Determination of reference genes for circadian studies in different tissues and mouse strains.

Abstract

Circadian rhythms have a profound effect on human health. Their disruption can lead to serious pathologies, such as cancer and obesity. Gene expression studies in these pathologies are often studied in different mouse strains by quantitative real time polymerase chain reaction (qPCR). Selection of reference genes is a crucial step of qPCR experiments. Recent studies show that reference gene stability can vary between species and tissues, but none has taken circadian experiments into consideration.
In the present study the expression of ten candidate reference genes (Actb, Eif2a, Gapdh, Hmbs, Hprt1, Ppib, Rn18s, Rplp0, Tbcc and Utp6c) was measured in 131 liver and 97 adrenal gland samples taken from three mouse strains (C57BL/6JOlaHsd, 129Pas plus C57BL/6J and Crem KO on 129Pas plus C57BL/6J background) every 4 h in a 24 h period. Expression stability was evaluated by geNorm and NormFinder programs. Differences in ranking of the most stable reference genes were observed both between individual mouse strains as well as between tissues within each mouse strain. We show that selection of reference gene (Actb) that is often used for analyses in individual mouse strains leads to errors if used for normalization when different mouse strains are compared. We identified alternative reference genes that are stable in these comparisons.
Genetic background and circadian time influence the expression stability of reference genes. Differences between mouse strains and tissues should be taken into consideration to avoid false interpretations. We show that the use of a single reference gene can lead to false biological conclusions. This manuscript provides a useful reference point for researchers that search for stable reference genes in the field of circadian biology.

Full text

Page 1

RESEARCH ARTICLEOpen Access
Determination of reference genes for circadian
studies in different tissues and mouse strains
Rok Kosir1, Jure Acimovic2, Marko Golicnik2, Martina Perse3, Gregor Majdic4, Martina Fink5, Damjana Rozman1*
Abstract
Background: Circadian rhythms have a profound effect on human health. Their disruption can lead to serious
pathologies, such as cancer and obesity. Gene expression studies in these pathologies are often studied in different
mouse strains by quantitative real time polymerase chain reaction (qPCR). Selection of reference genes is a crucial
step of qPCR experiments. Recent studies show that reference gene stability can vary between species and tissues,
but none has taken circadian experiments into consideration.
Results: In the present study the expression of ten candidate reference genes (Actb, Eif2a, Gapdh, Hmbs, Hprt1,
Ppib, Rn18s, Rplp0, Tbcc and Utp6c) was measured in 131 liver and 97 adrenal gland samples taken from three
mouse strains (C57BL/6JOlaHsd, 129Pas plus C57BL/6J and Crem KO on 129Pas plus C57BL/6J background) every
4 h in a 24 h period. Expression stability was evaluated by geNorm and NormFinder programs. Differences in
ranking of the most stable reference genes were observed both between individual mouse strains as well as
between tissues within each mouse strain. We show that selection of reference gene (Actb) that is often used for
analyses in individual mouse strains leads to errors if used for normalization when different mouse strains are
compared. We identified alternative reference genes that are stable in these comparisons.
Conclusions: Genetic background and circadian time influence the expression stability of reference genes.
Differences between mouse strains and tissues should be taken into consideration to avoid false interpretations.
We show that the use of a single reference gene can lead to false biological conclusions. This manuscript provides
a useful reference point for researchers that search for stable reference genes in the field of circadian biology.
Background
Circadian rhythms are oscillations in behaviour and phy-
siology whose function it is to anticipate environmental
changes associated with the solar day [1]. At the mole-
cular level, they consist of a network of transcriptional
and translational feedback loops that drive the 24 h
expression of core clock components [2,3]. Circadian
control is required for healthy life, thus disruption of
circadian cycle leads to pathologies such as cancer, obe-
sity, lipid disorders and type 2-diabetes [4-6]. Some of
these abnormalities were discovered in mouse models
lacking core clock genes Clock and Bmal1 [7-9]. Pheno-
types resulting from mutations of clock genes are highly
affected by genetic background [4,10,11]. Yoshiki
defined genetic background as the influence of all genes
of the genome that may interact with the gene of inter-
est and potentially influence the specific phenotype [12].
To date, a number of reports have shown that genetic
background affects the phenotype caused by gene dis-
ruption [12-15].
When studying gene expression, qPCR is the domi-
nant quantitative technique due to its broad dynamic
range, accuracy, sensitivity, specificity and speed [16].
Normalization in qPCR controls for variations in all
experimental steps, enabling comparison between differ-
ent samples [17]. Different normalization strategies are
available, where application of reference genes as inter-
nal controls seems to be the most appropriate [18,19].
Unfortunately, there is no universal reference gene that
would be stably expressed under all experimental condi-
tions. Hence, normalization to reference genes that are
validated in individual experiments is a prerequisite for
accurate interpretation of biological data [20-24].
* Correspondence: damjana.rozman@mf.uni-lj.si
1Center for Functional Genomics and Bio-Chips, Institute of Biochemistry,
Faculty of Medicine, University of Ljubljana, Zaloska 4, SI-1000 Ljubljana,
Slovenia
Full list of author information is available at the end of the article
Kosir et al. BMC Molecular Biology 2010, 11:60
http://www.biomedcentral.com/1471-2199/11/60
© 2010 Kosir et al; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons
Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in
any medium, provided the original work is properly cited.

