Vertebrate evolution is accompanied by a substantial conservation of transcriptional programs with more than a third of unique orthologous genes showing constrained levels of expression. Moreover, there are genes and exons exhibiting excellent expression stability according to RNA-seq data across a panel of eighteen tissues including the ovary (Human Body Map 2.0).
We hypothesized that orthologs of these exons would also be highly uniformly expressed across neonatal ovaries of the horse, which would render them appropriate reference genes (RGs) for normalization of reverse transcription quantitative PCR (RT-qPCR) data in this context. The expression stability of eleven novel RGs (C1orf43, CHMP2A, EMC7, GPI, PSMB2, PSMB4, RAB7A, REEP5, SNRPD3, VCP and VPS29) was assessed by RT-qPCR in ovaries of seven neonatal fillies and compared to that of the expressed repetitive element ERE-B, two universal (OAZ1 and RPS29) and four traditional RGs (ACTB, GAPDH, UBB and B2M). Expression stability analyzed with the software tool RefFinder top ranked the normalization factor constituted of the genes SNRPD3 and VCP, a gene pair that is not co-expressed according to COEXPRESdb and GeneMANIA. The traditional RGs GAPDH, B2M, ACTB and UBB were only ranked 3rd and 12th to 14th, respectively.
The functional diversity of the novel RGs likely facilitates expression studies over a wide range of physiological and pathological contexts related to the neonatal equine ovary. In addition, this study augments the potential for RT-qPCR-based profiling of human samples by introducing seven new human RG assays (C1orf43, CHMP2A, EMC7, GPI, RAB7A, VPS29 and UBB).
Citation: Scarlet D, Ertl R, Aurich C, Steinborn R (2015) The Orthology Clause in the Next Generation Sequencing Era: Novel Reference Genes Identified by RNA-seq in Humans Improve Normalization of Neonatal Equine Ovary RT-qPCR Data. PLoS ONE 10(11): e0142122. https://doi.org/10.1371/journal.pone.0142122
Editor: Hemachandra Reddy, Texas Tech University Health Science Centers, UNITED STATES
Received: July 17, 2015; Accepted: October 16, 2015; Published: November 4, 2015
Copyright: © 2015 Scarlet et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited
Data Availability: All relevant data are within the paper and its Supporting Information files.
Funding: The authors have no support or funding to report.
Competing interests: The authors have declared that no competing interests exist.
In recent years the mare has intensively been used as an animal model for age-associated changes in human reproductive medicine . Better approaches for the analysis of the molecular mechanisms behind ovarian development and maturation would help to prevent and treat infertility in both the human and the horse. The developing fetal equine ovary possesses several morphologic and developmental differences compared to other domestic species . After the interstitial cell hypertrophy and hyperplasia that results in marked gonadal enlargement during the time spent in utero, a period of regression and ovarian cortical migration follows . Recently, we have described the presence of tertiary follicles up to 4 mm in diameter in newborn fillies . However, the regulatory mechanisms behind these processes have not been elucidated yet.
Quantitative real-time RT-PCR (RT-qPCR) is the most commonly used method for measuring transcript copies being especially useful when samples contain small numbers of cells . The accuracy of RT-qPCR measurement is strongly dependent on appropriate normalization—usually accomplished using reference genes (RGs; ). RGs are mainly derived from the class of housekeeping genes required for the maintenance of basal cellular functions that are essential for the existence of a cell, regardless of its specific role in the tissue or organism . RGs are the major drawback for RT-qPCR, as their validation on an individual basis for all treatment and experimental conditions is essential for accurate data interpretation, especially when normalizing for inter-sample variability . Depending on the biological context, two [8,9] or three [8,10,11] of the most stably expressed RGs are sufficient to normalize the expression.
Over the past years, several mammalian genes have been intensively used as internal controls in gene expression studies, such as the traditional RGs RN18S (18S rRNA), GAPDH, ACTB, HPRT, B2M and ALB. Some of the latter genes (ACTB, GAPDH and RN18S) appear in 70% of the publications on RNA based qPCR as the solely normalizer, a fact that makes it highly unlikely that these have been validated . Traditional RGs were derived by educated guesswork in the era of Northern blotting [13–15]. In the absence of any transcriptome-wide expression information, the expression uniformity of these genes was supposed to be stably expressed not only in model species, but also in their genetic relatives and hence frequently used as normalizers across plants and animals [16–19]. Some years ago, the orthology clause was suggested for a particular target context, the liver of mice, cattle and pig . The clause indicated a similar expression of gene orthologs across related species. A huge amount of information, including novel aspects on human and mouse transcriptomes, was recently uncovered through RNA-seq [6,21]. This has opened a new dimension for studying mechanisms of molecular regulation. This data also indicated a substantial conservation of transcriptional programs between human and mouse by uncovering a distinct class of genes with levels of expression that have been constrained early in vertebrate evolution . In detail, it was found that evolutionary constraint in gene expression levels is associated with conserved epigenetic marking and a characteristic post-transcriptional regulatory program.
