Abstract

Measurement of tissue mRNA transcript levels is critically dependent upon the normalization strategy used. For real-time polymerase chain reaction (PCR) (RT–PCR) this commonly depends upon identification of stable (non-variable) endogenous reference genes, with housekeeping genes (HKGs) the most commonly used. In this report we describe the use of an external mRNA standard to identify stable HKGs in the human fetal gonad. Total RNA was extracted from second trimester human fetal gonads and a standard amount of luciferase mRNA was added at the start of the extraction process. Levels of luciferase were then measured relative to each of seven HKGs (SDHA, TBP, B2M, PMM1, SFRS4, HMBS and UBC) by RT–PCR. When normalized to tissue weight, HKG expression was constant across fetal ages. Measurement of overall variation in transcript expression showed that PMM1 was the most stable HKG in the ovary while B2M was most stable in the testis. Re-analysis of the data using GeNorm and NormFinder algorithms showed that two of the top three most stable HKGs were the same using all three methods. This study describes a method for identification of endogenous, stable reference genes for RT–PCR studies of transcript expression levels which is objective and not dependent on prior assumptions of HKG expression. This technique is likely to be applicable to most tissues and, in this case, identifies suitable HKGs for studies into human gonadal development.

Introduction

The relevance and meaning of much biochemical data is dependent upon the normalization strategy and this is particularly the case for mRNA quantification using real-time polymerase chain reaction (PCR) (RT–PCR). Most commonly, normalization is to an endogenous reference gene and the overwhelming majority of the reference genes used for normalization are housekeeping genes (HKGs). These HKGs carry out essential cell maintenance functions and are used as reference genes on the assumption that they will show stable expression—i.e. transcript levels will not show developmental-, treatment- or pathological-regulation. It has been shown, however, that many of the most commonly used HKGs do not fulfil these criteria (Huggett et al., 2005) and, indeed, an HKG which is universally stable across all tissues almost certainly does not exist. To identify the optimal normalization strategy for a particular tissue or cell type, it is necessary, therefore, to examine the stability of a number of reference genes using algorithms such as GeNorm and NormFinder (Vandesompele et al., 2002; Andersen et al., 2004). These algorithms make certain assumptions, however, that may not always be valid. For example, the most commonly used algorithm, GeNorm, is based on the hypothesis that the most stable HKGs will show the least variation relative to each other (Vandesompele et al., 2002). Low relative variation between HKGs may arise, however, for reasons such as co-regulation and may not necessarily indicate high tissue stability (Andersen et al., 2004; Ellefsen et al., 2008). An alternative strategy for normalization of RT–PCR data is to use an external ‘foreign’ mRNA as standard, added at the stage of mRNA extraction (Baker and O'Shaughnessy, 2001; Smith et al., 2003; Johnson et al., 2005; Ellefsen et al., 2008). This provides an objective measure of tissue transcript levels relative to the mass of the starting tissue without assumptions of reference gene stability. This technique is not always practical, however, if the extracted RNA is archival, tissue mass or cell number is unknown, or there has been partial RNA degradation before extraction. Nevertheless, the external standard method can be used with a characterized tissue set to assess the relative stability of potential HKGs in that tissue (O'Shaughnessy et al., 2002; Ellefsen et al., 2008). In this report, we have expanded the use of the external standard to rate the stability of HKGs in human fetal gonads without assumptions of co-variability or reliance on model-based estimates of expression variance (Vandesompele et al., 2002; Andersen et al., 2004). This approach has general applicability to other tissues and systems and can be used to identify the most appropriate HKG for studies in which an endogenous reference gene is necessary.

