You are on page 1of 11

Nucleocytoplasmic shuttling and phosphorylation of BMAL1

T
8?O
Circadian
riginal
Tamaru
Blackwell
Oxford,
Genes
GTC
©
1356-9597
2003 Article
BMAL1
et al.
Blackwell
toUK
Cells nucleocytoplasmic
Publishing
Publishing
Ltd.
Ltd shuttling

are regulated by circadian clock in cultured fibroblasts


Teruya Tamaru1,*, Yasushi Isojima2, Gijsbertus T. J. van der Horst3, Kohtaro Takei4,
Katsuya Nagai2 and Ken Takamatsu1
1
Department of Physiology, Toho University School of Medicine, 5-21-16 Ohmori-nishi Ohta-ku, Tokyo 143-8540, Japan
2
Division of Protein Metabolism, Institute for Protein Research, Osaka University, 3-2 Yamadaoka, Suita, Osaka 565-0871, Japan
3
Department of Cell Biology & Genetics, Erasmus MC, PO Box 1738, 3000 DR Rotterdam, the Netherlands
4
Department of Molecular Pharmacology and Neurobiology,Yokohama City University School of Medicine, 3-9 Hukuura, Kanazawa-ku,
Yokohama 236-0004, Japan

Abstract
Background: Recent discoveries of clock proteins of BMAL1 matched nuclear accumulation of
have unveiled an important part of the mammalian CLOCK and the peak of Per1 transcription. Nuclear
circadian clock mechanism. However, the molecular BMAL1 was gradually phosphorylated and then
clockwork that cause these fundamental feedback dephosphorylated in a temporally regulated manner,
loops to stably oscillate with a ∼24 h-periodicity although cytoplasmic BMAL1 was not. In serum-
remain unclear. shocked mCry1/mCry2 (CRY)-deficient fibroblasts,
which lack a functional clock, both the cytoplasmic
Results: Serum-shocked fibroblasts were used as a
and nuclear BMAL1 were only present as hyper-
cellular clock model. Circadian changes in the sub-
phosphorylated forms and their circadian nucleocy-
cellular localization and phosphorylation of BMAL1
toplasmic shuttling was absent.
protein in these cells were assessed by immunocyto-
chemistry and immunoblotting. A significant time Conclusions: We propose that the nucleocytoplasmic
lag between Bmal1 transcription and the cytoplasmic/ shuttling and phosphorylation states of BMAL1 are
nuclear accumulation of BMAL1 was observed. After regulated by circadian clock, and that this tempo-
its nuclear accumulation, BMAL1 accumulated in rally regulated and time-delayed nuclear entry of
the cytoplasm again, mainly by nucleoexport, before BMAL1 is important in the maintenance of a stably
the increase of Bmal1 transcripts. Nuclear accumulation oscillating clock.

genes in a variety of other organisms (Crosthwaite et al.


Introduction 1997; Darlington et al. 1998; Ishiura et al. 1998). How-
Recent developments have unveiled an important part ever, the molecular events that cause these fundamental
of the mechanism behind circadian clocks, namely, that feedback loops to oscillate stably with an approximate
a set of mammalian circadian clock genes, such as Per1, periodicity of 24 h remain unclear. It has been postu-
Per2, Cry1, Cry2, Clock and Bmal1, are expressed in an lated in the general delay models that the single most
oscillatory manner with a circadian rhythm in the central important parameter driving this periodicity is a ‘time-
circadian pacemaker in the hypothalamic suprachiasmatic delay’ event that regulates the activity and /or localization
nucleus (SCN) (Moore 1997; Morse & Sassone-Corsi of the core clock genes/proteins (Scheper et al. 1999;
2002; Reppert & Weaver 2002). It is believed that these Lema et al. 2000). For example, in Drosophila, PER and
mammalian clock genes may directly or indirectly regu- TIM first undergo cytoplasmic accumulation, after which
late their own expression by means of feedback loops, the PER:TIM complexes enter the nucleus.This sequence
which is similar to what has been observed for the clock of temporally spaced events is believed to provide the
time-delay between the transcription of PER and TIM
Communicated by : Eisuke Nishida and the inhibition of CLK:CYC transcriptional activity
*Correspondence: E-mail: tetamaru@med.toho-u.a.jp by the PER:TIM complex. (Edery et al. 1994; Seaz &
DOI: 10.1046/j.1365-2443.2003.00686.x
© Blackwell Publishing Limited Genes to Cells (2003) 8, 973–983 973
T Tamaru et al.

