Content uploaded by Diego F Calvisi
Author content
All content in this area was uploaded by Diego F Calvisi
Content may be subject to copyright.
Selenoprotein deficiency and high levels of selenium compounds can
effectively inhibit hepatocarcinogenesis in transgenic mice
Sergey V Novoselov
1
, Diego F Calvisi
2
, Vyacheslav M Labunskyy
1
, Valentina M Factor
2
,
Bradley A Carlson
3
, Dmitri E Fomenko
1
, Mohamed E Moustafa
3
, Dolph L Hatfield
3
and
Vadim N Gladyshev*
,1
1
Department of Biochemistry, University of Nebraska, N 151 Beadle Center, 1901 Vine Street, Lincoln, NE 68588, USA;
2
Laboratory
of Experimental Carcinogenesis, Center for Cancer Research, National Cancer Institute, National Institutes of Health, Bethesda, MD
20892, USA;
3
Molecular Biology of Selenium Section, Laboratory of Cancer Prevention, Center for Cancer Research, National
Cancer Institute, National Institutes of Health, Bethesda, MD 20892, USA
The micronutrient element selenium (Se) has been shown
to be effective in reducing the incidence of cancer in
animal models and human clinical trials. Selenoproteins
and low molecular weight Se compounds were implicated
in the chemopreventive effect, but specific mechanisms are
not clear. We examined the role of Se and selenoproteins
in liver tumor formation in TGFa/c-Myc transgenic mice,
which are characterized by disrupted redox homeostasis
and develop liver cancer by 6 months of age. In these mice,
both Se deficiency and high levels of Se compounds
suppressed hepatocarcinogenesis. In addition, both treat-
ments induced expression of detoxification genes, in-
creased apoptosis and inhibited cell proliferation. Within
low-to-optimal levels of dietary Se, tumor formation
correlated with expression of most selenoproteins. These
data suggest that changes in selenoprotein expression may
either suppress or promote tumorigenesis depending on
cell type and genotype. Since dietary Se may have
opposing effects on cancer, it is important to identify the
subjects who will benefit from Se supplementation as well
as those who will not.
Oncogene (2005) 24, 8003–8011. doi:10.1038/sj.onc.1208940;
published online 19 September 2005
Keywords: selenoprotein; selenocysteine; selenium;
cancer prevention; redox regulation
Introduction
The micronutrient element selenium (Se) has been
shown to be effective in reducing the incidence of cancer
in animal models and human clinical trials (Clark et al.,
1996; Yu et al., 1997; Yoshizawa et al., 1998; Milner
et al., 2001; Ip et al., 2002). Of the more than 100
reported studies that have examined the chemopreven-
tive potential of Se, approximately two-thirds demon-
strated a protective effect and the remaining studies
reported no effect. A double-blind clinical trial showed
that supplementing a diet with Se resulted in a
significant decrease in cancer mortality and overall
cancer incidence (Clark et al., 1996). In this study of
1312 individuals, supplementation of the normal diet
with 200 mg of Se daily resulted in decreases of 63, 58,
and 48% in the incidence of prostate, colon, and lung
cancers, respectively (Clark et al., 1996). However, in the
same study, a chemopreventive effect of this micronu-
trient on the incidence of melanoma was not observed.
The biochemical basis for the protective effect of Se
against cancer is not well understood. A number of
mechanisms have been proposed, which are based on
two major ideas. One considers that the chemopreven-
tive action of dietary Se is mediated by an increased
abundance in Se-containing proteins, while the other
that low molecular weight Se-containing compounds (or
selenocompounds) are responsible for the chemopre-
ventive effect (Ganther, 1999; Whanger, 2004). The term
of low molecular weight selenocompounds refers herein
to Se-containing organic and inorganic compounds that
have potent cancer prevention potential. Among the
best characterized Se compounds that decrease cancer
incidence in various rodent models or inhibit growth of
human cancer cells are methylselenenic acid, methylse-
lenocysteine, and selenomethionine. Whether selenopro-
teins may provide the protection against cancers at an
optimal nutritional concentration of Se, and low
molecular weight Se compounds at higher concentra-
tion, is not known.
The predominant biological form of Se in mammals
at low, suboptimal, and optimal dietary levels of this
element is selenocysteine in proteins (Sunde, 1994).
Under these conditions, selenoproteins contain >90%
of the Se pool in mammals. Selenocysteine is encoded by
the UGA codon and is recognized as the 21st amino acid
in protein (Low and Berry, 1996; Stadtman, 1996; Bock,
2000; Hatfield, 2001; Driscoll and Copeland, 2003).
There are 25 human and 24 mouse selenoproteins
(Kryukov et al., 2003).
In the present report, we used a well-characterized
transgenic mouse model of liver cancer, in which
hepatocyte-specific coexpression of c-Myc oncogene
and transforming growth factor alpha (TGFa) leads to
Received 22 February 2005; revised 20 May 2005; accepted 15 June 2005;
published online 19 September 2005
*Correspondence: VN Gladyshev; E-mail: vgladyshev1@unl.edu
Oncogene (2005) 24, 8003–8011
&
2005 Nature Publishing Group
All rights reserved 0950-9232/05 $30.00
www.nature.com/onc
rapid hepatocellular carcinoma (HCC) development
with high penetrance (Santoni-Rugiu et al., 1996). The
trademark features of liver oncogenesis in TGFa/c-Myc
mice are chronic oxidative stress resulting from
increased production of reactive oxygen species and
genomic instability (Factor et al., 1998, 2000; Hironaka
et al., 2003; Calvisi et al., 2004). In contrast to
expectations, we found that not only supplementation
of the diet with high levels of Se but also Se deficiency
suppressed tumorigenesis in this model. Our study
implicated selenoproteins in tumor formation in
TGFa/c-Myc mice and suggested that some cancers
may be promoted by dietary Se supplementation.