Materials and Methods

Tissues

The collection of fetal material was approved by the NHS Grampian Research Ethics Committees (REC 04/S0802/21). For a detailed explanation, see O'Shaughnessy et al. (2007). Women seeking elective, medical terminations of pregnancy were recruited with full written, informed consent by nurses working independently at Aberdeen Pregnancy Counseling Service. There was no change in patient treatment or care and women were able to withdraw from the study at any point. Only normally progressing pregnancies from women over 16 years of age and between 11 and 21 weeks of gestation were collected. Fetuses were transported to the laboratory within 30 min of delivery. The gonads were weighed, snap-frozen in liquid nitrogen and stored at −85°C. For this study gonads were collected from 22 male and 22 female fetuses.

RNA extraction and reverse transcription

To extract RNA, whole testes or ovaries were homogenized in Trizol (Invitrogen Ltd, Paisley, UK) according to the manufacturer's protocol. Immediately before homogenization, an aliquot of luciferase mRNA (5 ng, Promega UK, Southampton, UK) was added to the Trizol and extracted along with the endogenous RNA. The extracted RNA was stored in liquid N2. Nucleic acid concentration and quality was determined using a NanoDrop (Fisher Scientific UK Ltd, Loughborough, Leicestershire, UK) with expected 260/280 ratios between 1.9 and 2.1. First-strand cDNA was synthesized from 2 µg of total RNA using random hexamers and Moloney murine leukemia virus reverse transcriptase (Superscript III, Life Technologies, Paisley, UK) as previously described (Rajapaksha et al., 1997).

Real-time PCR

The SYBR green method was used for RT– PCR in a 96-well plate format using a Stratagene MX3000 cycler. Reactions contained 5 μl 2 × SYBR mastermix (Agilent Technologies, Wokingham, UK), primer (100 nM) and template in a total volume of 10 µl. At the end of the amplification phase, a melting curve analysis was carried out on the products formed. No template controls were run on each plate as appropriate for the primer sets under study. All primers were designed by Primer Express 2.0 (Applied Biosystems, Warrington, UK) with parameters set for amplicon length to be between 160 and 200 bp with a Tm less than 85°C and primer Tms to be between 64 and 66°C (Baker and O'Shaughnessy, 2001). All primers were designed so that the amplicon crossed an intron and genomic DNA would not amplify. This was tested for each primer pair using samples containing genomic DNA and with a positive control designed to amplify genomic DNA. The primers used are described in Table I. The amplification efficiencies of the primer pairs used were between 94 and 102% as determined using sample dilution curves.

Table 1

Primers.

Gene Genbank Forward primer Reverse primer 
TATA box binding protein (TBP) nm_003194 aggaaaaaattgaatagtgagacgagttcca tggactaaagatagggattccgggagt 
Phosphomannomutase 1 (PMM1) nm_002676 aacatctcgcccatcggcc tcaaagctgatcatgcctcctcg 
Hydroxymethylbilane synthase (HMBS) nm_000190 gctggcacaaccgggtgg ttcagcgatgcagcgaagca 
Ubiquitin C (UBC) nm_021009 ggcggtgaacgccgatgatt agggtgatggtcttaccagtcagagtctt 
Beta-2-microglobulin (B2M) nm_004048 tcgctccgtggccttagctg caatgtcggatggatgaaacccag 
Succinate dehydrogenase (SDHA) nm_004168 acgtcacgaaggagccgatcc atgtaccgaggcacaggcgg 
Splicing factor, arginine/serine-rich 4 (SFRS4) nm_005626 cgggaagatcctggaggtgga cgtaactgccatctcgccgtg 
Gene Genbank Forward primer Reverse primer 
TATA box binding protein (TBP) nm_003194 aggaaaaaattgaatagtgagacgagttcca tggactaaagatagggattccgggagt 
Phosphomannomutase 1 (PMM1) nm_002676 aacatctcgcccatcggcc tcaaagctgatcatgcctcctcg 
Hydroxymethylbilane synthase (HMBS) nm_000190 gctggcacaaccgggtgg ttcagcgatgcagcgaagca 
Ubiquitin C (UBC) nm_021009 ggcggtgaacgccgatgatt agggtgatggtcttaccagtcagagtctt 
Beta-2-microglobulin (B2M) nm_004048 tcgctccgtggccttagctg caatgtcggatggatgaaacccag 
Succinate dehydrogenase (SDHA) nm_004168 acgtcacgaaggagccgatcc atgtaccgaggcacaggcgg 
Splicing factor, arginine/serine-rich 4 (SFRS4) nm_005626 cgggaagatcctggaggtgga cgtaactgccatctcgccgtg 