Figure 1 Characterization of the anti-mBMAL1 antibody. (A) The anti-mBMAL1 antibody was used in immunoblot analysis of (a) the
crude lysate (10 µg protein) and anti-Flag immunoprecipitates of Cos7 cells transfected with pCMV-tag2B (Mock) or pCMV-tag2B/
Flag-mBMAL1b and (b, c) of the crude lysate (10 µg protein) of mouse brains and anti-BMAL1 immunoprecipitates prepared from
200 µg of brain protein. +Antigen, the signal was absorbed with the GST-BMAL1 protein to which the antibody was raised (20 µg protein/
mL). +λPPase, an aliquot of the immunoprecipitates was treated with λ protein phosphatase. (B) NIH-3T3 fibroblasts were stained using
the anti-mBMAL1 antibody. +Antigen, the signal was absorbed with the GST-BMAL1 protein to which the antibody was raised (20 µg
protein/mL).

Young 1996). However, while this may be the crucial (Oishi et al. 2000; Shearman et al. 2000). That BMAL1
event in the fly clocks, mammalian clocks appear to is a crucial component of the molecular clockwork
involve other time delay events, since it has been suggests that the regulatory events that control its local-
reported that in mammalian clocks, the time lags ization and /or activation may serve as the crucial time-
between Per1/Per2 transcription and PER1/PER2 pro- delay events that drive the stable 24-h periodicity of
tein accumulation in the nucleus are smaller (∼4 h, 6– the circadian clock. To test this notion, in this paper, we
9 h and ∼6 h in rat-1 fibroblasts, mouse liver and SCN, examined the temporal regulation of the subcellular
respectively). Moreover, it does not appear that there is localization and phosphorylation of the BMAL1 protein
an significant accumulation of cytoplasmic PER1/ in serum-shocked NIH-3T3 fibroblasts, which are a
PER2 before the nuclear entry during the mammalian cellular model of circadian clocks (Akashi & Nishida
circadian cycle (Hastings et al. 1999; Lee et al. 2001; 2000; Balsalobre et al. 1998).
Yagita et al. 2001a,b).
BMAL1 (MOP3) is a basic helix-loop-helix-PER-
ARNT-SIM (bHLH-PAS) protein that forms a hetero- Results
dimer with CLOCK and thereby acts as the positive
Characterization of the anti-mBMAL1 antibody
feedback driver for the transcription from the E box of
circadian-responsive genes (Gekakis et al. 1998). BMAL1 We generated a highly sensitive polyclonal antibody by
also behaves as a negative feedback regulator for its own immunizing rabbits with recombinant GST-mBMAL1
gene expression through hypothetical interlocked feed- N-terminal protein. The specificity of the affinity-
back loops (Yu et al. 2002). BMAL1 mRNA and protein purified anti-mBMAL1 antibody was characterized as
levels in the SCN and other peripheral clock cells oscil- follows. Immunoblot analysis revealed that the antibody
late robustly in a circadian manner (Abe et al. 1998; Oishi reacts with Flag-mBMAL1 in Flag-immunoprecipitates
et al. 2000; Tamaru et al. 2000; Yagita et al. 2001a,b). As of transfected COS7 cell lysates (Fig. 1A(a)). Immuno-
Mop3 (Bmal1) (−/−) mice immediately display com- blot analysis of anti-BMAL1 immunoprecipitates of the
plete arrhythmic wheel-running activity in constant mouse brain also showed that the antibody recognized a
darkness, it appears that BMAL1 is an essential and non- broad band spanning 70– 80 kDa and that this recognition
redundant component in the mammalian clock (Bunger could be eliminated by adding the antigen to which the
et al. 2000). In addition, lower levels of Bmal1 transcripts antibody was raised (Fig. 1A(b)). When the anti-BMAL1
are found in all mutant mice with arrhythmic clocks, immunoprecipitates of the mouse brain were treated
including the Clock mutant, the Period2Brdm1 mutant and with λ protein phosphatase, the broad band shifted into
the Cry1-Cry2 (−/−) mutant (mCRY-deficient mice) a narrower 70 kDa band (Fig. 1A(c)). Thus, it appears