Stability of RNA expression can be assessed by software tools such as GeNorm , NormFinder  or BestKeeper . These major computational programs are integrated by the tool RefFinder (, http://fulxie.0fees.us/?type=reference), thus compensating for their individual weaknesses.
No gene exhibits a stable pattern of expression across all tissues and conditions of an individual . Commonly used reference genes such as ACTB, GAPDH, UBC and RN18S can vary considerably depending on tissue types, developmental stage, sex, pathology, and experimental conditions . For example, in whole blood donated by healthy human individuals, expression of ACTB, GAPDH or HPRT differed by a factor of 20- to 25-fold . This level of variation considerably exceeds the strong threshold criterion of ΔCq ≤ ±0.5 (twofold expression alteration) recommended for RG suitability . Only few genes, termed “universal” reference genes, are relatively uniformly expressed across a wide range of conditions in men or mice . For each biological context, a subset of genes exists that is more stably expressed than traditional normalizers or RGs that were selected by meta-analysis based on their expression stability across a plenitude of experimental conditions .
So far, there is a lack of reliable RGs for the biological context of this study—the neonatal ovary of Equus caballus. Here we tested the application of the orthology clause in the next generation sequencing era for this experimental setup. We targeted the equine orthologous genes of the most stably expressed human counterparts identified by RNA-seq from a wide panel of human tissues (Illumina Human Body Map 2.0) including the ovary. We argued that the extended number of transcripts identified, a resolution down to the level of individual exons and the expression consensus obtained by profiling across a large set of target tissues would facilitate a successful application of the clause, hence improve the gene repertoire for normalization of RT-qPCR data in this context.
Material and Methods
In agreement with University Guidelines for Good Scientific practice, owners’ approval for scientific use of post-mortem material was obtained. In accordance with Austrian legislation on animal experimentation, collection of postmortem material from animal that had to be euthanized for veterinary medical reasons is not considered animal experimentation. Thus neither an animal experimentation license issued by the competent authority nor an internal approval by the University Animal Care and Use Committee was required.
Fourteen ovaries were collected from euthanized newborn female foals (n = 7). Euthanasia was performed due to various pathological conditions not related to the female genital tract using 10 mg/kg Thiopental (Inresa, Freiburg, Germany) followed by 4–6 ml/50kg T61 (MSD Animal Health, Vienna, Austria). The average age of foals was two days (range: 0 to 8 days).
Sample preparation and RNA extraction
Fresh ovary biopsies were embedded in Tissue-Tek O.C.T. compound (Sakura Finetek, Alphen aan den Rijn, Netherlands), snap-frozen in liquid nitrogen and then stored at -80°C for less than 12 months. Frozen sections of 5 × 100 μm tissue were prepared on the cryostat CryoStar NX70 (Thermo Scientific, Waltham, USA). After removal of excess embedding medium, the sections were placed into 2 ml-screw caps filled with 1.3 mm ceramic beads (Peqlab Biotechnologie GmbH, Erlangen, Germany) and 350 μl RLT lysis buffer supplemented with 3.5 μl β-mercaptoethanol (Qiagen, Hilden, Germany). Mechanical homogenization was performed on the MagNA Lyser instrument (Roche, Rotkreuz, Switzerland) at 6500 rpm for 20 s followed by cooling on ice for 1 min. If necessary, the homogenization step was repeated. Total RNA was extracted using the RNeasy Micro Kit (Qiagen) according to Qiagen’s RNeasy Micro Kit protocol for fibrous tissues except an extended time for Proteinase K incubation (30 min). The procedure included on-column DNase I digestion for 15 min to remove residual DNA that can generate a false positive signal in some of the RT-qPCR assays. RNA concentration was determined by UV spectrophotometry on the NanoDrop 2000c (Peqlab Biotechnologie GmbH). RNA integrity was assessed by capillary electrophoresis on the Agilent 2100 Bioanalyzer (Agilent Technologies, Santa Clara, USA) using the RNA 6000 Nano Kit (Agilent Technologies). Only samples with RNA integrity number (RIN) values of at least seven were used for expression analysis (Fig 1). To further reduce unspecific signals from the presence of genomic DNA as concluded from a RT-qPCR assay against a traditional RG, ACTB (data not shown), samples were again DNase digested (TURBO DNA-free Kit; Life Technologies, Austin, USA). Exemplarily, two of the seven samples were again assessed on the Agilent Bioanalyzer to exclude that the second DNase digestion had affected the integrity of the RNAs (RIN > 7, ΔRIN of 0 and 1.0).
