To identify potential downstream gene targets of FIGLA, total RNA was obtained from normal and Figla null gonads isolated from E12.5, E14.5, E17.5 and newborn female mice. Three independent biological samples obtained from each embryonic time point and four from newborn gonads were linearly amplified, labeled with Cy3 and Cy5 and hybridized to the National Institute of Aging (NIA) cDNA microarray consists of ~22K features enriched for transcripts from newborn ovaries, pre- and peri-implantation embryos 1213. After washing, the average statistically significant intensities for each element were analyzed with Gene Spring GX software. Features that varied more than 2-fold (with a coefficient of variance less than 30%) between normal and Figla null gonads were selected for further analysis.

The M (mean log ratio) vs. A (average log₂ signal value) scatter plots reflect the fold change of differentially expressed genes in Figla null and normal ovaries on the Y axis relative to their abundance on the X axis (Fig. 1). Thus, transcripts with low intensity ratios (blue) indicate genes that are potentially up-regulated by FIGLA and transcripts with high intensity ratios (red) represent genes that are potentially down-regulated by FIGLA. As expected, no genes were differentially expressed at E12.5, a point in development prior to the onset of Figla expression. Only a few differences were observed at E14.5 and E17.5 with 6 and 4 genes up-regulated ≥ 2-fold (ρ ≥ 0.05) in normal ovaries and 8 and 1 up-regulated in Figla null ovaries, respectively (Fig. 1A). In marked contrast, 176 transcripts were ≥ 2-fold more abundant in normal and 44 were ≥ 2-fold more abundant in Figla null newborn ovaries (ρ ≤ 0.05).

The average intensities of hybridization of the 176 genes that were less abundant (i.e. potentially up-regulated, Fig. 2A) and the 44 genes that were more abundant (i.e., potentially down-regulated, Fig. 2B) in Figla null newborn ovaries were compared by hierarchical cluster analysis. Almost all the 176 potentially up-regulated genes showed similar expression pattern wherein the expression of these genes commenced only at newborn stage of the ovary development. As expected, both Zp2 and Zp3, previously described direct downstream targets of FIGLA, were up-regulated in the normal newborn ovaries, although they were not closely clustered to each other. The four genes [NIA:551381, H330A03, H3134D03, H3058H02] that were first up-regulated E17.5 and persisted in the newborn also did not cluster together (dots, Fig. 2A). The corresponding positions of the genes which were further characterized by qRT-PCR and in situ hybridizations are marked by asterisks and labeled. The 44 genes which were potentially down regulated by FIGLA (i.e., more abundant in Figla null ovaries) all had similar expression patterns with the major change in expression occurring in the newborn ovary (Fig. 2B).

Of these, 165 of the potentially up-regulated and 38 of the potentially down-regulated transcripts were judged to differ with statistical significance after analysis of variance (ANOVA). There was no overlap of regulated genes (≥ 2-fold, ρ ≤ 0.05) among the various time points except for 2 genes [NIA:551381, NIA:H330A03] up-regulated at E17.5 that were also up-regulated in the newborn ovary (Fig. 1B). Genes that were potentially up- and down-regulated are provided (see Additional files 1 and 2). Using gene ontology software PANTHER 14, the 203 genes that were differentially-regulated in the newborn were analyzed based on their molecular function (Table 1). Of the 165 up-regulated genes, 25% were grouped in unknown molecular function category, 13% were nucleic acid binding proteins, 7% were transcription factors and 6% were genes with transferase activity. Of the 38 down-regulated genes, 23% were transcription factors, and 20% encoded proteins with nucleic acid binding functions. Two of these genes, Taf7l and Tia1, are normally expressed in the testes. Genes with unknown molecular function and transferase activity were comparable (25% and 7%, respectively) to the up-regulated genes.

