the potential effect of uNK cell depletion on uterine gene expression during decidualization, we compared the global mRNA levels between RNA samples in day 7.5 implantation segment tissues from Il15 wild-type (Il15^+/+) and knockout mice (Il15^−/−) using Illumina BeadChip microarray analysis. Day 7.5 of pregnancy was chosen because by this day decidual growth is well under way and uNK cells are very abundant in implantation segment (IS) tissue of the uterus (Herington & Bany 2007b). Of the 46,633 probes, signals for 14,148 were detected significantly above background in all samples. Analysis of this data suggested that 285 of these probes showed a significant difference in the level of hybridization signals between samples from Il15^−/− and Il15^+/+ mice and could be assigned a gene symbol. These probes represented the mRNAs for 267 different genes that were differentially expressed, with 150 and 117 having greater hybridization signals in IS tissue from Il15^+/+ and Il15^−/− mice, respectively. Genes listed in Tables 1 and 2 are those differentially expressed 2-fold or greater in the IS tissue of Il15^+/+ and Il15^−/− mice, respectively. Further analysis of the microarray data revealed that hybridization signals for 29 and 7 probes, respectively, were found in IS tissues of only Il15^+/+ (Table 3) or Il15^−/− (Table 4) mice. These represented probes representing 26 and 7 genes, respectively. Full annotation of the all 321 probes representing differential expression of 300 genes is shown in Supplemental Table 1. Finally, the hypogeometric tests for gene ontology (GO) terms of the list of all differentially-expressed genes revealed significant over- and under-representation of GO terms for molecular function (Supplemental Table 2). The top over-represented biological process was proteolysis and several immune cell-related processes were in the list.

Table 1.

Genes whose steady-state mRNA levels are significantly greater in IS tissue from Day 7.5 pregnant Il15^+/+ mice compared to that of Il15^−/− mice by at least an average of 2-fold. Probe ID, Illumina Probe identification number; Fold, average fold difference mRNA levels in IS tissue of Il15^+/+ relative to Il15^−/− mice.

Gene Symbol	Name	Fold	Probe ID
Spp1	secreted phosphoprotein 1	11.931	2470609
Gdpd3	glycerophosphodiester phosphodiesterase domain containing 3	6.423	6110292
Ear11	eosinophil-associated, ribonuclease A family, member 11	5.938	2260292
Tcrb-J	T-cell receptor beta, joining region	5.040	103800086
Rgs5	regulator of G-protein signaling 5	3.982	101410068
Aldh1a3	aldehyde dehydrogenase family 1, subfamily A3	3.775	105700162
Pdgfrl	platelet-derived growth factor receptor-like	3.749	870168
Tnfrsf9	tumor necrosis factor receptor superfamily, member 9	3.138	2510400
Csrnp1	cysteine-serine-rich nuclear protein 1	3.095	1230053
Tpsab1	tryptase alpha/beta 1	3.095	1400347
Pdpn	podoplanin	3.095	4640280
Pstpip1	proline-serine-threonine phosphatase-interacting protein 1	2.921	3190156
Degs2	degenerative spermatocyte homolog 2, lipid desaturase	2.915	1170056
Rac2	RAS-related C3 botulinum substrate 2	2.802	1580541
Irf8	interferon regulatory factor 8	2.751	610161
Necap2	NECAP endocytosis associated 2	2.726	6200048
Cldn15	claudin 15	2.682	4150270
Selplg	selectin, platelet (p-selectin) ligand	2.651	1770167
Fbln1	fibulin 1	2.585	2480059
Sprr2d	small proline-rich protein 2D	2.579	1340458
A430104N18Rik	RIKEN cDNA A430104N18 gene	2.537	106510711
Cinp	cyclin-dependent kinase 2 interacting protein	2.514	106770484
Ncf4	neutrophil cytosolic factor 4	2.474	610164
Tbc1d20	TBC1 domain family, member 20	2.440	7000286
Cnot7	CCR4-NOT transcription complex, subunit 7	2.428	101570114
Coro1a	coronin, actin binding protein 1A	2.417	3140609
Gnl2	guanine nucleotide binding protein-like 2	2.406	6620022
Myo1g	myosin IG	2.395	2970358
Fbln2	fibulin 2	2.395	3440215
Coro1a	coronin, actin binding protein 1A	2.367	3190020
Lat2	linker for activation of T cells family, member 2	2.362	5340440
Spink2	serine peptidase inhibitor, Kazal type 2	2.357	3990093
Wrn	Werner syndrome homolog	2.329	540692
Cd47	CD47 antigen (Rh-related antigen, integrin-associated signal transducer)	2.303	103390731
9130230L23Rik	RIKEN cDNA 9130230L23 gene	2.271	2190452
Pnpla3	patatin-like phospholipase domain containing 3	2.266	5050551
Rab44	RAB44, member RAS oncogene family	2.240	102230139
Farsa	phenylalanyl-tRNA synthetase, alpha subunit	2.235	2650593
Ocel1	occludin/ELL domain containing 1	2.204	2470088
Anxa8	annexin A8	2.194	4780022
Akr1b8	aldo-keto reductase family 1, member B8	2.184	3120288
Rap1gap	Rap1 GTPase-activating protein	2.184	101850672
Retnla	resistin like alpha	2.168	3840138
Pfpl	pore forming protein-like	2.149	6450670
Fam83g	family with sequence similarity 83, member G	2.144	103130735
Col5a2	collagen, type V, alpha 2	2.139	104280100
Doxl2	diamine oxidase-like protein 2	2.134	104780056
Ccdc114	coiled-coil domain containing 114	2.129	104610181
Pcsk5	proprotein convertase subtilisin/kexin type 5	2.109	5720600
Emid2	EMI domain containing 2	2.099	360400
Serhl	serine hydrolase-like	2.090	2470605
Ciapin1	cytokine induced apoptosis inhibitor 1	2.080	2900114
Selplg	selectin, platelet (p-selectin) ligand	2.056	101470563
Sft2d3	SFT2 domain containing 3	2.051	103060670
Oxsr1	oxidative-stress responsive 1	2.047	1500025
Alox5	arachidonate 5-lipoxygenase	2.037	105340280
Adamts7	a disintegrin-like and metallopeptidase (reprolysin type) with thrombospondin type 1 motif, 7	2.023	3360368
Gabbr1	gamma-aminobutyric acid (GABA) B receptor, 1	2.000	101780504

Table 2.

Genes whose steady-state mRNA levels are significantly less in IS tissue from Day 7.5 Il15^−/− pregnant mice compared to that of Il15^+/+ mice by 2-fold or greater. Probe ID, Illumina Probe identification number; Fold, average fold difference mRNA levels in IS tissue of Il15^+/+ relative to Il15^−/− mice.

