Reference gene selection for quantitative real-time PCR normalization: Application in the caprine mammary gland

Small Ruminant Research 95 (2011) 20–26

Contents lists available at ScienceDirect

Small Ruminant Research journal homepage: www.elsevier.com/locate/smallrumres

Reference gene selection for quantitative real-time PCR normalization: Application in the caprine mammary gland Laurence Finot a,b , Pierre-Guy Marnet a,b,c , Frederic Dessauge a,b,∗ a b c

INRA UMR 1080 Dairy Production, F-35590 Saint Gilles, France Agrocampus UMR 1080 Dairy Production, F-35000 Rennes, France Universite Europeenne de Bretagne, France

a r t i c l e

i n f o

Article history: Received 18 May 2010 Received in revised form 18 August 2010 Accepted 18 August 2010 Available online 20 November 2010 Keywords: Goat Gene expression GeNorm Puberty

a b s t r a c t In dairy animals, gene expression analysis has become increasing key to understand the biological processes occurring in mammary gland development that shape future milk potential. Selecting high-stability reference genes is crucial to interpret real-time qPCR data. This study investigated the expression stability of five top-ranked candidate reference genes in the goat mammary gland through three assays comparing different experimental conditions (physiological states, sample types and experimental treatments). The expression stability of genes including ˇ-actin, glyceraldehyde-3-phosphate dehydrogenase, 18S rRNA, cyclophilin A and ribosomal protein large P0 was analyzed. Normalization for each experimental condition expression data revealed a different reference gene. Nevertheless, in our various assays, genes encoding for ribosomal proteins, 18S rRNA and RPLP0 presented the best expression stability. This result has been confirmed using a combined analysis of stability on the three assays. All genes showed the same distribution within and among the three assays and a different distribution between Ct variability and GeNorm normalization. In addition, the application on Catenin B1 expression using an inappropriate reference gene confirmed erroneous variations in interpretation. To conclude, there is no single ideal reference gene for caprine mammary gland studies and we recommend using a panel of topranked reference genes, including RPLP0, at the beginning of each experiment to validate the most stable(s) gene(s). © 2010 Elsevier B.V. All rights reserved.

1. Introduction The shape of the lactation curve is determined by the number and secretory activity of mammary epithelial cells (Capuco et al., 2001, 2003). A thorough understanding of the mechanisms involved in mammary gland development is essential for increasing milk potential and lactation persistency and to achieve profitability of dairy farms. Many laboratories have investigated mammary gland function at a molecular level based on physiological models to

∗ Corresponding author at: INRA UMR 1080 Dairy Production, F-35590 Saint Gilles, France. Tel.: +33 0223485097. E-mail address: [email protected] (F. Dessauge). 0921-4488/$ – see front matter © 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.smallrumres.2010.08.008

understand the relations between mammary epithelial cell genetics and milk production (Capuco et al., 2001; Marti et al., 1997; Miller et al., 2005; Sorensen et al., 2008). The development of new powerful molecular biology tools allowing studies on RNA, such as real-time quantitative Polymerase Chain Reaction (RT-qPCR), has made it possible to elucidate gene regulation in the mammary gland. RT-qPCR is widely used in all kinds of mRNA quantification studies due to its high sensitivity, good reproducibility and dynamic quantification range. In this technique, numerous factors (quality and integrity of RNA, unspecific PCR products, sampling methods, etc.) are sources of variability and can affect interpretation of RT-qPCR results. To control inter-sample differences due to these factors, normalization of RT-qPCR data is required and ensures an