Thirteen genes that were ≥ 2-fold more highly expressed in normal than Figla null newborn ovaries were chosen for more detailed analysis. Three were members of the Nalp gene family that have oocyte-specific expression 15; five were functionally annotated genes with oocyte-specific expression; and five were functionally un-annotated. Two additional members of the Nalp family (Nalp4a, Nalp14) that were up-regulated by FIGLA, albeit to a lesser extent, were included in the analysis and two of the selected genes (Oas1d, [Genbank:BC052883]) missed the statistical cutoff because of single (out of eight) outlying data points (shaded backgrounds in Figs 3, 4, 5). Primers specific for each gene were designed (see Additional file 3) and the presence and absence of specific transcripts in newborn normal and Figla null ovaries were confirmed in 14 of the 15 by qRT-PCR (all but Nalp4a).

Total RNA was isolated from 9 normal newborn organs and assayed for gene expression which was normalized to HPRT and standardized to 100% in the ovary (Fig. 3A). As previously reported, all five Nalp genes were expressed in the ovary with low levels of transcripts (< 5% of ovarian expression) observed in the uterus for Nalp14 and in the liver for Nalp4a of newborn mice. The developmental profiles of transcript abundance from E12.5 to newborn were consistent with FIGLA regulating expression of Nalp4b, Nalp5, Nalp4f and Nalp14. No expression was detected in E12.5, E14.5, E17.5 or newborn gonads isolated from Figla null mice, but all five genes were expressed in normal newborn ovaries. However, the absence of FIGLA did not preclude expression of Nalp4a in newborn ovaries, although expression was ~60% of normal. Although the members of the Nalp gene family share structural motifs and are co-expressed in oocytes, they are not functionally redundant and the inactivation of one (Nalp5) is sufficient to arrest pre-implantation development 16.

Among the other annotated genes, each was preferentially expressed in the ovary with only low levels (< 5% of ovarian expression) of Oas1d, Serpinb6C and Arhgap20 transcripts in other tissues in newborn mice (Fig. 4A). Expression of Pdzk1 and Elavl2 was less tightly controlled with transcripts detected in kidney (~25% of ovarian expression) and brain (~40%), respectively. Each of these five genes was first detected in the newborn ovary where their expression was dependent on FIGLA (Fig. 4B). These observations are consistent with each gene being a direct downstream target of FIGLA, but transcripts of each are detected in adult tissues, suggesting that other transcription factors play a role in developmental and organ specific expression. Oas1d encodes an oocyte-specific dsRNA, 2'5'-oligoadenylate synthetase that is ~60% identical to OAS1A that has been implicated in the interferon defense response against viral infections. Genetically altered mice, lacking OAS1D, are fertile, but have reduced fecundity associated with lowered ovulatory capacity and defects in early embryonic development 17. Mice lacking PDZK1, implicated in controlling ion transport and cholesterol metabolism 18, had normal fertility and produced off-spring with the expected Mendelian inheritance of the Pdzk1 null allele 19.

Five additional genes that were more abundant in normal than in Figla null newborn ovaries and functionally un-annotated were also selected for further evaluation (Fig. 5). Three [E330009P21Rik, E3300017A01Rik, Genbank:BC052883] were detected only in the ovary by qRT-PCR and the remaining two [C330003B14Rik, E330034G19Rik] had only low levels of expression in other tissues. All five genes had similar developmental profiles as the aforementioned genes with little expression prior to birth and virtual absence of expression in Figla null ovaries.

In situ hybridizations were performed using the digoxigenin-labeled oligonucleotide probes designed specifically for transcripts of each gene (Additional file 4). All the functionally unknown genes showed predominant expression in oocytes (Fig. 6) with only background staining surrounding ovarian tissue. Although some binding was observed with sense probes for [E330017A01Rik, C330003B14Rik and Genbank:BC052883](Fig. 6D, F, J), it was minor compared to the strong signals observed with anti-sense probes (Fig. 6C, E, F).

The developmental microarray analysis suggested that significant differences in potential FIGLA gene targets were not observed prior to birth. Therefore, to complement these data and avoid the bias introduced by pre-selection of elements on the microarrays, a SAGE analysis was undertaken to provide identity and relative abundance of individual transcripts. Total RNA was isolated from newborn ovaries and used to construct a normal and a Figla null SAGE library in which transcripts were identified by a 10 nt tag and their relative abundance by the number of tags detected. The 79,095 quality tags sequenced from the normal library and the 77,851 from the null library represented 23,595 and 32,123 different transcripts, respectively. 15,838 tags were unique to the normal library; 24,366 were unique to the null library and 7,757 were present in both. To avoid sequence-error artifacts, only tags that were observed at least twice in the normal (7,715) and Figla null (9,477) SAGE libraries were included in subsequent analyses (Fig. 7A).