Gene Symbol	Name	Fold	Probe ID
Igkc	Immunoglobulin (Ig) and T-cell receptor	11.628	1230347
Igkc	immunoglobulin kappa constant	10.526	6370309
Igh-VJ558	immunoglobulin heavy chain (J558 family)	9.709	60025
Ndufb10	NADH dehydrogenase (ubiquinone) 1 beta subcomplex, 10	8.850	5900576
Igh-VJ558	immunoglobulin heavy chain (J558 family)	8.333	1690184
Man2b1	mannosidase 2, alpha B1	6.410	106520575
Rassf8	Ras association (RalGDS/AF-6) domain family (N- terminal) member 8	4.505	105690053
Glrx	glutaredoxin	4.444	3520471
Scoc	short coiled-coil protein	3.937	2230053
Prdx2	peroxiredoxin 2	3.891	4010619
Arl3	ADP-ribosylation factor-like 3	3.817	5690725
Trmt1	TRM1 tRNA methyltransferase 1 homolog	3.759	103780280
Pla2g12a	phospholipase A2, group XIIA	3.636	103390286
Sdhaf1	succinate dehydrogenase complex assembly factor 1	3.247	102230433
Rusc2	RUN and SH3 domain containing 2	3.125	460168
Syne2	synaptic nuclear envelope 2	2.924	102260348
Col1a2	collagen, type I, alpha 2	2.801	380364
Prcp	prolylcarboxypeptidase (angiotensinase C)	2.801	7050746
Gpx2	glutathione peroxidase 2	2.717	6940619
Gm3322	predicted gene 3322	2.710	6660128
Timm9	translocase of inner mitochondrial membrane 9	2.611	1450019
S100a9	S100 calcium binding protein A9 (calgranulin B)	2.611	7050528
BC003331	cDNA sequence BC003331	2.525	1980164
Mtap7d2	MAP7 domain containing 2	2.469	106860494
Rg9mtd1	RNA (guanine-9-) methyltransferase domain containing 1	2.457	4480368
Gch1	GTP cyclohydrolase 1	2.445	670364
Mmp3	matrix metallopeptidase 3	2.398	2650368
Trem2	triggering receptor expressed on myeloid cells 2	2.398	6290358
Trf	transferrin	2.381	1050193
Adamts9	a disintegrin-like and metallopeptidase (reprolysin type) with thrombospondin type 1 motif, 9	2.283	101580446
Mmp3	matrix metallopeptidase 3	2.232	2970056
H-T23	histocompatibility 2, T region locus 23	2.212	2850609
Sorcs2	sortilin-related VPS10 domain containing receptor 2	2.208	100460068
Wdr92	WD repeat domain 92	2.203	5570097
Fut10	fucosyltransferase 10	2.174	103120039
Trim16	tripartite motif-containing 16	2.151	6510068
Hps1	Hermansky-Pudlak syndrome 1 homolog (human)	2.114	2510026
Nipal1	NIPA-like domain containing 1	2.079	6130315
Aktip	thymoma viral proto-oncogene 1 interacting protein	2.066	6350575
Txnl4a	thioredoxin-like 4A	2.062	4120286
Gch1	GTP cyclohydrolase 1	2.058	6550358
Nupr1	nuclear protein 1	2.053	1990524
Gstt2	glutathione S-transferase, theta 2	2.053	106650102
Rad23a	RAD23a homolog	2.045	670082
Vps35	vacuolar protein sorting 35	2.045	7050687
Cdkn2b	cyclin-dependent kinase inhibitor 2B (p15, inhibits CDK4)	2.033	6020040
Hist1h4f	histone cluster 1, H4f	2.033	102450672
Dbt	dihydrolipoamide branched chain transacylase E2	2.028	106180717

Table 3.

Genes whose steady-state mRNA levels were only significantly above background in IS tissue from Day 7.5 Il15^+/+ mice. Probe ID, Illumina Probe identification number.

Gene Symbol	Name	Probe ID
2010011I20Rik	RIKEN cDNA 2010011I20 gene	102900270
6330403K07Rik	RIKEN cDNA 6330403K07 gene	2060504
Abca5	ATP-binding cassette, sub-family A (ABC1), member 5	1660110
Avil	advillin	5690452
Ctsg	cathepsin G	1190711
Ctsw	cathepsin W	360368
Fcho1	FCH domain only 1	1240411
Fgfr1op2	FGFR1 oncogene partner 2	60538
Fggy	FGGY carbohydrate kinase domain containing	7100026
Gm949	predicted gene 949	104480131
Gzmd	granzyme D	4920368
Gzme	granzyme E	1990112
Gzme	granzyme E	5080292
Gzmf	granzyme F	6200008
Gzmg	granzyme G	1230341
Klf5	Kruppel-like factor 5	104150152
Klrd1	killer cell lectin-like receptor, subfamily D, member 1	3190114
Klrg1	killer cell lectin-like receptor subfamily G, member 1	2360133
Ltb4r1	leukotriene B4 receptor 1	1770056
Nkg7	natural killer cell group 7 sequence	1780390
Nkg7	natural killer cell group 7 sequence	6370059
Nmu	neuromedin U	6400025
Prf1	perforin 1 (pore forming protein)	6660309
Rasgef1c	RasGEF domain family, member 1C	105420563
Sh2d2a	SH2 domain protein 2A	2970594
Tcrb-V8.2	T-cell receptor beta, variable 13-2	3610048
Ttc27	tetratricopeptide repeat domain 27	1090280
Xcl1	chemokine (C motif) ligand 1	3800504
Zbtb32	zinc finger and BTB domain containing 32	5900397

Table 4.

Genes whose steady-state mRNA levels were only significantly above background in IS tissue from Day 7.5 Il15^−/− mice. Probe ID, Illumina Probe identification number.

Gene Symbol	Name	Probe ID
6720422M22Rik	RIKEN cDNA 6720422M22 gene	104590600
Gfpt2	glutamine fructose-6-phosphate transaminase 2	102810242
Ighg2c	immunoglobulin heavy constant gamma 2C	360113
Prdx2	peroxiredoxin 2	5340577
Slpi	secretory leukocyte peptidase inhibitor	2120446
Sspn	sarcospan	4670091
Wdr46	WD repeat domain 46	6020601

Expression of Genes involved in Decidualization

Although it is clear that “decidual integrity” depends on the presence of uNK cells (Ashkar et al. 2003, Monk et al. 2005), it has not been solidly established whether or not these cells play a role in decidual cell differentiation. The expression of many genes in rodent endometrial stromal cells is known to increase during decidual cell differentiation. Notably, in this study, the mRNA levels of two of the more commonly used decidual markers, Prl8a2 and Akpl, were not different between the IS tissue of Il15^+/+ compared to Il15^−/− mice. Two major regulators of decidualization are bone morphogenic protein 2 (BMP2) and progesterone (Lee et al. 2007). The mRNA levels of Bmp2 and progesterone receptor (Pgr) were not found to be differentially expressed in the microarray analysis. As shown in Fig. 1A, we verified the lack of differential mRNA levels for Bmp2 as well as several BMP2-target genes in the uterus such as FK506 binding protein 3 (Fkbp3), 4 (Fkbp4), 5 (Fkbp5). In addition, mechanistic targets of rapamycin (Mtor/Frap1), follistatin (Fst) and inhibitor of DNA binding 1 (Id1) (Lee et al. 2007) were not significantly (P>0.05) different in IS tissue between Il15^+/+ and Il15^−/− mice (Fig. 1A). In a similar fashion, we verified Pgr mRNA levels did not differ in IS tissues between these mice (Fig. 1A). Finally, kruppel-like factor 5 (Klf5) is expressed in the uterus (Watanabe et al. 1999) and is known to be involved in vascular remodeling and angiogenesis (Nagai et al. 2005). Klf5 mRNA levels in the IS tissues of the uterus significantly (P<0.05) increased compared to non-implantation segment (NIS) tissues on days 5.5 to 8.5 of pregnancy in Il15^+/+ mice (Fig. 1B). Although the microarray analysis indicated differential expression of Klf5 in IS tissues of Il15^+/+ compared to Il15^−/− mice on Day 7.5 (Table 3), we did not detect a significant (P>0.05) difference between these tissues using qRT-PCR (Fig. 1C).