Data analysis

The threshold cycle (Cq) value for each sample was determined with each primer pair in duplicate. Levels of luciferase mRNA in each sample were then expressed relative to each HKG using the ΔCq method (2Cq(HKG)-Cq(luciferase)). The CV for each HKG was determined as the standard deviation/mean of all samples. To determine whether there were age-dependent changes in HKG expression, the data were analysed by linear regression using Prism (GraphPad, La Jolla, CA, USA), with P < 0.05 considered significant. Data were also analysed using GeNorm v3.5 and NormFinder algorithms as previously described (Vandesompele et al., 2002; Andersen et al., 2004).

Results

The HKGs that were assessed in this study were TATA box binding protein (TBP), phosphomannomutase 1 (PMM1), hydroxymethylbilane synthase (HMBS), ubiquitin C (UBC), beta-2-microglobulin (B2M), succinate dehydrogenase (SDHA) and splicing factor, arginine/serine-rich 4 (SFRS4). Levels of luciferase mRNA in extracted gonadal tissue were determined relative to each HKG by RT–PCR and the untransformed data, expressed as a function of age, are shown in Supplementary data, Figs 1 and 2. The weight of both testes and ovaries increased throughout the second trimester (Supplementary data, Figs 1 and 2) and, since a standard amount of luciferase mRNA had been added to each gonad at the extraction phase, the amount of luciferase, relative to each HKG, decreased significantly with age. The only exceptions to this were UBC in the testis and HMBS in the ovary, relative to which luciferase levels did not change significantly with age. To correct for tissue mass, the data shown in the Supplementary figures were multiplied by the weight of each individual gonad. Following this correction, and with one exception, the amount of luciferase measured in each gonad did not change significantly with age, relative to each HKG, throughout the second trimester (Figs 1 and 2). The one exception was when luciferase was expressed relative to HMBS in the ovary. In this case there was a significant increase in measured luciferase with age (Fig. 2), which is likely to be due to an actual decrease in HMBS with age. These data show that, with the exception of HMBS in the ovary, the levels of each HKG were constant throughout the second trimester, relative to the mass of the tissue.

Figure 1

Luciferase mRNA relative to seven different HKGs in the fetal human testis. Data show luciferase mRNA measured relative to each HKG in testes of different ages. The data have been corrected for tissue mass by dividing each data point by the relevant testis weight. Each point represents an individual testis and the significance of age-dependent changes was determined by linear regression. The dotted line is the mean value and the solid line is the regression line. The regression slope was not significantly different to zero for any HKG. The range of the Y-axis has been set at 6 times the mean value.

Figure 1

Luciferase mRNA relative to seven different HKGs in the fetal human testis. Data show luciferase mRNA measured relative to each HKG in testes of different ages. The data have been corrected for tissue mass by dividing each data point by the relevant testis weight. Each point represents an individual testis and the significance of age-dependent changes was determined by linear regression. The dotted line is the mean value and the solid line is the regression line. The regression slope was not significantly different to zero for any HKG. The range of the Y-axis has been set at 6 times the mean value.

Figure 2

Luciferase mRNA relative to seven different HKGs in the fetal human ovary. Data show luciferase mRNA measured relative to each HKG in ovaries of different ages. The data have been corrected for tissue mass by dividing each data point by the relevant ovarian weight. Each point represents an individual ovary and the significance of age-dependent changes was determined by linear regression. The dotted line is the mean value and the solid line is the regression line. The regression slope was not significantly different to zero for any HKG apart from HMBS. The range of the Y-axis has been set at 6 times the mean value.