*Input RNA might contain nucleotides or short DNA fragments as a result of DNase digestion of co-purified DNA. The same amount of RNA was used for RT as concluded from a second spectrophotometric measurement of RNA mass before the RT step and based on the subsequent determination of Cq values for the rRNA genes RN18S and RN28S (Fig 2).
Selection of RG candidates and design of RT-qPCR primers
Eleven genes showing the most ubiquitous and uniform expression across the ovary and 15 other normal human tissue types  were selected as novel RG candidates, namely C5H1orf43 (ortholog of human C1orf43), CHMP2A, EMC7, GPI, PSMB2, PSMB4, RAB7A, REEP5, SNRPD3, VCP, and VPS29. This expression data aiming at housekeeping gene detection in the era of massive parallel sequencing and RNA-seq was part of the Human BodyMap 2.0 project (www.ebi.ac.uk/arrayexpress/experiments/E-MTAB-513). This RNA-seq data delivered expressions at the exon level. This measure takes care of normalization for the two most obvious factors affecting the raw number of reads per gene, transcript, or exon: the total number of reads produced and their length. The variation of exon expression given as reads per kilobase per million mapped reads (RPKM) is estimated by the standard deviation of log2 (RPKM) over samples, abbreviated as SD(log2(RPKM)).
DNA and cDNA sequences including the transcript isoforms available for the horse were retrieved from NCBI. Aligning them with the orthologous human sequences allowed identifying primer target sequences that were conserved across the two species and less affected by individual polymorphisms.
A biological sample is composed of different types of cells mixed at different proportions. If an RT-qPCR assay is sensitive to alternative splicing, sample heterogeneity can lead to the identification of artificially deregulated transcript isoforms and in case of a RG to an increased expression variation. Therefore, our protocol was designed to target exons that are present in multiple tissue types.
As the exon/intron architecture of genes influences the way of mRNA processing, it was argued that selecting flanking introns of sizes around 1 to 2 kb would be an appropriate compromise to reduce the chance of alternative splicing caused by exon skipping and the alternative usage of 5’ or 3’ splice sites (based on human data from ). In general, the orthologous human exons targeted by the equine RT-qPCR assays were moderately to highly and uniformly expressed across the panel of human reference tissues (SD(log2(RPKM)) of 0.32 to 0.52 ). In a single case (SNRPD3), a less stable exon (SD(log2(RPKM)) = 0.70 ) had to be accepted as a target region in order to achieve the proposed size (1 to 2 kb) of the flanked intron.
To avoid co-amplification of contaminating genomic DNA due to the short elongation time, a critical size limit of at least 750 bp was adopted for the central intron, i.e. the one flanked by the target exons in genomic DNA. The limit was met for nine of the eleven novel putative RGs (Table 1). Two of the primer pairs flanked short introns of 73 and 412 bp (CHMP2A and PSMB4, respectively). In these cases, the minus RT control routinely performed for each experimental RNA confirmed specificity of mRNA amplification.
Each RT-qPCR assay targeted all of the two to four transcripts available at the NCBI Gene database for E. caballus and E. przewalskii (S1 File).
In order to exclude differences in the amount of co-isolated DNA, samples were analysed for expression of 18S und 28S ribosomal fractions.
For comparison four traditional RGs, two universal RGs and one expressed repetitive element were selected. The traditional RGs were selected just by educated guess-work and tradition (ACTB and GAPDH, ) or based on stable expression across a panel of equine tissues (UBB and B2M, [31,32]). The universal RGs, OAZ1 and RPS29, exhibited highly uniform expression in men and mice across a plenitude of conditions [33–35]. The assay for the expressed repetitive element targeted a consensus sequence deduced from all equine repetitive element B (ERE-B) sequences available at the SINEBase database (; http://sines.eimb.ru/).
Primers for the novel candidates, the universal RGs and the expressed repetitive element were designed in the Primer Express 2.0 program (Life Technologies, Foster City, USA). Primer dimerization was assessed by the NetPrimer web tool (Premier Biosoft, Palo Alto, USA) freely available at www.premierbiosoft.com. Amplicon specificity was evaluated by the Primer-BLAST tool of NCBI using the non-redundant database of the taxid “horses”. Amplicon folding was analyzed using the Mfold web server . Primer sequences and assay details are listed in Table 1.
