For the 14 genes used in this study, the sequences were referred from the databases (Table 1), and primers were designed from these sequences. After optimizing PCR conditions and amplification, the products were sequenced for confirmation. Based on the sequence analysis, all primers amplified the expected amplicon sizes (Table 2).

To amplify beta actin (Actb) gene, we designed primers from 3 different sequences (AY598932, AF404278 and X60733), optimized and compared them. However, the current primer (design from X60733 sequence) gave the best amplification results. It has also the best beta actin similarity hits with orthologs. Although the source sequence was identified as gamma non-muscle actin in the NCBI, and as actin, alpha skeletal in the Ensembl databanks, we assumed this to be a wrong labelling. The feedback we got from the NCBI staff also supports our view. Therefore we used it to design primer and further comparison.

Quality of the analyzed samples plays a major role in the correct interpretation of the results. Based on the morphological observations, in both in vivo derived and in vitro produced embryo samples, presumably good quality oocytes and preimplantation stage embryos were collected. In the subsequent procedures, in addition to our established quality RNA isolation procedures, the minus RT (reverse transcriptase) reaction during cDNA synthesis, and the design of most primers at the exon-exon junction enabled us to control the absence of contaminating genomic DNA. During the initial screening assays, similar cDNA dilutions from the pooled embryos were used and reaction conditions were optimised for each primer pair separately. Optimum primer concentrations were selected based on the absence of dimer and signal detection recognized by earlier C_T values. The C_T is defined as the number of cycles needed for the fluorescence to reach a specific threshold level of detection and is inversely correlated with the amount of template nucleic acid present in the reaction 35. The majority of the selected candidate reference genes were detected in most preimplantation stages, but with various signal intensities. Using a similar low concentration template for all, the ten references (Eef1e1, Polr2a, Ywhaz, Ppia, H2afz, Hprt1, Pgk1, G6pdx, Gapdh, and Actb) and POU5f1 genes were detected at C_T values below 33. The other three genes (B2m, Ubc, and Tbp) were detected at C_T values above 33 with irregularities in amplification (at some of the stages where the transcripts of that particular gene were at lower concentration). As embryo materials are scarce at the preimplantation stages, we excluded the three genes detected with the latest C_T values. Then, the ten candidate references and POU5F1 genes with the earlier signals (earlier C_T values) were selected for further comparisons.

For all the selected candidate genes, melt curve analyses were performed at the end of PCR reactions. The specificity and integrity of the PCR products were confirmed by the presence of a single peak (See additional files 1, 2 and 3 as representatives). For the selected genes, the standard curves were deduced from four-fold serial dilutions of the five-pooled embryo cDNA preparations measured at five points. To ensure the comparability of PCR assays, three serial dilutions were made independently that enabled us to determine the C_T values and PCR efficiencies of the individual assay and calculate the correlation between them. The assays for the selected candidate reference genes were characterized by a linear correlation coefficient (R²) that varied from 0.963 to 0.997 (average 0.986) and PCR efficiencies between 92.3% and 110.7% (average 97.5%), where as the assay for POU5F1 gene has 82.4% efficiency and 99.8% correlation (Table 3). Based on these results, the assays can be trusted and valid for the quantification of transcripts and further comparisons.

In order to select the best stable genes for normalization, expressions of the selected ten candidate reference genes were compared in different preimplantation stage embryos. For gene transcript quantification, five individual embryo cDNA preparations per stage were used in identical experimental procedures but optimised conditions for each gene amplification.

The ten candidate genes were detected in all developmental stages examined with various signal intensities. The oocyte stage was taken as a calibrator for all the genes and the relative expression levels at different developmental stages are shown in Figures 1 and 2.

Compared to the oocyte stage, the relative levels of most genes (Eef1e1, Ppia, H2afz, Hprt, Pgk1 and G6pdx) reduced at the 1-cell stage, while the relative levels of the remaining genes (Polr2a, Ywhaz, Gapdh, and Actb) increased in various intensities at this stage. The relative levels of some genes (Eef1e1, Polr2a, Ppia, Gapdh, Actb (in vitro) were transiently reduced to the minimum levels at the 8-cell stages, and increased thereafter. Despite differences in the embryo type (in vivo or in vitro), the above expression patterns remained similar (Figure 1 and Figure 2). In both in vivo derived and in vitro produced embryos, the relative levels of H2afz and Hprt1 genes showed a persistent increase in levels after fertilization. The gene Ywhaz (except 2-cell in vivo) also has the same profile.