Page 2

The broader scope of our research is to understand
the circadian expression of genes in different tissues of
the two commonly used laboratory mouse strains
(C57BL/6JOlaHsd and 129Pas plus C57BL/6J) and to
determine the tissue-specific effect of the targeted dis-
ruption of transcription factor Crem. In this manuscript,
we show that genetic background and the circadian time
are important factors influencing expression of com-
monly used qPCR reference genes. This should be taken
into consideration for accurate interpretation of biologi-
cal data.
Results
Selection and characteristics of candidate reference genes
Seven candidate reference genes (Ppib, Rplp0, Gapdh,
Actb, Hmbs, Hprt1 and Rn18s) were selected [20,23] and
their expression measured in 35 livers and 33 adrenals
from the inbred strain (C57BL/6JOlaHsd), 51 livers and
34 adrenals from the wild type mixed strain (129Pas plus
C57BL/6J) and 45 livers and 30 adrenals from the mixed
strain with a targeted disruption of the Crem gene (Crem
KO). For livers, three additional reference genes were
selected by RefGenes (Eif2a, Utp6c and Tbcc), based on
meta analysis of the most stably expressed liver genes in
various mouse strains as detected by Affymetrix chips
[25]. Gene symbols, full names, accession numbers and
gene functions are listed in Table 1. Intron spanning pri-
mers were designed wherever possible, with the excep-
tion of Rplp0 and Tbcc where primers lie within a single
exon. To determine primer specificity, melting curve ana-
lyses were performed on all primer pairs during the pri-
mer validation process as well as after each qPCR run.
The specificity of the amplicon was confirmed by the
presence of a single peak. Primer efficiencies were calcu-
lated based on slopes from standard curves by LightCy-
cler 480 software (Roche Diagnostics). Standard curves
were prepared with five-fold serial dilutions of the cDNA
pool. A negative control (without reverse transcriptase)
was also included to determine possible amplification
from genomic DNA. Only primers with single peaks and
good negative controls were used in the study. Primer
details are listed in Table 2.
Expression level of reference genes
Expression of measured reference genes is represented
as raw quantification cycle (Cq) in Figure 1. Rn18 s
shows highest expression in all samples with a mean Cq
of 8.8. This is not surprising since Rn18 s represents the
bulk of total RNA. The Cq values of other candidates
were between Cq 18-29. Tbcc shows lowest expression
in all samples with a mean Cq of 28.5. The largest varia-
tion across the studied 228 samples was observed in
Actb and Hmbs (12 cycles) and the smallest for Eif2a,
Tbcc and Utp6c (< 4 cycles).
Search for optimal reference genes by geNorm
geNorm ranks reference genes according to their aver-
age expression stability value (M), from the most stable
(lowest M value) to the least stable (highest M value).
An important advantage of geNorm is that it provides
the optimal number of reference genes required for
accurate normalization. This number is based on the
pairwise variation values (V(n/n+1)) [20].
We divided expression data into eight groups, accord-
ing to the tissue (liver or adrenal) and mouse strain
(C57BL/6JOlaHsd; 129Pas plus C57BL/6J; CremKO in
mixed strain or including all strains; Table 3). Groups A
to F thus represent stably expressed reference genes of
each tissue in each mouse strain. Groups G and H con-
sider genetic differences between mouse strains because
samples of all strains are joined for each tissue. The
time of sacrifice (Ct) cannot be included as a variable by
geNorm analyses. The same groups have been applied
as well in NormFinder analyses.
Figure 2 summarises the average expression stabilities
and ranking of candidate reference genes in individual
Table 1 Official symbols, accession numbers, full names and functions of candidate reference genes evaluated in this
study
Symbol Accession
number
Full nameFunction
Rplp0
Ppib
NM_007475.4
NM_011149.2
ribosomal protein, large, P0
peptidylprolyl isomerase B
Structural constituent of ribosome
Associated with the secretory pathway and released in
biological fluids
Cytoskeletal structural protein
Heme synthesis, porphyrin metabolism
Purine synthesis in salvage pathway
Ribosomal RNA
Protein translation
Rn18 s biogenesis
Actb
Hmbs
Hprt1
Rn18s
Eif2a
Utp6c
NM_007393.3
NM_013551.2
NM_013556.2
NR_003278.1
NM_001005509.1
NM_144826.3
actin, beta
Hydroxymethyl-bilane synthase
hypoxanthine guanine phosphoribosyl transferase 1
18 S RNA
eukaryotic translation initiation factor 2a
small subunit (SSU) processome component, homolog
(yeast)
tubulin-specific chaperone c
Tbcc
NM_178385.3Protein folding
Kosir et al. BMC Molecular Biology 2010, 11:60
http://www.biomedcentral.com/1471-2199/11/60
Page 2 of 11