Results
Regulation of selenoprotein expression in TGFa/c-Myc
mice
By performing metabolic labeling of mice with
75
Se, we
analysed the distribution of selenoproteins by SDS–
PAGE in TGFa/c-Myc mice at the early dysplastic stage
(10 weeks) and tumor stage (27 weeks) (Supplementary
Figure S1). There were no significant differences in the
labeling pattern between dysplastic livers at 2 and 6
months of age. However, in the tumors derived from
6-month-old mice, expression of glutathione peroxidase 1
(GPx1; a major 25 kDa
75
Se-labeled band) was signifi-
cantly reduced, whereas expression of thioredoxin
reductase 1 (TR1; a 55 kDa
75
Se-labeled band) appeared
to be slightly elevated (Supplementary Figure S1). These
data were consistent with the possibility that tumorigen-
esis alters selenoprotein expression in these mice. Thus, it
is also possible that changes in selenoprotein levels might
influence the incidence of cancer in TGFa/c-Myc mice.
Se diets
Regulation of selenoprotein expression in TGFa/c-Myc
mice was further analysed by changing dietary levels of
Se. Torula yeast Se-deficient diet (designated as the
0 ppm Se diet) was used. We also supplied this diet with
0.1, 0.4, and 2.25 ppm Se (0.1, 0.4, and 2.25 ppm Se
diets, respectively) (see Materials and methods for the
rationale regarding the use of diets). A fifth diet was the
0 ppm Se diet supplemented with 30 ppm triphenylsele-
nonium (TPS) chloride (designated the TPS diet). This
diet served as a control diet that achieved selenoprotein
deficiency in the presence of high levels of Se in the diet.
Regulation of selenoprotein expression by dietary Se
Expression and/or catalytic activity of several seleno-
proteins were assessed in livers of TGFa/c-Myc mice fed
the five Se diets. Glutathione peroxidase activity was
reduced approximately 25- and 50-fold in mice on the
0 ppm Se diet relative to 0.1 and 0.4 ppm Se diets,
respectively (Figure 1a). The increase in Se above
0.4 ppm (the 2.25 ppm Se diet) did not result in an
additional increase in GPx activity. GPx activity in
transgenic mice maintained on the TPS diet was similar
to that observed in mice fed the 0 ppm Se diet. Although
this enzyme activity assay measured total glutathione
peroxidase activity, GPx1 is by far the most active (and
abundant) glutathione peroxidase. Therefore, this assay
primarily measured GPx1 activity.
The activity of selenoprotein R (SelR), which is a
major methionine-R-sulfoxide reductase in mammals,
was regulated by Se similarly to the GPx activity
(Figure 1b), with a dramatic decrease in 0 ppm Se and
TPS diets and maximal activity in 0.4 and 2.25 ppm
diets. Expression of GPx1 and SelR were also assessed
by immunoblot assays. Both proteins were undetectable
in mice fed 0 ppm Se and TPS diets and were
significantly reduced in the 0.1 ppm Se dietary group
(relative to the 0.4 ppm Se group) (Figure 1d). Two
additional selenoproteins, 15 kDa selenoprotein (Sep15)
and TR1 were less responsive to changes in Se levels in
the diets (Figure 1d, e, respectively). There was little
difference in their expression between 0.1, 0.4, and
2.25 ppm Se diets. However, Se deficiency (0 ppm Se) as
well the TPS diet resulted in a significant reduction in
their expression levels (although less dramatic than that
of GPx1 and SelR). A fifth protein analysed, mitochon-
drial thioredoxin reductase 3 (TR3), did not respond
significantly to Se availability (Figure 1e).
Since selenoprotein deficiency was expected to cause
oxidative stress, we reasoned that certain compensatory
antioxidant pathways could be upregulated in the 0 ppm
Se group. We assayed catalase activity and found that
0 ppm Se and TPS diets resulted in increased catalase
activity (Figure 1c). However, there were no obvious
differences in levels of oxidized glutathione or in the
ratio of oxidized to reduced glutathione in mice
maintained on the five diets (data not shown).
Regulation of the liver/body mass ratio by Se
We observed no significant differences in body weights
between different groups of TGFa/c-Myc mice,
although mice fed 0.4 and 2.25 ppm Se diets were about
7–10% larger (39.470.2 and 38.770.6 g, respectively).
The effect of dietary Se supplementation on liver mass
was more pronounced. Mice, which received 0.1 ppm Se,
displayed a transgene-induced hepatomegaly character-
istic of fast-growing TGFa/c-Myc livers. However, the
liver/body weight ratios progressively decreased with
increase in dietary Se (6.6970.26%, P ¼ 0.013 and
5.9570.26%, P ¼ 0.06 in mice fed 0.4 and 2.25 ppm Se
diets, respectively, as compared to 8.1870.58% in the
0.1 ppm Se group). Similar decrease in liver mass was
found in mice maintained on the TPS diet
(6.0770.59%, P ¼ 0.037 as compared to the 0.1 ppm
Se group), whereas Se deficiency (0 ppm Se) did not
significantly affect liver mass (P ¼ 0.247).
Regulation of hepatic tumor formation in TGFa/c-Myc
mice by Se
TGFa/c-Myc mice develop hepatic tumors by 6–8
months of age. To determine whether and how changes
Mechanisms of cancer prevention by selenium in mice
SV Novoselov et al
8004
Oncogene
in selenoprotein expression and concentration of Se
compounds affected TGFa/c-Myc-mediated hepato-
carcinogenesis, we maintained mice on the 0 ppm Se
(n ¼ 11), 0.1 ppm Se (n ¼ 10), 0.4 ppm Se (n ¼ 12),
2.25 ppm Se (n ¼ 8), and TPS diets (n ¼ 4) starting from
weaning until 27 weeks of age. The highest incidence of
both preneoplastic and neoplastic lesions was found in
mice maintained on the 0.4 and 0.1 ppm Se diets.