974 Genes to Cells (2003) 8, 973–983 © Blackwell Publishing Limited


Circadian BMAL1 nucleocytoplasmic shuttling

anti-mBMAL1 antibody and the immunoprecipitates


were subjected to immunoblotting with the same anti-
body.The broad 70– 80 kDa band was detected with both
fractions (Fig. 2A). λ protein phosphatase treatment
caused this band to shift and narrow into a 70 kDa band
(Fig. 2A), which indicates that the proteins in the upper
parts of the broad 70– 80 kDa signal are highly phosphor-
ylated BMAL1 proteins. When the serum-shocked NIH-
3T3 cells were subjected to immunocytostaining with
Figure 2 Phosphorylation of the BMAL1 protein in NIH-3T3
fibroblasts. (A) Anti-BMAL1 immunoprecipitates of serum-shocked
the anti-BMAL1 antibody, staining in the cytoplasm and
NIH-3T3 fibroblasts lysed 30 h after the serum shock (200 µg the nucleus was also observed (Fig. 2B), which is consistent
protein) were subjected to immunoblot analysis using the anti- with the signals detected in the immunoprecipitates.
mBMAL1 antibody. +λPPase, aliquots of the immunoprecipitates We then used immunoblot analysis to assess the changes
were treated with λ protein phosphatase. The phosphorylated in the protein levels of BMAL1 and other clock-related
(p-BMAL1) and unphosphorylated (arrow) forms of BMAL1 are proteins in the nuclear and cytosolic fractions of serum-
indicated. (B) Immunocytostaining of serum-shocked NIH-3T3 shocked NIH-3T3 fibroblasts over the 48 h after serum
fibroblasts 30 h after the serum shock with the anti-mBMAL1 shock (Fig. 3). Before serum shock, considerable amounts
antibody. of BMAL1 were detected in the nuclear and cytosolic
fractions. However, after serum shock, the total and
that the BMAL1 protein in the mouse brain is highly phosphorylated BMAL1 levels in the nuclear fraction
phosphorylated. When we performed immunocyto- showed robust circadian oscillatory patterns that peaked
staining of NIH-3T3 fibroblasts with the anti-mBMAL1 at 24 h in the first cycle after serum shock. This pattern
antibody, clear nuclear-positive signals that were specifi- was repeated in the second cycle (these data are all
cally absorbed with the antigen were observed (Fig. 1B). plotted in Fig. 5A). The proportion of phosphorylated
BMAL1 in the nucleus increased progressively until
the peaking time (∼24 h) (Fig. 3). Interestingly, in the
BMAL1 nucleocytoplasmic localization and
cytosolic fraction, the total and phosphorylated BMAL1
phosphorylation in NIH-3T3 fibroblasts show a
levels showed a different pattern. First, the total cytoplasmic
circadian pattern
levels peaked much earlier at ∼16 h (plotted in Fig. 5A).
We investigated whether the nucleocytoplasmic localiza- Second, the ratios between the hyperphosphorylated
tion and phosphorylation of BMAL1 follows a circadian and unphosphorylated forms in the cytosol are almost
rhythm by using serum-shocked NIH-3T3 fibroblasts constant.
as a model of circadian clock cells. Thus, we treated the The nuclear levels of CLOCK changed in a similar
cells with 50% serum for 2 h and then 30 h later sepa- circadian pattern to that of nuclear BMAL1 but in
rated their lysates into nuclear and cytosolic fractions. contrast showed unaltered high constitutive expression in
Each fraction was then immunoprecipitated with the the cytosol (Fig. 3). We observed that in the nucleus, the

Figure 3 Alteration of the levels of


BMAL1 and other clock-related proteins
in NIH-3T3 fibroblasts after serum shock.
NIH-3T3 fibroblasts were treated with
50% serum for 2h and then the nuclear and
cytosolic fractions were harvested at the
indicated times and subjected to immuno-
blot analysis using antibodies specific for
mBMAL1, CLOCK, CRY1, CRY2,
CKIe, active MAPK (pERK), ERK1/2
(ERK) and phospho-CREB (pCREB).
The phosphorylated (dotted line) and un-
phosphorylated (arrow) forms of BMAL1
are indicated.

© Blackwell Publishing Limited Genes to Cells (2003) 8, 973–983 975


T Tamaru et al.