L. Finot et al. / Small Ruminant Research 95 (2011) 20–26

accurate measurement of gene expression. Actually, the most used method of normalization involves the analysis of target gene expression relative to a reference gene. A good reference gene is assumed to be expressed abundantly at a constant level in most tissues, at all stages of development, without being affected by the experimental treatments. Historically, genes encoding for ␤-actin (ACTB), a cytoskeletal protein, and glyceraldehyde-3-phosphate dehydrogenase (GAPDH), a glycolytic enzyme, have been widely used as reference genes. However, numerous studies have highlighted that the expression of these genes may be modulated during cellular and hormones processes (Revillion et al., 2000). It has become clear that data normalization using unstable reference genes can result in erroneous interpretation (Dheda et al., 2005). Working on mammary gland development in ruminants is hampered by the lack of information on the expression stability of reference genes for RT-qPCR. There are a few published studies on the expression stability of reference genes in cows (De Ketelaere et al., 2006; Goossens et al., 2005; Lisowski et al., 2008; Robinson et al., 2007). Interestingly, Bionaz and Loor (2007) studied the expression stability of reference genes in the bovine mammary gland during the lactation cycle, and identified two genes encoding for ribosomal proteins, Ribosomal Protein S9 (RPS9) and Ribosomal Protein S15 (RPS15) as the most stable genes (Bionaz and Loor, 2007). However, very little is known on goats, and especially on the caprine mammary gland. The objective of the present study was to evaluate the expression stability of five potential reference genes, i.e. ␤actin (ACTB), glyceraldehyde-3-phosphate dehydrogenase (GAPDH), 18S rRNA, cyclophilin A (PPIA) and ribosomal protein large P0 (RPLP0), in goat mammary gland across different experimental conditions to identify a stable reference gene; and to illustrate the impact of choosing an appropriate reference gene by running data normalization on Catenin b1 expression (CNNTB1), a key gene in our research on mammary gland development, using candidate reference genes. 2. Materials and methods All animal experimentation was conducted in accord with the relevant guidelines and licensing requirements of animal care (defined by the French Ministry for Agriculture) and approved by the French National Institute of Agricultural Research INRA).

2.1. Animals and tissue preparation 2.1.1. Assay 1: lactating vs. prepubertal goats Two experiments were conducted on 12 lactating and 24 prepubertal dairy Alpine goats from the experimental farm herd of the French National Institute for Agricultural Research at Le Rheu. All 12 dairy goats were milked twice daily for 2 wk postpartum. Mean stage of lactation was 25 ± 10 d at the beginning of the experiments. Samples of mammary tissue were collected by biopsy (Farr et al., 1996). The 24 prepubertal goats were slaughtered at 7 months of age for mammary gland sampling.

2.1.2. Assay 2: tissue sample vs. digested tissue (cells) Mammary tissues of goats came from two different sample preparations: dissociation of mammary explants by single-cell suspension using enzymatic digestion (see the tissue dissociation section) or directly extracted mammary explants.

21

2.1.3. Assay 3: shammed vs. ovariectomized goats 24 alpine goats were assigned to one of two treatments: shammed (sham) or ovariectomized (ovx). Ovary resection concerned half the goats in the experiment, while the other half was only opened and stitched. Samples of mammary tissue were obtained from animals 1 month after surgery, at autopsy, under general anesthesia with subsequent euthanasia (Rompun i.v. 1 cc; Dolethal, i.v. 25 cc). The goat mammary glands were removed within 20 min from slaughter (Dessauge et al., 2009). 2.2. Sampling Freshly-dissected tissue segments were cut into small pieces (1 g) and immediately frozen in liquid nitrogen for RNA extraction. Small pieces were then stored at −80 ◦ C. 2.3. Tissue dissociation Pieces of fresh tissue (5 g) were immediately washed after collection in vetedine solution, ethyl alcohol 70%, sterile water baths and sterile Hank’s balanced Salt solution (HBSS). The tissues were then ground down minced with a scalpel and transferred to a sterile Erlenmeyer containing 20 ml Tissue Dissociation Solution (HBSS (antibiotic 1×, fungizone 1×, gentamicin 1×, 20 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES) complemented with 10 mg/ml collagenase and hyaluronidase. The tissues were gently dissociated on a rotary shaker at 37 ◦ C for 2 h. After discarding the fat layer, the dissociated tissue was filtered through a nylon filter (0.45 ␮m) and centrifuged at 450 × g for 5 min at 4 ◦ C to pellet the cells. The supernatant was eliminated and the cell pellet was washed twice by resuspending the cells in 15 ml of HBSS followed by a repeat centrifugation step (450 × g for 5 min at 4 ◦ C). The cells were then resuspended in 1 ml sterile phosphate buffer saline (PBS) to determine cell concentration and viability using a ViCell apparatus (Beckman Coulter, Roissy, France). Cells were finally pelleted and stored at −80 ◦ C until analysis. 2.4. RNA extraction and cDNA synthesis Total RNA was extracted from tissue samples (50 mg) using Trizol (Invitrogen Life Technologies, Germany) and RNeasy Mini kit (Qiagen, Courtaboeuf, France) according to the manufacturer’s instructions. Tissue powder was ground in liquid nitrogen with a mortar and pestle and homogenized in 1 ml of Trizol reagent followed by aspiration of the mixture 10 times through a 21-gauge syringe needle. After 5 min incubation at room temperature, 200 ␮l of chloroform were added to each sample. The mixture was centrifuged at 12,000 × g for 15 min at 4 ◦ C and the upper aqueous phase containing total RNA was recovered. The RNeasy Mini kit was used to precipitate and purify the RNA. A DNase digestion step was carried out to remove genomic DNA using an RNase-free DNase kit (Qiagen). The RNA was eluted from spin columns with 30 ␮l of sterile RNase-free water. Total RNA concentration and the 260/280 nm and 260/230 nm absorbance ratios were measured using a NanoDrop® ND-1000 spectrophotometer (NanoDrop Technologies, Wilmington, DE, USA). RNA quality was determined on an Agilent 2100 Bioanalyzer (Agilent Technologies, Massy, France) by RNA profile and measurement of the RNA Integrity Number (RIN) using Agilent 2100 Expert Software, version B.02 (Agilent Technologies). Then, 1 ␮g of RNA was reverse-transcribed at 42 ◦ C for 1 h using a ThermoScript RT-PCR System (Invitrogen, Germany) according to the manufacturer’s protocol for oligo(dT) 15-primed cDNA synthesis. Prior to use in RT-qPCR, cDNA was diluted in diethyl pyrocarbonate (DEPC) water (1:50). 2.5. Quantitative PCR Real-time PCR analysis was performed on an ABI Prism 7000 sequence detection system (Applied Biosystem, Courtaboeuf, France) using SYBR Green PCR master mix. Reactions were carried out in 96-well optical reaction plates with 200 nM of each specific primer and 2.5 ␮l of diluted cDNA. Each sample was run in technical triplicates with a non-templated control included. The RT-qPCR program consisted of an initial denaturation step at 95 ◦ C for 10 min followed by 40 cycles of denaturation at 95 ◦ C for 10 s and a combined primer annealing–extension step at 60 ◦ C for 1 min in which fluorescence was measured. A melting curve was produced after completion