In order to account for variations during the RT step, two replicates of each RNA sample were converted into cDNA using the High Capacity Reverse Transcription Kit (Life Technologies) (Fig 1). For each replicate, 2 μg total RNA input were reverse-transcribed with random hexamers in a reaction volume of 40 μl following the recommended protocol. Since sensitivity towards genomic DNA contamination differs greatly between assays [38,39], controls in which no reverse transcriptase was added were run for each assay and each experimental RNA. RT-qPCR was performed in duplicates for each cDNA replicate (equals four technical replicates for each RNA sample) using the Power SYBR Green Master Mix (Life Technologies). The 20-μl reaction included 150 nM of each primer and 2 μl of five-fold diluted cDNA. All qPCR runs were performed on a Viia7 Real-Time PCR System (Life Technologies). After an initial denaturation at 95°C for 10 min, 40 amplification cycles of 95°C for 15 s and 60°C for 1 min were performed. Amplicon dissociation was monitored over a temperature range of 60°C to 95°C. Additionally, PCR products were separated on a 2% agarose gel stained with GelGreen (Biotium, Hayward, USA) and visualized by blue light excitation (S2 File). Absence of unspecific amplification derived from the mastermix was confirmed by melting curve analysis of the no template control product. Absence of PCR inhibitors was assessed by RT-qPCR against RN18S using eightfold dilution of all experimental cDNAs as template.
Reverse transcription was repeated if the Cq values of two cDNA replicates differed by more than 0.5 cycles.
A standard curve was generated for each assay for calculating PCR efficiency (E) from the term 10−1/s—1, were s is the slope of the graph obtained by plotting the Cq value against the log10 of the cDNA input . For Cq values generated at E < 1, efficiency corrected Cq values were obtained from the term Cq * log10(E + 1)/log10(2) .
The efficiency-corrected Cq values were processed using the RefFinder tool integrating the currently available major computational programs geNorm, Normfinder, BestKeeper, and the comparative ΔΔCq method to compare and rank the tested candidate reference genes (; http://fulxie.0fees.us/?type=reference). Based on the rankings from each program, RefFinder assigns an appropriate weight to an individual gene or set of genes and calculates the geometric mean of their weights for the overall final ranking. A normalization factor, i.e. the geometric mean of several top-ranking genes, was considered suitable as soon as it was ranked ahead of all single genes.
In order to exclude the association of highly coexpressed genes which could affect the stability of the normalization factor, coexpression was identified with COXPRESdb  software, which calculates Pearson’s correlation coefficient for each gene pair, and then transfers these values to the mutual rank value, which is the geometric average of asymmetric ranks in coexpressed gene lists.
To confirm COXPRESdb results, the predicted coexpression of the novel RGs was analysed with GeneMANIA , a software based on functional association data such as protein and genetic interactions, pathways, co-expression, co-localization, and protein domain similarity. The default approach, the automatically selected weighting method, was used when combining all networks into the final composite. In case of our search list with more than five genes, GeneMANIA assigns weights to maximize connectivity between all input genes using the ‘assigned based on query gene’ strategy. According to this strategy the weights are chosen automatically using linear regression, to make genes on the search list interact as much as possible with each other and as little as possible with genes not on this list.
As E. caballus species is neither reviewed by COEXPRESdb, nor by GeneMANIA, Homo sapiens databases were used in both cases.
The RT-qPCR data of this work comply with the Minimum Information for Publication of Quantitative Real-Time PCR Experiments (MIQE) guidelines .
Statistical analysis was performed using SPSS Statistics 22 (IBM-SPSS, Armonk, NY, USA). The Kolmogorov–Smirnov test was used to demonstrate a normal distribution of efficiency corrected and log transformed Cq values of the candidate RGs. Correlations between parameters were analyzed using the Pearson’s correlation test. A P < 0.05 was considered significant.
Results and Discussion
This study addressed the orthologous use of exons ubiquitously and stably expressed across a wide set of human tissues for improving RT-qPCR normalization in the context of the equine ovary. If genome-wide transcript data is missing for a specific biological context in a related species, for example the neonatal ovary of the horse, this assumption could be applied to select RGs for improving RT-qPCR data normalization.
Choosing the most suitable RGs for one experiment is being complicated by the fact that the majority of mRNAs undergoes alternative polyadenylation . This is more likely to occur when the target exons are located in the three prime parts of a gene. In addition, most human multi-exon genes are known to be alternatively spliced. Out of these, 85% have a minor isoform frequency of at least 15% . For the human genome, alternative splicing by exon skipping is more likely to occur when exons are flanked by long introns, whereas the activation of alternative splice sites is less likely if the flanking introns are long [29,46]. Therefore, for the first time we considered these crucial principles when selecting appropriate exons for targeting by the RT-qPCR primers.
The eleven novel RGs showed high to low expression levels (efficiency-corrected Cq values < 15 to > 30, respectively; S3 File) and span a wide range of biological functions in general (S4 File). According to the GeneMANIA prediction server, they are not part of the same biochemical process (“high” false discovery rate of 5.44e-2) with exception of PSMB2, PSMB4 and VCP which belong to the proteasome complex.
