Normalizing Gene Expression Levels in Mouse Fetal Germ Cells

Abstract

Real-time PCR has become a popular method to analyze transcription of genes that are developmentally regulated during organogenesis of the testes and ovaries. However, the heterogenous cell populations and commitment to strikingly different developmental pathways of the germ and somatic cells in these organs complicate analysis of this process. The selection of suitable reference genes for quantifying gene expression in this system is essential, but to date it has not been sufficiently addressed. To rectify this problem, we have used fluorescence-activated cell sorting to purify germ cells from mouse fetal testes and ovaries and examined 16 common housekeeping genes for their suitability as reference genes. In pure populations of germ cells isolated from Embryonic Day 12.5 (E12.5) to E15.5 male and female gonads, Mapk1 and Sdha were identified as the most stable reference genes. Analysis of the heterogenous fraction of gonadal somatic cells revealed that Canx and Top1 were stable in both sexes, whereas a comparative analysis of germ and somatic cell populations identified Canx and Mapk1 as suitable reference genes through these developmental stages. Application of these reference genes to quantification of gene expression in developing gonads revealed that past assays, which employed nonverified reference genes, have in some cases provided misleading gene expression profiles. This study has identified suitable reference genes to directly compare expression profiles of genes expressed in germ and somatic cells of male and female fetal gonads. Application of these reference genes to expression analysis in fetal germ and somatic cells provides a more accurate system in which to profile gene expression in these tissues.

Identification of transcriptionally stable genes allows accurate normalization of gene expression in developing gonads.

INTRODUCTION

Testes and ovaries provide complex developmental and functional niches for the production of gametes from the resident germ cells. The cellular and molecular processes involved in formation of testes and ovaries and the development of the germ cells within these organs have been the subject of study for many years. Aberrant development of either the somatic or germ cell components of these organs can lead to pathologies, such as infertility and cancer. Organogenesis of the testis and ovary, as well as the regulation of proliferation and differentiation of the germ cells within these organs, is a complex process that involves several bipotential progenitor lineages, including the germ cells, supporting cells, and various interstitial cell types. The germ cells enter the gonads at approximately Embryonic Day 10.5 (E10.5), and under the influence of the surrounding somatic tissues, they enter the spermatogenic (male) or oogenic (female) developmental pathways. This results in mitotic arrest of the developing male germ cells but arrest of the female germ cells in meiotic prophase. At the same time, other processes occur within the male and female germ cells, including erasure of existing epigenetic patterns and their replacement with new patterns that are compatible with development in the following generation.

The gonadal primordia of both males and females are initially bipotential. The decision to commit to the male developmental pathway is determined by expression of the transcription factor Sry in XY gonads at E11.5 [1–3]. SRY induces differentiation of Sertoli cells, which in turn produce signals crucial for the differentiation, migration, and proliferation of the various somatic cell types that constitute the testis [4]. By E12.5, Sertoli cells are organized into cords that surround the germ cells and direct their development. Through an unknown signaling mechanism that is likely mediated via a secretory pathway involving SDMG1 [5], germ cells are committed to spermatogenic development between E11.5 and E12.5 [6] and arrest in G₀ of mitosis between E12.5 and E14.5 [6–8].

By contrast, in the absence of Sry, the bipotential gonad develops as an ovary, with the bipotential supporting cells differentiating as granulosa cells, which support germ cell development. The genetic pathway leading to ovarian development is poorly understood. However, disruption of Wingless-related MMTV integration site 4 (Wnt4) [9, 10], follistatin (Fst) [11], R-spondin 1 (Rspo1) [12], and forkhead box L2 (Foxl2) [13, 14] function in the somatic cells results in aberrant ovarian differentiation. Furthermore, stabilization of beta-catenin results in male-to-female sex reversal [15]. Commitment of the germ cells to oogenesis occurs between E12.5 and E13.5 [4, 6]. Coincident with this commitment, retinoic acid-dependent induction of Stra8 expression is required for passage of the germ cells through premeiotic S phase, which is followed by entry of the germ cells into meiotic prophase, where they arrest [16–18].