Figure 4 Circadian nucleocytoplasmic


localization of BMAL1 in NIH-3T3
fibroblasts after serum shock. At various
time points after serum shock, NIH-3T3
fibroblasts were immunostained with the
anti-mBMAL1 antibody. The antibody
binding was detected using Alexa Fluor
488-labelled goat anti-rabbit IgG (Green)
as a secondary antibody. The cells were
also stained with DAPI to reveal the
nucleus (Blue).

anti-CLOCK antibody recognizes a band that includes similar circadian pattern to that of CRY1 and CRY2.
heavier proteins as well as the predicted ∼100 kDa CLOCK Thus, nuclear CKIε levels do not follow the same timing
protein (Fig. 3).These may be the phosphorylated forms as the progressive phosphorylation of nuclear BMAL1.
of the CLOCK protein, as has been proposed earlier The activated ERK protein levels showed a similar cir-
(Lee et al. 2001). However, unlike nuclear BMAL1, the cadian pattern to the profile of phosphorylated CREB
phosphorylated CLOCK levels in the nucleus did (Fig. 3), a transcription factor that is a nuclear target of
not change progressively during circadian cycle (Fig. 3). the MAP kinase pathway and that has been implicated
Moreover, cytosolic CLOCK appeared to be in the in circadian clock and photic entrainment (Gau et al.
unphosphorylated form and showed high constitutive 1996).The nuclear and cytosolic levels of activated ERKs
expression. With regard to nuclear CRY1 and CRY2, showed only weak changes during the circadian cycle.
both proteins exhibited a slightly time-delayed circadian Relative to the first cycle, the circadian patterns of the
pattern as compared with the profile of nuclear BMAL1 protein levels of activated and nuclear-translocated ERKs
(Fig. 3). and phosphorylated CREB became attenuated in the
Since CKIε and MAP kinases (ERKs) are known to second cycle. Thus, the nuclear and cytosolic levels of
phosphorylate BMAL1 in vitro (Sanada et al. 2002; Eide activated ERKs and phosphorylated CREB do not follow
et al. 2002), we also examined the circadian profiles of the same timing as the progressive phosphorylation of
these kinases. The nuclear profile of CKIε changed in a nuclear BMAL1.

976 Genes to Cells (2003) 8, 973–983 © Blackwell Publishing Limited


Circadian BMAL1 nucleocytoplasmic shuttling

Figure 5 Time-delayed accumulation of


BMAL1 in the nucleus in NIH-3T3
fibroblasts after serum shock. Total RNA
harvested from NIH-3T3 fibroblasts at
various time points after serum-shock was
subjected to fluorescence-based real-time
PCR technology (TaqMan Real-Time
PCR) to quantify Bmal1 (A) and Per1 (B
(a)) mRNA.The reactions were performed
with 40 ng total RNA/reaction. Bmal1
(A) and Per1 mRNA (B (a)) mRNA values
from triplicate experiments are shown as
means ± SEM (Blue rectangle). The levels
of BMAL1 protein in the immunoblot
shown in Fig. 3 were also quantified by
densitometry of the anti-BMAL1 positive
signal in the immunoblot using a com-
puterized image analyzer (A). The values
from triplicate experiments are shown as
means ± SEM (Nuclear; Green circle,
Cytosolic; Red triangle). Staining with
anti-PER1 antibody (Green) and DAPI
(Blue) are shown (B (b)). The values of
nuclear PER1-positive cells (%) from
triplicate experiments are plotted as means
± SEM (B (a)).

BMAL1 levels decreased rapidly while at the same time


BMAL1 nuclear accumulation shows a 12–16 h time
the total and BMAL1 levels in the cytoplasm increased
lag after Bmal1 transcription
(Figs 3, 4, 5A; 28, 29, 30, 31 and 32 h after serum-shock).
To further characterize the temporal relationships between In contrast to BMAL1, PER1 nuclear accumulation
the transcription, nucleocytoplasmic shuttling and phos- in serum-shocked fibroblasts shows only a ∼4 h time lag
phorylation of BMAL1 in serum-shocked NIH-3T3 after Per1 transcription. Per1 mRNA levels oscillated
fibroblasts, we examined Bmal1 transcription by quanti- with a peak at 28 h after the serum shock (Fig. 5B(a))
tative PCR and BMAL1 nucleocytoplasmic localization and nuclear PER1 levels showed a peak at 32 h during
by immunocytochemistry. Bmal1 mRNA levels oscillated the circadian cycle without any evidence of slow accu-
with a peak at 8 h after the serum shock in the first cycle mulation in the cytoplasm (Fig. 5B(a,b)).
(Fig. 5A). As the mRNA levels decreased, BMAL1 protein
progressively accumulated in the cytoplasm (Figs 3, 4, 5A;
The circadian subcellular localization and
12, 13, 14, 15 and 16 h after serum shock).With regard to
phosphorylation of BMAL1 are blunted in
BMAL1 protein in the nucleus, these levels also began to
CRY-deficient MEFs
increase progressively and peaked 12–16 h (at the 1st and
2nd cycles) after the Bmal1 mRNA levels peaked (Figs 3,4 We examined whether the nucleocytoplasmic localization
and 5A; ∼24 h after serum shock).Thereafter, the nuclear and phosphorylation of BMAL1 also follows a circadian