Page 3

groups. Ranking is different for livers of different mouse
strains. In 129Pas plus C57BL/6J and Crem KO, which
has the Crem gene disrupted on the same mixed back-
ground, Hprt1 is always most stable, together with either
Rn18 s (Figure 2E) or Rplp0 (Figure 2C). In the C57BL/
6JOlaHsd strain, the most stable genes are Eif2a and
Hmbs (Figure 2A), which were, interestingly, among the
least stable in the mixed strain (Figure 2C,2E). In adre-
nal glands of individual mouse strains, the ranking is
more consistent, with Hprt1 and Actb both ranking last
and Rn18 s and Ppib always ranking among the top
three (Figure 2B,2D,2F).
When evaluating reference genes within each strain,
relatively small M values are observed (Figure 2A to 2F),
indicating a greater degree of expression stability between
samples. However, when genetic differences between
mouse strains are included in evaluation, a larger degree
of variation (larger M value) is observed (Figure 2G and
2H), which supports the notion that genetic variability
importantly influences expression. In this case, Actb is the
least stable in both liver and adrenal glands. It also displays
a far greater M value compared to other genes, showing
that it is indeed not a good gene for normalization in
mouse tissues. Eif2a and Tbcc in the liver (Figure 2G) and
Gapdh and Hmbs in adrenals (Figure 2H) are most stable
if all strains are considered.
We also determined the optimal number of reference
genes for normalization (Figure 3). In all groups (A - H),
a pairwise variation value (V2/3) of less than 0.15
Table 2 Primer sequences, exon location (where possible), efficiency and amplicon length for candidate reference
genes and the test circadian gene
SymbolPrimerSequenceAmplicon length Primer location1[exon]Primer efficiency
Reference genes
Rplp0
fw
rv
fw
rv
fw
rv
fw
rv
fw
rv
fw
rv
fw
rv
fw
rv
fw
rv
fw
rv
CACTGGTCTAGGACCCGAGAAG
GGTGCCTCTGGAGATTTTCG
GGAGATGGCACAGGAGGAAA
CCGTAGTGCTTCAGTTTGAAGTTCT
CCAATGTGTCCGTCGTGGATCT
GTTGAAGTCGCAGGAGACAACC
CTTCCTCCCTGGAGAAGAGC
ATGCCACAGGATTCCATACC
TCCCTGAAGGATGTGCCTA
AAGGGTTTTCCCGTTTGC
TCCTCCTCAGACCGCTTTT
CCTGGTTCATCATCGCTAATC
CGCCGCTAGAGGTGAAATTC
TTGGCAAATGCTTTCGCTC
CAACGTGGCAGCCTTACA
TTTCATGTCATAAAGTTGTAGGTTAGG
TTTCGGTTGAGTTTTTCAGGA
CCCTCAGGTTTACCATCTTGC
GACTCCTTCCTGAACCTCTGG
GGAGGCCATTCAAAACTTCA
734
4
3
4
5
6
4
5
7
8
1
2
1.98
Ppib
731.93
Gapdh
2391.94
Actb
1241.98
Hmbs
731.64
Hprt1
901.89
Rn18s
62n.a.
n.a.
2
3
17
18
n.a.
n.a.
1.79
Eif2a
74 1.95
Utp6c
75 1.82
Tbcc
621.90
Circadian gene
Dbp
fw
rv
AATGACCTTTGAACCTGATCCCGCT
GCTCCAGTACTTCTCATCCTTCTGT
1757
7
1.93
Figure 1 Expression levels of candidate reference genes.
Expression is represented as quantification cycle values (Cq)
obtained from qPCR. Variability is displayed as medians (line), 25th
percentile to 75thpercentile (box) and min to max (whiskers). Gene
symbols are explained in Table 1. All experiments (three mouse
strains, two tissues) were considered in this analysis.
Kosir et al. BMC Molecular Biology 2010, 11:60
http://www.biomedcentral.com/1471-2199/11/60
Page 3 of 11

Page 4

Table 3 Distribution of RT-qPCR data into groups
Mouse Strain Tissue Time Point [Ct]Biological replicates Dataset
[num. of samples]
C57BL/6JOlaHsdLiver05A
[35]
4
8
5
5
5
5
5
5
5
12
16
20
24
0Adrenal glandB
[33]
4
8
4
5
5
5
5
4
12
16
20
24
129Pas plus C57Bl/6JLiver06C
[45]
Crem knock-out4
8
6
7
6
6
7
7
3
12
16
20
24
0Adrenal glandD
[30]
4
8
4
4
5
4
5
5
12
16
20
24
129Pas plus C57Bl/6JLiver08E
[51]
Wild type4
8
7
7
7
7
8
7
5
12
16
20
24
0Adrenal glandF
[34]
4
8
5
4
5
5
5
5
12
16
20
24
All strains Liver 7 time points131G
All strains Adrenal 7 time points97H
Total number of samples 228
Kosir et al. BMC Molecular Biology 2010, 11:60
http://www.biomedcentral.com/1471-2199/11/60
Page 4 of 11

Page 5

Figure 2 Expression stabilities and ranking of reference genes in individual groups according to geNorm. Groups A to F enable
determination of reference genes for circadian experiments within separate tissues and mouse strains. Data is divided according to tissue type
(liver and adrenal gland) and mouse strain. The remaining two groups (G and H) were used to identify reference genes when comparing
different mouse strains. Figures are divided according to tissues (liver and andrenal glands) and mouse strain. Crem KO is on the 129Pas plus
C57BL/6J background. Graphs G and H represent reference genes from joint data of all strains.
Kosir et al. BMC Molecular Biology 2010, 11:60
http://www.biomedcentral.com/1471-2199/11/60
Page 5 of 11