At present, the most widely used techniques to examine transcriptional expression of genes through time or between sexes during fetal gonad development are whole-mount in situ hybridization and quantitative real-time RT-PCR (qRTPCR). Although whole-mount in situ hybridization is useful to determine temporal and spatial gene expression patterns, it is not a quantitative technique. By contrast, qRTPCR can be highly quantitative and allows rapid expression profiling of many genes in numerous samples.

Many variables must be considered when conducting gene expression analyses by qRTPCR. Common factors taken into account include the amount of starting material, the quality of RNA, amplification efficiencies, and the use of endogenous reference genes (or “housekeeping” genes) to normalize data [19–21]. An important issue with validating qRTPCR assays is the choice of method used to normalize differences in the amount of RNA between samples. One method is to measure the amount of starting RNA in the sample and then spike in a known amount of exogenous target. This exogenous target is then used to normalize expression between samples [22]. A more common approach to normalizing gene expression between samples is to use an endogenously expressed reference gene. A good reference gene should be expressed stably in the tissue or treatment investigated. Because transcription of some common reference genes, such as Gapdh and Actb, is regulated in various tissues, it cannot be assumed that these genes are suitable for normalization [23]. Further, Vandesompele et al. [21] have shown that normalization using a single reference gene can be inaccurate. To circumvent this problem, they devised software that allows geometric averaging of multiple reference genes to systematically identify genes that are stably expressed in the model of interest. Several similar tools, such as NormFinder and Bestkeeper, are also available to identify suitable reference genes [24, 25].

Cell populations within developing male and female gonads are heterogeneous and fluid in terms of differentiation, proliferation, relative cell abundance, and gene expression. Changes in the relative proportion of various cell types during development will affect expression profiles of genes expressed in a specific cell type within the tissue examined. This problem can be overcome by isolating the cell type of interest and analyzing gene expression with reference to genes that are stably expressed in that cell type.

To accurately quantify transcript abundance in germ cells by qRTPCR, we used fluorescence-activated cell sorting (FACS) to separate germ cells from the somatic cells of E12.5–E15.5 male and female gonads. The purified RNA from these samples was used in qRTPCR to analyze expression of 16 commonly used reference genes in these tissues. We then applied GeNorm to these data to identify appropriate stably expressed genes for normalization in this system.

MATERIALS AND METHODS

Mouse Preparation and Tissue Collection

Embryos used for all germ cell-sorting experiments were derived from OG2 (Oct4 [Pou5f1]-eGFP) transgenic male (C57Bl6) × CD1 female matings. Mating was detected by the presence of a vaginal plug in the morning and recorded as E0.5. All animal procedures were carried out in accordance with the guidelines stipulated by the Murdoch Children's Research Institute (MCRI) Animal Ethics Committee.

RNA Isolation and Amplification

The E12.5–E15.5 embryos were obtained from OG2/CD1 matings and carefully sexed and staged according to counting tail somites or morphological criteria [26, 27]. Gonads were cleanly separated from the mesonephric tissue, and 6–18 gonads (the number depending on age) of the same sex were pooled for each stage and were dissociated using trypsin. Tissues were collected in biological triplicate for each sampling point tested. Enhanced green fluorescent protein (EGFP)-positive and EGFP-negative cells were isolated using FACS. Cells were collected by centrifugation, and RNA was isolated using an RNeasy Micro Purification Kit (Qiagen) according to the manufacturer's instructions. RNA quantity and RNA integrity were assessed using a Nanodrop ND-1000 spectrophotometer and an RNA 6000 Pico LabChip with an Agilent 2100 Bioanalyzer (Agilent Technologies, Palo Alto, CA), respectively. RNA integrity was evaluated using Agilent 2100 Expert software (Agilent Technologies), which generated an RNA integrity number (RIN) for each sample. The RIN scale ranges from a value of 1 (RNA that is completely degraded) to 10 (RNA exhibiting no degradation). RNA samples with a high RIN (>9.0) were used in single-round amplification reactions using a MessageAmp II antisense RNA (aRNA) Amplification Kit (Ambion) according to the manufacturer's instructions. The quantity and quality of the aRNA were assessed by spectrophotometry and gel electrophoresis.