Figure 2

Luciferase mRNA relative to seven different HKGs in the fetal human ovary. Data show luciferase mRNA measured relative to each HKG in ovaries of different ages. The data have been corrected for tissue mass by dividing each data point by the relevant ovarian weight. Each point represents an individual ovary and the significance of age-dependent changes was determined by linear regression. The dotted line is the mean value and the solid line is the regression line. The regression slope was not significantly different to zero for any HKG apart from HMBS. The range of the Y-axis has been set at 6 times the mean value.

To determine which HKG showed the most stability between individual gonads, the CV for each HKG was determined as the standard deviation/mean using data normalized to gonad weight (Figs 1 and 2). The HKGs with the lowest CVs (i.e. most stable) were B2M in the testis and PMM1 in the ovary (Figs 1–3). To determine whether the overall CV was reduced by combining data from different HKGs, the results from each HKG were combined in the order of stability (i.e. most stable HKG, combined with second most stable, etc.). The results in Fig. 3 show that combining the most stable HKGs slightly reduced the CV in the testis but had no effect in the ovary. For both ovary and testis, addition of the least stable HKGs led to an increase in CV.

Figure 3

CV for each HKG in the testis and the ovary (A) and effect of combining data from different HKGs on the overall CV (B). Data in (A) are taken from Figs 1 and 2. To combine data in (B) the geometric mean of different HKGs for each sample was calculated and used to determine the CV.

Figure 3

CV for each HKG in the testis and the ovary (A) and effect of combining data from different HKGs on the overall CV (B). Data in (A) are taken from Figs 1 and 2. To combine data in (B) the geometric mean of different HKGs for each sample was calculated and used to determine the CV.

An overall analysis of HKG transcript levels in all samples was carried out to determine whether any samples had markedly reduced levels of all HKGs, which would be an indication of mRNA degradation prior to tissue collection and storage. None of the samples showed apparently high levels of luciferase across all HKGs (which would mean that HKG levels were low) (Supplementary data, Fig. 3)

To compare these results with the other methods of identifying stable HKGs, the data were analysed using GeNorm v3.5 and NormFinder algorithms as previously described (Vandesompele et al., 2002; Andersen et al., 2004) (Fig. 4). The order of HKG stability, in both testis and ovary, was very similar using GeNorm and NormFinder (Table II). The most stable two HKGs identified by GeNorm and NormFinder in both ovary and testis also ranked highly using the external standard method described here, with PMM1 ranked the most stable HKG in the ovary by all three methods. It is worth noting, however, that B2M was ranked as the most stable HKG in the testis using the external standard method but was ranked second or third least stable using GeNorm or NormFinder (Table II).

Table II

Reference gene stability.

Stability
 
Method
 
Ranking NormFinder GeNorm External standard 
(A) Testis 
1. SDHA SDHA/TBPa B2M 
2. TBP PMM1 TBP 
3. PMM1 SFRS4 SDHA 
4. SFRS4 B2M SFRS4 
5. HMBS HMBS PMM1 
6. B2M UBC HMBS 
7. UBC  UBC 
(B) Ovary 
1. PMM1 PMM1/TBPa PMM1 
2. TBP SFRS4 B2M 
3. SFRS4 SDHA TBP 
4. SDHA B2M HMBS 
5. B2M UBC SDHA 
6. UBC HMBS SFRS4 
7. HMBS  UBC 
Stability
 
Method
 
Ranking NormFinder GeNorm External standard 
(A) Testis 
1. SDHA SDHA/TBPa B2M 
2. TBP PMM1 TBP 
3. PMM1 SFRS4 SDHA 
4. SFRS4 B2M SFRS4 
5. HMBS HMBS PMM1 
6. B2M UBC HMBS 
7. UBC  UBC 
(B) Ovary 
1. PMM1 PMM1/TBPa PMM1 
2. TBP SFRS4 B2M 
3. SFRS4 SDHA TBP 
4. SDHA B2M HMBS 
5. B2M UBC SDHA 
6. UBC HMBS SFRS4 
7. HMBS  UBC 

aGeNorm does not differentiate between the two most stable genes.