22

L. Finot et al. / Small Ruminant Research 95 (2011) 20–26

Table 1 Information on genes and primers used for real-time qPCR including gene name and function, GenBank accession numbers used for primer design, forward and reverse primer sequences (indicated as “F” and “R”, respectively) and size of amplicon. Gene symbol

Full gene name

Reference genes GAPDH Glyceraldehyde-3phosphate dehydrogenase

GenBank accession number

Function

Primers (5 –3 )

Amplicon size (bp)

BC 102589.1

Oxidoreductase in glucose metabolism

F: GTCTTCACTACCATGGAGAAGG R: TCATGGATGACCTTGGCCAG F: TCATCACCATCGGCAATGAG R: GGTAGTTTCGTGAATGCCGC F: CAAATTACCCACTCCCGACCC R: AATGGATCCTCGCGGAAGG F: GGATTTATGTGCCAGGGTGGTGA R: CAAGATGCCAGGACCTGTATG F: CAACCCTGAAGTGCTTGACAT R: AGGCAGATGGATCAGCCA

197

ACTB

␤-Actin

BC 102948.1

Cytoskeletal structural protein

18S rRNA

18S ribosomal RNA

DQ 066896.1

Proteins synthesis

PPM

Peptidylpyrolyl isomerase A (cyclophilin A)

AY 247029.1

Immunoregulation

RPLP0

Ribosomal protein large, P0

NM 001012682.1

Protein biosynthesis

Catenin (cadherin-associated protein), ␤1

NM 001076141

Wnt/Ctnnb1 signaling pathways

Target genes CTNNB1

F: GGATGTGGATACCACCCAAG R: CCCTCATCTAGCGTCTCAGG

95

114

120

227

153

of the thermal PCR program to check the presence of one gene-specific peak and the absence of primer dimer. Data, raw cycle threshold (Ct), obtained from ABI Prism 7000 Software, version 1.1 (Applied Biosystems) were used for the comparative Ct method (delta delta cycle threshold (DDCT)) using specific efficiencies (Pfaffl, 2001).

tion of a goat mammary gland template. Confirmation of PCR product length was obtained through melting curve analysis and 1.5% agarose gel electrophoresis

2.6. Evaluation of reference gene expression stability

Data of relative gene expression (DDCt values) for CNNTB1 expression were expressed as the mean ± standard error (SE). The data were subjected to an analysis of variance (ANOVA) with a Student’s t-test to compare the mean values resulting from both treatments (sham and ovx). Effects were considered to be statistically significant at P < 0.05. Statistical analyses were carried out using the SAS MIXED procedure of the SAS software package (SAS Institute Inc., Cary, NC).