Reverse Transcription and Real-Time PCR Analysis

A total of 100 ng of aRNA was reverse transcribed using random hexamers and Superscript III (Invitrogen) according to the manufacturer's instructions. Real-time PCR was performed using the mouse Universal Probe Library (UPL) system and LightCycler480 Probe Master mix (Roche). A total of 1 ng of cDNA was subject to amplification using a LightCycler480 Real-Time PCR Instrument (Roche). Samples were run in biological and technical triplicate, and experiments were performed twice. The following PCR reaction parameters were used: 40 cycles of 95°C for 15 sec and 60°C for 1 min in a two-step thermal cycle, preceded by an initial 10-min step at 95°C to activate the polymerase. All qRTPCR primer/UPL probe set combinations used are included in Supplemental Table S1 (all Supplemental Tables are available online at www.biolreprod.org). For all qRTPCR data presented here, no-template controls exhibited no amplification. In addition, each amplification set was performed with four-point standard curves in duplicate to confirm primer/probe efficiency. Specificity of all real-time PCR assays was confirmed by running products on a 2% agarose gel. All assays showed a single expected product size (data not shown).

Determination of Gene Stability and Expression Levels

Raw-data Ct mean values were converted to relative quantities using the comparative Ct method, where highest relative quantity (Q) for each of the genes studied was set to 1. These values were then submitted to the GeNorm Excel VBA applet for further analysis [28]. Briefly, the GeNorm program determines the gene-stability measure M as the average pairwise variation of one reference gene compared with all other reference genes analyzed. GeNorm ranks the reference genes accordingly; stepwise exclusion of genes that are not stably expressed results in a lower M value for the remaining genes. Relative expression profiles were generated using the comparative Ct method (ΔΔCt). The geometric mean was used when multiple reference genes were considered. Expression levels are relative to highest-expressing sample where the mean biological fold change and SEM were graphed for each gene of interest.

RESULTS

Quality Control of Samples and Choice of Reference Genes for qRTPCR Analysis

Initially, we were interested in analyzing the expression profiles of genes in fetal germ cells. Germ cells were purified from the somatic cells of the gonad by subjecting pooled, cleanly dissected male gonad tissue to FACS based on expression of an Oct4 (Pou5f1)-eGFP transgene, which is expressed only in germ cells (Fig. 1A) [29]. Purity of the isolated germ and somatic cell fractions was 99% according to subsequent assessment of each sorted fraction using FACS. In support of this analysis, qRTPCR revealed expression of the germ cell-specific markers Mouse Vasa Homologue (Mvh, also known as Ddx4) and Oct4 transcription factor (Oct4, also known as Pou5f1) in the germ cell fraction but negligible expression in the somatic cell fraction (100-fold and 80-fold lower expression, respectively, in somatic cells than in germ cells). In the reciprocal experiment, expression of the somatic cell marker Sry box gene 9 (Sox9) was detected in the somatic fraction but not in the germ cell fraction (120-fold lower expression in germ cells than in somatic cells; Fig. 1B). Finally, immunoblotting using an antibody specific for the germ cell marker MVH (DDX4) revealed strong expression in protein samples from the germ cell fraction but no expression in the somatic fraction (Fig. 1C and Western et al. [8]). These experiments confirmed that the FACS-isolated cell populations were of very high purity.

FIG. 1.

Analysis of purified germ and somatic cell fractions from developing E13.5 male gonads. A) Fluorescence-activated cell sorting, based on expression of an Oct4 (Pou5f1)-eGFP transgene, was used to isolate germ cell and somatic cell fractions from male and female fetal gonads. B) Purity of the sorted cell fractions was assessed using qRTPCR by analyzing expression of the germ cell-specific markers Mvh (Ddx4) and Oct4 (Pou5f1) and the somatic cell-specific marker Sox9. Y-axis represents expression levels relative to the lowest-expressing sample, set at 1.0. Expression levels were normalized to Mapk1/Hmbs. C) Immunoblotting of protein samples from E12.5–E15.5 male germ cell and somatic cell fractions using an antibody specific to the germ cell-specific marker MVH (DDX4). ACTIN provides a loading control. (From Western et al. [8]; reprinted with permission from AlphaMed Press Inc.)