Page 6

was determined, confirming that 2 stable reference genes
might be sufficient for accurate normalization, as pro-
posed by Vandesompele et al [20]. Addition of further
genes in groups A to F did not influence the V value sig-
nificantly. However, in groups G and H, addition of least
stable genes did raise the V value above 0.15 (Figure 3).
Determination of reference genes by NormFinder
NormFinder uses a model based approach to calculate
gene stability value for either the most stable reference
gene or the best combination of two genes. It has the
advantage to allow estimation of variations between
time points, which is crucial in circadian experiments.
NormFinder gene stability values and rankings are
shown in Table 4. In this analysis, samples in all groups
(A - H) were divided further into 7 subgroups according
to the time of scarification (time point).
Similarly to geNorm, NormFinder also showed differ-
ences in ranking of reference genes in different mouse
strains. In liver samples of 129Pas plus C57BL/6J and
Crem KO strains Hprt1 always ranks among the top two
reference genes (Table 4C and 4E), however in the
C57BL/6JOlaHsd strain it is among the least stable
genes (Table 4A). The opposite is seen for Hmbs, which
is most stable in the C57BL/6JOlaHsd strain (Table 4A)
and among least stable in 129Pas plus C57BL/6J and
Crem KO strains (Table 4C and 4E). When genetic dif-
ferences between strains are included in the evaluation,
Figure 3 Optimal number of reference genes used for normalization. geNorm determination of the optimal number of reference genes
based on the pairwise variation value (Vn/n+1) that is calculated between two sequential normalization factors [20]. The optimal number of
reference genes was calculated for liver and adrenal glands where samples from all mouse strains are included.
Table 4 Stability value and ranking of reference genes based on NormFinder
Circadian
C57BL/6JOlaHsd
liver
A
129Pas plus C57Bl/6J-KO
liveradrenal gland
C
129Pas plus C57Bl/6J-WT
liveradrenal gland
E
Mouse strain
adrenal gland Gene
Dataset
adrenal gland
B
liver
GDFH
Rn18s
Actb
Gapdh
Hmbs
Hprt1
Ppib
Rplp0
Eif2a
Tbcc
Utp6c
0.118
0.196
0.134
0.096*
0.150
0.124
0.187
0.127
0.124
0.190
0.070
0.090
0.088
0.058
0.156
0.057*
0.081
0.078
0.093
0.123
0.114
0.062
0.090
0.056*
0.112
0.107
0.060
0.068
0.104
0.113
0.080
0.166
0.062*
0.099
0.063
0.087
0.106
0.095
0.058
0.070
0.077
0.106
0.081
0.057*
0.062
0.060
0.058
0.063
0.187
0.025*
0.073
0.020
0.555
0.112
0.309
0.024
0.028
0.021
0.017*
0.021
0.027
0.054
0.506
0.080
0.088
0.199
0.028*
0.064
Best two
Ppib
Hmbs
0.061
Ppib
Hmbs
0.046
Hprt1
Ppib
0.044
Rn18s
Ppib
0.048
Hprt1
Utp6c
0.037
Gapdh
Ppib
0.032
Rn18s
Eif2a
0.013
Rn18s
Ppib
0.031
In groups A to F, data was grouped according to tissue, mouse strain and the time point. In groups G and H, data was grouped only according to tissue and
time point. *-best candidate reference gene according to NormFinder. (Best two) - the best combination of two reference genes together with their stability
value are calculated only if group identifiers (in our case time of sacrifice) are included in the analysis. “Best two” is not always equal to the first and second
ranking genes, but represent the two genes with minimal combined inter-and intragroup expression variation [22].
Kosir et al. BMC Molecular Biology 2010, 11:60
http://www.biomedcentral.com/1471-2199/11/60
Page 6 of 11

Page 7

Actb is the least stable gene (Table 4G), followed by
Hmbs and Gapdh.
Ranking is again more consistent for adrenal glands.
Ppib ranks as the most stable reference gene both in
each individual mouse strain (Table 4B,D and 4F) and
also irrespective of the mouse genetic background
(Table 4H). Hprt1 is the least stable gene in adrenals of
each investigated mouse strain (Table 4B, D and 4F),
while Actb ranks last. Again, this result is obtained only
if genetic differences between strains are taken into con-
sideration (Table 4H).
In individual mouse strain groups (A - E), with the
exception of group F, gene stability value for the best
combination of two reference genes is lower than the
value of the most stable reference gene (Table 4), sug-
gesting that normalization on a single reference gene
may not be sufficient.
Normalization of biological data by reference genes of
different stability
Dbp (D-box binding protein) is one of the most robust
circadian genes in the liver, with the peak of expression
between CT10-CT14 [26]. We monitored the Dbp hepa-
tic expression (Cq values) in mouse strains C57BL/
6JOlaHsd, 129Pas plus C57BL/6J and Crem KO. To test
the role of reference gene selection on interpretation of
the hepatic Dbp expression, we applied different nor-
malization procedures: with a) Actb which is commonly
used for normalization, but was determined as the least
stable gene by geNorm and NormFinder; b) the average
of the two most stable reference genes according to
geNorm (Eif2a and Tbcc) and c) the average of the two
most stable reference genes according to NormFinder
(Rn18 s and Eif2a).
Figure 4 shows that normalization to Actb inserts bias
into the data. A seemingly large difference in Dbp
expression is detected between the C57BL/6JOlaHsd
and the 129Pas plus C57BL/6J and Crem KO strains
(Figure 4A). This leads to a false conclusion that Dbp
has a very high expression in the C57BL/6JOlaHsd
strain, with almost none in the 129Pas plus C57BL/6J
and Crem KO strains. However, when normalization is
carried out with the most stable reference genes deter-
mined by either geNorm or NormFinder, the difference
is greatly reduced (Figure 4B and 4C). In this case, Dbp
is equally expressed in both mouse strains, which can be
expected for a gene with a high expression level and
robust circadian rhythm [27].
Discussion
qPCR is a method of choice for quantitative gene
expression analysis. Due to its high sensitivity, normali-
zation with stable reference genes is important for accu-
rate analysis of the biological variation in the data
Figure 4 Liver expression profile of the mouse circadian gene
Dbp. The Dbp gene with known circadian transcription in mouse
liver was normalized with an unstable reference gene Actb (A) or
with the two most stable reference genes as calculated by geNorm
(B) or NormFinder (C). Due to low values not seen in panel A, an
insert represents the expression profile of Dbp in 129Pas plus
C57BL/6J and Crem KO mouse strains.
Kosir et al. BMC Molecular Biology 2010, 11:60
http://www.biomedcentral.com/1471-2199/11/60
Page 7 of 11