Surprisingly, both Se deficiency (0 ppm Se) and high
doses of Se (2.25 ppm Se and TPS) significantly inhibited
all stages of neoplastic development, including foci,
adenomas, and carcinomas, as compared to the 0.4 ppm
Se group (Figure 2a). In addition, the tumor multiplicity
was also significantly reduced in these three groups of
Figure 1 Regulation of selenoprotein expression and antioxidant enzyme activities by dietary selenium in livers of TGFa/c-Myc mice.
Glutathione peroxidase (a), SelR (methionine-R-sulfoxide reductase) (b) and catalase (c) activities were analysed in crude extract
samples (n ¼ 8 for each activity) prepared from mice maintained on five different Se diets. (d) Expression of Sep15 (upper panel), SelR
(second panel) and GPx1 (third panel) was analysed by immunoblot assays. The lower panel represents a control and shows a pattern
of Coomassie Blue-staining for samples used in the upper two panels. Expression is shown for two mice fed the 0 ppm Se diet, two mice
the 0.1 ppm Se diet,two mice the 0.4 ppm Se diet, one mouse the 2.25 Se diet and two mice the TPS diet. Molecular weights of Sep15,
SelR and GPx1 based on SDS–PAGE gel migration are shown on the right. (e) Regulation of expression of TR1 and TR3. Thioredoxin
reductases present in liver homogenates were enriched on ADP-Sepharose columns and their levels were determined by immunoblot
assays. For each dietary group, homogenates from two different mice were combined. (f) Expression of 3b hydroxysteroid
dehydrogenase (3bHSD) and pancreatitis-associated protein (PAP) was assayed by immunoblot assays using the samples shown in
panel (d)
Mechanisms of cancer prevention by selenium in mice
SV Novoselov et al
8005
Oncogene
mice (Figure 2b). Thus, in mice fed the 0 ppm Se diet, on
average, only one tumor per mouse was found
compared to an average of 7.5 tumors in the 0.4 ppm
Se group (P ¼ 0.007). Morphologically, Se supplementa-
tion did not affect the phenotypes of TGFa/c-Myc
tumors. The vast majority of double transgenic mice
developed hepatocellular adenomas and carcinomas
with small/clear cell phenotype.
Hepatic dysplasia was evident in all livers from
TGFa/c-Myc transgenic mice regardless of the diet.
However, the degree of dysplasia was strikingly different
between the five dietary groups. TGFa/c-Myc transgenic
mice maintained on 0.1 and 0.4 ppm Se diets showed
highly proliferative diffuse small cell dysplasia resulting
in a marked disorganization of the lobular architecture
(Figure 3a). In contrast, mice receiving 0 or 2.25 ppm Se
displayed mainly large dysplastic cells undergoing a
frequent apoptotic death (Figure 3b–d). Furthermore,
large areas of necrosis and inflammatory infiltrates
within preneoplastic and neoplastic hepatic lesions were
detected in TGFa/c-Myc mice fed the 2.25 ppm Se diet
(Figure 3d). Transgenes probably caused or contributed
to the observed lesions and infiltrates as inflammation
and neoplasia were not evident in control (wild type)
mice receiving 2.25 ppm Se (data not shown).
To better define the effect of Se treatment on tumor
development, we next determined mitotic and apoptotic
indices in TGFa/c-Myc livers (Figure 2c, d). The highest
rate of cell proliferation was found in mice receiving
0.4 ppm Se when compared with that in either 0 ppm,
2.25 ppm or TPS groups (P ¼ 0.001, 6.25E-05, 0.049,
respectively) (Figure 3a). In contrast, the rate of
apoptosis was significantly higher in mice fed 2.25 ppm
Se, 0 ppm Se, and TPS diets (P ¼ 4.6E-06, 3E-06, 0.0001
vs the 0.4 ppm Se group, respectively) (Figures 2d, 3b).
The resulting mitosis/apoptosis ratios were considerably
reduced in the 2.25 ppm Se, 0 ppm Se and TPS dietary
groups coincidently with decreased tumor growth. Of
importance, the most frequent apoptotic cell death was
found in association with extensive inflammatory
infiltrates found in mice fed 2.25 ppm Se diet suggesting
that high levels of Se might promote immune surveil-
lance (Figure 3d). Taken together, these data indicate
that both low and high levels of dietary Se decrease the
severity of dysplastic lesions and tumor formation in
TGFa/c-Myc transgenic mice. In particular, high levels
of Se might inhibit liver tumor development by inducing
a proapoptotic state and favoring immune response
against the tumor.
Gene expression changes in response to Se deficiency and
high Se in TGFa/c-Myc mice
To gain further insights into the mechanism of
suppressed hepatocarcinogenesis in TGFa/c-Myc mice
maintained on the Se-deficient (0 ppm Se) and high Se
(2.25 ppm. Se) diets, we compared gene expression
profiles using Affymetrix mouse oligonucleotide chips
representing B60% of mouse genes (Supplementary
Figure S2). The analysis was performed at the stage
preceding major tumor formation (21 weeks on the diet)
on two mice from each (0, 0.4 and 2.25 ppm Se) dietary
groups (the full data set is at http://genomics.unl.edu/
CANCER). Changes in gene expression in response to
Figure 2 Selenium-deficient diet and diets with high levels of selenium inhibit hepatocarcinogenesis in TGFa/c-Myc transgenic mice.