© Blackwell Publishing Limited Genes to Cells (2003) 8, 973–983 977


T Tamaru et al.

Figure 6 The circadian phosphorylation


and nucleocytoplasmic localization of
BMAL1 are blunted in CRY-deficient MEFs.
(A) Wild type [W] and CRY-deficient
MEFs [(−/−)] were treated with 50% serum
and their nuclear and cytosolic fractions
were obtained at the indicated times there-
after and subjected to immunoblot analysis
using antibodies specific for mBMAL1,
CLOCK, CRY1, CRY2, CKIε, active
MAPK (pERK) and ERK1/2 (ERK). The
phosphorylated (dotted line) and unphos-
phorylated (arrow) forms of BMAL1 are
indicated. (B) At various time points after
serum shock, wild type [Wild] and CRY-
deficient MEFs [Cry (−/−)] were immuno-
stained with the anti-mBMAL1 antibody.

rhythm in serum-shocked mCry1/mCry2-deficient mouse hyperphosphorylated forms (Fig. 6A). Again, this is sup-
embryonic fibroblasts (CRY-deficient MEFs), which ported by previous observations of the liver of the CRY-
lack a functional biological clock (Yagita et al. 2001a,b). deficient mouse, as hyperphosphorylated BMAL1 is
Previous studies of CRY-deficient MEFs have shown detected in the nuclear fraction in the liver, although not
that the RNA levels of BMAL1 are arrhythmic and are in the cytoplasm (Lee et al. 2001).
maintained at moderate levels during the circadian cycle. The nuclear levels of CLOCK in CRY-deficient MEFs
We found that cytoplasmic and nuclear BMAL1 protein were at substantially higher levels, which is due to increase
levels were also arrhythmic and maintained at moderate of the unphosphorylated forms (Fig. 6A). Cytosolic acti-
levels (Fig. 6A). Moreover, unlike what is observed in the vated ERKs showed lower levels in CRY-deficient MEFs
wild-type MEFs, circadian subcellular BMAL1 localiza- before serum shock but slightly higher levels after serum
tion changes was not observed in CRY-deficient MEFs, shock (Fig. 6A). Nuclear total ERKs and nuclear/cytosolic
even at unicellular levels, as revealed by immunocyto- CKIε proteins were at higher levels in CRY-deficient
chemistry that showed staining in nuclei and cytoplasms MEFs, but the activated ERKs did not differ significantly
at moderate intensities (Fig. 6B). Supporting these between the wild-type and CRY-deficient MEFs (Fig. 6A).
observations is that it has been previously reported that
the cytoplasmic and nuclear BMAL1 levels in the liver of
The circadian nuclear export of BMAL1 is inhibited
the CRY-deficient mouse are also arrhythmic and are
with Leptomycin B
maintained at substantially reduced levels (Lee et al. 2001).
In addition, in the CRY-deficient MEFs, the cytoplasmic To demonstrate that the nucleocytoplasmic shuttling of
and nuclear BMAL1 proteins were constantly detected as BMAL1 in the cultured fibroblasts is controlled by the

978 Genes to Cells (2003) 8, 973–983 © Blackwell Publishing Limited


Circadian BMAL1 nucleocytoplasmic shuttling

nuclear export of BMAL1 was likely to be occurring,


dramatically increased the dominant nuclear BMAL1
signals in NIH-3T3 cells (Fig. 7A(a–c)) and wild-type
MEFs but not in CRY-deficient MEFs (Fig. 7B).Without
cycloheximide, BMAL1 also accumulated in the nucleus
after Leptomycin B treatment (Fig. 7A(c)). CLOCK
protein also accumulated in the nucleus after Leptomycin
B treatment with a similar pattern to BMAL1. These
data clearly demonstrate that BMAL1 and CLOCK are
exported from the nucleus via the CRM-1-mediated
pathway in a circadian clock-dependent manner.We also
addressed the effect of cycloheximide on the expression
pattern of BMAL1. Only the highly phosphorylated forms
of BMAL1 remained in cycloheximide-treated cells
(Fig. 7A(c)).