Page 8

[18,23,28]. The selection of appropriate reference genes
is, however, far from trivial. It has been shown that
application of non-validated reference genes can lead to
inaccurate data interpretation [23,29,30].
Understanding the tissue-specific circadian behaviour
of genes and proteins is often required in drug-treated
mouse models, including knockout models from differ-
ent mouse strains. In recent years, the circadian aspects
of metabolism and drug detoxification became more
important for proper understanding of physiology,
pathophysiology, drug metabolism, etc. [5,31-34]. Even
though a number of studies discuss the selection of
reference genes [35-39] no study discussed circadian
experiments and only one past study evaluated different
mouse strains [40]. The majority of circadian studies
still perform normalization using a single reference gene
[41-44].
We evaluate ten candidate reference genes for their
expression stability in a circadian experiment with
mouse liver and adrenal glands in three mouse strains
(C57BL/6JOlaHsd, 129Pas plus C57BL/6J and Crem
KO). Seven genes from our study have been evaluated
previously under a variety of experimental conditions
[20,45,46]. Three additional genes (Eif2a, Tbcc and
Utp6c) were selected by RefGene as the most stably
expressed after the meta analysis of liver Affymetrix
expression profile [25]. Publicly available microarray
data have been successfully used before for reference
gene selection [47,48]. We show that irrespective of the
analysis applied (geNorm or NormFinder), different
mouse strains show different rankings of reference
genes. Utp6c was indicated as unstable reference gene in
C57BL/6JOlaHsd livers by both programs (Figure 2A,
Table 4A), whereas in the 129Pas plus C57BL/6J and
Crem KO it ranked among the top three (Figure 2C and
2E; Table 4C, E).
When comparing gene expression between different
mouse strains, three reference genes from the Affyme-
trix meta analyses ranked among the most stable, with
Eif2a being first by both programs (Figure 2G and Table
4G). This indicates that RefGenes tool is useful in nar-
rowing down the number of candidate reference genes,
especially when comparing mouse strains.
Differences in ranking of reference genes were
observed not only between mouse strains, but also
between tissues, as confirmed by both programs. Genes
most stable in livers of 129Pas plus C57BL/6J and Crem
KO strains ranked among the least stable in C57BL/
6JOlaHsd mice (Figure 2 and Table 4). Adrenals show
similar ranking between strains. Ppib is usually the most
stable and Hprt1 the least stable gene in all strains by
both programs. Our study shows that reference genes
suitable for one mouse strain should not automatically
be used for normalization in another strain. The same
applies for different tissues.
The average expression stability value M determined
by geNorm shows little variability in gene expression
between samples taking into account circadian effect
both in liver and adrenals within each strain. This is in
line with the pairwise variation value (V2/3), which is
well below the 0.15 threshold value set by Vandensom-
pele [20], indicating that normalization of target genes
with a combination of the two best genes is sufficient.
The M value of reference genes was substantially higher
when searching for stable genes between mouse stains.
Here Actb was identified as the least stable in both liver
and adrenals (Figure 2G,2H and Table 4G, H). Similar
results were obtained when comparing human tissue
samples [20].
To test whether selection of the least stable reference
gene (Actb) affects normalization when compared to
normalization with the two most stable genes selected
by geNorm and NormFinder, a known and robust liver
circadian gene Dbp has been normalized to three factors
(Figure 4). Normalization to Actb leads to a large differ-
ence in the expression pattern between the C57BL/
6JOlaHsd strain and the 129Pas plus C57BL/6J and
Crem KO strains (Figure 4A). This can lead to a false
impression that expression of Dbp is substantially higher
in C57BL/6JOlaHsd. This difference, however, almost
disappears when normalisation is performed on the two
most stable genes Eif2a and Tbcc (geNorm) or Eif2a
and Rn18 s (NormFinder) (Figure 4B and 4C). Even
though the programs did not select identical reference
genes, a similar Dbp expression profile was obtained in
both cases. However, the ability of NormFinder to dis-
tinguish between different time points provides an
advantage over geNorm when sufficient samples and
genes are evaluated [22], as is the case in circadian
experiments.
Conclusions
In this study we investigated the most reliable refer-
ence genes for normalization of circadian studies
within or between mouse strains in livers and adrenal
glands. The study is unique in its analysis of 3 mouse
strains, 2 tissues and circadian sampling and in the
magnitude of samples and genes tested (10 candidate
reference genes in 228 samples by two programs). We
show that differences in the reference genes exist
between mouse strains as well as between tissues of
the same strain. We also show that selection of a refer-
ence gene that appears stable in each mouse strain
separately, can lead to interpretation errors, when used
for normalization in different mouse stains. We identi-
fied altrernative reference genes that are stable in
mouse strain comparisons.
Kosir et al. BMC Molecular Biology 2010, 11:60
http://www.biomedcentral.com/1471-2199/11/60
Page 8 of 11