(a) Frequencies of foci, adenomas and carcinomas in TGFa/c-Myc mice maintained on different Se diets. All mice fed the 0.4 ppm Se
diet developed foci and adenomas, and 11 out of 12 developed carcinomas. The frequencies of foci, adenomas and carcinomas were
significantly reduced in mice on 0 ppm Se (P ¼ 0.008, 0.001, 0.01), 2.25 ppm Se (P ¼ 7.24E-05, 0.04, 0.0003), and TPS (P ¼ 0.007, 0.003,
0.06) as compared to corresponding lesion frequencies in the 0.4 ppm Se group. (b) Multiplicity of tumors in mice maintained on the
five Se diets. (c) Mitotic indices in TGFa/c-Myc livers. (d) Apoptosis indices in TGFa/c-Myc livers
Mechanisms of cancer prevention by selenium in mice
SV Novoselov et al
8006
Oncogene
dietary Se deficiency (0 ppm Se) or high levels of Se
(2.25 ppm Se) were compared vs the 0.4 ppm Se group.
Subsequently, genes affected in 0 and 2.25 ppm Se
samples were compared against each other to identify
unique and common alterations in gene expression
between these two groups.
As expected, we found that several selenoprotein
genes showed decreased mRNA levels in mice fed the
0 ppm Se diet (Table 1), but these did not change when
0.4 and 2.25 ppm Se samples were compared. Additional
unique responses of 0 and 2.25 ppm Se groups included
genes involved in cell cycle, apoptosis, transcription,
redox regulation, fatty acid metabolism, protein folding
and immune/inflammatory response (Supplementary
Tables S1 and S2).
Surprisingly, we also observed a very significant
coregulation of genes in response to 0 and 2.25 ppm Se
diets (Supplementary Figure S3 and Table S3). Among
the genes with elevated expression in both 0 and
2.25 ppm Se groups, most prominent were genes
responsible for detoxification and repair, as well as the
proteins involved in redox regulation (Table 2). In
particular, a number of glutathione-S-transferases and
cytochrome P450s manifested dramatically elevated
gene expression.
These observations are consistent with the idea
that repair/detoxification systems are upregulated
during both Se deficiency and exposure to high levels
of Se, thereby contributing to suppression of tumor-
igenesis observed in TGFa/c-Myc transgenic mice.
Although a possible contribution of variable amounts
of neoplastic tissue to some of the observed effects
in gene expression cannot be excluded, we also
observed increased expression of certain detoxification
enzymes in other animal models subjected to Se diets
(unpublished data).
To verify the microarray data, we assayed expression
of two proteins, 3b hydroxysteroid dehydrogenase
(3bHSD) and pancreatitis-associated protein (PAP)
(Figure 1f). Expression of 3bHSD and PAP mRNAs
was strongly affected by both 0 and 2.25 ppm Se diets
(relative to the 0.4 ppm Se diet) (Table 2 and Supple-
mentary Table S3), but they responded in an opposite
manner to the Se diets. Consistent with these mRNA
Figure 3 Liver histopathology in TGFa/c-Myc mice fed different Se diets. (a) Widespread small cell liver dysplasia in transgenic
mouse receiving the 0.4 ppm Se diet. Note the presence of frequent mitosis (arrows). (b,c) Large cell dysplasia limited to pericentral
areas of hepatic lobules in mice from the 0 ppm Se (b) and 2.25 ppm Se groups (c). Note the presence of multiple apoptotic bodies in c
(arrows). (d) Diffuse liver necrosis accompanied by inflammatory infiltrate in TGFa/c-Myc mice fed the 2.25 ppm Se diet. Inset,
Eosinophilic focal lesion displaying inflammatory infiltrate leading to extensive apoptosis as indicated by brown ApoTag staining
Mechanisms of cancer prevention by selenium in mice
SV Novoselov et al
8007
Oncogene
changes, 3bHSD expression increased in response to
both the 0 and 2.25 ppm Se diets, whereas expression of
PAP decreased in both diets, relative to the 0.4 ppm Se
diet (Figure 1f).
Discussion
The goal of this study was to examine the cancer
preventive effect of selenoproteins and low molecular
weight Se compounds in the same animal model. The
TGFa/c-Myc double transgenic mice, which exhibit a
high penetrance of liver cancer, were chosen as a well
characterized model relevant to human liver disease
(Santoni-Rugiu et al ., 1996; Factor et al., 1998;
Hironaka et al., 2003). In these mice, constitutive
coexpression of c-Myc and TGFa in the liver generates
an oxidative stress environment prior to tumor devel-
opment and predisposes mice to a 100% incidence of
HCC by 6–8 months of age (Calvisi et al., 2004).
Importantly, vitamin E, a potent free radical scavenging
antioxidant, was able to protect liver tissue against
oxidative stress and suppress hepatic tumor formation in
TGFa/c-Myc mice (Factor et al., 2000).