Fig. 1.

qRT-PCR assessment of the expression of genes involved in decidualization. Bmp2 mRNA and the mRNAs of BMP2 downstream target genes plus progesterone receptor (Pgr) mRNA levels in IS tissue from Day 7.5 pregnant Il15^+/+ and Il15^−/− mice (A). Relative Klf5 mRNA levels in non-implantation (NIS) and implantation (IS) segments of the uterus (B) or IS segments from Il15+/+ and Il15^−/− mice (C) on Days 5.5–8.5 and Days 6.5–8.5 of pregnancy, respectively. Bars represent the mean ± SEM (N=4) and statistically significant difference in mRNA levels is denoted by asterisks (*P<0.05).

Expression of Genes involved in Angiogenesis

Previous work suggests that uNK cells express genes involved in angiogenesis and thus may play a role in endometrial angiogenesis during implantation (Lash et al. 2006, Li et al. 2008, Kalkunte et al. 2009). Indeed, current dogma suggests that activated uNK cells are critical for endometrial angiogenesis that occurs in early implantation site development (Santoni et al. 2008, Petitbarat et al. 2010). Therefore, we expected abnormalities in uterine angiogenesis during decidualization in IS tissue of the Il15^−/− mice due to the lack of uNK cells. Expression of disintegrin-like and metallopeptidase 9 (Adamts9) increases in the mouse uterus during implantation and is localized mainly to a subpopulation of vascular endothelial cells in the endometrium (Jungers et al. 2005). Since ADAMTS9 is an anti-angiogenic factor (Koo et al. 2010), we verified that Adamts9 mRNA levels were significantly (P<0.01) greater in IS tissues from Il15^−/− compared to that of Il15^+/+ mice (Fig. 2A). On the other hand, given that uNK cells are believed to play a role in angiogenesis, it was surprising that the microarray analysis did not reveal differential expression of any other genes that may be involved in angiogenesis or that are expressed in endothelial cells. Therefore, we assessed the expression of several of these genes using qRT-PCR. Vegfa, Vegfb, Vegfc, vascular endothelial growth factor D/FOS-induced growth factor (Figf) as well as prostaglandin-endoperoxide synthase 2 (Ptgs2) and inhibin betaB (Inhbb) mRNA levels were not significantly (P>0.05) different in IS tissues from Il15^+/+ compared to Il15^−/− mice on Day 7.5 of pregnancy (Fig. 2A). To examine this in more detail, we measured mRNA levels of several other angiogenesis-related genes as well as endothelial cell markers in the IS tissue of Il15^+/+ and Il15^−/− mice on days 6.5 to 10.5 of pregnancy. Gja1, angiopoietin-1 (Angpt1), Angpt2, and runt-related transcription factor 1 (Runx1) may play a role in uterine angiogenesis during implantation (Bany & Cross 2006, Hess et al. 2006, Ma & Zhu 2007, Laws et al. 2008, Dong & Chen 2009, Sur 2009). We verified that the expression of all four of these genes were not significantly (P>0.05) different between IS tissue from Il15^+/+ and Il15^−/− mice on each day of pregnancy examined (Fig. 2B–E). Endothelial cells strongly express CD34 antigen (Cd34), platelet/endothelial cell adhesion molecule 1 (Pecam1), endothelial-specific receptor tyrosine kinase (Tek), FMS-like tyrosine kinase 1 (Flt1), kinase insert domain protein receptor (Kdr), and FMS-like tyrosine kinase 4 (Flt4) in the endometrium during decidualization (Wong et al. 1997, He et al. 1999, Halder et al. 2000, Matsumoto et al. 2002, Douglas et al. 2009). We verified no significant (P>0.05) effect of genotype on the expression of all six of these genes between IS tissue from Il15^+/+ and Il15^−/− mice on each day of pregnancy examined except on Day 9.5 for Pecam1 and Flt4 (Fig. 2F–K).

Fig. 2.

qRT-PCR assessment of the expression of genes potentially involved in angiogenesis or vascular markers in the uterus. Relative mRNA levels in IS tissue from Day 7.5 pregnant Il15^+/+ and Il15^−/− mice (A). Relative mRNA levels in IS tissues from Il15^+/+ and Il15^−/− mice on Days 6.5–10.5 of pregnancy for Gja1 (B), Angpt1 (C), Angpt2 (D), Runx1 (E), Cd34 (F), Pecam1 (G), Tek (H), Flt1 (I), Kdr (J) and Flt4 (K). Bars represent the mean ± SEM (N=4) and statistically significant difference in mRNA levels is denoted by asterisks (*P<0.05, **P<0.01).

Angiogenesis and Vascular Density Changes during Decidualization

Although it is clear that the characteristic of the spiral arteries by mid-pregnancy in mice depends on the presence of uNK cells, it is not solidly established that these cells play a role in angiogenesis of the uterus during decidualization. We speculated that vascular density and endothelial cell proliferation does not differ between the IS tissues of Il15^+/+ and Il15^−/− mice given the results of the microarray and qRT-PCR analyses above. To assess this, we measured vascular density and endothelial cell proliferation in IS tissues of Il15^+/+ and Il15^−/− mice on Days 6.5, 7.5 and 8.5 of pregnancy. As shown in Fig. 3A and 3B we used CD34 staining to assess the localization and numbers of endothelial cells. We also used bromodeoxyuridine (BrdU) to identify those endothelial cells undergoing proliferation. We assessed 3 different regions including the antimesometrial (AM) region as well as the central (CM) plus lateral (LM) mesometrial regions (Fig. 3AB). There was no effect (P>0.05) of genotype on endothelial cell density for each day examined in the central (Fig. 3C) and lateral (Fig. 3D) mesometrial plus antimesometrial (Fig. 3E) areas of the endometrium. Similar findings were found for endothelial cell proliferation (Fig. 3F,G,H) except for the antimesometrial region in Day 8.5 Il15^−/− mice where endothelial cells were not undergoing proliferation while a few were in the Il15^+/+ mice.

Fig. 3.

Assessment of endothelial density and proliferation in the antimesometrial (AM), central mesometrial (CM) and lateral mesometrial (LM) regions of the endometrium of IS tissues of Il15^+/+ and Il15^−/− mice during decidualization on days 6.5–9.5 of pregnancy. Photomicrogaphs of sections from IS tissue of Il15^+/+ (A) and Il15^−/− (B) mice stained for CD34 (red fluorescence), BrdU (green fluorescence) and counterstained nuclei (DAPI, blue fluorescence). Endothelial cell density in CM (C), LM (D) and AM (E) areas on days 6.5–8.5 of pregnancy. Endothelial cell proliferation in CM (F), LM (G) and AM (H) areas on days 6.5–8.5 of pregnancy. Bars represent the mean ± SEM (N=4).