Using the Audic and Claverie algorithm that incorporates Bayesian and false alarm analyses 20, 977 genes more highly expressed in the normal library and 1308 genes more highly expressed in the Figla null library were identified (Fig. 7B). The data also were analyzed with other statistical tests commonly used for SAGE 21 and Fisher's exact test provided similar results (> 90% overlap) with 810 genes more highly expressed in normal and 1318 more highly expressed in null ovaries. However, only Pou5f1 (Oct4) and Zp2 were observed as up-regulated in both newborn microarrays and SAGE libraries and no commonly down regulated genes were observed among those identified by SAGE and by microarray analyses. Although the microarray and SAGE screens were selected to complement one another in identifying potential downstream targets of FIGLA, the lack of overlap in targets was unanticipated. The data from the newborn normal and Figla null microarray data were reexamined by False Discovery Analysis 22. Of 2,233 identified features (FDR threshold of 5%) from the microarray data, 1,550 had single unigene annotations 23, 202 mapped to multiple clusters and 248 were not mapped. Of the annotated genes, 768 were up-regulated (see Additional file 5) and 782 were down-regulated (see Additional file 6). Eleven genes (marked in Additional files 5, 6), including Pou5fl and Zp2 from our initial analysis, were also present in our SAGE analysis. Eight were up-regulated (Pou5f1, Zp2, Ivns1abp, Vbp1, Padi6, Rbpms2, Genbank:EG626058, 6430591E23Rik) and three were down-regulated (Sp3, Hdac2, Ogt) in newborn normal ovaries. Because alternative statistical analyses 242526 may be useful for the comparison of these data sets, we have made the original microarray and SAGE data available at Gene Expression Omnibus 27 under accession GSE5558 and GSE5802, respectively.

Of the genes identified in the SAGE screen, 64 (see Additional file 7) of the potentially up-regulated genes had ≥ 10 tags in normal and none in the null and 312 (see Additional file 8) of the potentially down-regulated genes had ≥ 10 tags in the null and none in normal ovaries. These 376 tags were reliably mapped to genes using the NCBI SAGE database 28 and ontology of 112 was determined by PANTHER (Table 1). The greatest percentage (26%) of genes were nucleic-acid binding proteins (including transcription factors) followed by oxidoreductases (13%), cytoskeletonal proteins (10%) and ligases (8%). Ten of the down-regulated genes (Tnp2, Hils1, Clgn [Calmegin], Tekt1, Fscn3, Dnahc8, Ldh3, Adam3 [Cyritestin], Oaz3, Akap3), scattered among the categories, have sperm-associated patterns of expression (Table 1).

Eight transcripts for which there were ≥ 10 tags in the normal and 0 tags in the Figla null newborn SAGE libraries (Pou5f1, Dppa3, Oas1h, Padi6, [Genbank:AK087784, 2410146L05Rik, Genbank:BG074389, Genbank:AK139812]) were analyzed further. Transcripts for each were detected by semi-quantitative RT-PCR in normal, but not Figla null newborn ovaries. Total RNA from 11 tissues was analyzed and transcripts for each gene were detected only in ovarian tissue, with the exception of Pou5f1, which was also present in testis (Fig. 8A).

Pou5f1 (Oct4) is expressed in pluripotent cells during mouse development before becoming restricted to germ cells 29 and regulates a significant number of downstream target genes by itself and in tandem with other transcription factors 30. The complex pattern of POU5F1 transcripts in normal and Figla null mice (Fig 8B) indicates participation of other transcription factors in controlling its expression. Expression of the other seven genes was not detected in Figla null mice (except for a very modest accumulation of [Genbank:BG084789] at E15.5) and was first observed in normal ovaries at E15.5 (Dppa3, [Genbank:AK139812] or E19.5 (Oash1, Padi6, [Genbank:AK087784, 2410146L05Rik, Genbank:AK139812])(Fig. 8B).