Page 9

Methods
Animals
54 wild type (129Pas plus C57BL/6J) and 45 Crem
knock-out (Crem KO) mice of the mixed strain and 35
mice of the inbred strain (C57BL/6JOlaHsd) were used.
Animals had free access to food (Harland Tekland 2916)
and water and were maintained under a 12:12 h light
cycle (light on at 7:00 am, light off at 7:00 pm). The
experiment was approved by the Veterinary Administra-
tion of the Republic of Slovenia (license number 34401-
9/2008/4) and was conducted in accordance with the
European Convention for the protection of vertebrate
animals used for experimental and other scientific pur-
poses (ETS 123) as well as in accordance with National
Institutes of Health guidelines for work with laboratory
animals.
Tissue samples
Mice were sacrificed with cervical dislocation under dim
red light every 4 h over a 24 h period starting on the
second day after being transferred to dark: dark (DD)
conditions. Immediately after they were sacrificed, liver
and adrenal glands were excised, snap frozen in liquid
N2and stored at -80°C for subsequent analysis. 96 liver
and 58 adrenal glands from the mixed strain and 35
liver and 34 adrenal glands from the inbred strain were
used for cDNA synthesis. The lower number of adrenal
glands is due to insufficient amount of material or due
to small RNA isolation yield. RNA was isolated from 30
mg of liver tissue and one adrenal gland per animal.
RNA extraction and cDNA preparation
Liver and adrenal gland samples were homogenized in
1000 μl and 500 μl of TRI reagent (Sigma) respectively
and total RNA was isolated according to the manufac-
tures instructions. RNA from liver samples of the inbred
strain was isolated using QuickGene-810 (FujiFilm)
according to manufacturer’s instructions. RNA quantity
and quality were assessed with NanoDrop and Agilent
2100 Bioanalyzer instruments. DNAse treatment was
performed on all samples using DNAse I (Roche
Applied Bioscience) according to the manufacturer’s
instructions. cDNA synthesis was carried out using
SuperScript III reverse transcriptase (Invitrogen). 3 μg of
liver RNA was mixed together with 20 μl of reverse
transcriptase master mix which contained 8 μl of 5 ×
first strand buffer, 2 μl of 100 mM DTT, 2 μl of 10 mM
dNTP mix, 1 μl of random primers (Promega 500 ng/
μl), 0.75 μl of SuperScript III (200 U/μl), 0.75 μl of
RNAse OUT (Invitrogen) and 5.5 μl of RNAse free
water in a final volume of 40 μl. 1 μg of adrenal gland
RNA was mixed together with 10 μl of reverse tran-
scriptase master mix which contained 5 μl of 5x first
strand buffer, 1.25 μl of 100 mM DTT, 1.25 μl of
10 mM dNTP mix, 0.65 μl of random primers (Promega
500 ng/μl), 0.5 μl of SuperScript III (200 U/μl), 0.5 μl of
RNAse OUT (Invitrogen) and 0.85 μl of RNAse free
water in a final volume of 25 μl. The reaction mixtures
were incubated at 25°C for 5 minutes, 50°C for 60 min-
utes and 70°C for 10 minutes.
Primer design
Wherever possible, intron spanning primers for ten can-
didate reference genes were designed based on publicly
available sequences (Table 2). Genes chosen belong to
different functional classes, which reduces the chance of
co-regulation. Primer specificity and amplification effi-
ciency were also validated empirically with melting
curve and standard curve analysis of a six fold dilution
series.
Quantitative qPCR
Real time quantitative PCR was performed in a 384 well
format on LightCycler 480 (Roche Applied Science)
using LightCycler 480 SYBR Green I Master (Roche
Applied Science). The PCR reaction consisted of 2.5 μl
of SYBR Green I Master, 1.15 μl of RNAse free water,
0.6 μl of 300 nM primer mix and 0.75 μl of cDNA in a
total volume of 5 μl. Three technical replicates were
performed for each sample. Cycling conditions were as
follows: 10 min at 95°C followed by 40 rounds of 10 s at
95°C, 20 s at 60°C and 20 s at 72 °C. Melting curve ana-
lysis for determining the dissociation of PCR products
was performed from 65°C to 95°C.
Analysis of expression stability
Expression stabilities of selected reference genes were
evaluated by two publicly available programs, geNorm
VBA and NormFinder applets for Microsoft Excel. geN-
orm calculates the stability of selected reference genes
according to the similarity of their expression profile by
pair-wise comparison and calculates M value, where the
gene with the highest value is the least stable one. All
calculations were performed on quantities, which were
transformed from Cq values based on gene specific effi-
ciencies [20]. NormFinder also requires quantities input,
but calculates a gene-stability value with a mathematical
model based on separate analysis of the sample sub-
groups and estimation of both intra-and intergroup var-
iation in expression levels [22]. Relative quantities (Q)
needed for the input were calculated via the delta-Ct
method with the formula Q = (E)ΔCq. ΔCq equals Cq of
the sample with the lowest Cq value (highest abun-
dance) minus Cq of a sample. Efficiency (E) corrected
relative amount were calculated. Normalization factors
were calculated based on geometric averaging of the
Kosir et al. BMC Molecular Biology 2010, 11:60
http://www.biomedcentral.com/1471-2199/11/60
Page 9 of 11