Expression of Genes Involved in Proteolysis

Many of the differentially expressed genes found in the microarray work are involved in proteolysis so we verified the differential expression of several of these genes. Cathepsins are serine proteases, some of which have been shown to be expressed in uNK cells (Croy et al. 2010). The microarray results indicated a significant (P<0.05) decrease in cathepsin G (Ctsg) and cathepsin w (Ctsw) expression in the IS tissue of Day 7.5 Il15^−/− mice. As determined by qRT-PCR, we found Ctsg mRNA levels were significantly (P<0.05) greater in IS compared to NIS tissues on Days 7.5–8.5 of pregnancy from Il15^+/+ mice (Fig. 4A). Ctsg mRNA levels were significantly (P<0.05) greater in IS tissue of Il15^+/+ mice compared to that of Il15^−/− mice on Days 5.5, 6.5, 7.5 and 8.5 by approximately 40-, 200-, 600- and 2000-fold, respectively (Fig. 4B). Also known to be expressed in NK cells (Wex et al. 2001), Ctsw mRNA levels differed significantly (P<0.05) between NIS and IS tissue only on Day 8.5 (Fig. 4C). However, Ctsw mRNA levels were significantly (P<0.05) greater in IS tissue of Il15^+/+ mice compared to that of Il15^−/− mice on Days 5.5, 6.5, 7.5 and 8.5 by approximately 30-, 15-, 19- and 34-fold, respectively (Fig. 4D). Granzymes are another group of serine proteases, some of which have been shown to be expressed in uNK cells (Allen & Nilsen-Hamilton 1998). The mRNA levels of several granzymes including granzyme D (Gzmd), E (Gzme), F (Gzmf) and G (Gzmg) were verified to be expressed at a significantly (P<0.001) greater level in Day 7.5 IS tissues from IL15^+/+ compared to that of Il15^−/− mice (Fig. 4E). Finally, expression of two protease genes, HtrA serine peptidase 1 (Htra1) and prolylcarboxypeptidase/angiotensinase C (Prcp), was significantly (P<0.05) greater in IS tissues from Il15^+/+ and Il15^−/− mice, respectively (Fig. 4E).

Fig. 4.

qRT-PCR assessment of Ctsg and Ctsw expression in the mouse pregnant mouse uterus. Relative Ctsg and Ctsw mRNA levels in non-implantation (NIS) and implantation (IS) segments of the uterus (A, C) or IS segments from Il15^+/+ and Il15^−/− mice (C, D) on Days 5.5–8.5 of pregnancy. Relative mRNA levels of 12 genes in the IS tissues of Day 7.5 pregnant Il15^+/+ and Il15^−/− mice (E). Bars represent the mean ± SEM (N=4–6) and statistically significant difference in mRNA levels is denoted by asterisks (*P<0.05, **P<0.01, ***P<0.001).

Expression of Genes Found Previously to be Expressed in uNK or NK Cells

Killer cell lectin-like receptor subfamily D, member 1 (Klrd1) (Soderstrom et al. 1997, Verma et al. 1997, King et al. 2000, McGrath et al. 2009), killer cell lectin-like receptor subfamily G member 1 (Klrg1) (Li et al. 2009), natural killer cell group 7 (Nkg7), perforin (Prf1) (Parr et al. 1990), secreted phosphoprotein 1 (Spp1) (Herington & Bany 2007a) and tumor necrosis factor receptor superfamily member 9 (Tnfrsf9) (Eckstrum & Bany 2011) genes have previously been shown to be expressed in uNK or NK cells. Since uNK cells are missing from the IS tissues of Il15^−/− mice (Barber & Pollard 2003, Croy et al. 2003b) and the microarray analysis identified the above-mentioned genes to be differentially expressed, we verified differential expression of these genes using qRT-PCR in the IS tissue from Day 7.5 pregnant Il15^−/− compared to Il15^+/+ mice (Fig. 4E).

Advillin (Avil) Expression

Since AVIL is involved in cell morphogenesis (Shibata et al. 2004) and its expression was only seen in the IS tissue of Il15^+/+ mice in the microarray analysis, we further characterized its expression in the uterus. Avil mRNA levels significantly (P<0.05) increased in the IS compared to NIS tissues of the uteri of Il15^+/+ mice on Days 6.5, 7.5 and 8.5 of pregnancy during decidualization (Fig. 5A). Previously, we found uNK cell numbers are significantly decreased in uteri undergoing bead-induced decidualization by days 7.5 and 8.5 (Herington & Bany 2007b) but in a recent microarray study, differences in Avil mRNA levels were not seen between the decidua and deciduoma on Day 7.5 (McConaha et al. 2011). Therefore, we examined Avil mRNA levels in IS tissue of pregnant mice and bead-induced deciduomas (BID) of pseudopregnant mice on days 4.5–8.5. There was a significant (P<0.05) decrease in Avil mRNA levels in BID compared to IS tissues only on Day 8.5 (Fig. 5B). Next we assessed differential expression of Avil between IS tissues from Il15^+/+ and Il15^−/− mice on Days 4.5–8.5 of pregnancy. As shown in Fig. 5C, Avil mRNA levels were significantly (P<0.05) greater in the IS tissues of Il15^+/+ compared to Il15^−/− mice on Days 6.5, 7.5 and 8.5 of pregnancy. Finally, to localize where Avil is expressed we carried out in situ hybridization on sections from Day 7.5 (Fig. 5D,E) and 8.5 (Fig. 5F,G) pregnant uteri from Il15^+/+ mice. Avil mRNA was localized to a subgroup of cells in the mesometrial decidual region. To determine if this expression was localized to uNK cells we used dolichos biflorus agglutinin (DBA) lectin histochemistry coupled with Avil in situ hybridization. DBA lectin stains a subset of uNK cells in the mouse uterus during decidualization. As shown in Fig. 5H–J, the Avil mRNA was localized to a subset of DBA-positive uNK cells in the mesometrial decidua of Day 7.5 mice. Not all DBA lectin-positive cells were Avil mRNA-positive but all Avil mRNA-positive cells were DBA lectin-positive. Finally, sections from Il15^−/− mice showed neither DBA-positive uNK cells nor Avil mRNA-positive cells in the mesometrial decidua (Fig. 5K).

Fig. 5.

Avil expression in the mouse uterus during decidualization and its localization to uNK cells using qRT-PCR and in situ hybridization. Relative Avil mRNA levels between NIS and IS tissue of pregnant mouse uterus on Days 4.5–8.5 of pregnancy (A). Relative Avil mRNA levels between IS tissue and BID tissue of the mouse uterus on Days 4.5–8.5 of pregnancy and pseudopregnancy, respectively (B). Relative Avil mRNA levels IS tissue of from Il15^+/+ and Il15^−/− mice on Days 4.5–8.5 of pregnancy (C). Bars represent the mean ± SEM (N=3–4) and * denotes a statistically significant (P<0.05) difference between bars on a given day. Localization of Avil mRNA in IS tissue of Il15^+/+ mice on Day 7.5 (D, E) and 8.5 (F, G) of pregnancy. Localization of Avil mRNA and DBA lectin-positive uNK cells in IS tissue of Il15^+/+ (H–J) and Il15^−/− (K) mice on Day 7.5 of pregnancy. Scale bars are in microns.

Small Proline-Rich Protein 2D (Sprr2d) Expression

The expression of the large family of small proline-rich protein 2 (Sprr2) genes has been examined in the mouse endometrium and most are estrogen-regulated genes (Hong et al. 2004). Our microarray data suggested that Sprr2d mRNA levels are significantly (P< 0.05) reduced in the IS tissue of Day 7.5 pregnant uteri from Il15^−/− mice compared to those from Il15^+/+ mice. To characterize the expression of the Sprr2d gene during decidualization we examined expression in the uterus on Days 4.5 to 8.5 of pregnancy. As shown in Fig. 6A, Sprr2d mRNA levels significantly (P< 0.05) increased in IS compared to NIS tissues from the uteri of pregnant Il15^+/+ mice on days 6.5, 7.5 and 8.5 of pregnancy by approximately 16-, 29- and 214-fold, respectively. Next, we examined the differences in Sprr2d mRNA levels in IS tissues from Il15^+/+ compared to Il15^−/− mice. As shown in Fig. 6B, mRNA levels were significantly greater in IS tissues from Il15^+/+ compared to Il15^−/− mice on days 7.5–8.5. We then carried out in situ hybridization to localize Sprr2d mRNA. As shown in Fig. 6C, Sprr2d mRNA was localized to the antimesometrial decidua in the IS tissue of Il15^+/+ mice. Some mRNA was also localized to the same area in Il15^−/− mice, but to a lesser extent (Fig. 6C).