Intactness of experimental RNAs was confirmed by RIN values above the required threshold of at least seven (range: 7.4 to 8.9; Fig 2). Co-purification of traces of DNA during RNA extraction is inevitable  with some RNA preparations yielding virtually pure RNA and others virtually pure DNA and this is independent of tissue type or operator . Therefore, after determination of RIN values, samples were digested by DNase I to include also primer pairs from previous publications that were not intron-spanning or that flanked smaller introns (< 750 bp). Considering that in solution digestion did not remove the degradation products of carry-over DNA, we ruled out that the experimental cDNAs were derived from different proportions of total RNA due to differences in the amount of co-extracted genomic DNA. This was concluded from similar Cq values of RT-qPCRs targeting the main contributors to total RNA mass, RN18S and RN28S (Fig 2). In other words, the similarity of Cq values indicated that the selection of novel uniformly expressed genes for normalization of RT-qPCR data in the context of postnatal equine ovaries would not be affected by differences in the amount of co-isolated DNA. We argued that the approximately 50% change in the amount of input RNAs concluded from a ΔCq value of approximately 0.5 for the RN28S transcript (Fig 2) did not affect the aim of this study—selection of novel RGs, since Cq values of RN28S and putative RGs were not correlated (P > 0.1).
Absence of PCR inhibitors from the experimental cDNAs was confirmed by RT-qPCR profiling of experimental cDNAs and their eightfold dilutions with an assay against RN18S. The low standard deviation of Cq values determined for the eightfold dilutions of the cDNAs indicated homogeneity regarding sample purity (mean ΔCq of 2.7 ± 0.1 corresponding to E = 1). The slight deviation from the theoretical ΔCq of three was attributed to inter-assay variation considering that the sets of diluted and undiluted samples were run on different plates.
Expression stability rankings produced by analytic algorithms like NormFinder and BestKeeper are often in overall agreement . However, a discrepancy in stability rankings and outcomes between different analytical methods has been reported [24,25,48]. To increase the chance for deducing a “consensus” in the expression uniformity ranking of the novel candidate reference genes we used the comprehensive tool RefFinder (S5 File). The comprehensive approach identified SNRPD3—a core component of the spliceosome and VCP—an ATPase of the endoplasmic reticulum that prevents mutations caused by DNA damage, as the most stably expressed pair of genes in the biological context studied. Traditional RGs like GAPDH, B2M, ACTB and UBB were less uniformly expressed as documented by their ranks 3 and 12 to 14, respectively. If normalization should be performed with more than one RG, the pair SNRPD3 and VCP is recommended for calculating a normalization factor (Fig 3).
The individual genes or the normalization factor calculated from the best-ranking, not co-expressed pair of genes are depicted. Only the “expressors” of highest and lowest ranks are presented.
Finally, co-expression analysis of traditional and universal reference genes as well as the novel normalization candidates was performed using COEXPRESdb using all tissues available. For the best-performing gene pair, VCP and SNRPD3, absence of a significant co-expression was found (mutual rank: 716.7; Pearson’s coefficient: 0.33; S6 File). The latter values were far from being biologically significant (respective thresholds: < 30  and > 0.55 ). While the Pearson’s correlation coefficient is affected by the function of the target gene and the method of gene expression database construction, the alternative use of correlation ranks helps to retrieve only genes that are functionally related . Likewise, identification of co-expressed genes by GeneMANIA, containing peer-reviewed data collected mostly from the Gene Expression Omnibus (GEO), also failed to detect a co-expression for this pair of genes (Fig 4).
Two genes are linked if their expression levels are similar across conditions in a gene expression study, mostly derived from peer-reviewed publication data publicly accessible via the Gene Expression Omnibus (GEO). A thicker line indicates a higher co-expression. Note that ERE-B represents a repetitive element specific to the horse.
For seven genes (C1orf43, CHMP2A, EMC7, GPI, RAB7A, VPS29 and UBB; S7 File) RT-qPCR primers could be designed based on a consensus sequence obtained by aligning the equine transcripts with all human orthologs available in the GenBank. The design of cross-species primers considerably reduces the likelihood of mismatches in their priming sites and hence improves quantification accuracy. For experimental validation of these seven RT-qPCR assays for human samples, RNA isolated from HeLa and Jurkat cells was used as template. The specificity of amplicons was demonstrated by agarose gel electrophoresis (S8 File).
This study assayed the expression stability of highly conserved human exon orthologs in the neonatal horse ovary. Novel housekeeping genes identified in the related species Homo sapiens including the best-ranking gene pair of our study, SNRPD3 and VCP, have been confirmed as more stable RGs for RT-qPCR profiling compared to traditional ones. The superiority of the novel RG set is characterized by high expression uniformity and obviously no co-expression. We attributed the successful application of the orthology clause in the next generation sequencing era to the extended number of transcripts identified, a resolution down to the level of individual exons and to the expression consensus obtained by profiling across a large set of target tissues.
The functional diversity of the novel orthologous RGs likely facilitates expression studies over a wide range of physiological and pathological contexts related to the neonatal equine ovary. However, for each biological context a specific subset of RGs needs to be validated from this extended pool of genes. With regard to context-specific RG validation in equine research, the ethical aspect needs to be strongly considered, as sample collection at different stages of development would involve inducing abortion in late pregnant mares or euthanizing young healthy foals.