The software GeNormTM , version 3.4 (Visual Basic application tool for Microsoft Excel) was used to calculate the stability of the candidate reference genes. For each candidate reference gene, the GeNorm application generates a stability measure value (M) which can be used to rank the reference genes. Genes with the lowest M values offer the most stable expression. In addition, by successive exclusion of the least stable genes and recalculation of the new M values, the two most stable reference genes (pairwise) are finally determined, and the reference genes are ranked according to their M value (Vandesompele et al., 2002).

2.8. Statistical analysis

3. Results and discussion 3.1. RT-qPCR efficiencies

2.7. Primer design and testing The potential reference genes chosen and evaluated in this study were: ˇ-actin (ACTB), glyceraldehyde-3P-dehydrogenase (GAPDH), cyclophilin A (PPIA), 18S ribosomal RNA (18S rRNA), and ribosomal protein large P0 (RPLP0). Gene sequences for primer design were based on bovine mRNA sequences and were obtained from the National Center for Biotechnology Information Genebank. Only primers for RPLPO were taken from the literature (Robinson et al., 2007). Primer sequences were determined in the coding regions of the target mRNAs using Primer Express software v 2.0 (Applied Biosystems) with selection settings (size 18–22 base pairs (bp); annealing temperature (Tm ) 57–63 ◦ C; GC percentage 50–60%) and a PCR product length of 75–250 bp. To prevent genomic DNA amplification, primer pairs were selected to bind to different exons. Then, the specificity of each primer pair was verified using Basic Local Alignment Search Tool (Altschul et al., 1990). Table 1 gives sequence data for the primers used for real-time PCR. Efficiency of RT-qPCR, slope values and determination coefficients (R2 ) were determined for each primer (Table 2). Amplification efficiency, E, was calculated from the slope of the standard curve using the formula E = 10(−1/slope) . The standard curve was generated using a 10-fold dilu-

First, we focused on RT-qPCR product size. All our RTqPCR products were less than 250 bp, which also makes them easily distinguishable it from any primer-dimers that Table 2 Real-time PCR efficiency data for reference and target genes. Linear regression analyze of standard curve are characterized by the determination coefficient (R2 ) and the slopes value. Amplification efficiency is calculated by the equation E = 10(−1/slope) . Gene symbol Reference genes GAPDH ACTB 18S rRNA PPIA RPLP0 Target genes CTNNB1

Slope

R2

Efficiency

−3.3595 −3.4478 −3.3908 −3.4023 −3.4405

0.993 0.989 0.990 0.983 0.993

1.98 1.95 1.97 1.97 1.95

−3.3532

0.997

1.99

L. Finot et al. / Small Ruminant Research 95 (2011) 20–26

23

Table 3 Mean, standard error (SE), minimum (min) and maximum (max) of A260/A280 and A260/A230 ratios measurement using nanodrop ND-40 as indicators of RNA purity. Mean, standard error (SE), minimum (min) and maximum (max) RNA integrity number (RIN) determination by the Bioanalyzer 2100 as indicator of RNA integrity. Factor

Mean SE Min Max

Assay 1 Lactating vs. young goats

Assay 2 Tissue vs. cells

Assay 3 Sham vs. ovariectomized goats

260/280

260/230

RIN

260/280

260/230

RIN

260/280

260/230

RIN

1.92 0.11 1.72 2.08

2.23 0.11 1.88 2.34

8.94 0.52 8.00 9.60

1.99 0.03 1.94 2.08

2.23 0.11 1.88 2.34

8.86 0.59 7.40 9.30

2.00 0.04 1.94 2.08

2.19 0.15 1.72 2.34

8.87 0.51 7.90 9.40

might form. In this study, primer efficiencies were calculated for the genes using serial diluted cDNA of goat mammary gland (Table 2). Results showed that the determination coefficient (R2 ) of the linear regression models for the five reference genes ranged from 0.98 to 1. Based on the slopes of standard curves, the amplification efficiencies ranged from 1.95 to 1.99. In addition, specificity was investigated by gel electrophoresis, confirming an expected single band at the reported amplicon size, as shown in Table 1. All melting curve analyses described a single distinctive peak, indicating the formation of one specific amplicon (data not shown). A successful RT-qPCR reaction requires efficient and specific amplification of the product (Fleige and Pfaffl, 2006). 3.2. RNA quality assessment Using intact and pure RNA is critical for a successful RT-qPCR, and therefore the determination of RNA quality