Fig. 6.

Sprr2d (A–C), Degs2 (D–F) and Ciapin1 (G–I) expression and localization in the mouse uterus during decidualization using qRT-PCR and in situ hybridization, respectively. Relative mRNA levels between NIS and IS tissue of pregnant mouse uterus on Days 4.5–8.5 of pregnancy (A, D, G). Relative mRNA levels IS tissue of from Il15^+/+ and Il15^−/− mice on Days 4.5–8.5 of pregnancy (B, F, H). Bars represent the mean ± SEM (N=3–4) and * denotes a statistically significant (P<0.05) difference between bars on a given day. Localization of mRNA in IS tissue of Il15^+/+ (left panel) and Il15^−/− (right panel) (C, F, I) mice on Day 7.5 of pregnancy.

Degenerative Spermatocyte Homolog 2 (Degs2) Expression

Our microarray data suggested that Degs2 mRNA levels are significantly reduced in the IS tissue of Day 7.5 pregnant uteri from Il15^−/− mice compared to those from Il15^+/+ mice. Since the levels of one of the products of DEGS2 action, ceramide, increases in the uterus during implantation and may play a role in decidualization (Kaneko-Tarui et al. 2007), we characterized the expression of Degs2 in the uterus during implantation. As shown in Fig. 6D, Degs2 mRNA levels were significantly (P<0.05) greater in NIS compared to IS tissue from the uteri of pregnant Il15^+/+ mice on days 5.5 and 6.5 of pregnancy. Next, we examined the differences in Degs2 mRNA levels in IS tissues from Il15^+/+ compared to Il15^−/− mice. As shown in Fig. 6E, mRNA levels were significantly greater in IS tissues from Il15^+/+ compared to Il15^−/− mice on days 7.5 and 8.5 of pregnancy. Finally, we carried out in situ hybridization to localize Degs2 expression in the uteri on Day 7.5 of pregnancy. As shown in Fig. 6F, Degs2 mRNA was found in mesometrial decidual cells just adjacent to the conceptus in IS tissue from Il15^+/+ mice. but not that of Il15^−/− mice.

Cytokine-Induced Apoptosis Inhibitor 1 (Ciapin1) Expression

Our microarray data suggested that Ciapin1 mRNA levels are significantly reduced in the IS tissue of Day 7.5 pregnant uteri from Il15^−/− mice compared to those from Il15^+/+ mice. Since RAS plays a role in decidualization (Cho et al. 2011) and Ciapin1 operates downstream of the receptor tyrosine kinase-Ras signaling pathway (Hao et al. 2006), we characterized Ciapin1 mRNA levels in the mouse uterus during decidualization. As shown in Fig. 6G, Ciapin1 mRNA levels increased in IS compared to NIS tissues from the uteri of pregnant Il15^+/+ mice on days 6.5–8.5 of pregnancy. Next, we examined the differences in Ciapin1 mRNA levels in IS tissues from Il15^+/+ compared to Il15^−/− mice. As shown in Fig. 6H, mRNA levels were significantly greater in IS tissues from Il15^+/+ compared to Il15^−/− mice on days 5.5–8.5 of pregnancy. Finally, we carried out in situ hybridization to localize Ciapin1 expression in the uteri on Day 7.5 of pregnancy. As shown in Fig. 6I, Ciapin1 mRNA was found in mesometrial decidual cells just adjacent to the ectoplacental cone and in the antimesometrial decidua adjacent to the myometrium in IS tissue from Il15^+/+ mice. A similar signal for Ciapin1 mRNA in tissues from Il15^−/− mice was only seen in the antimesometrial stromal cells in IS tissues.

Annexin A8 (Anxa8) Expression

Anxa8 encodes a protein that plays a role in late endosome organization and function (Goebeler et al. 2008). Our microarray data suggested that Anxa8 mRNA levels are significantly reduced in the IS tissue of Day 7.5 pregnant uteri from Il15^−/− mice compared to those from Il15^+/+ mice. To characterize the expression of this gene during decidualization we examined expression in the uterus on Days 3.5 to 8.5 of pregnancy. As shown in Fig. 7A, Anxa8 mRNA levels increased in IS compared to NIS tissues from the uteri of pregnant Il15^+/+ mice on days 5.5–8.5 of pregnancy. Next, we examined the differences in Anxa8 mRNA levels in IS tissues from Il15^+/+ compared to Il15^−/− mice. As shown in Fig. 7B, mRNA levels were significantly greater in IS tissues from Il15^+/+ compared to Il15^−/− mice on day 7.5.

Fig. 7.

Anxa8 (A, B), Rap1gap (C, D) and Zbtb32 (E, F) expression and localization in the mouse uterus during decidualization using qRT-PCR and in situ hybridization, respectively. Relative mRNA levels between NIS and IS tissue of pregnant mouse uterus on Days 4.5–8.5 of pregnancy (A, C, E). Relative mRNA levels IS tissue of from Il15^+/+ and Il15^−/− mice on Days 4.5–8.5 of pregnancy (B, D, F). Bars represent the mean ± SEM (N=3–4) and * denotes a statistically significant (P<0.05) difference between bars on a given day.

RAP1 GTPase-Activating Protein (Rap1gap) Expression

Although little is known about the function of Rap1gap expression in the uterus during implantation, we characterized its expression in the uterus since the microarray data suggested that its mRNA levels are significantly reduced in the IS tissue of Day 7.5 pregnant uteri from Il15^−/− mice compared to those from Il15^+/+ mice. As shown in Fig. 7C, relative Rap1gap mRNA levels significantly (P<0.05) increased in IS compared to NIS tissues on Days 6.5–9.5 of pregnancy. As shown in Fig. 7D, mRNA levels were also significantly (P<0.05) greater in IS tissues from Il15^+/+ compared to Il15^−/− mice on these days.

Zinc Finger and BTB Domain Containing 32 (Zbtb32) Expression

Zbtb32 is a transcriptional repressor that negatively regulates T-cell activation (Kang et al. 2005). Although little is known about the function of Zbtb32 expression in the uterus during implantation, we characterized its expression in the uterus since the microarray data suggested that its mRNA levels are significantly reduced in the IS tissue of Day 7.5 pregnant uteri from Il15^−/− mice compared to those from Il15^+/+ mice. Relative mRNA levels of Zbtb32 were found to be significantly (P<0.05) increased in IS compared to NIS tissues of the pregnant uterus on Days 6.5, 7.5 and 8.5 of pregnancy by an average of 7-, 15- and 35-fold, respectively (Fig. 7E). As shown in Fig. 7F, mRNA levels were also significantly (P<0.05) greater in IS tissues from Il15^+/+ compared to Il15^−/− mice on days 6.5–8.5.