Figure 4

HKG stability as assessed using GeNorm and NormFinder algorithms in ovary (A) and testis (B). Data plotted in Supplementary data Figs 1 and 2 were subjected to GeNorm and NormFinder analysis.

Figure 4

HKG stability as assessed using GeNorm and NormFinder algorithms in ovary (A) and testis (B). Data plotted in Supplementary data Figs 1 and 2 were subjected to GeNorm and NormFinder analysis.

Discussion

Appropriate normalization of RT–PCR data is essential to generate meaningful results. A number of groups have described strategies based on use of external standards (Baker and O'Shaughnessy, 2001; Smith et al., 2003; Johnson et al., 2005; Bower et al., 2007; Ellefsen et al., 2008), but the method appears to have been adopted for routine use by only a very small minority. This is surprising since it overcomes many of the inherent problems associated with use of endogenous genes. There are a number of situations in which it is not practical to use an external standard, as outlined in the Introduction, but the most likely reason why HKGs remain so popular is that historically they have been routinely used in studies of transcript levels (e.g. for Northern blots) and there may be an inherent inertia to the adoption of other methods. Whatever the reason, since reference to an HKG is likely to remain the normalization strategy in the overwhelming majority of real-time studies, it is essential that the most appropriate HKG is identified for each tissue, age and pathology of interest.

The external standard method for assessing HKG stability is straightforward and generally applicable, although it is not assumption-free. Firstly, it is implicit that the exogenous mRNA will be extracted with either the same efficiency as the endogenous mRNA or, at least, at a constant ratio. In soft tissues such as the human fetal gonad, this is likely to be valid as they can be homogenized quickly and completely with rapid denaturation of proteins and mixing of exogenous and endogenous RNA. It is possible that variation in extraction efficiency between endogenous and exogenous RNA may contribute to overall variance, but if it was a major component then it would be expected to affect all HKGs in any one sample in the same way. There is little evidence from Supplementary data, Fig. 3 to suggest a consistent pattern within samples of high deviation either above or below the mean. The second assumption is that there is a linear relationship between tissue mass and total mRNA. This relationship is likely to hold true unless changes in tissue mass are due to accumulation or loss of fluid or extracellular matrix composition. In a tissue like the developing gonad, increased tissue mass is due largely to cell proliferation and is likely to reflect changes in mRNA content. Overall, therefore, for soft tissues like the gonads, the assumptions are probably valid but they may need to be considered more carefully for some other tissues or pathologies.

Recognition that the traditionally used HKGs such as β-actin and GAPDH show differential expression (Huggett et al., 2005) led to the development of techniques such as GeNorm and NormFinder which objectively compare HKG stability in a given tissue. Any technique which attempts to identify stable HKG from endogenous expression levels alone is caught in the conundrum that to identify the most stable HKG you have to already know which HKG is most stable. To circumvent this, assumptions are made about the likely co-variance patterns of stable HKGs or model-based strategies are used to estimate expression variability (Vandesompele et al., 2002; Andersen et al., 2004). From our data, these assumptions appear to be largely valid for the developing human gonad in that two of the top three most stable transcripts were the same in all three methods. Interestingly, however, in both ovary and testis, B2M was ranked among the least stable HKGs by GeNorm and NormFinder but was the most stable in the testis and the second most stable in the ovary as assessed using the external standard method. This indicates that some apparently stable HKGs may not fulfil the criteria of GeNorm and NormFinder and may be erroneously excluded from normalization strategies. Erroneous exclusion of a stable HKG will reduce the normalization power of an experiment, but is far less damaging than including an unstable gene in the normalization algorithm.