should be monitored prior to downstream analysis. Sample quality was checked by determining RNA integrity and purity (Table 3). RNA integrity was given by using the Agilent Bioanalyzer 2100 to measure the RNA Integrity Number (RIN). The average RIN in this study was higher than 8.8 (Assay 1: 8.87 ± 0.52, Assay 2: 8.86 ± 0.59, Assay 3: 8.87 ± 0.51). Sample purity was subsequently evaluated using the NANODROP ND-40 to measure the absorbance ratios A260/A280 and A260/A230. A260/A280 ratio always comprised between 1.9 and 2 whereas A260/A230 ratio was greater than 2.19 (range 2.19–2.23) for the three assays. Based on these results, all samples were pure and free from proteins and phenolic pollutants. Globally, RNA obtained from frozen tissue showed good quality and integrity (Table 3). It is now well established that the accuracy of gene expression is influenced by RNA purity and integrity, which govern the reliability and reproducibility of RTqPCR. Working with low-quality RNA may compromise the experimental results (Fleige and Pfaffl, 2006; Fleige et al.,

Table 4 Expression stability values (M) and rankings of the five reference genes determined by the GeNorm software in (A) lactating and prepubertal goats (Assay 1), (B) tissue and enzyme-digested tissue (cells) from goats (Assay 2) and (C) shammed and ovariectomized goats (Assay 3). (A) Assay 1 rank

All data

Lactating goats

Young goats

Gene name

M value

Gene name

M value

Gene name

M value

1 2 3 4 5 Pairwise

18S rRNA PPIA GAPDH RPLP0 ACTB RPLPO/18S rRNA

0.464 0.482 0.498 0.507 0.536 0.366

PPIA 18S rRNA RPLP0 ACTB GAPDH PPIA/18S rRNA

0.391 0.416 0.497 0.503 0.528 0.331

RPLP0 PPIA 18S rRNA GAPDH ACTB RPLPO/18S rRNA

0.434 0.435 0.444 0.445 0.543 0.274

(B) Assay 2 rank

All data Gene name

M value

Gene name

M value

Gene name

M value

1 2 3 4 5 Pairwise

RPLP0 GAPDH 18S rRNA PPIA ACTB RPLPO/18S rRNA

0.544 0.568 0.571 0.624 0.656 0.404

RPLP0 PPIA 18S rRNA GAPDH ACTB RPLPO/18S rRNA

0.434 0.435 0.444 0.445 0.543 0.274

RPLP0 PPIA GAPDH 18S rRNA ACTB PPIA/RPLPO

0.477 0.514 0.596 0.61 0.652 0.429

(C) Assay 3 rank

1 2 3 4 5 Pairwise

Tissue

All data

Enzyme-digested tissue (cells)

Shammed goats

Ovariectomized goats

Gene name

M value

Gene name

M value

Gene name

M value

PPIA GAPDH RPLP0 ACTB 18S rRNA PPIA/GAPDH

0.603 0.692 0.702 0.787 0.81 0.497

RPLP0 PPIA 18S rRNA GAPDH ACTB RPLPO/18S rRNA

0.434 0.435 0.444 0.445 0.543 0.274

PPIA RPLP0 GAPDH ACTB 18S rRNA PPIA/GAPDH

0.642 0.705 0.805 0.858 0.963 0.642

24

L. Finot et al. / Small Ruminant Research 95 (2011) 20–26

Fig. 1. Distribution of real-time PCR cycle threshold (Ct) values for the five reference genes investigated in caprine mammary gland. Each boxplot is based on Ct mean value from (A) lactating and prepubertal goat (Assay 1), (B) tissue and enzyme-digested tissue (cells) (Assay 2), and (C) shammed and ovariectomized goats (Assay 3). Boxes represent the lower and upper quartile ranges, medians are represented by black dashes bisecting the box, and the bottom and top “whiskers” represent minimum and maximum values, respectively.

