After a survey of the recent literature we curated a short list of fourteen commonly used reference genes and either designed new primers or pulled existing ones from references cited in the materials and methods
. These genes are listed in and represent several distinct functional classes so as to minimize the possibility of co-regulation. For each gene, a minimum of two independent primer sets were tested and of these, the primer set exhibiting the greatest efficiency was selected. To conduct an accurate survey of candidate gene expression levels between ES, TS and XEN stem cells we isolated RNA from at least three independent stem cell lines, representing at least two different genotypes. We postulate that utilizing lines derived from diverse genotypes will more accurately identify stable reference genes to be used in future studies contrasting patterns of gene expression.
Descriptions of the fourteen candidate reference genes studied.
Previous studies in our laboratory have utilized stem cell lines derived from Mus musculus castaneus x mus musculus (
C57Black6) F1 embryos. Polymorphisms between these genetic strains allow the examination of mono-allelic patterns of epigenetic marks and gene expression within loci regulated by genomic imprinting 
. For ES, TS and XEN stem cell analysis we utilized lines derived from F1 embryos of reciprocal crosses between these stains (C57Black6×Castaneous and Cast7×Black6) 
. For analysis of ES and TS cells we also utilized the previously described R1 ES and TS3.5 lines derived from 129 stain mice 
. Each of these different lines demonstrated cellular morphology consistent with their cell type and expressed unique cohorts of transcription factors characteristic of their lineage 
(). Cell lines were cultured to 80% confluence, RNA isolated and seeded into five independent qRT-PCR reactions measuring our fourteen candidate genes. Results presented below are the combined analysis of all genetic backgrounds tested.
Characteristic cellular morphology and marker gene expression for ES, TS and XEN Stem Cells.
Of the candidate genes tested Rn7sk
demonstrated the most robust expression averaging expression levels 125 fold higher than the remaining candidates; which were all readily detectable in each of the cell lines tested. To measure the relative stability of each of the candidate genes between the ES, TS and XEN lines, the CT values for the measured transcripts were compiled and run through the NormFinder, GENorm, and BestKeeper software packages 
. Each of these algorithms utilize slightly different methods of estimating both the intra- and the intergroup expression variation, and allow the ranking of candidate genes based on the calculation of a “stability value”. While there was variation amongst the mid-range to least stable genes, all three software packages identified Pgk1
as the most consistently stable reference genes between ES, TS and XEN stem cells (). Similar to previous studies by Mamo et al., we observed the classic “housekeeping genes” Actb
and to a lesser extent Gapdh
were comparatively unstable and by our analysis would not be the best choice to normalize qPCR expression levels 
Candidate reference genes ranked in order of their stability.
We next chose to make pair-wise comparisons between ES and TS, ES and XEN as well as TS and XEN to see which candidates emerged as the most stable in contrasts between any two cell types. A consensus of all three software packages can be seen in . As with the comparisons between all three lines, Pgk1
remained in the top five most stable genes indicating no one cell type was biasing our analysis and that these five reference genes represent the best normalization controls for qPCR-based analysis of gene expression. Utilizing the geometric mean of Pgk1
, and Tbp
we normalized the CT values for each of the fourteen candidates and graphed their relative expression levels as described previously 
. As can bee seen in , Rn7sk
is expressed at a drastically higher level than any of the other candidates tested and therefore does not represent a viable reference gene. Similarly, analysis of Actb
all yielded significant differences in measurements of TS cell expression as compared to both ES and XEN cells eliminating their candidacy. Our results indicate normalizing quantitative RT-PCR measurements using the geometric mean CT values obtained for the Pgk1, Sdha and Tbp mRNAs, offers the most reliable method to assess differing patterns of gene expression between the three founding stem cell lineages present within the mammalian preimplantation embryo.
Consensus of the stability rankings for pair-wise comparisons between the stem cell types.
Relative expression of the fourteen candidate genes between all three genetic backgrounds of ES, TS and XEN stem cells.
We next sought to determine which of the candidate genes remained the most stable throughout the process of differentiation. Therefore we chose to differentiate our ES and TS cell lines by removal of the key growth factors LIF and FGF4 respectively 
. To this end ES cells were cultured in LIF - ES cell medium, allowed to form embryoid bodies on untreated plastic dishes and then plated on regular tissue culture plastic to differentiate into fibroblast like cells. Similarly, TS cells were plated on tissue culture treated plastic at low density in FGF4- medium which promoted the formation of TS giant-like cells. We chose not to investigate the process of XEN cell differentiation as reliable protocols for the induction of differentiation have not yet been established. In contrast to both ES and TS cell lines, when XEN cells are plated on plastic many cells simply senesce, while the remainder do not uniformly differentiate into one cell type, thus complicating our analysis.
RNA samples were collected from ES cells on Day 0, Day 4 (embryoid body) and Day 8 and RNA seeded into five independent qPCR reactions measuring each of the fourteen candidate genes. Using a similar experimental design as described above, we identify Sdha, Tbp and Ywhaz as the three most stable transcripts (). To examine relative changes in gene expression, we utilized the geometric mean of these three most stable candidates to normalize CT values and graphed the relative expression of all fourteen candidate genes though ES cell differentiation (). We then chose to examine the expression of the cell lineage marker fibroblast-specific protein-1 (FSP-1) which is active in fibroblasts but not in epithelium, mesangial cells or embryonic endoderm 
. In accordance with previous studies this maker demonstrated increasing expression in differentiating cell cultures, indicating our three candidate genes provided a valid reference point () 
. In contrast, transcripts encoding Pgk1
) and Gapdh
all demonstrate a significant down-regulation and therefore are not suitable reference genes for this experimental time course. Similar to results reported by Willems et al., examining ES cell differentiation induced by both DMSO and Retinoic acid, we also identify B2m
as among the most unstable transcripts 
. Using similar methodologies, we identified the Ywhaz
transcripts as the most stable during TS cell differentiation (). After applying the geometric mean of these three candidates to normalize CT values we observed massive changes in transcripts encoding Actb
(). Previous studies have identified increased actin mobilization as a key feature of trophectoderm stem cell differentiation, validating our identified reference genes 
. Taken together our data indicate Sdha
represent the most stable of our fourteen candidate reference genes for use as qPCR normalization controls during ES and TS cell differentiation respectively.
Candidate reference genes ranked in order of their stability throughout ES cell differentiation using the NormFinder, GENorm and BestKeepr software tools.
A. Relative expression of the fourteen candidate genes throughout differentiation of all three genetic backgrounds of ES cells examined.
Candidate reference genes ranked in order of their stability throughout TS cell differentiation using the NormFinder, GENorm and BestKeepr software tools.
Relative expression of the fourteen candidate genes throughout differentiation of all three genetic backgrounds of TS cells.