Further Independent Verification of Microarray Data

We verified differential expression of several additional genes using qRT-PCR. As shown in Fig. 8, we attempted to verify differential levels of mRNAs between IS tissue from Il15^+/+ compared to Il15^−/− mice for an additional 18 genes. As shown in Fig. 8A, we verified a significantly greater mRNA level in IS tissue from Il15^+/+ mice for coronin actin binding protein 1A (Coro1a, P<0.05), diamine oxidase-like protein 2 (Doxl2, P<0.05), glycerophosphodiester phosphodiesterase domain containing 3 (Gdpd3, P< 0.005), NECAP endocytosis associated 2 (Necap2, P<0.001), neuromedin U (Nmu, P<0.05), platelet-derived growth factor receptor-like (Pdgfrl, P<0.001), phospholipase C gamma 1 (Plcg1, P<0.001), proline-serine-threonine phosphatase-interacting protein 1 (Pstpip1, P<0.001), selectin platelet ligand (Selplg, P<0.001), serine peptidase inhibitor Kazal-type 2 (Spink2, P< 0.05), chemokine (C motif) ligand 1 (Xcl1, P<0.001) and RIKEN cDNA 6330403K07 (6330403KO7Rik, P<0.05). As shown in Fig. 8B, we verified significantly greater mRNA level in IS tissue from Il15^−/− compared to Il15^+/+ mice for GTP cyclohydrolase 1 (Gch1, P<0.05), glutathione peroxidase 2 (Gpx2, 0.001), MAP7 domain containing 2 (Mtap7d2, P<0.001), regulator of G-protein signaling 5 (Rgs5, P<0.05), calcium binding protein A9 (S100a9, P<0.05), short coiled-coil protein (Scoc, P<0.005), serine peptidase inhibitor Kazal-type 3 (Spink3, P<0.05), and RIKEN cDNA 6720422M22 (6720422M22Rik, P<0.01).

Fig. 8.

Independent verification of the differential expression of 20 additional genes using qRT-PCR. Relative mRNA levels of 12 and 8 genes found to be expressed at higher levels in IS tissue of I115^+/+ (A) and Il15^−/− (B) mice by microarray analysis. Bars represent the mean ± SEM (N=3–6) and statistically significant difference in mRNA levels is denoted by asterisks (*P<0.05, **P<0.01, ***P<0.005 and ****P<0.001).

Discussion

The developing decidual tissue provides a critical nutritive environment for the conceptus before the establishment of the functional placenta. The major immune cells present in the uterus during decidualization are uNK cells and they are present in the mesometrial decidua during decidualization (Zhang et al. 2009). However, the role of the immune system in the process of decidualization is currently not well understood. The availability of several genetic models, where uNK cells are absent during pregnancy, makes such an assessment possible. Il15-deficient mice lack uNK cells in the uterus during implantation and this genetic model has been used previously to study uNK cell function in the uterus during implantation in mice (Ashkar et al. 2003, Barber & Pollard 2003, Croy et al. 2003a). This work clearly supports a role for uNK cells in “maintenance of decidual integrity” based on the observation that the decidual basalis after mid-pregnancy is acellular in uNK-deficient mice (Ashkar et al. 2003). Notably, the use of this and other similar models of uNK cell deficiency have not revealed if uNK cells play a role in decidualization itself, which occurs in uterus during the 4–5 days of pregnancy leading up to mid-pregnancy. Thus, the present study was initially undertaken to determine the differences in uterine gene expression in IS tissues between Il15^+/+ and Il15^−/− mice during decidualization using whole-genome Illumina Beadarray analysis. The microarray results suggested that the expression of 300 genes were differentially expressed and the differential mRNA levels of approximately one-sixth of these were verified independently using qRT-PCR. Interestingly, the present study did not reveal any differential expression of key genes known to play roles in endometrial stromal cell decidualization or of genes that are decidual cell markers. Therefore, the results of this study support the hypothesis that uNK cells do not play a critical role in decidualization and the loss of decidual integrity seen after mid-pregnancy in Il15^−/− mice is not due to a defect in decidual cell differentiation.

Decidualization involves a rapid growth of the endometrial tissue mass and is accompanied by angiogenesis. Several genetic models, including Il15^−/−, have established a critical role for uNK cells in the spiral arteriole modification seen after mid-pregnancy. Further, it has been speculated that uNK cells may play a critical role in endometrial angiogenesis during decidualization or implantation in rodents and humans due to the observations that they are a source of angiogenic factors (Lash et al. 2010, Zhang et al. 2011). The expression of several genes such as Vegfa (Halder et al. 2000), Ptgs2 (Lim et al. 1997, Matsumoto et al. 2002), Gja1 (Laws et al. 2008), Angpt2 (Matsumoto et al. 2002, Hess et al. 2006), Tie2 (Matsumoto et al. 2002), Flt1, Kdr (Chakraborty et al. 1995, Douglas et al. 2009) are believed to play a role in angiogenesis in the endometrium during decidualization. Further, several endothelial vascular endothelial cell markers such as Flt4 (Douglas et al. 2009), Cd34 (Morison et al. 2007), Pecam1 (Ma et al. 1997), Adamts9 (Jungers et al. 2005) and Inhbb (McConaha et al. 2011) have been shown to be expressed in the endometrial endothelial cells during decidualization. Notably, with the exception of Adamts9 which is an anti-angiogenic factor, none of the other genes were differentially expressed in IS tissues between Il15^+/+ and Il15^−/− mice. This included genes that encode angiogenic factors such as Vegfa and Vegfc which have previously been shown to be expressed by mouse and human uNK cells, respectively (Wang et al. 2000, Wang et al. 2003, Lash et al. 2006). Therefore, uNK cells appear not to be the major source of Vegfa expression in the mouse uterus during decidualization. Finally, vascular morphometric analysis in the present study revealed very little difference in the endothelial cell densities and endothelial cell proliferation indices between IS tissues from Il15^+/+ and Il15^−/− mice during decidualization. Taken together, the data of this study suggest that uNK cells do not play a major role in angiogenesis nor neovascularization in the endometrium during decidualization. Therefore, the lack of decidual integrity in genetic models that lack uNK cells is likely not due to decreased angiogenesis and impaired blood flow to the developing decidua during decidualization.

The cathepsin family of genes encodes either serine- or cysteine- proteases found in lysosomes within cells. Previous work has showed that the cathepsin D gene (Ctsd) is expressed in mature uNK cells in the mouse uterus by mid-pregnancy (Croy et al. 2010). However, whether uNK cells are a major source of other cathepsins remains to be determined. Therefore, it was interesting that the present study did not reveal a significant decrease in Ctsd expression in the IS tissue of IL15^−/− mice but did for Ctsg and Ctsw. CTSG is found in neutrophil lysozymes and upon secretion it can enhance the activity of interleukin-8, decrease the bioavailability of interleukin-6, activate cytokines (interleukin-1 beta, tumor necrosis factor alpha) and activate certain receptors (epidermal growth factor, proteinase activated receptors) (Conus & Simon 2010). Although it still needs to be proven, we speculate that the reduction of Ctsg expression found in the IS tissue of Il15^−/− mice may be due to a decrease in neutrophil numbers. Ctsw is exclusively expressed in NK cells as well as CD8-positive T cells but is not involved in cytotoxicity or interferon gamma production (Stoeckle et al. 2009). Unlike Ctsg, where there is an increased expression in IS tissues during decidualization, Ctsw expression in pregnancy appears to be complex because expression levels are similar in NIS and IS tissue until Day 8.5. The present study shows that Ctsg and Ctsw expression in the mouse uterus during decidualization are both dependent on IL15 and further work to localize expression and determine their function in the uterus during this process is warranted.