In the near future RNA-seq analysis putatively performed for the neonatal and the adult equine ovary will help to further extend the repertoire of RGs for RT-qPCR normalization by adding context-optimized RGs with superior expression uniformity as shown before for the context of the mouse jejunum .
We also note an extension of the toolkit for human molecular biology by introducing novel validated RG assays for human samples that are in the truest sense in accordance with the MIQE guidelines regarding full transparency of oligonucleotide sequences .
S1 File. The RT-qPCR assays for the novel reference genes detect all transcript isoforms currently known.
S2 File. RT-PCR product size of equine ovary sample determined on a 2% agarose gel stained with GelGreen.
S3 File. Estimated transcript abundance of all reference gene candidates.
S4 File. Names and functions of reference gene candidates.
S5 File. Complete list of expression uniformity analysis by the RefFinder software.
S6 File. Co-expression analysis for all pairs of gene candidates using COEXPRESdb.
S7 File. Validation of RT-qPCR assays for novel human reference genes.
Conceived and designed the experiments: DS RS. Performed the experiments: RE. Analyzed the data: DS RE CA RS. Wrote the paper: DS RE CA RS. Study initiation: DS CA RS. Supervised the study: CA RS.
- 1. Carnevale EM (2008) The mare model for follicular maturation and reproductive aging in the woman. Theriogenology 69 (1): 23–30. pmid:17976712
- 2. Walt ML, Stabenfeldt GH, Hughes JP, Neely DP, Bradbury R (1979) Development of the equine ovary and ovulation fossa. Journal of reproduction and fertility. Supplement (27): 471–477. pmid:289826
- 3. Scarlet D, Walter I, Hlavaty J, Aurich C (2015) Expression and immunolocalisation of follicle-stimulating hormone receptors in gonads of newborn and adult female horses. Reproduction, fertility, and development.
- 4. Cai J, Li T, Huang B, Cheng H, Ding H, Dong W et al. (2014) The use of laser microdissection in the identification of suitable reference genes for normalization of quantitative real-time PCR in human FFPE epithelial ovarian tissue samples. PloS one 9 (4): e95974. pmid:24776823
- 5. Bustin SA, Benes V, Garson JA, Hellemans J, Huggett J, Kubista M et al. (2009) The MIQE guidelines: minimum information for publication of quantitative real-time PCR experiments. Clinical chemistry 55 (4): 611–622. pmid:19246619
- 6. Eisenberg E, Levanon EY (2013) Human housekeeping genes, revisited. Trends in genetics: TIG 29 (10): 569–574. pmid:23810203
- 7. Ferguson BS, Nam H, Hopkins RG, Morrison RF (2010) Impact of reference gene selection for target gene normalization on experimental outcome using real-time qRT-PCR in adipocytes. PloS one 5 (12): e15208. pmid:21179435
- 8. Pan H, Yang X, Siegfried BD, Zhou X (2015) A Comprehensive Selection of Reference Genes for RT-qPCR Analysis in a Predatory Lady Beetle, Hippodamia convergens (Coleoptera: Coccinellidae). PLoS ONE 10 (4): e0125868. pmid:25915640
- 9. Saint-Marcoux D, Proust H, Dolan L, Langdale JA (2015) Identification of reference genes for real-time quantitative PCR experiments in the liverwort Marchantia polymorpha. PLoS ONE 10 (3): e0118678. pmid:25798897
- 10. Thomas KC, Zheng XF, Garces Suarez F, Raftery JM, Quinlan KG, Yang N et al. (2014) Evidence based selection of commonly used RT-qPCR reference genes for the analysis of mouse skeletal muscle. PLoS ONE 9 (2): e88653. pmid:24523926
- 11. Sun M, Lu M, Tang X, Du Y (2015) Exploring valid reference genes for quantitative real-time PCR analysis in Sesamia inferens (Lepidoptera: Noctuidae). PLoS ONE 10 (1): e0115979. pmid:25585250
- 12. Dijkstra JR, van Kempen LC, Nagtegaal ID, Bustin SA (2014) Critical appraisal of quantitative PCR results in colorectal cancer research: can we rely on published qPCR results. Mol Oncol 8 (4): 813–818. pmid:24423493
- 13. Goldsworthy SM, Goldsworthy TL, Sprankle CS, Butterworth BE (1993) Variation in expression of genes used for normalization of Northern blots after induction of cell proliferation. Cell Prolif 26 (6): 511–518. pmid:9116118