2006). It has been shown that there is a significant negative relationship between RIN and Ct values (Fleige and Pfaffl, 2006) and authors pointed out that data normalization by a reference gene is RIN-dependant. Assessing RNA quality is a prerequisite before any gene expression analysis and subsequent publication according to recent recommendations (Bustin et al., 2009). 3.3. Gene expression stability analysis using GeNorm We selected classical constitutive genes commonly used in RT-qPCR, encoding for the cytoskeletal protein ␤-actin (ACTB), a glucogenesis enzyme (GAPDH) and the ribosomal protein (18S rRNA), in order to evaluate their expres-

sion stability. In addition, Ribosomal protein (RPLPO) and cyclophilin A (PPIA) genes were chosen for their increasing use as reference genes in ruminant mammary gland studies (Boutinaud et al., 2004; Robinson et al., 2007). We tested the expression stability of our reference genes by comparing three parameters: physiological stage, sample type and surgical treatment. For each assay, we analyzed (separately and pooled) the expression stability of the reference genes for the sub-groups constituting the assays using GeNorm, a powerful normalization software. Assay 1 was designed to compare the expression stability of reference genes between two different physiological stages. Thus, we compared mammary gland data from prepubertal and milking goats. The results presented in

L. Finot et al. / Small Ruminant Research 95 (2011) 20–26

25

Table 5 Expression stability values (M) and ranking of the five reference genes determined by the GeNorm software for all data of the 3 assays (lactating and prepubertal goats (Assay 1), tissue and enzyme-digested tissue (cells) (Assay 2) and shammed and ovariectomized goats (Assay 3)). All data rank

1 2 3 4 5 Pairwise

Combined assays 1, 2 and 3 Gene name

M value

RPLP0 18S rRNA ACTB PPIA GAPDH RPLPO/18S rRNA

1.414 1.505 1.612 1.677 2.256 0.476

Table 4A showed that 18S rRNA gene is the most stable for all animals. We then analyzed the two sub-groups separately; the genes showing higher stability in the mammary gland of prepubertal and lactating goats were RPLP0 and PPIA, respectively. In the Assay 2, we compared data from enzyme-digested tissue (single-cell suspension) and tissue explants from the same animal at an equivalent physiological stage (prepuberty) (Table 4B). RPLP0 gene was found to be the most stable gene independently from cell and tissue type. Assay 3 focused on experimental conditions by comparing data on mammary gland tissue from ovariectomized goats and shammed goats (Table 4C). We obtained the best stability for PPIA gene using pooled data. When we analyzed the two sub-groups separately, the most stable genes in the mammary gland from ovariectomized and sham goats were PPIA and RPLP0, respectively. In addition, GeNorm analysis was run pooling all the data (3 assays) to identify the most suitable candidate gene(s) (Table 5). In this multicriterion analysis, RPLP0 and 18S rRNA are highlighted as the most stable genes. We noted that using a combined approach, pairwise RPLP0/18S rRNA seems to be an optimal internal control for caprine mammary gland gene expression profiling. Globally, in our assays, genes encoding for ribosomal proteins, i.e. 18S rRNA and RPLP0, presented the best expression stability. Another study reinforced the use of ribosomal protein genes such as RPLP0 as internal controls for quantitative applications of RT-qPCR (Lyng et al., 2008). Surprisingly, GAPDH gene, which many studies have found not very stably expressed (Etschmann et al., 2006; Nygard et al., 2007) was often well-ranked in our study. In summary, the genes presenting the highest stability were different in each experimental assay within our 3 assays in mammary gland studies. Taken together, these results reinforced and confirmed the necessity to validate the expression stability of a candidate reference gene from a selected set of potential reference genes for each experimental design. Using a single reference gene in RT-qPCR normalization without validating it stability expression is now unviable. 3.4. Absolute expression levels of reference genes The abundance and cycle threshold (Ct) variability of the five reference genes used in this study is illustrated in box plot graphs (Fig. 1). Ct dispersion was illustrated through

Fig. 2. Catenin B1 relative expression in ovariectomized (black bars) and shammed (control) goats (shaded bars) normalized to reference genes from most stable (left) at least stable (right). Error bars represent SE and p values indicate statistical significance.

