The low success rate of in vitro embryo production procedures in cattle, together with early embryonic mortality, lead to the loss of a large number of potential calves, retarded genetic progress and consequently the loss of money and time for the cattle breeding industry. Despite improvements on the culture media and culture conditions only 30–40% of the fertilised oocytes reach the blastocyst stage and it is generally accepted that the quality and developmental competence of in vitro produced (IVP) bovine embryos has failed to keep up with those of their in vivo counterparts 1.

Normal preimplantation embryo development in cattle is characterised by several cleavage divisions of the fertilised egg, activation of the embryonic genome around the 8–16 cell stage, compaction and blastocoel formation leading to the blastocyst. Before the major activation of the embryonic genome, the bovine preimplantation embryo is controlled by maternal genomic information that is accumulated during oogenesis 2. Both the maternal and embryonic gene expression occurs in a stage- and time-dependent manner 3.

The first important differentiation events take place at the blastocyst stage, resulting in the generation of two distinct cell lineages: the trophectoderm cells (TE) and the inner cell mass (ICM). The ICM gives rise to the embryo, whereas the TE forms the placenta. Since these two lineages are divergent in morphological and biochemical aspects, it is reasonable to hypothesise that many differentiation-related genes are expressed at this stage. Genes being expressed in the blastocyst stage but not in earlier stages are therefore functional candidates for the regulative processes that take place at the onset of differentiation 4. The identification of novel genes involved in blastocyst formation and the analysis of their expression patterns may help us to understand the mechanisms that control blastocyst formation and may help us to understand why certain embryos do not make it through this stage. As in vitro culture conditions do not fully mimic the in vivo situation and as it has been shown that morphological and physiological differences 567 as well as differences in gene expression 891011 exist between in vivo and in vitro cultured embryos, the RNA expression levels of in vitro produced embryos should be compared with those of in vivo "golden standard" embryos. Changes in transcript abundance between in vitro and in vivo embryos may help to assess the normality of in vitro produced embryos and may help to optimise the in vitro culture conditions.

In previous studies, genes involved in preimplantation embryo development were identified using techniques such as differential display reverse transcription PCR 41213, large scale cDNA library construction 141516, cDNA microarray 171819 and suppression subtractive hybridisation (SSH) 4202122. Subtractive cDNA cloning 23 is a powerful tool for the detection of stage-specific transcribed genes. In contrast to e.g. microarray analysis, this PCR-based method can be accomplished without prior knowledge of the genes being expressed, and yields subtracted cDNA pools that are differentially expressed. However, further confirmation of the outcome is required by means of real-time PCR, which is a perfect technique not only to validate the subtractive cDNA cloning results, but also to gather quantitative data concerning the selected transcripts. In the present study subtractive cDNA cloning was used to detect genes that are differentially expressed in the bovine blastocyst compared to genes present in 2-cell and 8-cell stage embryos. The differential RNA expression status of 12 subtracted cDNA clones was validated throughout 3 stages of preimplantation embryo development (2-cell, 8-cell and blastocyst) and compared with the RNA expression levels of their in vivo "golden standard" counterparts using real-time PCR. By immunofluorescent labelling, the protein expression of 3 genes with high differences in RNA expression levels between the developmental stages was examined during preimplantation embryo development to check whether the protein expression patterns were comparable with the RNA expression levels.

In the present study a subtractive cDNA library was constructed between blastocyst embryos as the tester population and a pool of 2-cell and 8-cell embryos as the driver population, in order to enrich for genes differentially expressed at the blastocyst stage.

Sequence information was obtained for 65 randomly picked clones. Out of these 65 ESTs representing the blastocyst embryos, 10 were homologous to mitochondrial genes and 14 to ribosomal genes. Mitochondria play an essential role in many events during early development. In cattle, the mtDNA copy number increases during blastocyst expansion and hatching, consistent with an increase in mitochondrial RNAs. This indicates that the transcriptional activation of the mitochondrial genome coincides with the pronounced structural and functional differentiation of the mitochondria during preimplantation development 26. The largest group of ESTs was identified as genes involved in protein synthesis and included several ribosomal proteins. The high degree of ribosomal protein mRNAs corresponds with previous data showing that between the 2-cell stage and the blastocyst stage, the level of ribosomal protein mRNA is said to increase about 20-fold 27. The protein content of in vivo derived preattachment cattle embryos increases 2-fold from the 2-cell through to the blastocyst stage 28. This rise in protein content is correlated with rapid cell divisions at the late morula and early blastocyst stage 2930.