Discussion
We found that in cells with a functional clock, BMAL1
undergoes progressive cytoplasmic accumulation followed
by a much later nuclear entry. Thus, in serum-shocked
NIH 3T3 cells and wild-type MEFs, we found that
BMAL1 protein progressively accumulates in the cyto-
plasm with peaks observed at 12–16 h after serum shock.
This is followed by a diminishment in the cytoplasmic
protein levels and a concomitant increase of nuclear
BMAL1 protein with a peak at ∼24 h. Thereafter, the
nuclear BMAL1 levels exhibit a dramatic reduction at
28– 32 h. This is due in part to nuclear export since at
Figure 7 Effect of Leptomycin B on the circadian this time point, the cytoplasmic BMAL1 levels start to
nucleocytoplasmic localization of BMAL1:CLOCK in cultured increase again while Bmal1 gene expression in the second
fibroblasts. After adding cycloheximide (50 µg/mL; Calbiochem- cycle has not yet begun. Strongly supporting this notion
Novabiochem) or methanol (as a control) 18hr after serum shock, is that this increase of cytoplasmic BMAL1 is dramati-
NIH-3T3 fibroblasts (A), wild type [Wild] and CRY-deficient cally inhibited by the nuclear export inhibitor Leptomycin
MEFs [Cry (−/−)] (B) were treated with Leptomycin B (LMB) B. Proteolysis may also contribute to the rapid decline
(10 ng/mL; Calbiochem-Novabiochem) or methanol (Control) of nuclear BMAL1 because cycloheximide-treated cells
from 24–30 hr after serum shock. Samples of the fibroblasts were that lack de novo protein synthesis showed significantly
analysed with immunocytochemistry (A (a, b), B) or immuno-
lower nuclear levels of BMAL1. It may be that certain
blotting (A (c); nuclear [N] and cytosolic [C] fractions) using the
anti-mBMAL1 and anti-CLOCK antibody. Staining with anti-
temporally specific dephosphorylation events may con-
BMAL1 antibody (Green) and DAPI (Blue) are shown (A (a), B). tribute to the destabilization of BMAL1 because only the
The values of nuclear BMAL1-dominant cells (%) in NIH-3T3 highly phosphorylated forms of BMAL1 remain in both
cells from triplicate experiments are plotted as means ± SEM (A the nucleus and the cytoplasm at this time point. There-
(b)). (***P < 0.001). after, cytoplasmic BMAL1 levels reach a second peak at
∼32 h.
These observations indicate that the BMAL1 protein
circadian clock, we examined the effect of Leptomycin undergoes temporally regulated subcellular localization
B, a potent inhibitor of CRM-1-mediated nucleoex- in mammalian cells that have a functional circadian clock.
port, on the nucleocytoplasmic shuttling of BMAL1. To This temporal distance between transcription and nuclear
eliminate de novo synthesized BMAL1 protein, the cells localization is larger than has been reported previously
were pretreated with cycloheximide for 6 h (18–24 h for any other mammalian clock protein. The temporal
after serum shock). Treatment with Leptomycin B dur- regulation of BMAL1 localization is supported by the
ing the 24– 30 h after serum shock, when the circadian fact that the subcellular localization of BMAL1 protein

© Blackwell Publishing Limited Genes to Cells (2003) 8, 973–983 979


T Tamaru et al.