whiskers representing minimum and maximum values. Results confirmed that 18S rRNA and RPLP0 are highly abundant whereas ˇ-actin showed the lowest abundance. This visual representation highlights that all genes shared the same distribution within and among the three assays. However, normalization data by GeNorm showed differential distribution of these genes, confirming the necessity of this analysis step. 3.5. Normalization of Catenin B1 (CNNTB1) expression relative to different reference genes The accuracy of gene expression analysis with RT-qPCR is highly dependent on the reference gene used to normalize data. To illustrate this point, we normalized the data (DDCt) of CNNTB1 expression in the mammary gland with some previous candidate reference genes (Fig. 2) in order to compare normalization strategy. CNNTB1 expression was chosen for its involvement in the mammary gland development in dairy goats. Indeed, during the mammary gland development Wnt/␤-catenin signaling pathway is one of the pivotal pathways for cellular dynamics (proliferation and apoptosis) and differentiation (Dessauge et al., 2009; Teuliere et al., 2004). The DDCT results normalized with the pairwise of the two most stable reference genes (pairwise GADPH/PPIA) or by GAPDH, a well-ranked reference gene, determined by GeNorm showed a significant decrease (p ≤ 0.05) in CNNTB1 transcripts levels in the mammary gland of ovariectomized goats. However, when data were normalized with 18S rRNA, the least stable gene, (M value = 0.81), CNNTB1 expression tended to decrease (0.05 ≤ p ≤ 0.1). Next, we normalized CNNTB1 expression data with RPLPO, which has an acceptable M value (=0.702) but not well-ranked, and the data showed no difference in CNNTB1 expression between ovariectomized and shammed goats. Through this analysis, we underlined that the difference in CNNTB1 expression between shammed and ovariectomized goats had been hidden by using the wrong reference gene, leading to erroneous interpretation of results. If we used RPLPO or 18S rRNA as reference genes,

26

L. Finot et al. / Small Ruminant Research 95 (2011) 20–26

we could have concluded that CNNTB1 expression was not different in the mammary gland of ovariectomized goats. In our study, normalization strategy consisting in using the geometric average of multiple reference genes (Etschmann et al., 2006; Vandesompele et al., 2002) gave more accurate results. 4. Conclusion Hence, our results demonstrate the importance of validating reference genes under experimental conditions in ruminant mammary gland studies. Through this study, we recommend using a comprehensive panel of top-ranked reference genes including ribosomal protein genes as RPLP0 and 18S rRNA which are to be relevant in caprine mammary gland gene expression profiling. We concluded that choice of the reference gene in data normalization impacts on interpretation of the results, as it can lead to deep misinterpretation. Acknowledgements Authors would like to thank Michel Chorho, Eric Siroux and Jean-Marc Aubry for assisting in animal handling and knowledge about goats rearing. References Altschul, S.F., Gish, W., Miller, W., Myers, E.W., Lipman, D.J., 1990. Basic local alignment search tool. J. Mol. Biol. 215, 403–410. Bionaz, M., Loor, J.J., 2007. Identification of reference genes for quantitative real-time PCR in the bovine mammary gland during the lactation cycle. Physiol. Genomics 29, 312–319. Boutinaud, M., Shand, J.H., Park, M.A., Phillips, K., Beattie, J., Flint, D.J., Allan, G.J., 2004. A quantitative RT-PCR study of the mRNA expression profile of the IGF axis during mammary gland development. J. Mol. Endocrinol. 33, 195–207. Bustin, S.A., Benes, V., Garson, J.A., Hellemans, J., Huggett, J., Kubista, M., Mueller, R., Nolan, T., Pfaffl, M.W., Shipley, G.L., Vandesompele, J., Wittwer, C.T., 2009. The MIQE guidelines: minimum information for publication of quantitative real-time PCR experiments. Clin. Chem. 55, 611–622. Capuco, A.V., Ellis, S.E., Hale, S.A., Long, E., Erdman, R.A., Zhao, X., Paape, M.J., 2003. Lactation persistency: insights from mammary cell proliferation studies. J. Anim. Sci. 81 (Suppl. 3), 18–31. Capuco, A.V., Wood, D.L., Baldwin, R., Mcleod, K., Paape, M.J., 2001. Mammary cell number, proliferation, and apoptosis during a bovine lactation: relation to milk production and effect of bST. J. Dairy Sci. 84, 2177–2187. De Ketelaere, A., Goossens, K., Peelman, L., Burvenich, C., 2006. Technical note: validation of internal control genes for gene expression analysis in bovine polymorphonuclear leukocytes. J. Dairy Sci. 89, 4066–4069.