Admittedly, we initially anticipated that optimally
increased levels of Se (0.4 ppm) would similarly normal-
ize redox homeostasis by elevating expression of certain
selenoproteins and hypothesized that it would result in
suppression of the tumorigenesis process in the TGFa/c-
Myc mice. On the contrary, we found that selenoprotein
deficiency significantly suppressed hepatic tumor forma-
Table 1 Changes in selenoprotein gene expression in response to low and high selenium in livers of TGFa/c-Myc mice
Selenoprotein Fold change (0 vs
0.4 ppm Se diet)
Fold change (2.25 vs
0.4 ppm Se diet)
Glutathione peroxidase 1 (GPx1) 2.2 k No change
Glutathione peroxidase 2 (GPx2) No change No change
Glutathione peroxidase 3 (GPx3) No change No change
Phospholipid hydroperoxide glutathione peroxidase (PHGPX; GPx4) No change No change
Thioredoxin reductase 1 (TR1, TxnRd1, TrxR1) 2.17 k No change
Thioredoxin/glutathione reductase (TGR) No change No change
Thyroid hormone deiodinase 1 (D1) 3.2 k No change
15 kDa selenoprotein (Sep15) No change No change
Selenoprotein M (SelM) No change No change
Selenoprotein K (SelK) No change No change
Selenoprotein R (methionine-R-sulfoxide reductase 1; MsrB1; SelR; SelX) No change No change
Selenoprotein W (SelW) 2.2 k No change
Fold change is shown for mice fed the 0 ppm Se diet relative to the 0.4 ppm Se diet and for mice fed the 2.25 ppm Se diet relative to the 0.4 ppm Se
diet. TGFa/c-Myc mice (24 months old) were used, which were fed Se diets from weaning. Each dietary group had two independent samples and the
average fold change in mRNA levels is shown
Table 2 Changes in expression of repair and detoxification genes in liver in response to low and high selenium in livers of TGFa/c-Myc mice
Accession number 2.25 ppm Se 0 ppm Se Annotation
NM_008295 m46.85 m28.84 3b hydroxysteroid dehydrogenase
NM_010357 m18.7 m29.86 Glutathione S-transferase, alpha 4
NM_053262 m8.28 m10.2 Dehydrogenase/reductase, SDR family
NM_023617 m6.73 m8.28 Aldehyde oxidase 3
NM_007825 m4.14 m3.86 Cytochrome P450, family 7, subfamily b, polypeptide 1
BC018344 m3.19 m4.29 Cytochrome P450, family 2, subfamily d, polypeptide 13
BC026757 m2.93 m2.22 Hydroxysteroid dehydrogenase-2, delta
NM_008182 m2.22 m6.73 Glutathione S-transferase, alpha 2, Yc2
BC010973 m2 m2.83 Cytochrome P450, family 8, subfamily b, polypeptide 1
BC012707 m2 m3.14 Glutathione S-transferase, theta 2
AF276917 m2 m2.55 Glutaredoxin 1, thioltransferase
NM_010000 NC m7.73 Cytochrome P450, family 2, subfamily b, polypeptide 9
NM_007812 NC m3.61 Cytochrome P450, family 2, subfamily a, polypeptide 4
AF128849 NC m6.28 Cytochrome P450, family 2, subfamily b, polypeptide 20
AB041034 NC m2.46 NADPH oxidase 4
BE952632 NC m2.38 2-4-dienoyl-Coenzyme A reductase 2, peroxisomal
BC020001 NC m2.22 Aldehyde dehydrogenase 1 family, member B1
NM_010191 NC m2.17 Farnesyl diphosphate farnesyl transferase 1
NM_134072 NC m2.07 3-alpha-hydroxysteroid dehydrogenase type 1
BC025822 NC m2.07 Cytochrome P450, family 2, subfamily c, polypeptide 70
Fold change is shown for mice fed the 0 ppm Se diet relative to the 0.4 ppm Se diet and for mice fed the 2.25 ppm Se diet relative to the 0.4 ppm Se
diet. TGFa/c-Myc mice (24 months old) were used, which were fed Se diets from weaning. Each dietary group had two independent samples and
the average fold change in mRNA levels is shown
Mechanisms of cancer prevention by selenium in mice
SV Novoselov et al
8008
Oncogene
tion. To our knowledge, this is the first demonstration of
a lower HCC incidence in response to Se deficiency.
The increase in dietary Se levels from 0 to 0.4 ppm of
Se resulted in elevated expression of various selenopro-
teins, as assessed by Affymetrix chips, enzyme assays,
and immunoblot analyses. Antioxidant selenoproteins
GPx1 and SelR were particularly responsive to Se
supplementation. Sep15 and TR1 also responded,
although the changes in expression were less pro-
nounced. Expression of TR3 remained constant regard-
less of the Se diet. At low levels of dietary Se,
selenoproteins account for the major part of Se in the
body. Thus, there was a clear correlation between
overall selenoprotein expression and increased tumor-
igenesis in the TGFa/c-Myc mice.
Although both tumor incidence and tumor multi-
plicity were similarly decreased in mice receiving high
levels of Se in the diet, this decrease did not correlate
with selenoprotein expression, as the 0.4 ppm Se already
saturated selenoprotein expression. These data suggest
that low molecular weight Se compounds are respon-
sible for the chemopreventive effect of higher doses of
dietary Se in the TGFa/c-Myc mice. Thus, both
selenoprotein deficiency and high levels of Se com-
pounds independently inhibited tumorigenesis in TGFa/
c-Myc mice. Such a correlated effect of Se deficiency and
high levels of Se (as compared to optimal Se levels) has
not been previously reported.
Moreover, both Se deficient and high Se diets resulted
in decreases in cell proliferation and increases in
apoptosis. A likely possibility is that high doses of Se
(2.25 ppm Se and TPS) are cytotoxic, particularly for
preneoplastic and cancer cells. It appears that decreased
expression of selenoproteins observed in mice fed the
0 ppm Se diet was also cytotoxic as it promoted
apoptosis and decreased the frequency of mitosis
concomitantly with inhibition of tumorigenesis. In
addition, we observed upregulation of mRNAs for
repair and detoxification proteins, such as glutathione-
S-transferases, cytochrome P450s, and glutaredoxin,
under conditions of both Se deficiency (0 ppm Se)
and high levels of Se (2.25 ppm Se). Although our study
did not employ chemicals to induce tumorigenesis, it
seems possible that the low levels of antioxidant
selenoproteins may have increased the demand for
detoxification of endogenous toxic compounds and for
repair of damaged macromolecules. Thus, cytotoxicity
may be a common feature associated with both Se
deficiency and high levels of Se suggesting that
detoxification and repair mechanisms may contribute
to or be responsible for suppression of tumorigenesis in
this animal model.