temporal regulation of endogenous BMAL1 protein


localization.We also demonstrated that the nucleoexport
of BMAL1 in these cells involves the CRM-1-mediated
pathway In CRY-deficient MEFs that lack a functional
biological clock, the nucleoexport-blocking effect of
Leptomycin B was not observed. This suggests that
nuclear export of BMAL1 is largely dependent on the
circadian clock. It may be that circadian time-specific
BMAL1 phosphorylation events may be required before
its CRM-mediated nucleoexport can take place. These
observations of the nucleocytoplasmic shuttling of BMAL1
are supported by the presence in the mBMAL1 protein
Figure 8 Nuclear-dominant localization of BMAL1 in the of several well-conserved motifs that are involved in
mouse liver throughout the circadian cycle. (A) The nuclear and nucleocytoplasmic localization (Seaz & Young 1996;
cytosolic fractions of mouse livers harvested at the indicated Yagita et al. 2001a,b), namely, two well-conserved nuclear
circadian time points were subjected to immunoblot analysis using localization signals (NLS) in the N-terminal region
anti-mBMAL1 antibody. The phosphorylated (dotted line) and (N36RKRK) and the bHLH region (K82RRRR), two
unphosphorylated (arrow) forms of BMAL1 are indicated. (B) At
well-conserved nucleoexport signals (NES) in the PAS B
circadian time 10 and 22, the BMAL1 in the mouse liver was
region (L311SCLVAIGRL and I360LAYLPQELL), and a
immunohistochemically detected with the anti-mBMAL1 anti-
body. The antibody binding was detected using Alexa Fluor 488- partially conserved cytoplasmic localization domain (CLD)
labelled goat anti-rabbit IgG as a secondary antibody. The cells in amino acids 398– 460.
were also stained with PI to reveal the nucleus. Nuclear BMAL1 protein in serum-shocked fibroblasts
was gradually phosphorylated and dephosphorylated in
a temporal manner, but in contrast cytoplasmic BMAL1
in CRY-deficient MEFs, which lack a functional circa- protein exhibited the same phosphorylation pattern
dian clock, do not change over time. A recent report throughout the circadian cycle. This indicates that the
has suggested that in the mouse liver, BMAL1 does phosphorylation of nuclear and cytoplasmic BMAL1
not accumulate significantly in the cytoplasm before proteins may be differently regulated and that the phos-
the nuclear entry (Lee et al. 2001). We also observed this phorylation of nuclear BMAL1 protein may be regulated
when we subjected mouse liver to immunohistochemical by circadian clock mechanisms. Supporting this notion
and immunoblot analysis (Fig. 8). This difference sug- is that in CRY-deficient MEFs, which lack a functional
gests that BMAL1 may be subjected to different post- biological clock, both nuclear and cytoplasmic BMAL1
translational regulatory mechanisms in the molecular proteins consisted of moderate levels of hyperphosphory-
clockworks of different cell types. Nevertheless, another lated forms throughout the entire circadian cycle. As
recent report (Yagita et al. 2001a,b) has suggested that reported recently, BMAL1 is also temporally phosphory-
cultured fibroblasts exhibit similar circadian gene expres- lated in the mouse liver (Lee et al. 2001). However, the
sion profiles to the SCN. This supports the importance circadian profile of BMAL1 phosphorylation in the liver
of cultured fibroblasts as a model for the central circadian is different from that in NIH-3T3 cells. Moreover, in
clock. Moreover, according to a recent report that was contrast to wild-type and CRY-deficient fibroblasts,
published during the preparation of this paper, it suggests cytoplasmic BMAL1 in the liver of both wild-type and
that the similar post-translational regulatory mechanisms CRY-deficient mice do not show hyperphosphorylated
that control subcellular localization of CLOCK, which forms. It may be that the BMAL1 protein in different
is shown to undergo BMAL1-dependent nuclear entry, cells is subjected to different phosphorylation events.
in fibroblasts may also occur in the SCN (Kondratov Further analyses will be required to adequately define
et al. 2003). However, further analyses will be required the circadian phosphorylation of BMAL1 in different
to adequately define the circadian regulatory mecha- cell types, especially SCN neurones.
nisms that drive the subcellular localization of BMAL1 in cBMAL1 can be phosphorylated in vitro by ERK at
different cell types, especially neurones in the SCN. Ser-527, Thr-534 and Ser-599, which are largely con-
Our immunocytochemical analysis shows that in cells served in mBMAL1 (Sanada et al. 2002). Moreover,
with a functional clock, BMAL1 slowly accumulates in CKIε can also phosphorylate BMAL1 in vitro (Eide et al.
the cytoplasm before its nuclear entry.These data consti- 2002) and we found a variety of predicted Ser/Thr
tute the first immunocytochemical evidence for the phosphorylation motifs in BMAL1. However, we found