Real-time qPCR was used to verify the SSH results and at the same time to compare the gene expression between in vitro produced embryos and in vivo golden standard embryos. The geometric mean of GAPD, YWHAZ and SDHA was used for normalisation. Vandesompele and colleagues 31 demonstrated that the conventional use of a single gene for normalisation leads to relative large errors and validated the geometric mean of multiple carefully selected housekeeping genes as an accurate normalisation factor. In Goossens and colleagues 25 it was demonstrated that GAPD, YWHAZ and SDHA were stably expressed reference genes, suitable for normalisation of quantitative data within different preimplantation embryo stages. The problem of measuring RNA expression levels throughout preimplantation development is confounded by the fact that cell numbers and cell sizes are constantly changing during this developmental interval. To allow ontogenic analysis, the embryos were compared as a single unit and the reference genes will correct for the differences between the embryos. Other normalisation strategies, like normalisation against total cell number, normalisation against RNA mass quantity or the use of exogenous control RNA (spike) 3233 are not assumption free. Spikes correct for differences in RNA extraction and differences in enzymatic efficiencies but do not account for the quality and quantity of input sample.

In most studies about gene expression in embryos, only the RNA expression was considered, but the differences in RNA expression do not always imply differences on protein level, as there may be regulation at the translation level as well. By immunofluorescent labelling using mouse antibodies, the protein expression of KRT18, FN1 and MYL6 was evaluated for the first time during bovine preimplantation embryo development. Furthermore, the location of the protein expression in the embryo can tell us more about the specific function of the protein during the process of blastocyst formation.

Both KRT18 mRNA and protein expression were found to be absent at the 2-cell and 8-cell stage and abundant at relatively high levels at the blastocyst stage, demonstrating the embryonic origin of this transcript. KRT18 was only expressed in the cell-cell contact sites of the TE, but not in those of the ICM. This makes that KRT18 can be seen as a marker for differentiation, adding to previous described markers for trophoblast differentiation in mice as there are Cdx2 and Eomes 343536. As TE is a kind of specialized epithelium, genes involved in de novo differentiation of an epithelium, including gene families encoding for cell polarity, cell junctions, cytoskelet formation and ion transport 3738 are involved in TE differentiation. KRT18 is a cytoskeletal protein and is found, together with KRT8, in simple epithelia where they heterodimerise to form the intermediate filaments. They influence the 3-D formation of cell-cell or cell-substrate contacts in embryonic visceral endoderm 394041. Stanton and Green 42 reported in mouse embryos a progressive rise for both KRTs from the 2-cell to the blastocyst stage, paralleling the development of the TE. Targeted deletion of KRT18 in the mouse leads to trophoblast fragility and early embryonic lethality 43. Other cytoskeletal proteins such as α-catenin, β-catenin, occludin, zonula occludens protein-1 and connexin43 are also expressed at the cell-cell contact sides of the TE cells 44, suggesting a co-localisation with KRT18.

FN1 was found to be significantly higher expressed in blastocysts compared to the earlier stages, and this difference was even 3.5× higher in the in vivo produced embryos compared to the in vitro produced ones. FN1 protein was predominantly expressed in the ICM and formed filamentous structures between the TE and the ICM. The differential expression of FN1 in blastocyst stage embryos of several mammalian species, including bovine, has previously been reported by other authors 422454647 and this gene has been associated with blastocyst formation. It is an adhesive extracellular matrix component and performs a vital role during cell proliferation, cell adhesion and cell mobility. It exists as a homodimer and is composed of several domains such as a heparin binding domain, a fibrin binding domain, a collagen binding domain and a cell recognition domain. This provides FN1 with the opportunity to interact with and bind to several ligands such as cells, heparin, fibrin, collagen, immunoglobulins and DNA 48. Because events such as cell proliferation, cell adhesion and cell mobility are essential during early embryogenesis, FN1 does qualify as an important participant during this phase of embryo development. Besides, Aplin and colleagues 46 reported that FN1 acts as a bridging ligand between the collagen matrix and integrins at the cytotrophoblast surface, mediating anchorage and/or migratory activity in the process of implantation. The vital importance of FN1 for normal embryo development has been demonstrated convincingly since mouse embryos lacking this protein die shortly after gastrulation from mesodermal defects 49. The significant difference in FN1 mRNA expression levels which we noticed between in vivo and in vitro produced bovine embryos has also been reported by Mohan and colleagues 10. Lower RNA levels exhibited by in vitro produced embryos may be responsible for the poor quality of these embryos compared to their in vivo counterparts.

Myosin light chains associate with the motor protein myosin and are believed to play a role in the regulation of its actin-based ATPase activity. RNA levels of MYL6, the smooth muscle isoform of myosin light chain, were significantly higher at the blastocyst stage compared to the earlier stages. No significant differences were seen between in vivo and in vitro produced embryos. MYL6 is involved in the cytoskeletal organisation 50 which is in agreement with the observed protein expression in the TE cells surrounding the blastocoel cavity. The expression of MYL6 proteins around the blastocoel may also contribute to the statement of Niimura 51 who inferred that contractions in mouse blastocysts occur by activation of myosin light chain kinase resulting in the phosphorylation of myosin light chains and causing the efflux of blastocoelic fluid.