Our findings are consistent with the known inhibitory
effect of antioxidant selenoproteins on apoptosis (Fu
et al., 2001; Gouaze et al., 2002). Previous studies
reported elevated expression of detoxification and repair
enzymes in Se-deficient mouse livers (Reiter and
Wendel, 1984; Carlson et al., 2004), as well as in cells
treated with high levels of Se (El-Bayoumy and Sinha,
2004; Zhao et al., 2004). In addition, Se deficiency had a
protective effect against acetaminophen and aflatoxin
injuries in rat livers, presumably due to stimulation
of GSH synthesis and induction of expression of
detoxification enzymes (Burk and Lane, 1983; Hill and
Burk, 1984).
Our data also agree with the recent report that Se
deficiency abrogated peritoneal plasmacytomas in
BALB/c mice, the experimental model of inflamma-
tion-dependent plasma cell transformation (Felix et al.,
2004). Thus, the role of Se and selenoproteins in cancer
and the other findings reported in our study are likely
not limited to strain, organ, and type of tumor that were
examined.
When used in physiological levels, Se is often viewed
as antioxidant that works in concert with vitamin E.
However, vitamin E decreased tumorigenesis in TGFa/
c-Myc mice (Factor et al., 2000), suggesting that at least
in this mouse model, these compounds may have
opposite effects. Besides, different selenoproteins may
influence carcinogenesis in various ways (Gladyshev
et al., 1998). Whereas some, such as GPx1, are
antioxidant proteins with well-defined functions, the
contribution of other selenoproteins may be more
complex. For example, TR1, the enzyme controlling
the redox state of thioredoxin, is implicated in a myriad
of cellular functions (Arner and Holmgren, 2000).
Thioredoxin serves as an antioxidant and antitumor
protein by reducing thioredoxin peroxidase and con-
trolling the redox state of tumor suppressor p53.
However, it also reduces various transcription factors
and provides electrons for deoxyribonucleotide bio-
synthesis. Thus, Se deficiency, by suppressing TR1
expression, may reduce the antioxidant and tumor
suppressor capacity of the thioredoxin system as well
as inhibit growth control functions. It is possible that in
the TGFa/c-Myc model, growth inhibitory functions of
Se deficiency prevail over antioxidant functions thereby
inhibiting carcinogenesis.
Our data have important (although indirect) implica-
tions for the use of dietary Se in cancer prevention in
humans. Optimal levels of dietary Se, which are slightly
above those needed for maximal expression of seleno-
proteins, were thought to provide health benefits,
particularly in reducing cancer incidence (Rayman,
2000; Hatfield, 2001; Burk, 2002). However, our new
findings suggest a possibility that elevated Se might also
promote tumorigenesis in certain genetic backgrounds.
Thus, whereas the overall effect of Se, when assessed in
large population groups (Clark et al., 1996; Yu et al.,
1997), may be positive, there might be individuals, in
which Se supplementation may promote tumor forma-
tion. Although our studies cannot be directly applied to
humans (e.g. severe Se deficiency is unlikely to occur in
humans unless there is a disruption in the pathway of Se
insertion), they establish a precedent for a similar effect
of Se deficiency and high levels of Se on cancer incidence
in the same model system.
Given the promising results of previous clinical trials
involving dietary Se (Clark et al., 1996, Yu et al., 1997),
a number of new trials have been initiated, including the
Se and Vitamin E Cancer Prevention Trial (SELECT)
that involves 32 400 male subjects (Klein et al., 2000).
Mechanisms of cancer prevention by selenium in mice
SV Novoselov et al
8009
Oncogene
Despite the extent of ongoing clinical trials, basic
knowledge of the types of cancer that are prevented by
Se and the mechanism of the chemopreventive effect is
lacking. If the effect of Se on hepatocarcinogenesis that
we observed in the TGFa/c-Myc mice can be seen in
other mouse models or in subsets of the human
population, it would mean that certain human cancers
may be promoted by Se supplementation. Thus, rather
than generally increasing the intake of Se, it may be
important to identify segments of the human population
that can benefit from Se supplementation as well as
those in which Se supplementation can increase cancer
incidence.
Materials and methods
Mice and diets
Double transgenic TGFa/c-Myc mice were generated as
described (Santoni-Rugiu et al., 1996). Five different diets
(Harland TekLad, Madison, WI, USA) were used: 0 ppm Se
(Se-deficient), 0.1 ppm Se, 0.4 ppm Se, 2.25 ppm Se, and
30 ppm TPS. The diets were based on the Torula yeast Se-
deficient diet. Se was provided in the form of sodium selenite
(except for the TPS diet).
Previous reports indicated that B0.1 ppm Se is a minimal
amount of Se that is sufficient for maximal expression of
GPx1. Since maximal expression of plasma GPx is achieved at
B55 mg/day Se in humans (Burk, 2002), the 0.1 ppm Se mouse
diet may be viewed as approximately corresponding to the
human Recommended Dietary Allowance for adults. By
analogy, the 0.4 ppm Se diet would correspond to supplemen-
tation of the human diet with 200 mg Se/day, which is the
common amount of Se used in clinical trials (Clark et al.,
1996). The rationale for the 2.25 ppm Se diet was based on the
average amount of Se that was required for the chemopreven-
tive effect of Se against chemically induced carcinogenesis in
rodents (Ganther, 1999; Ip et al., 2002; Whanger, 2004).