Page 10

most stable reference genes. Normalization was carried
out by dividing relative quantities with the normaliza-
tion factor [20].
Acknowledgements
The work was supported by the Slovenian Research Agency Grants J1-9438
and P1-0527. Rok Kosir was supported by the fellowship from the Slovenian
Research Agency. We thank Jernej Ule for helpful discussions and careful
reading of the manuscript.
Author details
1Center for Functional Genomics and Bio-Chips, Institute of Biochemistry,
Faculty of Medicine, University of Ljubljana, Zaloska 4, SI-1000 Ljubljana,
Slovenia.2Institute of Biochemistry, Faculty of Medicine, University of
Ljubljana, Vrazov trg 2, SI-1000 Ljubljana, Slovenia.3Medical Experimental
Centre, Institute of Pathology, Faculty of Medicine, University of Ljubljana,
Zaloska 4, SI-1000 Ljubljana, Slovenia.4Center for Animal Genomics,
Veterinary Faculty; University of Ljubljana, Gerbiceva 60, SI-1000 Ljubljana,
Slovenia.5University Medical Center Ljubljana, Department of Haematology,
Zaloska cesta 7, SI-1000 Ljubljana, Slovenia.
Authors’ contributions
RK carried out qPCR and data analysis and wrote the manuscript. JA and MG
participated in the planning, sampling and data analysis. MP and GM are
responsible for animal experiments and participated in sampling, design of
the study and provided useful discussions. MF participated in study design,
data analysis and provided useful discussion.DR supervised the study,
participated in study design and coordination. All authors read and
approved the final manuscript.
Received: 13 April 2010 Accepted: 16 August 2010
Published: 16 August 2010
References
1.Ko CH, Takahashi JS: Molecular components of the mammalian circadian
clock. Hum Mol Genet 2006, 15(Spec No 2):R271-277.
2.Lowrey PL, Takahashi JS: Mammalian circadian biology: elucidating
genome-wide levels of temporal organization. Annu Rev Genomics Hum
Genet 2004, 5:407-441.
3.Reppert SM, Weaver DR: Coordination of circadian timing in mammals.
Nature 2002, 418(6901):935-941.
4.Ishida N: Circadian clock, cancer and lipid metabolism. Neurosci Res 2007,
57(4):483-490.
5.Sahar S, Sassone-Corsi P: Metabolism and cancer: the circadian clock
connection. Nat Rev Cancer 2009, 9(12):886-896.
6.Li S, Lin JD: Molecular control of circadian metabolic rhythms. J Appl
Physiol 2009, 107(6):1959-1964.
7.Turek FW, Joshu C, Kohsaka A, Lin E, Ivanova G, McDearmon E, Laposky A,
Losee-Olson S, Easton A, Jensen DR, Eckel RH, Takahashi JS, Bass J: Obesity
and metabolic syndrome in circadian Clock mutant mice. Science 2005,
308(5724):1043-1045.
8.Antoch MP, Gorbacheva VY, Vykhovanets O, Toshkov IA, Kondratov RV,
Kondratova AA, Lee C, Nikitin AY: Disruption of the circadian clock due to
the Clock mutation has discrete effects on aging and carcinogenesis.
Cell Cycle 2008, 7(9):1197-1204.
9.Rudic RD, McNamara P, Curtis AM, Boston RC, Panda S, Hogenesch JB,
Fitzgerald GA: BMAL1 and CLOCK, two essential components of the
circadian clock, are involved in glucose homeostasis. PLoS Biol 2004,
2(11):e377.
10. Vitaterna MH, King DP, Chang AM, Kornhauser JM, Lowrey PL,
McDonald JD, Dove WF, Pinto LH, Turek FW, Takahashi JS: Mutagenesis
and mapping of a mouse gene, Clock, essential for circadian behavior.
Science 1994, 264(5159):719-725.
11.Oishi K, Miyazaki K, Ishida N: Functional CLOCK is not involved in the
entrainment of peripheral clocks to the restricted feeding: entrainable
expression of mPer2 and BMAL1 mRNAs in the heart of Clock mutant
mice on Jcl:ICR background. Biochem Biophys Res Commun 2002,
298(2):198-202.
12. Yoshiki A, Moriwaki K: Mouse phenome research: implications of genetic
background. ILAR J 2006, 47(2):94-102.
Fleming RE, Holden CC, Tomatsu S, Waheed A, Brunt EM, Britton RS,
Bacon BR, Roopenian DC, Sly WS: Mouse strain differences determine
severity of iron accumulation in Hfe knockout model of hereditary
hemochromatosis. Proc Natl Acad Sci USA 2001, 98(5):2707-2711.
Montagutelli X: Effect of the genetic background on the phenotype of
mouse mutations. J Am Soc Nephrol 2000, 11(Suppl 16):S101-105.
Wahlsten D, Bachmanov A, Finn DA, Crabbe JC: Stability of inbred mouse
strain differences in behavior and brain size between laboratories and
across decades. Proc Natl Acad Sci USA 2006, 103(44):16364-16369.
Bustin SA, Nolan T: Pitfalls of quantitative real-time reverse-transcription
polymerase chain reaction. J Biomol Tech 2004, 15(3):155-166.
Huggett J, Dheda K, Bustin S, Zumla A: Real-time RT-PCR normalisation;
strategies and considerations. Genes Immun 2005, 6(4):279-284.
Bustin S: Quantification of mRNA using real-time reverse transcription
PCR (RT-PCR): trends and problems. J Mol Endocrinol 2002, 29(1):23-39.
Bustin SA, Benes V, Nolan T, Pfaffl MW: Quantitative real-time RT-PCR–a
perspective. J Mol Endocrinol 2005, 34(3):597-601.
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, 3(7):RESEARCH0034.
Pfaffl MW, Tichopad A, Prgomet C, Neuvians TP: Determination of stable
housekeeping genes, differentially regulated target genes and sample
integrity: BestKeeper–Excel-based tool using pair-wise correlations.
Biotechnol Lett 2004, 26(6):509-515.
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, 64(15):5245-5250.
Radonic A, Thulke S, Mackay IM, Landt O, Siegert W, Nitsche A: Guideline
to reference gene selection for quantitative real-time PCR. Biochem
Biophys Res Commun 2004, 313(4):856-862.
Dheda K, Huggett JF, Bustin SA, Johnson MA, Rook G, Zumla A: Validation
of housekeeping genes for normalizing RNA expression in real-time PCR.
Biotechniques 2004, 37(1):112-114, 116, 118-119.
Hruz T, Laule O, Szabo G, Wessendorp F, Bleuler S, Oertle L, Widmayer P,
Gruissem W, Zimmermann P: Genevestigator V3: a reference expression
database for the meta-analysis of transcriptomes. Advances in
Bioinformatics 2008, 420747:5.
Wuarin J, Schibler U: Expression of the liver-enriched transcriptional
activator protein DBP follows a stringent circadian rhythm. Cell 1990,
63(6):1257-1266.
Lopez-Molina L, Conquet F, Dubois-Dauphin M, Schibler U: The DBP gene
is expressed according to a circadian rhythm in the suprachiasmatic
nucleus and influences circadian behavior. EMBO J 1997,
16(22):6762-6771.
VanGuilder HD, Vrana KE, Freeman WM: Twenty-five years of quantitative
PCR for gene expression analysis. Biotechniques 2008, 44(5):619-626.
Bustin SA: Real-time, fluorescence-based quantitative PCR: a snapshot of
current procedures and preferences. Expert Rev Mol Diagn 2005,
5(4):493-498.
Dheda K, Huggett JF, Chang JS, Kim LU, Bustin SA, Johnson MA, Rook GA,
Zumla A: The implications of using an inappropriate reference gene for
real-time reverse transcription PCR data normalization. Anal Biochem
2005, 344(1):141-143.
Takahashi JS, Hong HK, Ko CH, McDearmon EL: The genetics of
mammalian circadian order and disorder: implications for physiology
and disease. Nat Rev Genet 2008, 9(10):764-775.
Green CB, Takahashi JS, Bass J: The meter of metabolism. Cell 2008,
134(5):728-742.
Levi F, Okyar A, Dulong S, Innominato PF, Clairambault J: Circadian timing
in cancer treatments. Annu Rev Pharmacol Toxicol 50:377-421.
Levi F, Schibler U: Circadian rhythms: mechanisms and therapeutic
implications. Annu Rev Pharmacol Toxicol 2007, 47:593-628.
Cook NL, Kleinig TJ, van den Heuvel C, Vink R: Reference genes for
normalising gene expression data in collagenase-induced rat
intracerebral haemorrhage. BMC Mol Biol 11:7.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
35.
Kosir et al. BMC Molecular Biology 2010, 11:60
http://www.biomedcentral.com/1471-2199/11/60
Page 10 of 11