All RNA samples used in this study had an RIN score greater than 9.0, and therefore they were high-quality RNA samples [30]. We then used qRTPCR to assess expression of 16 commonly used reference genes, including hexose-6-phosphate dehydrogenase (H6pd), aminolevulinic acid synthase 1 (Alas1), tyrosine 3-monooxygenase/tryptophan 5-monooxygenase activation protein, zeta polypeptide (Ywhaz), eukaryotic translation elongation factor (Eef1e1), transferrin receptor (Tfrc), TATA box-binding protein (Tbp), hypoxanthine guanine phosphoribosyl transferase 1 (Hprt), hydroxymethylbilane synthase (Hmbs), mitogen-activated protein kinase 1 (Mapk1) and Mapk7, ATP synthase, H+ transporting mitochondrial complex, beta subunit (Atp5b), calnexin (Canx), cytochrome c-1 (Cyc1), Succinate dehydrogenase complex, subunit A, flavoprotein (Sdha), topoisomerase (DNA) I (Top1), and polymerase (RNA) II (DNA-directed) polypeptide A (Polr2a) in germ and somatic cell fractions purified from E12.5, E13.5, E14.5, and E15.5 (E12.5–E15.5) male and female gonads (Supplemental Table S2). Care was taken to avoid reference genes that are involved in similar functional pathways because these genes could be coregulated. In addition, each assay was designed to specifically amplify the transcript of interest. To this end, the sequences included for each assay were compared to the mouse genome using the Blat search engine (University of California Santa Cruz Genome Browser; http://genome.ucsc.edu/), and amplicons with homology to other genomic sequences were avoided. This ensured that pseudogenes were not included in the assays. Other commonly used reference genes, such as glyceraldehyde-3-phosphate dehydrogenase (Gapdh) and actin beta (Actb), were considered but excluded because of the presence of pseudogenes and because these genes have been found to be unstable in other systems [22]. Phosphoglycerate kinase 1 (Pgk1) was excluded because suitably specific primer/probe sets could not be found. Several other genes were also considered but were not tested, because expression microarray evidence suggested they were either regulated or not significantly expressed in fetal gonads (Supplemental Table S3). Finally, assays more than 1.2 kb upstream of the 3′ end of the transcript were avoided (for example, for Eef2) because the samples to be analyzed were derived from amplified RNA samples. All qRTPCR assays were performed in biological triplicate for each time point assayed to ensure reproducibility of relative expression levels.

Expression Levels of Candidate Reference Genes in Fetal Gonads

The raw expression data (Ct values) of the 16 reference genes initially tested showed that message RNAs for most of these genes were reasonably abundant, with Ct values between 20 and 26 in all of the samples and biological replicates tested (E12.5–E15.5 male and female, germ and somatic cell fractions; Fig. 2). One exception to this was H6pd, which was detected with Ct values of 30–37, indicating it was expressed at very low to insignificant levels. As expected, some reference genes (e.g., Alas1, Tfrc, and H6pd) showed similar variation in both somatic and germ cell fractions. Interestingly, most reference genes appeared to show slightly more variation in the purified germ cell samples than in their age-matched somatic cell counterparts (Fig. 2). This may reflect the divergent developmental pathways (mitotic arrest and meiosis) of the male and female germ cells.

FIG. 2.

Box plot representations of qRTPCR Ct values for 16 reference genes assayed in male and female E12.5–E15.5 purified germ cells (A) and male and female E12.5–E15.5 purified somatic cells (B). Each box plot is based on the biological triplicate Ct mean value for each of the male and female time points analyzed. Boxes represent the lower and upper quartile ranges, medians are represented by black dashes within boxes, and whiskers indicate the upper and lower data value ranges for the time points and samples tested.

Suitable Reference Genes for Normalizing Expression in Fetal Germ Cells

Mean Ct values were calculated on the basis of three biological replicates for each sample. These values were converted to relative quantities using the comparative Ct method, and the highest relative quantity for each of the genes studied was set to 1. These values were then submitted to the GeNorm Excel VBA applet for further analysis [28]. The GeNorm program determines the gene-stability measure (M) as the average pairwise variation of one reference gene compared with all other reference genes analysed. GeNorm ranks the reference genes accordingly and uses a stepwise exclusion process to eliminate genes, such as H6pd, that are not expressed stably. This results in a lower M value for the remaining genes. Following this process, GeNorm ultimately ranked Mapk1 and Sdha (M = 0.4) as the two most transcriptionally stable genes in germ cells isolated from E12.5 to E15.5 of both male and female gonads (Fig. 3A and Table 1), although stability of most of the reference genes in these samples was relatively high (M range, 0.4–1.1).