- 14. Bhatia P, Taylor WR, Greenberg AH, Wright JA (1994) Comparison of glyceraldehyde-3-phosphate dehydrogenase and 28S-ribosomal RNA gene expression as RNA loading controls for northern blot analysis of cell lines of varying malignant potential. Anal Biochem 216 (1): 223–226. pmid:8135355
- 15. Savonet V, Maenhaut C, Miot F, Pirson I (1997) Pitfalls in the use of several "housekeeping" genes as standards for quantitation of mRNA: the example of thyroid cells. Anal. Biochem. 247 (1): 165–167. pmid:9126387
- 16. Smits K, Goossens K, van Soom A, Govaere J, Hoogewijs M, Vanhaesebrouck E et al. (2009) Selection of reference genes for quantitative real-time PCR in equine in vivo and fresh and frozen-thawed in vitro blastocysts. BMC Res Notes 2: 246. pmid:20003356
- 17. Reddy DS, Bhatnagar-Mathur P, Cindhuri KS, Sharma KK (2013) Evaluation and validation of reference genes for normalization of quantitative real-time PCR based gene expression studies in peanut. PLoS ONE 8 (10): e78555. pmid:24167633
- 18. Hao X, Horvath DP, Chao WS, Yang Y, Wang X, Xiao B (2014) Identification and Evaluation of Reliable Reference Genes for Quantitative Real-Time PCR Analysis in Tea Plant (Camellia sinensis (L.) O. Kuntze). International Journal of Molecular Sciences 15 (12): 22155–22172. pmid:25474086
- 19. Omondi BA, Latorre-Estivalis JM, Rocha Oliveira IH, Ignell R, Lorenzo MG (2015) Evaluation of reference genes for insect olfaction studies. Parasit Vectors 8: 243. pmid:25896676
- 20. Hruz T, Wyss M, Docquier M, Pfaffl MW, Masanetz S, Borghi L et al. (2011) RefGenes: identification of reliable and condition specific reference genes for RT-qPCR data normalization. BMC genomics 12: 156. pmid:21418615
- 21. Yue F, Cheng Y, Breschi A, Vierstra J, Wu W, Ryba T et al. (2014) A comparative encyclopedia of DNA elements in the mouse genome. Nature 515 (7527): 355–364. pmid:25409824
- 22. Pervouchine DD, Djebali S, Breschi A, Davis CA, Barja PP, Dobin A et al. (2015) Enhanced transcriptome maps from multiple mouse tissues reveal evolutionary constraint in gene expression. Nat Commun 6: 5903. pmid:25582907
- 23. Vandesompele J, de Preter K, Pattyn F, Poppe B, van Roy N, De Paepe A et al. (2002) Accurate normalization of real-time quantitative RT-PCR data by geometric averaging of multiple internal control genes. Genome Biol 3 (7): RESEARCH0034. pmid:12184808
- 24. Andersen CL, Jensen JL, Orntoft TF (2004) Normalization of real-time quantitative reverse transcription-PCR data: a model-based variance estimation approach to identify genes suited for normalization, applied to bladder and colon cancer data sets. Cancer Res 64 (15): 5245–5250. pmid:15289330
- 25. Pfaffl MW, Tichopad A, Prgomet C, Neuvians TP (2004) Determination of stable housekeeping genes, differentially regulated target genes and sample integrity: BestKeeper—Excel-based tool using pair-wise correlations. Biotechnol Lett 26 (6): 509–515. pmid:15127793
- 26. Chen D, Pan X, Xiao P, Farwell MA, Zhang B (2011) Evaluation and identification of reliable reference genes for pharmacogenomics, toxicogenomics, and small RNA expression analysis. Journal of cellular physiology 226 (10): 2469–2477. pmid:21412770
- 27. Svingen T, Letting H, Hadrup N, Hass U, Vinggaard AM (2015) Selection of reference genes for quantitative RT-PCR (RT-qPCR) analysis of rat tissues under physiological and toxicological conditions. PeerJ 3: e855. pmid:25825680
- 28. Dheda K, Huggett JF, Bustin SA, Johnson MA, Rook G, Zumla A (2004) Validation of housekeeping genes for normalizing RNA expression in real-time PCR. BioTechniques 37 (1): 112–4, 116, 118–9. pmid:15283208
- 29. Fox-Walsh KL, Dou Y, Lam BJ, Hung S, Baldi PF, Hertel KJ (2005) The architecture of pre-mRNAs affects mechanisms of splice-site pairing. Proc Natl Acad Sci USA 102 (45): 16176–16181. pmid:16260721
- 30. Herrera-Luna CV, Budik S, Aurich C (2012) Gene expression of ACTH, glucocorticoid receptors, 11βHSD enzymes, LH-, FSH-, GH receptors and aromatase in equine epididymal and testicular tissue. Reproduction in domestic animals 47 (6): 928–935. pmid:22335522
- 31. Ahn K, Bae J, Nam K, Lee C, Park K, Lee H et al. (2011) Identification of reference genes for normalization of gene expression in thoroughbred and Jeju native horse(Jeju pony) tissues. Genes Genom 33 (3): 245–250.