Several genes known to be expressed by uNK cells of the uterus during the establishment of pregnancy or in NK cells in general were expected to be expressed at significantly lower levels in IS tissue from Il15^−/− mice compared to that of Il15^+/+ mice. This included some granzymes (Allen & Nilsen-Hamilton 1998) and killer cell lectin-like receptor (Soderstrom et al. 1997, Verma et al. 1997, King et al. 2000, McGrath et al. 2009) genes along with Ctsd (Croy et al. 2010), Prf1 (Parr et al. 1990), Spp1 (Herington & Bany 2007a), and Tnfrsf9 (Eckstrum & Bany 2011). Such a result was expected as uNK cells would be a major source of this expression in the uterus during decidualization. Indeed, many of these genes (but not all) were actually shown to be differentially expressed. However, it is notable that of the four genes examined further using in situ hybridization, only Avil was shown to be expressed in uNK cells. Since expression of Sprr2d, Degs2 and Ciapin1 were not found in uNK cells, the control of their expression by IL15 may be more complex. Whether or not all of the other genes found to be differentially expressed at higher levels in IS tissues from Il15^+/+ compared to Il15^−/− mice are expressed in uNK cells will require further work on each gene.

AVIL is an intracellular protein that belongs to the gelsolin superfamily of actin regulatory proteins (Silacci et al. 2004). The present study demonstrated that the expression of Avil in the endometrium during decidualization is located to uNK cells. Although the precise function of AVIL in uNK cell biology is not understood at present, it is known to interact with a protein called scavenger receptor member 1 (SCARF1) (Shibata et al. 2004). AVIL is believed to play a role in cell morphogenesis through its interaction with SCARF1 (Shibata et al. 2004). Since uNK cells undergo a great deal of changes in cell size and shape as they mature in the endometrium during decidualization (He et al. 1999), we speculate AVIL may be playing a role in uNK maturation. However, SCARF1 is also responsible for binding to and internalization of extracellular heat shock proteins, which may be important in cell signaling and immunity (Calderwood et al. 2007a, Calderwood et al. 2007b). Thus, the precise function of AVIL in the endometrium during decidualization remains to be determined.

Ceramide levels increase in the uterus during decidualization where it may play a role in apoptosis. Apoptosis occurs in the epithelia and stromal compartments of the endometrium during implantation but little is known how this is regulated (Pampfer & Donnay 1999, Joswig et al. 2003). Ceramide levels are increased by a diverse number of stress stimuli and ceramide causes suppression in cell growth and promotion of apoptosis (Ruvolo 2003). It appears the balance of ceramide production and breakdown pathways seems to favor an increased accumulation in the uterus since ceramide levels in the mouse uterus increase during decidualization (Kaneko-Tarui et al. 2007). One major pathway of ceramide production is via serine where the last step involves the dihydroceramide Δ4-desaturase and C-4 hydroxylase activities of DEGS2, which produces ceramide from dihydroceramide (Enomoto et al. 2006). The present study shows that Degs2 mRNA is localized to the mesometrial decidua and levels are significantly reduced in IS site of Il15^−/− mice compared to that of Il15^+/+ mice. Therefore, it is possible that ceramide-mediated apoptosis of the mesometrial decidua may be occurring to make place for the developing placenta. Notably, we also characterized the expression of another apoptosis-related gene in this study, Ciapin1. Ciapin1 encodes a protein originally called anamorsin that also plays roles in apoptosis as a downstream effector of RAS-signaling in the uterus (Hao et al. 2006). Ciapin1 was also expressed in a similar region to Degs2 in the mesometrial decidua. Its expression levels dramatically increased in the IS tissue during implantation. In addition, expression of Ciapin1 in IS tissue of Il15^−/− mice was dramatically reduced during decidualization compared to that of Il15^+/+ mice. Therefore, it appears that the expression of both Degs2 and Ciapin1 in the uterus is dependent on IL15. Whether or not this involves the loss of uNK cells or occurs through another mechanism remains to be studied further, as do the potential roles of these genes in apoptosis in the mesometrial decidua during implantation.

In conclusion, this study provides the identity of many genes whose expression in the uterus during decidualization is altered between Il15^+/+ and Il15^−/− mice. However, only speculation on the potential function of many of these genes in the uterus during decidualization may be derived from the literature because the work was conducted on non-uterine tissues or cell types. For example, the present study shows the transcriptional repressor Zbtb32 is expressed at high levels in the IS tissues of the uterus during decidualization in Il15^+/+ mice and this expression is mainly lost in Il15^−/− mice. Since it has been shown in other tissues/cells that ZBTB32 acts as a repressor of the activity of GATA transcription factors, one can speculate it may play a similar role in the uterus. However, currently, very little is known about uterine GATA transcription factor gene expression and function in the uterine compartment during decidualization. Another example of a gene found to be differentially expressed in this study was Rap1gap. RAP1GAP is known to activate mitogen-activated or extracellular (MAP/ERK) kinase signaling cascades in other tissues/cell types by increasing the rate of GTP hydrolysis, which in turn causes the inactivation of RAP GTPases (Ika et al. 2009, Tsygankova et al. 2010). However, to the best of our knowledge, nothing is known about the function of Rap1gap expression in the uterus during decidualization. In summary, the functions of many of the other genes found to be differentially expressed in this study still remain to be determined and can only be speculated at this time. However, this study should form the basis of future research on these genes.

Methods and Materials

Mice and Sample Collection

All procedures involving mice were approved by the Southern Illinois University Institutional Animal Care and Use Committee. They were maintained under controlled light conditions (lights on from 0700 h to 2100 h), and allowed free access to food and water. CD1 mice with a targeted deletion of the Il15 gene (Il15^−/−) as well as wild-type (Il15^+/+) and Il15^−/− virgin females (8–12 weeks) were mated with Il15^−/− and wild-type male CD1 mice, respectively. Wild-type CD1 mice were obtained from Charles River Laboratories (Wilmington, MA) and generation of the Il15^−/− mice on a CD1 background was previously described (Eckstrum & Bany 2011). The morning a vaginal plug was detected was considered to be day 0.5 of pregnancy. These pregnant mice, all of which carried heterozygous embryos/conceptuses, were killed at 09:00 h on Days 4.5–9.5 of pregnancy. The bead-induced decidualization model was used exactly as described previously (McConaha et al. 2011).

RNA Isolation

For Day 4.5–8.5 uterine horns, NIS IS segments were separated. For IS segments, tissues of the conceptus (embryo and trophoblast cells) were carefully dissected out and discarded. All the remaining tissue were then pooled and considered IS segment tissue. Tissues were stored in RNA Later (Ambion). Total RNA was isolated using Trizol Reagent as recommended by the manufacturer (Invitrogen). The RNA was then subjected to DNase digestion using Turbo DNase as recommended by the manufacturer (Ambion), followed by re-extraction with Trizol.