To examine the effects of embryo type (in vivo and in vitro) on gene expression stability, expression profiles of the selected candidate genes were compared in the in vivo derived and in vitro produced embryo samples.

Major pattern differences were not observed due to change in the embryo type (in vivo or in vitro) (Figures 1 and 2). However, the stage-by-stage comparisons revealed differential transcript levels between the two embryo types. At the 2-cell stage, in vivo produced embryos showed higher transcript levels for majority of the genes (except Pgk1 and Actb) compared to their in vitro counterparts, and differences were significant (p < 0.05) for some genes (Eef1e1, Ywhaz, Ppia and G6pdx). However, at the 8-cell stage, the in vitro samples showed higher transcript levels for majority of the genes, with significant difference (p < 0.05) for some (Eef1e1, Ywhaz, Gapdh and Actb). At the morula and blastocyst stages, the relative transcript levels of in vivo derived embryo samples were higher for most genes compared to the in vitro samples. These differences were significant (p < 0.01 for Ywhaz and Ppia, and p < 0.05 for H2afz, Gapdh and Actb) at the morula stage and (p < 0.05 for Polr2a and Hprt1) at the blastocyst stage (Figure 2).

The profiles of candidate genes were analysed and the expression stability measure values (M) were calculated using the geNorm software 36. Following the procedures of the software, the least stable genes were determined (by higher M values) and continuously excluded to recalculate the M values for the rest. Finally all genes were ranked based on the calculated M values and the three most stable genes (with the lowest M values) were selected. Accordingly, for the in vivo derived samples, the genes H2afz and Hprt1 were found to be the most stable, followed by the genes Ywhaz, Actb, Ppia, G6pdx, Gapdh, Eef1e1, Polr2a and Pgk1 in their order of appearance. Similar analyses for the in vitro samples selected the genes H2afz and Ywhaz as the most stable genes, followed by Hprt1, G6pdx, Ppia, Gapdh, Eef1e1, Polr2a, Actb and Pgk1 in their order of appearance. Generally, differences in the embryo type (in vivo or in vitro) had a minor rearrangement effect on the stability order for majority of the genes examined. The gene Actb showed wider stability range between the embryo types with better stability in the in vivo samples compared to the in vitro (Figure 3).

To evaluate the performances of the newly selected reference genes, the normalization factor (calculated from the geometric averages of the three selected genes) was calculated. The expression of POU5F1 gene was quantified in the in vitro produced rabbit oocytes and embryo samples and the results were normalized by the newly calculated factor. The expression at the oocyte stage was used as a calibrator to calculate the relative expression levels in the different developmental stages (Figure 4). Based on analyses, the POU5F1 gene was expressed in all developmental stages with various signal intensities. Comparatively higher expression levels were observed at the oocyte and zygote stages, and declined gradually to the lowest levels at the 8-cell stage. However, starting from the morula stage, the levels increased continuously to the blastocyst stage.

Hycole hybrid female rabbits were induced to superovulate by administration of 120 IU pregnant mare serum gonadotropin (PMSG, Folligon^® Intervet, The Netherlands) i.m. and 72 hours later, these animals were injected i.v. with 170 IU human chorionic gonadotropin (hCG, Choragon^® Richter Gedeon Rt., Hungary). Donor females (does) were slaughtered at 16 hours post hCG administration and the oocytes were flushed from the ampullae of the oviducts with M2 medium. The cumulus layer was then removed using 0.1% hyaluronidase in M2 medium, and the denuded metaphase 2 stage oocytes (confirmed by the presence of a single polar body) were collected individually for RNA isolation.

Hycole hybrid female rabbits were induced to superovulate by administration of PMSG, and 72 hrs later, and shortly before mating, with hCG hormones as described earlier. Each injected female was mated with a male of proven fertility (buck) from the same breed. Donor females were slaughtered at 20 hours post hCG administration and the zygotes were flushed from the oviduct with M2 medium. The flushed zygotes were evaluated morphologically under the microscope and those with two pronuclei, two polar bodies and compact cytoplasm were selected for the experiment.