In conclusion, this study describes a straightforward method for objective identification of stable reference genes which does not rely upon variability models or assumptions of co-variance. Further data from a number of tissues are needed, but results from the developing human gonad also confirm that GeNorm and NormFinder will identify stable HKGs although they may also erroneously exclude suitable candidates. The stable HKGs identified in this study will be of use in future studies of transcript levels in the developing human gonad.

Authors’ roles

P.OS. conceived the study, analysed data and drafted the manuscript. A.M. carried out reverse transcription and RT–PCR studies. P.F. provided samples, participated in the design of the study and helped draft the manuscript. All authors read and approved the final manuscript.

Supplementary data

Supplementary data are available at http://molehr.oxfordjournals.org/.

Funding

This work was supported by a grant from the Chief Scientist Office, number CZB/4/742.

References

Andersen
CL
Jensen
JL
Orntoft
TF
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
 , 
2004
, vol. 
64
 (pg. 
5245
-
5250
)
Baker
PJ
O'Shaughnessy
PJ
Expression of prostaglandin D synthetase during development in the mouse testis
Reproduction
 , 
2001
, vol. 
122
 (pg. 
553
-
559
)
Bower
NI
Moser
RJ
Hill
JR
Lehnert
SA
Universal reference method for real-time PCR gene expression analysis of preimplantation embryos
Biotechniques
 , 
2007
, vol. 
42
 (pg. 
199
-
206
)
Ellefsen
S
Stenslokken
KO
Sandvik
GK
Kristensen
TA
Nilsson
GE
Improved normalization of real-time reverse transcriptase polymerase chain reaction data using an external RNA control
Anal Biochem
 , 
2008
, vol. 
376
 (pg. 
83
-
93
)
Huggett
J
Dheda
K
Bustin
S
Zumla
A
Real-time RT-PCR normalisation; strategies and considerations
Genes Immun
 , 
2005
, vol. 
6
 (pg. 
279
-
284
)
Johnson
DR
Lee
PK
Holmes
VF
Alvarez-Cohen
L
An internal reference technique for accurately quantifying specific mRNAs by real-time PCR with application to the tceA reductive dehalogenase gene
Appl Environ Microbiol
 , 
2005
, vol. 
71
 (pg. 
3866
-
3871
)
O'Shaughnessy
PJ
Willerton
L
Baker
PJ
Changes in Leydig cell gene expression during development in the mouse
Biol Reprod
 , 
2002
, vol. 
66
 (pg. 
966
-
975
)
O'Shaughnessy
PJ
Baker
PJ
Monteiro
A
Cassie
S
Bhattacharya
S
Fowler
PA
Developmental changes in human fetal testicular cell numbers and messenger ribonucleic acid levels during the second trimester
J Clin Endocrinol Metab
 , 
2007
, vol. 
92
 (pg. 
4792
-
4801
)
Rajapaksha
WRAKJS
McBride
M
Robertson
L
O'Shaughnessy
PJ
Sequence of the bovine HDL-receptor (SR-BI) cDNA and changes in receptor mRNA expression during granulosa cell luteinization in vivo and in vitro
Mol Cell Endocrinol
 , 
1997
, vol. 
134
 (pg. 
59
-
67
)
Smith
RD
Brown
B
Ikonomi
P
Schechter
AN
Exogenous reference RNA for normalization of real-time quantitative PCR
Biotechniques
 , 
2003
, vol. 
34
 (pg. 
88
-
91
)
Vandesompele
J
De Preter
K
Pattyn
F
Poppe
B
Van Roy
N
De Paepe
A
Speleman
F
Accurate normalization of real-time quantitative RT-PCR data by geometric averaging of multiple internal control genes
Genome Biol
 , 
2002
, vol. 
3
 pg. 
RESEARCH0034