Page 11

36. Maroufi A, Van Bockstaele E, De Loose M: Validation of reference genes
for gene expression analysis in chicory (Cichorium intybus) using
quantitative real-time PCR. BMC Mol Biol 11:15.
Langnaese K, John R, Schweizer H, Ebmeyer U, Keilhoff G: Selection of
reference genes for quantitative real-time PCR in a rat asphyxial cardiac
arrest model. BMC Mol Biol 2008, 9:53.
Kessler Y, Helfer-Hungerbuehler AK, Cattori V, Meli ML, Zellweger B,
Ossent P, Riond B, Reusch CE, Lutz H, Hofmann-Lehmann R: Quantitative
TaqMan real-time PCR assays for gene expression normalisation in feline
tissues. BMC Mol Biol 2009, 10:106.
Saviozzi S, Cordero F, Lo Iacono M, Novello S, Scagliotti GV, Calogero RA:
Selection of suitable reference genes for accurate normalization of gene
expression profile studies in non-small cell lung cancer. BMC Cancer
2006, 6:200.
Frericks M, Esser C: A toolbox of novel murine house-keeping genes
identified by meta-analysis of large scale gene expression profiles.
Biochim Biophys Acta 2008, 1779(12):830-837.
Asher G, Gatfield D, Stratmann M, Reinke H, Dibner C, Kreppel F,
Mostoslavsky R, Alt FW, Schibler U: SIRT1 regulates circadian clock gene
expression through PER2 deacetylation. Cell 2008, 134(2):317-328.
Canaple L, Rambaud J, Dkhissi-Benyahya O, Rayet B, Tan NS, Michalik L,
Delaunay F, Wahli W, Laudet V: Reciprocal regulation of brain and muscle
Arnt-like protein 1 and peroxisome proliferator-activated receptor alpha
defines a novel positive feedback loop in the rodent liver circadian
clock. Mol Endocrinol 2006, 20(8):1715-1727.
Liu AC, Tran HG, Zhang EE, Priest AA, Welsh DK, Kay SA: Redundant
function of REV-ERBalpha and beta and non-essential role for Bmal1
cycling in transcriptional regulation of intracellular circadian rhythms.
PLoS Genet 2008, 4(2):e1000023.
Ptitsyn AA, Zvonic S, Conrad SA, Scott LK, Mynatt RL, Gimble JM: Circadian
clocks are resounding in peripheral tissues. PLoS Comput Biol 2006, 2(3):
e16.
Shi G, Zhang Z, Feng D, Xu Y, Lu Y, Wang J, Jiang J, Li X, Ning G: Selection
of reference genes for quantitative real-time reverse transcription-
polymerase chain reaction in concanavalin A-induced hepatitis model.
Anal Biochem .
Tatsumi K, Ohashi K, Taminishi S, Okano T, Yoshioka A, Shima M: Reference
gene selection for real-time RT-PCR in regenerating mouse livers.
Biochem Biophys Res Commun 2008, 374(1):106-110.
Popovici V, Goldstein DR, Antonov J, Jaggi R, Delorenzi M, Wirapati P:
Selecting control genes for RT-QPCR using public microarray data. BMC
Bioinformatics 2009, 10:42.
Shulzhenko N, Yambartsev A, Goncalves-Primo A, Gerbase-DeLima M,
Morgun A: Selection of control genes for quantitative RT-PCR based on
microarray data. Biochem Biophys Res Commun 2005, 337(1):306-312.
37.
38.
39.
40.
41.
42.
43.
44.
45.
46.
47.
48.
doi:10.1186/1471-2199-11-60
Cite this article as: Kosir et al.: Determination of reference genes for
circadian studies in different tissues and mouse strains. BMC Molecular
Biology 2010 11:60.
Submit your next manuscript to BioMed Central
and take full advantage of:
• Convenient online submission
• Thorough peer review
• No space constraints or color figure charges
• Immediate publication on acceptance
• Inclusion in PubMed, CAS, Scopus and Google Scholar
• Research which is freely available for redistribution
Submit your manuscript at
www.biomedcentral.com/submit
Kosir et al. BMC Molecular Biology 2010, 11:60
http://www.biomedcentral.com/1471-2199/11/60
Page 11 of 11