The in vitro embryos were produced by further culturing the zygotes until the required developmental stages. For this, the zygotes were cultured in 50-μl EBSS (Earle's Balanced Salt Solution) complete drops 56 under mineral oil and incubated at 38.5°C in humidified atmosphere containing 5% CO₂ in air. The embryos were cultured until the required developmental stages, and five embryos each at the 2-cell (26 hrs), 8-cell early (44 hrs), 8-cell late (54 hrs), morula (68 hrs), and blastocyst (103 hrs) stages were collected individually.

The procedures of superovulation and mating were as described earlier. For collection of in vivo derived embryos, donor females were slaughtered at respective times (hours post hCG administration) and the embryos at the 2-cell (26 hrs), 8-cell early (41 hrs), 8-cell late (47 hrs), morula (62 hrs), and blastocyst (98 hrs) stages were collected individually.

During sample collection, the denuded oocytes and different developmental stage embryos were washed three times with and collected individually in 2-μl RNase-free water for storage at -80°C until RNA extraction.

The procedures of RNA isolation and cDNA synthesis were as described in detail earlier 15. Briefly, messenger RNA was extracted individually from five embryos per developmental stage and embryo type (in vivo or in vitro), using Dynabeads^® mRNA DIRECT™ Micro Kit (Dynal A.S, Oslo, Norway), following the manufacturer's instructions. The individually frozen embryos were lysed and incubated with pre-washed magnetic Dynabeads that can base pair with poly (A) tails of mRNA molecules. After hybridisation and subsequent repeated washes with buffers, the RNA was eluted in RNase-free water and reverse transcribed into cDNA, using M-MLV RT kit (Invitrogen, Carlsbad, USA) in a final 20-μl reaction volume. Minus RT reactions were performed to check the absence of contaminating residual DNA. The cDNA synthesis reactions were carried at 42°C for 1 hour, followed by heat inactivation of the enzyme at 75°C for 10 minutes.

A total of 14 genes, most commonly used and considered as reference for normalization by the earlier studies, were selected for evaluation throughout the different embryo developmental stages and embryo types (in vivo and in vitro). Moreover, POUF1 gene was also used for evaluating the selected reference genes. Primers were designed and optimised prior to initial screening and quantitation experiments. Comparison and screening of primers were carried out using optimised protocol for each primer. Template cDNA from the same source was used, and finally those with earlier signals were selected for further quantification. For these genes (Table 1), the expected sizes of the products were confirmed by gel electrophoresis on a 2% agarose gel. The PCR products were also cloned (TOPO TA cloning kit, Invitrogen), and sequenced for confirmation.

The details of real time PCR conditions for each gene were as described in Table 2. During quantification of the transcripts, the assay for each gene consisted of five replicates per stage, both in vivo derived and in vitro produced oocytes and preimplantation stage embryo samples, negative and positive controls. All genes were compared from the same stock to avoid inter-assay template variations, and all quantifications were performed consecutively without interruption. Each sample in a run consisted of 0.08 embryo equivalent cDNA template, 200–500 nM of each primer (Table 2), and 50% SYBR^® Green JumpStart™ Taq ReadyMix™ in 15-μl reaction volume. The reaction conditions were template denaturation and polymerase activation at 95°C for 2 min followed by 45 cycles of 95°C denaturation for 15 sec, 57°C to 68°C annealing and extension for 45 sec. All reactions were carried out using the Rotor-Gene™ 3000 real time PCR machine (Corbett Research, Mortlake, Australia), and the results were analysed with the integrated Rotor-Gene software (version 6.1). At the end of PCR reactions, melt curve analyses were performed for all the genes. For calculating PCR efficiencies, standard curves were generated from assays made with four fold serial dilutions of 5-pooled blastocyst cDNA preparations. To ensure the compatibility of PCR assays, three independent serial dilutions were made that enabled us to determine the C_T values and PCR efficiencies of the individual assay and calculate the correlation between them. PCR efficiencies (E) were calculated with the equation

E = (10 [-1/slope]-1) × 100.

Analyses of the gene expression stability over the different embryonic stages and types (in vivo vs. in vitro) were performed using the geNorm software 36. The analyses relies on the principle that the expression ratio of two ideal internal control genes is identical in all samples, regardless of the experimental condition or cell type, and determined as the standard deviation of the logarithmically transformed expression ratios 36. The internal control gene stability measure value (M) was calculated as the average pair-wise variation of a particular gene with respect to the rest of the genes, and ranking was made based on these values. The lower the M value, the more stable the expression of the gene under consideration. The most stable reference genes were identified by stepwise exclusions of the least stable gene and recalculating the M values with the rest genes.