Dppa3 (Stella), originally implicated in germ cell lineage specification 31, is a maternal effect gene required for pre-implantation development 3233. Oas1h encodes a 2'5'-oligoadenylate synthetase and clusters with Oas1d (detected by microarray analysis) along with 6 other closely related synthetases on mouse Chr5:119,938,130-120,073,460 34. Padi6 (Padi5) is one of 4 similar genes (Padi1–4) clustered on mouse Chr4:139,787,689-139,623,878 that encodes a peptidylarginine deiminase which converts arginine residues into citrulline 35. It is expressed during oogenesis where it is associate with cytoplasmic sheets and the protein persists in the early embryo up to the blastocyst stage 36.

Each of the remaining four SAGE tags matched either cDNA or spliced EST that was expressed in ovarian tissue and/or the early mouse embryo. [Genbank:AK087784] is a full-length cDNA from a 2 days pregnant adult female ovary [E330020D12Rik] that maps to mouse Chr1:153,290,975-153,292,514 and [2410146L05Rik] is a female germline specific cDNA that encodes a hypothetical 166 amino acid protein mapping to mouse Chr9:78,577,285-78,578,468. [Genbank:BG074389], a 532 bp spliced EST (Chr11:7,077,950-7,079,635), is part of the NIA15K cDNA clone set and was up-regulated in the microarray screen, albeit not to a statistically significant extent. [Genbank:AK139812] was reliably matched by a SAGE tag that also matched two other sequences, neither of which was ovary specific. [Genbank:AK139812] corresponds to a full-length cDNA [B020018122Rik] obtained from a 2-cell embryonic library and maps as a spliced EST on mouse Chr12:107,907,672-107,914,726.

CF1 mice were obtained commercially and Figla homozygous null mice 9 were identified by genotyping tail DNA using three primers (see Additional file 9) in a PCR reaction (95°C 5 min, 94°C 30 sec, 60°C 30 sec, 72°C 30 sec for 28 cycles). RNA was extracted from gonads isolated at embryonic day 12.5 (E12.5), E13.5, E14.5, E15.5, E17.5, E19.5, newborn (NB), 2 days post-partum (2dpp) and 7dpp females; all other tissues were obtained from newborn mice. All experiments were conducted in compliance with the guidelines of the Animal Care and Use Committee of the National Institutes of Health under a Division of Intramural Research, NIDDK approved animal study protocol.

RNA was extracted from tissue using Absolutely RNA RT PCR Miniprep Kit (Stratagene, La Jolla, CA) and its integrity was determined with RNA 6000 Nano Lab-on-Chip (Agilent Technologies, Waldbronn, Germany). For hybridization to microarrays, 50 μg of total RNA (20 ovaries) was linearly amplified with the Aminoallyl RNA Amplification and Labeling System (NuGen Technologies, San Carlos, CA) and labeled with ester-linked Cy3 and Cy5 dyes (Amersham Biosciences, Piscataway, NJ).

Labeled cDNA probes were hybridized to NIA cDNA microarrays on glass slides containing 20,996 features 1213 according to Vanderbilt Microarray Shared Resources protocols 77. In brief, arrays were washed (0.2% SDS) at RT and incubated in a pre-hybridization solution (5 × SSC, 0.1% SDS, 1% BSA) at 55°C for 45 min 78. After 5 rinses in Milli Q water and 1 in iso-propanol, arrays were air dried and overlaid with coverslips. Utilizing a Maui Hyb Station (BioMicro Systems, Salt Lake City, UT), the arrays were hybridized (25% formamide, 5 × SSC, 0.1% SDS, 1 μg poly(A)⁺ RNA) at 42°C for 16 h and then washed (5 min, 55°C) sequentially with 2 × SSC, 1 × SSC and 0.1 × SSC, each with 0.1% SDS After drying by centrifugation (20 × g), the arrays were scanned using the Axon 4000 B (Axon Instruments, Union City, CA).