FIG. 3.

GeNorm average expression stability values for 16 reference genes tested in male and female E12.5–E15.5 purified germ cell fractions (A), male and female E12.5–E15.5 purified germ and somatic cells (B), and male and female E12.5–E15.5 purified somatic cells (C). Higher M values represent genes that have low transcriptional stability, whereas low M values represent genes with high transcriptional stability.

TABLE 1.

Reference genes ranked according to stability.

Separate GeNorm analyses were also performed to determine the most stably expressed reference genes when considering only male or female germ cell populations through fetal development. This revealed Sdha/Ywhaz (M = 0.13) and Tbp/Tfrc (M = 0.15) as the most stable reference gene pair in male and female germ cells, respectively (data not shown).

Suitable Reference Genes for Normalizing Expression in Fetal Gonadal Somatic and Germ Cells

Analysis of gonad development during sex determination often involves distinguishing whether gene expression is confined to the germ or somatic fraction of the gonad and/or whether gene expression is sex specific. Therefore, we next endeavored to identify similar stably expressed genes in germ and somatic cells isolated from E12.5–E15.5 in male and female gonads. The same set of reference genes were subject to analysis, which revealed that Mapk1 and Canx exhibited the highest average stability (M = 0.38); however, the average stability for most reference genes in the combined germ and somatic sample set was lower than those observed in germ or somatic samples alone (M range, 0.38–1.2; Fig. 3B and Table 1). Finally, analysis of the somatic cell samples (separate from the germ cells) showed that Canx and Top1 were the most stable genes in both sexes (M = 0.052) although, like the germ cell set, the stability of all of the reference genes in these samples was relatively high (M range, 0.052–0.63, Fig. 3C and Table 1).

Application of Suitable Reference Genes Provides More Accurate Expression Profiles

Previous qRTPCR studies of fetal germ cell development have used the germ cell-specific marker Mvh (Ddx4) to normalize gene expression levels. Mvh (Ddx4) was chosen for this purpose because it is expressed exclusively by germ cells and was thought to be expressed relatively constantly through this developmental period. It was thought that the germ cell specificity would serve to control for changing relative germ cell numbers within the developing gonad [31]. However, because of the lack of suitable reference genes, the expression stability of Mvh (Ddx4) in E12.5–E15.5 male and female germ cells had never been verified systematically. Because we had purified germ cell samples and identified appropriate reference genes, it was possible to determine whether Mvh (Ddx4) expression was maintained at a stable level in E12.5–E15.5 germ cells. We therefore examined expression of Mvh (Ddx4) in sorted germ cell samples and normalized its expression to the most stable reference genes, Mapk1 and Sdha, identified in our GeNorm analysis. This analysis revealed that Mvh (Ddx4) expression increased approximately 2.5-fold in both male and female germ cells between E12.5 and E15.5 (Fig. 4A). In addition, we compared the GeNorm-predicted expression stability of Mvh (Ddx4) to the other 16 reference genes in the purified germ cell samples. In this case, Mvh (Ddx4) was excluded as the least stably expressed gene, as evident by the clear decrease of M for the remaining reference genes (M including Mvh [Ddx4] = 1.35, and M excluding Mvh [Ddx4] = 0.4; Fig. 4B).

FIG. 4.

A) Expression profile of Mvh (Ddx4) in germ cells isolated from E12.5–E15.5 male and female gonads, which was normalized using reference genes Sdha and Mapk1. Error bars represent SEM of biological triplicate values. Y-axis represents expression levels presented relative to the highest-expressing sample, set at 1.0. B) GeNorm average expression stability of reference genes, including Mvh (Ddx4), in E12.5–E15.5 male and female purified germ cells. Higher M values represent genes that are least transcriptionally stable, whereas low M values represent genes with high transcriptional stability.

Previously, we examined expression of the pluripotency marker Oct4 (Pou5f1) in E12.5–E15.5 male and female gonads and normalized expression to Mvh (Ddx4). To determine how the changing expression pattern of Mvh (Ddx4) affected this quantification, we next analyzed the expression profile of Oct4 (Pou5f1) in purified germ cells isolated from E12.5–E15.5 male and female gonads using various reference gene strategies. Oct4 (Pou5f1) expression data from the sorted cell populations was normalized using Mvh (Ddx4) or a common housekeeping gene (Hprt), or else a single verified reference gene (Mapk1), two verified reference genes (Mapk1 and Sdha), or three verified reference genes (Mapk1, Sdha, and Canx), as recommended by GeNorm (Fig. 5).