Finally, TPS was recently introduced as a cancer preventive
compound (Ip et al., 1998). Even at very high (30 ppm) levels,
TPS-derived Se was not accessible for insertion into seleno-
proteins, and thus, high levels of low molecular weight Se
compounds were achieved under conditions of selenoprotein
deficiency (Ip et al., 1998). Se-deficient and high Se diets have
been used in numerous previous studies involving rodents.
These diets, as well as the TPS diet, provide excellent tools for
the studies on the mechanisms of cancer prevention by Se in
animal model systems. One limitation of these diets, however,
is that the data can only partially be extrapolated to humans,
as severe Se-deficient and Se-enriched conditions are unlikely
to occur in humans.
Prior to the TGFa/c-Myc dietary study, we tested the Se
diets using wild-type mice, which were housed in groups of
three. After 2 months on the diets, the mice were killed and
GPx1 activity was determined in each dietary group. The mice
fed the 0 ppm Se diet had 3.6% of the GPx1 activity of mice on
the 0.4 ppm Se diet. Likewise, the mice on the TPS diet had
4.5% of the GPx1 activity of the 0.4 ppm Se control. In
addition, the diets were documented by analysing the amount
of Se in the livers of wild-type mice fed 0.1 and 0 ppm Se diets.
We found that the 0 ppm Se livers had 10 times lower total
Se content than the 0.1 ppm Se livers. The TGFa/c-Myc mice
were housed in groups of 2–5 and had free access to food and
water. The dietary interventions started from weaning.
Histology and tumor sampling
The mice were killed at 27 weeks of age by CO
2
inhalation. The
liver and body weights as well as the number and the size of all
grossly visible tumors were recorded. Representative samples
of liver taken from each lobe were fixed in 10% neutral
formalin at 41C overnight or frozen in liquid nitrogen and
stored at 801C. Paraffin sections (5 mm) were stained with
hematoxylin–eosin, and hepatocellular lesions were graded as
foci, adenomas and carcinomas as described (Calvisi et al .,
2004). The data are presented as mean7s.e. For statistical
comparison the Student’s t-test was used.
Mitotic and apoptotic indices
Mitotic and apoptotic indices were scored in TGFa/c-Myc
livers from at least four mice per diet group. Mitotic index was
determined by counting mitotic figures on H&E-stained slides.
For the apoptotic index, sections were stained with the
ApoTag peroxidase in situ apoptosis detection kit (Serologicals
Corporation, Norcross, GA, USA), and positive cells were
counted under a light microscope at a magnification 200.
Apoptotic cells, single, and closely clustered apoptotic bodies
were all scored as single events. Mitotic and apoptotic indices
were represented as a percentage (mean7s.e.) of the total cells
counted (at least 2000 cells per animal).
Metabolic labeling
To label selenoproteins in TGFa/c-Myc mice, animals were
injected with freshly neutralized
75
Se[selenite] (Research
Reactor Facility, University of Missouri, Columbia, MO,
USA), and after 48 h, the mice were killed, tissues removed,
and proteins extracted, electrophoresed, and transferred onto
PVDF membranes. Resulting transblots were analysed with a
PhosphorImager.
Enzyme assays
GPx1 and catalase activities were assayed using BIOXITECH
GPx-340 and BIOXITECH Catalase-520 kits (OxisResearch,
Portland, OR, USA), respectively, according to the manufac-
turer’s protocol. The methionine-R-sulfoxide reductase activ-
ity of SelR was assayed as described previously (Kim and
Gladyshev, 2004). Liver homogenates from eight animals from
each group (except the TPS dietary group that had four
animals) were used and each sample was analysed in triplicate.
Selenoprotein expression analyses
TR1 (cytosolic enzyme) and TR3 (this mitochondrial enzyme
is also called TnxRd2 and TrxR2) expression was assessed by
immunoblot assays with antibodies specific for each isozyme,
after the enzymes were enriched by binding to ADP-Sepharose
as described previously (Moustafa et al., 2001). Expression of
SelR, GPx1, and Sep15 was analysed by immunoblot assays
with antibodies specific for these proteins.
Gene expression analyses
Affymetrix GeneChip System (Affymetrix, Santa Clara, CA,
USA) was used to carry out gene expression analyses. Total
RNA was extracted from two 0 ppm Se, two 0.4 ppm Se, and
two 2.25 ppm Se frozen mouse livers using Trizol Reagent
(Invitrogen, Carlsbad, CA, USA) and RNeasy columns
(Qiagen, Valencia, CA, USA). Double-stranded cDNA synth-
esis and biotinylation was carried out as described in the
Expression Analysis Technical Manual (Affymetrix). Mouse
15K arrays were hybridized and scanned using Affymetrix
Mechanisms of cancer prevention by selenium in mice
SV Novoselov et al
8010
Oncogene
GeneChip Scanner 3000. To compare gene expression, the data
generated from the 0 ppm Se samples were averaged and
analysed against averaged 0.4 ppm Se data. Likewise, the
2.25 ppm Se data were averaged and analysed against the
0.4 ppm Se data. Expression of 3bHSD and PAP were assayed
by immunoblot assays with antibodies specific for each protein
(kindly provided by Dr Ian Mason and purchased from R&D
Systems, Minneapolis, MN, USA, respectively).
Abbreviations
GPx1, glutathione peroxidase 1; HCC, hepatocellular carci-
noma; Se, selenium; SelR, selenoprotein R; Sep15, 15 kDa
selenoprotein; TGFa, transforming growth factor a; TPS,
triphenylselenonium chloride; TR1, thioredoxin reductase 1
(cytosolic enzyme* also called TxnRd1 and TrxR1); TR3,
thioredoxin reductase 3 (mitochondrial enzyme* also called
TxnRd2 and TrxR2).
Acknowledgements
Services of the University of Nebraska-Lincoln genomics
facility, headed by Dr Yuannan Xia, and the bioinformatics
facility, headed by Dr Guoqing Lu, are acknowledged.