- 32. Beekman L, Tohver T, Dardari R, Léguillette R (2011) Evaluation of suitable reference genes for gene expression studies in bronchoalveolar lavage cells from horses with inflammatory airway disease. BMC molecular biology 12: 5. pmid:21272375
- 33. de Jonge HJ, Fehrmann RS, de Bont ES, Hofstra RM, Gerbens F, Kamps WA et al. (2007) Evidence based selection of housekeeping genes. PloS one 2 (9): e898. pmid:17878933
- 34. Meyer FR, Grausgruber H, Binter C, Mair GE, Guelly C, Vogl C et al. (2013) Cross-platform microarray meta-analysis for the mouse jejunum selects novel reference genes with highly uniform levels of expression. PloS one 8 (5): e63125. pmid:23671661
- 35. Kwon MJ, Oh E, Lee S, Roh MR, Kim SE, Lee Y et al. (2009) Identification of novel reference genes using multiplatform expression data and their validation for quantitative gene expression analysis. PloS one 4 (7): e6162. pmid:19584937
- 36. Vassetzky NS, Kramerov DA (2013) SINEBase: a database and tool for SINE analysis. Nucleic acids research 41 (Database issue): D83–9. pmid:23203982
- 37. Zuker M (2003) Mfold web server for nucleic acid folding and hybridization prediction. Nucleic acids research 31 (13): 3406–3415. pmid:12824337
- 38. Laurell H, Iacovoni JS, Abot A, Svec D, Maoret J, Arnal JF et al. (2012) Correction of RT-qPCR data for genomic DNA-derived signals with ValidPrime. Nucleic Acids Res. 40 (7): e51. pmid:22228834
- 39. De Keyser E, Desmet L, Van Bockstaele E, De Riek J (2013) How to perform RT-qPCR accurately in plant species? A case study on flower colour gene expression in an azalea (Rhododendron simsii hybrids) mapping population. BMC Mol. Biol. 14: 13. pmid:23800303
- 40. Higuchi R, Fockler C, Dollinger G, Watson R (1993) Kinetic PCR analysis: real-time monitoring of DNA amplification reactions. Biotechnology (N Y) 11 (9): 1026–1030.
- 41. Kenkel CD, Aglyamova G, Alamaru A, Bhagooli R, Capper R, Cunning R et al. (2011) Development of gene expression markers of acute heat-light stress in reef-building corals of the genus Porites. PloS one 6 (10): e26914. pmid:22046408
- 42. Okamura Y, Aoki Y, Obayashi T, Tadaka S, Ito S, Narise T et al. (2015) COXPRESdb in 2015: coexpression database for animal species by DNA-microarray and RNAseq-based expression data with multiple quality assessment systems. Nucleic acids research 43 (Database issue): D82–6. pmid:25392420
- 43. Warde-Farley D, Donaldson SL, Comes O, Zuberi K, Badrawi R, Chao P et al. (2010) The GeneMANIA prediction server: biological network integration for gene prioritization and predicting gene function. Nucleic acids research 38: W214–20. pmid:20576703
- 44. Hoque M, Ji Z, Zheng D, Luo W, Li W, You B et al. (2013) Analysis of alternative cleavage and polyadenylation by 3' region extraction and deep sequencing. Nature methods 10 (2): 133–139. pmid:23241633
- 45. Hallegger M, Llorian M, Smith Christopher W J (2010) Alternative splicing: global insights. The FEBS journal 277 (4): 856–866. pmid:20082635
- 46. Sterner DA, Carlo T, Berget SM (1996) Architectural limits on split genes. Proc Natl Acad Sci U S A 93 (26): 15081–15085. pmid:8986767
- 47. Bustin SA (2002) Quantification of mRNA using real-time reverse transcription PCR (RT-PCR): trends and problems. J Mol Endocrinol 29 (1): 23–39. pmid:12200227
- 48. Martínez-Beamonte R, Navarro MA, Larraga A, Strunk M, Barranquero C, Acin S et al. (2011) Selection of reference genes for gene expression studies in rats. J. Biotechnol. 151 (4): 325–334. pmid:21219943
- 49. Mutwil M, Klie S, Tohge T, Giorgi FM, Wilkins O, Campbell MM et al. (2011) PlaNet: combined sequence and expression comparisons across plant networks derived from seven species. The Plant cell 23 (3): 895–910. pmid:21441431
- 50. Aoki K, Ogata Y, Shibata D (2007) Approaches for extracting practical information from gene co-expression networks in plant biology. Plant & cell physiology 48 (3): 381–390.
- 51. Obayashi T, Kinoshita K (2009) Rank of Correlation Coefficient as a Comparable Measure for Biological Significance of Gene Coexpression. DNA Res 16 (5): 249–260. pmid:19767600