Microarray Analysis

RNA samples isolated from the uterine tissue of Day 7.5 IS segments in Il15^+/+ and Il15^−/− mice were used for the microarray analysis (N=3). The RNA quality was verified using an Agilent Bioanalyzer (Agilent Technologies, Santa Clara, CA, USA) before microarray analysis as previously described (McConaha et al. 2011). Briefly, total RNA isolated from the tissue for each experiment was used to create biotin-labeled cRNA target samples (N=3) using the Illumina RNA amplification kit (Applied Biosystems, Carlesbad, CA). The cRNA target preparations were then hybridized to the arrays for 16–20 h at 55 °C, and the bound biotin-labeled targets were then detected using streptavidin-Cy3 and an Illumina BeadArray Reader with BeadStudio software (Illumina, Inc.). Probe data was filtered to include only those with detection scores >0.98 (P<0.02) for all samples. The data were then exported to Chipster for further analysis (Gentleman et al. 2004). The quantile normalization (Smyth & Speed 2003) tool of Chipster, which uses the limma R package, was used to normalize the data. Significance of differential expression was determined with the Chipster tool that uses the Bioconductor package limma Empirical Bayes method (Smyth 2004). The default Benjamini–Hochberg method was used to control the false discovery rate. Differences were considered significant if the adjusted P values were <0.05. Finally, the raw data was filtered to find probe signals which were significantly above background in all 3 samples from all three Il15^+/+ or Il15^−/− mice. This was defined as a detection score > 0.98, a signal level > 150 plus an average of 4-fold or greater signal above background. The lists of probes and relative signal levels were combined and annotated using the Bioconductor annaffy package of Chipster. Probes were removed from further analysis and were removed from the lists of differentially expressed genes if they did not map to a gene or were mapped to either more than one gene or a pseudogene. Finally, the hypogeometric test for GO tool of Chipster was used to determine over-and under-represented GO terms of the list of differentially-expressed genes using a cutoff of P<0.01 for statistical significance.

Quantitative RT-PCR (qRT-PCR) Analyses

qRT-PCR was carried out as described previously (McConaha et al. 2011) using the oligonucleotide primers (IDT Technologies, Coralville, IA, USA) shown in Supplemental Table 3. Briefly, High Capacity RNA-to-cDNA kits (Applied Biosystems) were used to generate the cDNA. PCR was performed using 2x iQ SYBR Green Supermix and a CFX real-time PCR machine (Bio-Rad). PCR conditions were 3 min at 94 °C followed by 40 cycles of 94 °C, 62–65 °C for 20 s, and 72 °C for 1 min for melting, annealing, and extension steps, respectively. The PCR primers were used at a final concentration of 200 nM. Cycle thresholds (Ct) were determined using the CFX software, and relative mRNA levels were determined as described previously after normalization to 18S rRNA levels (McConaha et al. 2011). The efficiencies of the PCR were linear between the mRNAs and 18S rRNA and were >85%. The effects of day of pregnancy and site of sampling on mean relative steady-state mRNA levels between NIS and IS tissue were determined statistically using 2-way repeated measures ANOVA. Likewise, the effect of day of pregnancy and mouse genotype on mean relative steady-state mRNA levels in IS tissue from Il15^+/+ and Il15^−/− mice was determined statistically using 2-way repeated measures ANOVA. ANOVA analysis was followed by the use of Duncan multiple range tests to determine differences between the means when warranted. Data collected on a single day of pregnancy were analyzed using T-tests to determine the effect of genotype on mRNA levels in IS tissues. All statistical analysis was conducted using SigmaPlot software (Systat Software Inc., San Jose, CA).

In situ Hybridization

To collect tissue for in situ hybridization, mice were anesthetized, perfused with PBS, and then fixed with paraformaldehyde as described previously (McConaha et al. 2011). Processing of tissue into paraffin was done using routine methods. Five micron sections were mounted onto HistoBond glassslides (Statlab, McKinney, TX, USA). Templates for sense and antisense riboprobe synthesis were prepared by PCR and cloned into pGEM-T Easy as described previously (McConaha et al. 2011). PCR primers used are shown in Supplemental Table 3. After sequencing the cDNAclones (UIUC DNA Sequencing Lab), the plasmids were used to generate transcription templates by PCR as described previously (McConaha et al. 2011). After purification of the DNA using a QIAquick PCR purification kit (Qiagen), the template was used to generate digoxigenin (DIG)-labeled riboprobes using MAXIscript T7 and SP6 kits (Applied Biosystems)along with dixoxigenin-11-UTP (Roche). After riboprobe purification using MEGAclear kits (Applied Biosystems), in situ hybridization was performed using antisense DIG-labeled riboprobes exactly as described in detail elsewhere (Simmons et al. 2007). Both antisense and control sense DIG-labeled riboprobes were used for in situ hybridization. Use of sense probes for all target genes studied in the in situ hybridization work of this study did not result in any positive hybridization signals (see Supplemental Fig. 1). In some cases, after completion of the in situ staining, the slides were processed for DBA lectin histochemistry exactly as described elsewhere (Eckstrum & Bany 2011).

Immunohistochemistry and Vascular Morphometry

In order to visualize those cells undergoing proliferation, mice were injected with BrdU 4 h prior to tissue collection as previously described (Herington & Bany 2007b). Uterine sections were de-waxed and hydrated using routine techniques then exposed to 0.2% trypsin in PBS at 37°C for 10 min for antigen retrieval. After washing sections in PBS, they were incubated with 1.5M HCl for 15 min and then washed with water. To neutralize the sections they were covered with borate buffer (0.1M, pH 8.5) for 10 min, and then washed in PBS. Subsequent incubations were carried out using antibody amplifier trays from ProHisto (Columbia, SC) and a horizontal shaker. Sections were next incubated in blocking buffer containing 2% donkey serum in PBS with 0.05% tween (PBST) for 60 min. Sections were then incubated with 0.3 ug/ml sheep anti-BrdU (Biodesign International, Saco, ME) and rat anti-mouse CD34 IgG (0.1 μg/ml)(Cedarlane Laboratories, Hornby, ON) for 1 hour. After washing in PBST, the sections were incubated with 0.5 ug/ml biotinylated donkey anti-rat IgG and 1 ug/ml Dye-light 488 donkey anti-sheeep IgG in blocking medium (Jackson ImmunoResearch Laboratories, West Grove, PA) for 60 min. After washing sections in PBST, they were incubated with 0.5 ug/ml Dye-light 549-conjugated Streptavidin (Jackson Immunoresearch) in PBS for 30 min. After washing with PBST, the slides were incubated with 4′6-diamidino-2-phenylindole (DAPI) in PBS for 10 minutes then washed in PBS. Finally, coverslips were placed over the sections using Fluoromount G mountant (Southern Biotech, Birmingham, AL).

Slides were examined using a Leica CTR5000 fluorescence microscope equipped with a Retiga 2000JR QImaging Camera and using Qcapture Pro software (QImaging, Burnaby, BC). Endothelial cells were counted as CD34 cells lining the lumens of blood vessels, and BrdU-positive cells were counted as proliferating cells. Photomicrographs were taken with a 20x objective in central plus lateral mesometrial and antimesometrial regions of the endometrium, and total BrdU-positive and –negative endothelial cells were manually counted in enlarged prints of the photomicrographs from antimesometrial and lateral and central mesometrial uterine regions (from 4 independent samples). Endothelial cell density was calculated as the number of endothelial cells normalized to area while proliferation index was calculated as the percent of BrdU-positive endothelial cells or vessels. A minimum of 200 endothelial cells per area were counted in each independent sample. Two-way ANOVA was performed to detect overall differences in the uNK cell subtype densities for each individual day and sub-region using SigmaPlot software (Systat Software, Inc., Chicago, IL). This was followed by the use of Duncan multiple range tests to determine differences between the means.