We thank Dr Ian Mason for providing hydroxysteroid
dehydrogenase antibodies and Dr Hwa-Young Kim for help
with methionine sulfoxide reductase assays. Supported by
NIH CA080946 and in part by GM061603 and AG021518
(to VNG).
References
Arner ES and Holmgren A. (2000). Eur. J. Biochem., 267,
6102–6109.
Bock A. (2000). Biofactors, 11, 77–78.
Burk RF. (2002). Nutr. Clin. Care, 5, 75–79.
Burk RF and Lane JM. (1983). Fundam. Appl. Toxicol., 3,
218–221.
Calvisi DF, Factor VM, Ladu S, Conner EA and Thorgeirsson
SS. (2004). Gastroenterology, 126, 1374–1386.
Carlson BA, Novoselov SV, Kumaraswamy E, Lee BJ, Anver
MR, Gladyshev VN and Hatfield DL. (2004). J. Biol. Chem.,
279, 8011–8017.
Clark LC, Combs GF, Turnbull BW, Slate EH, Chalker DK,
Chow J, Davis LS, Glover RA, Graham GF, Gross EG,
Krongrad A, Lesher Jr JL, Park HK, Sanders Jr BB,
Smith CL and Taylor JR. (1996). J. Am. Med. Assoc., 276,
1957–1963.
Driscoll DM and Copeland PR. (2003). Annu. Rev. Nutr., 23,
17–40.
El-Bayoumy K and Sinha R. (2004). Mutat. Res., 551, 181–197.
Factor VM, Kiss A, Woitach JT, Wirth PJ and Thorgeirsson
SS. (1998). J. Biol. Chem., 273, 15846–15853.
Factor VM, Laskowska D, Jensen MR, Woitach JT, Popescu
NC and Thorgeirsson SS. (2000). Proc. Natl. Acad. Sci.
USA, 97, 2196–2201.
Felix K, Gerstmeier S, Kyriakopoulos A, Howard OM, Dong
HF, Eckhaus M, Behne D, Bornkamm GW and Janz S.
(2004). Cancer Res., 64, 2910–2917.
Fu Y, Sies H and Lei XG. (2001). J. Biol. Chem., 276,
43004–43009.
Ganther HE. (1999). Carcinogenesis, 20, 1657–1666.
Gladyshev VN, Factor VM, Housseau F and Hatfield DL.
(1998). Biochem. Biophys. Res. Commun., 251, 488–493.
Gouaze V, Andrieu-Abadie N, Cuvillier O, Malagarie-
Cazenave S, Frisach MF, Mirault ME and Levade T.
(2002). J. Biol. Chem., 277, 42867–42874.
Hatfield DL (ed) (2001). Selenium: Its Molecular Biology and
Role in Human Health. Kluwer Academic Publishers:
Norwell, MA.
Hill KE and Burk RF. (1984). Toxicol. Appl. Pharmacol., 72,
32–39.
Hironaka K, Factor VM, Calvisi DF, Conner EA and
Thorgeirsson SS. (2003). Lab. Invest., 83, 643–654.
Ip C, Dong Y and Ganther HE. (2002). Cancer Metastasis
Rev., 21, 281–289.
Ip C, Thompson HJ and Ganther HE. (1998). Anticancer Res.,
18, 9–12.
Kim HY and Gladyshev VN. (2004). Mol. Biol. Cell, 15,
1055–1064.
Klein EA, Thompson IM, Lippman SM, Goodman PJ,
Albanes D, Taylor PR and Coltman C. (2000). Prostate
Cancer Prostatic Dis., 3, 145–151.
Kryukov GV, Castellano S, Novoselov SV, Lobanov AV,
Zehtab O, Guigo R and Gladyshev VN. (2003). Science, 300,
1439–1443.
Low SC and Berry MJ. (1996). Trends Biochem. Sci., 21,
203–208.
Milner JA, McDonald SS, Anderson DE and Greenwald P.
(2001). Nutr. Cancer, 41, 1–16.
Moustafa ME, Carlson BA, El-Saadani MA, Kryukov GV,
Sun QA, Harney JW, Hill KE, Combs GF, Feigenbaum L,
Mansur DB, Burk RF, Berry MJ, Diamond AM, Lee BJ,
Gladyshev VN and Hatfield DL. (2001). tRNA. Mol. Cell.
Biol., 21, 3840–3852.
Rayman MP. (2000). Lancet, 356, 233–241.
Reiter R and Wendel A. (1984). Biochem. Pharmacol., 33,
1923–1928.
Santoni-Rugiu E, Nagy P, Jensen MR, Factor VM and
Thorgeirsson SS. (1996). Am. J. Pathol., 149, 407–428.
Stadtman TC. (1996). Ann. Rev. Biochem., 65, 83–100.
Sunde RA. (1994). Selenium in Biology and Human Health.RF
Burk (ed). Springer-Verlag: New York, pp 45–59.
Whanger PD. (2004). Br. J. Nutr., 91, 11–28.
Yoshizawa K, Willett WC, Morris SJ, Stampfer MJ, Spiegel-
man D, Rimm EB and Giovannucci E. (1998). J. Natl.
Cancer Inst., 90, 1219–1224.
Yu SY, Zhu YJ and Li WG. (1997). Biol. Trace Elements Res.,
56, 117–124.
Zhao H, Whitfield ML, Xu T, Botstein D and Brooks JD.
(2004). Mol. Biol. Cell., 15, 506–519.
Supplementary Information accompanies the paper on Oncogene website (http://www.nature.com/onc)
Mechanisms of cancer prevention by selenium in mice
SV Novoselov et al
8011
Oncogene