FIG. 5.

Oct4 (Pou5f1) and Mvh (Ddx4) expression profiles in germ cells isolated from E12.5–E15.5 male and female gonads using various normalization approaches. A and B) Oct4 (Pou5f1) expression was normalized against (A) Mvh (Ddx4) and (B) Mapk1, Sdha, and Canx (i), Mapk1 and Sdha (ii), and Mapk1 (iii). C) Oct4 (Pou5f1) (i) and Mvh (Ddx4) (ii) expression normalized against Hprt. D) Hprt normalized against Mapk1 and Sdha in germ cells isolated from E12.5–E15.5 male and female gonads. The geometric mean was used to normalize data when multiple reference genes were used. Y-axes represent expression levels presented relative to the highest-expressing sample, set at 1.0. Error bars represent SEM of biological triplicate values.

Normalization using Mvh (Ddx4) indicated that Oct4 (Pou5f1) was initially expressed at similar levels in male compared with female germ cells. Oct4 (Pou5f1) expression in females then decreased dramatically between E12.5 and E14.5 to less than 10% of its initial E12.5 level. This included a 60% reduction in transcript levels between E12.5 and E13.5. In males, a 60% reduction in Oct4 (Pou5f1) levels was observed between E12.5 and E14.5 in germ cells, and this level was maintained at E15.5 (Fig. 5A).

We next normalized Oct4 (Pou5f1) expression levels using the verified reference genes (one, two, or three genes). In general, the Oct4 (Pou5f1) expression trends observed when using one, two, or three verified reference genes were quite similar (Fig. 5B). Oct4 (Pou5f1) expression in females showed strong downregulation (more than 75%) between E12.5 and E14.5 for all reference genes tested. In males, expression levels of Oct4 (Pou5f1) were similar in E12.5–E15.5 germ cells. Thus, in male germ cells, the expression pattern assessed using the verified reference genes showed that Oct4 (Pou5f1) was maintained at relatively constant levels between E12.5 and E15.5, whereas the pattern observed using the germ cell marker Mvh (Ddx4) indicted that an approximate 60% decrease in Oct4 (Pou5f1) expression occurred over the same period.

We examined the germ cell expression patterns of Oct4 (Pou5f1) and Mvh (Ddx4) with reference to the commonly used housekeeping gene Hprt. Interestingly, the expression profiles obtained in E12.5–E15.5 male germ cells were similar to the GeNorm-verified reference genes for both Mvh (Ddx4) and Oct4 (Pou5f1). However, a large discrepancy was observed in the male versus female expression patterns for Oct4 (Pou5f1) and Mvh (Ddx4) when normalized with Hprt compared with the expression patterns obtained using the GeNorm reference genes. This resulted in apparently much higher levels of Mvh (Ddx4) and Oct4 (Pou5f1) in male germ cells than in female germ cells when Hprt was used, but relatively similar levels when the GeNorm reference genes were used (compare Fig. 4A with Fig. 5Cii, and Fig. 5A with Fig. 5Ci, respectively). When Hprt expression was normalized against the GeNorm-verified reference genes Mapk1 and Sdha, its levels were consistently higher in female germ cells than in male cells (Fig. 5D).

Finally, the somatic cell fraction was also normalized using the verified reference genes (in order of stability: Canx, Mapk1, and Sdha) or Hprt (Fig. 6). In general, the profiles of Amh and Sox9 were similar for two or three verified reference genes but showed minimal variation with only one (Canx) or an unverified common reference gene (Hprt).

FIG. 6.

Amh and Sox9 expression profiles in somatic cells isolated from E12.5–E15.5 male and female gonads using various normalization approaches. Canx, Mapk1, and Sdha (i), Canx and Mapk1 (ii), Canx (iii), and Hprt (iv). Error bars represent SEM of biological triplicate values. The geometric mean was used to normalize data when multiple reference gene were used. Y-axes represent expression levels presented relative to the highest-expressing sample, set at 1.0.