ATP1B3 shows a significant difference in RNA levels between the 2/8-cell stage and the blastocyst stage of in vitro produced embryos, but remarkably this difference is negligible for in vivo produced embryos. The Na/K ATPase consists of a catalytic α subunit and a noncatalytic, glycosylated β subunit, each encoded by multigene families 52. The β subunit is required for structural and functional maturation of the α subunit 53. Blastocyst formation involves the establishment of a transtrophectoderm ion gradient mediated by the Na/K ATPase which pumps water through water channels called aquaporins 28. Accumulation of fluid in the blastocoel is essential for differentiation of the ICM and TE cell types 54. The gene expression patterns of several Na/K ATPase subunits have been reported previously in mouse 55 and bovine 56, and Adjaye and colleagues 57 identified ATP1B3 as a marker specific for the TE in human blastocysts using a cDNA microarray but as far as we know, this is the first study in which the RNA expression levels of ATP1B3 in bovine embryos are described. The higher ATP1B3 RNA levels in in vitro produced blastocysts might be an explanation for the faster blastulation of in vitro produced embryos compared to in vivo produced ones. Van Soom and colleagues 58 have demonstrated that in vivo morulae display a more firm and prolonged compaction and that they start blastulation at a later embryonic age and cell number, moreover the addition of serum to the culture medium seemed to enhance blastocyst development 59. In vivo morulae develop more gradually to the blastocyst stage and this might go together with a slower rise in ATP1B3 expression. Another explanation for the higher ATP1B3 expression in vitro might be a compensation for the lower expression of another NA/K ATPase subunit or another functional analogue, but this was not further investigated.

The differences between the blastocysts and the 2- and 8-cell stage embryos for FTH1 were smaller but significant and no differences were observed between in vivo and in vitro produced embryos. FTH1 has an important role in the control of intracellular iron distribution and the constitution of long term iron stores. The ferroxidase activity associated with the H subunit is necessary for iron uptake by the ferritin molecule. FTH1-/- mice embryos die between 3.5 and 9.5 days of development 60 demonstrating that the ferritin H subunit is indispensable for embryonic development. FTH1 was also present in the human blastocyst SSH library as reported by Morozov et al. 61.

The mRNA levels for ATP6V0B, HINT1, RPL10 and SLC25A5 were higher at the blastocyst stage compared to the earlier stages for in vitro produced embryos, but no significant differences existed between the in vivo produced 8-cell and in vivo produced blastocyst embryos. The qPCR results of those genes for in vitro produced embryos confirmed the results of the subtractive cDNA library.

The RNA expression patterns of the remaining genes (ACTN1, COPE and EEF1A1) were not in agreement with the subtractive cDNA cloning results. Those genes were abundant at an equal amount or higher at the earlier stages. Those false positives prove the requirement for the verification of the subtractive cDNA outcome and are in agreement with the reported 30% false positives obtained by subtractive cDNA cloning in previous studies 10. The significantly higher RNA levels in vivo produced 8-cell embryos for COPE but also for EEF1A1 is an important observation. COPE is a subunit of the coatomer protein complex and is involved in the retrograde Golgi-to-ER transport of dilysine-tagged proteins, whereas EEF1A1 is responsible for the enzymatic delivery of aminoacyl tRNAs to the ribosome. Both genes are involved in protein synthesis. Around the 8–16 cell stage the embryonic genome becomes active, resulting in the synthesis of embryonic proteins. The lower expression of genes involved in protein synthesis (COPE, EEF1A1 and RPL10) in in vitro 8-cell embryos might indicate that the embryonic protein synthesis is affected or delayed in in vitro embryos.

Differences in gene expression between in vivo and in vitro produced embryos reflect the effects of the in vitro culture system on the transcriptional activity 11. Preimplantation embryos are capable of developing in a wide range of culture conditions. However, this adaptation of the embryo to suboptimal conditions may result in a lower embryo quality. It will be a challenge for the future to determine the importance of the affected genes for the process of embryo development and to change the culture conditions in vitro in order to induce the expression of those genes to levels identical to those under in vivo conditions, and in this way improve the quality of the embryo.

The results of this study add to the knowledge of the process of bovine blastocyst formation. Several genes were identified as candidate markers involved in blastocyst formation. The fact that some of those candidate genes were differentially expressed between in vivo and in vitro produced embryos confirms the belief that culture conditions influence embryonic gene expression. Further functional analyses of these candidate genes may help to optimise in vitro embryo culture systems in order to improve the quality of in vitro produced bovine embryo.

KG performed all the experimental procedures and was the primary author of the manuscript. AVS contributed to the design of the IVF experiments. MVP participated in the study design and provided real-time support. LV helped with the production of the in vivo embryos and the evaluation of the immunofluorescent experiments. JV provided expert input in data analysis and statistics. AVZ and LJP participated in the design of the project, helped to draft the manuscript and supervised the study. All authors read and approved the final manuscript.