Dessauge, F., Finot, L., Wiart, S., Aubry, J.M., Ellis, S.E., 2009. Effects of ovariectomy in prepubertal goats. J. Physiol. Pharmacol. 60 (Suppl. 3), 127–133. Dheda, K., Huggett, J.F., Chang, J.S., Kim, L.U., Bustin, S.A., Johnson, M.A., Rook, G.A., Zumla, A., 2005. The implications of using an inappropriate reference gene for real-time reverse transcription PCR data normalization. Anal. Biochem. 344, 141–143. Etschmann, B., Wilcken, B., Stoevesand, K., von der, S.A., Sterner-Kock, A., 2006. Selection of reference genes for quantitative real-time PCR analysis in canine mammary tumors using the GeNorm algorithm. Vet. Pathol. 43, 934–942. Farr, V.C., Stelwagen, K., Cate, L.R., Molenaar, A.J., McFadden, T.B., Davis, S.R., 1996. An improved method for the routine biopsy of bovine mammary tissue. J. Dairy Sci. 79, 543–549. Fleige, S., Pfaffl, M.W., 2006. RNA integrity and the effect on the real-time qRT-PCR performance. Mol. Aspects Med. 27, 126–139. Fleige, S., Walf, V., Huch, S., Prgomet, C., Sehm, J., Pfaffl, M.W., 2006. Comparison of relative mRNA quantification models and the impact of RNA integrity in quantitative real-time RT-PCR. Biotechnol. Lett. 28, 1601–1613. Goossens, K., Van Poucke, M., Van Soom, A., Vandesompele, J., Van Zeveren, A., Peelman, L.J., 2005. Selection of reference genes for quantitative real-time PCR in bovine preimplantation embryos. BMC Dev. Biol. 5, 27. Lisowski, P., Pierzchala, M., Goscik, J., Pareek, C.S., Zwierzchowski, L., 2008. Evaluation of reference genes for studies of gene expression in the bovine liver, kidney, pituitary, and thyroid. J. Appl. Genet. 49, 367– 372. Lyng, M.B., Laenkholm, A.V., Pallisgaard, N., Ditzel, H.J., 2008. Identification of genes for normalization of real-time RT-PCR data in breast carcinomas. BMC Cancer 8, 20. Marti, A., Feng, Z., Altermatt, H.J., Jaggi, R., 1997. Milk accumulation triggers apoptosis of mammary epithelial cells. Eur. J. Cell Biol. 73, 158–165. Miller, P.J., Finucane, K.A., Hughes, M., Zhao, F.Q., 2005. Cloning and expression of bovine glucose transporter GLUT12. Mamm. Genome 16, 873–883. Nygard, A.B., Jorgensen, C.B., Cirera, S., Fredholm, M., 2007. Selection of reference genes for gene expression studies in pig tissues using SYBR green qPCR. BMC Mol. Biol. 8, 67. Pfaffl, M.W., 2001. A new mathematical model for relative quantification in real-time RT-PCR. Nucleic Acids Res. 29, e45. Revillion, F., Pawlowski, V., Hornez, L., Peyrat, J.P., 2000. Glyceraldehyde3-phosphate dehydrogenase gene expression in human breast cancer. Eur. J. Cancer 36, 1038–1042. Robinson, T.L., Sutherland, I.A., Sutherland, J., 2007. Validation of candidate bovine reference genes for use with real-time PCR. Vet. Immunol. Immunopathol. 115, 160–165. Sorensen, B.M., Chris, K.E., Murdoch, G.K., Keating, A.F., Cruz-Hernandez, C., Wegner, J., Kennelly, J.J., Okine, E.K., Weselake, R.J., 2008. Effect of CLA and other C18 unsaturated fatty acids on DGAT in bovine milk fat biosynthetic systems. Lipids 43, 903–912. Teuliere, J., Faraldo, M.M., Shtutman, M., Birchmeier, W., Huelsken, J., Thiery, J.P., Glukhova, M.A., 2004. Beta-catenin-dependent and independent effects of DeltaN-plakoglobin on epidermal growth and differentiation. Mol. Cell. Biol. 24, 8649–8661. Vandesompele, J., De Preter, K., Pattyn, F., Poppe, B., Van Roy, N., De Paepe, A., Speleman, F., 2002. Accurate normalization of real-time quantitative RT-PCR data by geometric averaging of multiple internal control genes. Genome Biol. 3 (RESEARCH0034).