DISCUSSION

Gene expression analysis using qRTPCR combined with data normalization against internal reference genes has become a popular way to quantify transcript levels. A number of methods, such as GeNorm, allow identification of suitable reference genes [21, 24, 25]. Since these tools have become available, many papers have highlighted the need to identify suitable reference genes in various biological systems, whether they compare treated versus untreated samples or examine developmentally regulated gene expression [32–35]. However, suitable reference genes that are stably expressed in various cell types of the developing male and female gonads have not yet been identified.

To accurately quantify gene expression during germ and somatic cell development in E12.5–E15.5 male and female fetal gonads, we have systematically examined candidate reference genes for expression stability in these tissues. After testing 16 candidate genes, we identified the most suitable reference genes for this time in gonadal development. Analyses comparing the GeNorm-verifed reference genes with previously used housekeeping genes, such as Mvh (Ddx4) and Hprt, revealed substantially different expression profiles. Significantly, these differences were not substantiated when the samples were normalized to GeNorm-verified reference genes. This analysis shows that the choice of reference gene is very important for obtaining accurate qRTPCR normalization.

Previously, Mvh (Ddx4) has been used as a germ cell-specific marker to normalize transcript levels of germ cell-expressed genes from whole-gonad samples [31]. This was based on an assumption that male and female germ cells stably express Mvh (Ddx4) through the developmental stages studied (E12.5–E17.5). However, because of the heterogenous population of cells and differences in cell proliferation rates in the developing gonads, it was not possible to verify the stability of Mvh (Ddx4) expression through the developmental stages studied (E12.5–E17.5). This problem was overcome in the current study by purifying germ cells from the rest of the gonadal somatic tissue using FACS.

Having successfully identified suitable reference genes in male and female germ cells, for the first time we were able to generate accurate profiles of germ cell specific genes, such as Mvh (Ddx4) and Oct4 (Pou5f1). After normalization to suitable reference genes, we have shown that the expression of Mvh (Ddx4) is not stable through these developmental stages. Mvh (Ddx4) was upregulated approximately 2.5-fold in both male and female germ cells between E12.5 and E15.5.

Sdha and Ywhaz proved to be most stable in male germ cells, whereas Polr2a and Tfrc proved to be most stable in female germ cells when the sexes were considered independently (data not shown). The difference in expression levels of the various reference genes between male and females was not unexpected, because from E12.5 the developmental pathways of male and female germ cells diverge.

The gonadal somatic fractions from which germ cells had been purified were included in a second analysis in order to provide an internal negative control for genes that are exclusively expressed in germ cells of the gonad and to provide data on sex specificity of gene expression in the somatic compartment of the gonad. GeNorm identified Mapk1 and Canx as the most suitable reference genes for normalizing expression in both germ and somatic cells isolated from E12.5–E15.5 male and female gonads (Table 1). However, consideration of only the male and female somatic cells identified Canx and Top1 as the most stably expressed reference genes in this mixed population of gonadal cells.

An interesting difference was observed when the common housekeeping gene Hprt was used to normalize Sox9 and Amh expressions in somatic cells or Mvh (Ddx4) in germ cells. In somatic cells, the Sox9 and Amh profiles were similar when normalized using Hprt or verified reference genes (Fig. 6). However, expression of Mvh (Ddx4) was significantly reduced in female germ cells when normalized with Hprt (Fig. 5Cii) compared with the GeNorm-verified reference genes (Fig. 4A). The reason for this variability in the accuracy of Hprt as a normalization gene is evident in its increased Ct range (i.e., increased expression instability) in germ cells compared with somatic cells (Fig. 2) and its expression profile in germ cells (Fig. 5D). This once again shows that expression of commonly used housekeeping genes can vary and emphasizes the importance of identifying stable reference genes in the system to be analyzed.

We have identified reference genes suitable for the quantification of gene expression in midgestation male and female fetal germ cells using qRTPCR. Other reference genes are also appropriate for comparison of gene expression levels in the germ cells and the somatic cells, with the caveat that in the analysis of somatic gene expression examined here, a mixed gonadal cell population was used. Identification of suitable reference genes in other gonadal cell types, such as developing supporting cells, using different tissue-specific transgenes (e.g., Sf1-eGFP) would further enhance the power to accurately quantify genes important in embryonic gonad development.