Hybridizations were performed in triplicate (E12.5, E14.5, E17.5) or quadruplicate (newborn) with dye-swapped pairs of cDNA. The raw data were analyzed with GeneSpring GX 7.2 (Agilent Technologies, Waldbronn, Germany) to remove data that did not meet the following criteria: 1) Cy5 signal to background intensity ratio less than 1.5; 2) Cy3 signal to background intensity ratio less than 1.5; 3) Cy5 signal less than 200; and 4) Cy3 signal less than 200. Data were then normalized by the LOWESS sub-grid method and features with a coefficient of variance of less than 30% (across replicate arrays) and ≥ 2-fold changes between normal and Figla null gonads were identified. Analysis of Variance (ANOVA) was performed on the set of genes identified in newborn ovaries. Hierarchical clustering of transcripts during development was determined by GeneSpring GX 7.2.

Alternatively, all eight replicates of newborn ovaries were used in a one-sample Student's t-test to determine whether the mean normalized expression value for each element was statistically different from 1. False Discovery Rate 37 at a threshold of 5% was used as a multiple testing correction to decrease the number of false positives.

Total RNA (~50 μg) was obtained from newborn normal and Figla null ovaries with an RNeasy mini kit (Qiagen, Valencia, CA). The presence of MSY2 and the absence of ZP2 transcripts was assayed by one-step qRT-PCR according to the manufacturer's protocol (Qiagen, Valencia, CA) using oligonucleotide primers (see Additional file 3) to confirm the identity of Figla null ovaries. Poly(A)⁺ RNA was isolated with oligo d(T)-conjugated magnetic beads (Dynal-Invitrogen, Carlsbad, CA) and 5 μg was used to construct libraries using a modification 79 of the original SAGE protocol 80. The final concatenated ditags were cloned into Bluescript pKS (Stratagene, La Jolla, CA) using the BamHI site for blue-white screening. Plating, picking, DNA preparation and sequencing were performed at the NIH Intramural Sequencing Center (NISC). Low quality sequence data were eliminated by Phred analysis.

eSAGE 81 was used to extract tags and remove duplicate ditags from the raw sequencing data. 79,095 tags from the normal were compared to 77,851 tags from the Figla null libraries to determine statistically significant differences either by the Audic and Claverie test 20 using eSAGE 82 or Fisher's exact test 83 using IDEG6 21. Tags with ρ ≤ 0.05 were extracted from both lists and further analysis was performed on tags present at least 10 times in normal and absent from Figla null ovaries. These lists of tags were then mapped to Unigene clusters using reliable mapping from NCBI Unigne build #138. A group of 13 genes from the microarray data and 8 from the SAGE analysis that were up-regulated in normal newborn ovaries were selected for more detailed analysis based on their expression profile in the SOURCE data base at the Genetics Department at Stanford University 23.

Transcript abundance was determined by quantitative real-time polymerase chain reaction (qRT-PCR) on an ABI 7700 HT Sequence Detection System (Applied Biosystems, Foster City, CA) using QuantiTect SyBR green PCR kit (Qiagen, Valencia, CA) and primers (see Additional file 3) designed by Primer Express 2.0 (Applied Biosystems, Foster City, CA). Independently obtained biological samples (2–6), analyzed in triplicate were normalized to HPRT (averaged for tissue or developmental time point) and expressed as an average (± s.e.m). PCR products from the SAGE analysis were separated by agarose gel electrophoresis and stained with ethidium bromide (0.1 ug/ml). Tissue-specific expression for the in situ hybridization was performed on ovarian sections obtained from normal, 6–8 week old mice. Ovaries were fixed (4% paraformaldehyde, PBS, pH 7.4), and processed for paraffin sections (5 μm) which were then re-hydrated stepwise through alcohol washes (100%, 70%, 50%, 30%) and distilled water prior to PBS. Sections were hybridized with sense or antisense probes (200 ng/ml) (see Additional file 4) labeled with digoxigenin (DIG) according to the manufacturer's instructions (GenPoint, Dako, Carpenteria, CA) except for the use of rabbit anti-DIG-HRP antibodies (1:200, Dako, Carpenteria, CA). Sections were then counter-stained with hematoxylin and de-hydrated stepwise in ethanol (30%, 50%, 70%, 100%) prior to permount mounting (Fisher Scientific, Fair Lawn, NJ). Images were acquired with an Axioplan microscope equipped with a CCD camera using AxioVision software (Carl Zeiss, Gottingen, Germany).