980 Genes to Cells (2003) 8, 973–983 © Blackwell Publishing Limited


Circadian BMAL1 nucleocytoplasmic shuttling

in this paper that the temporal changes of nuclear ERKs CRY2, the major negative regulator of BMAL1:CLOCK-
and CKIε do not significantly match the timing of the driven gene expression, is slightly delayed compared to
progressive phosphorylation of nuclear BMAL1. These the timing of the nuclear accumulation of BMAL1
data do not necessarily exclude the possibility that ERKs and CLOCK and peaks when Per1 gene expression is
and CKIε phosphorylate BMAL1 in vivo, but they do repressed (Akashi & Nishida 2000). These observations
suggest that other kinases or phosphatases may be strongly suggest that the BMAL1:CLOCK directly drives
responsible for the circadian phosphorylation states of the circadian expression of Per1 in NIH-3T3 cells that
BMAL1. However, in CRY-deficient cells, higher levels have been stimulated to enter a circadian rhythm. It is
of nuclear and cytoplasmic CKIε protein are observed also possible that nuclear phosphorylated BMAL1 plays
along with hyperphosphorylation of the nuclear and an important role in the transcriptional activation by
cytoplasmic BMAL1 proteins in these cells.This suggests BMAL1:CLOCK because the phosphorylation pattern
that CKIε may be involved in these hyperphosphoryla- of nuclear BMAL1 also matches the transcriptional
tion events in CRY-deficient cells. Notably, in the pattern of Per1. Supporting this notion is that a repor-
mouse liver (Lee et al. 2001), the lack of nuclear accu- ter assay employing the Per1-promoter revealed that
mulation of CKIε correlates with the lack of nuclear BMAL1 phosphorylation by MAP kinase and CKIε
PER1/PER2. As reported previously, in CRY-deficient affects BMAL1:CLOCK-mediated transcriptional activity
MEFs, PER1 protein can enter the nucleus (Yagita et al. (Eide et al. 2002; Sanada et al. 2002).
2000). These observations together suggest that CKIε In summary, our observations suggest that the subcel-
may also be involved in the nuclear localization of PER1. lular localization and phosphorylation states of BMAL1,
Thus, CKIε may play different roles in the nucleocyto- a positive limb in the circadian core feedback loops
plasmic localization of clock proteins in different cell involved in the transcription from the E box, are impor-
types. tant events in the stably oscillating molecular clockwork
Unlike CKIε, the levels of activated ERKs are not sig- in cultured fibroblasts, as well as in negatively regulating
nificantly elevated in CRY-deficient cells. This suggests limbs such as PER and CRY. Supporting this notion is
that these kinases are not responsible for the constantly that the temporal Per1 transcription pattern matches the
hyperphosphorylated nuclear and cytoplasmic BMAL1 timing of the time-delayed nuclear entry, accumulation
proteins in these cells. and phosphorylation of BMAL1, which suggests that
The changing levels of nuclear CLOCK protein over BMAL1 acts as a positive driver in the molecular circa-
circadian time matched the circadian pattern of nuclear dian clockwork.
BMAL1 levels. This suggests that the subcellular locali-
zation of BMAL1 affects the localization of CLOCK.
Supporting this is that the subcellular localization of Experimental procedures
CLOCK in CRY-deficient MEFs does not change
Preparation of antibodies
throughout the entire experimental period. Moreover,
according to a recent report that was published during Glutathione S-transferase (GST)-mBMAL1 N-terminal (amino
the preparation of this paper, the circadian clock- acids 1–73; this peptide shows no homology to BMAL2) (Hogenesch
controlled nuclear entry of CLOCK is impaired in et al. 2000) and GST-mCRY2 C-terminal (amino acids 524–592)
BMAL1-deficient MEFs, and hyperphosphorylation of antigens were produced and then used to immunize rabbits as
BMAL1 is dependent on its interaction with CLOCK previously described (Tamaru et al. 2000). After removing possible
GST-recognizing antibodies by using Hitrap-NHS conjugated
(Kondratov et al. 2003). We demonstrated in our paper
with GST, the antisera were subjected to affinity purification using
here that BMAL1 and CLOCK undergo clock-controlled Hitrap-NHS conjugated with the antigen. Both the anti-mBMAL1
nucleoexport via the CRM-1-mediated pathway. From antibody and the anti-mCRY2 antibody specifically recognize
these findings, we propose that the nucleocytoplasmic their target proteins in immunochemical analysis (Fig. 1 and data
shuttling and interaction of BMAL1:CLOCK is controlled not shown).
by circadian clock-dependent post-transcriptional events
such as temporal phosphorylation.
The timing of the nuclear accumulation of BMAL1 Cell culture, preparation of tissue samples,
and CLOCK fits well with the expression of the clock- immunoprecipitation and immunoblot analysis
responsive genes Per1, which are the targets of the Cell culture and the serum-shocking of NIH-3T3 fibroblasts were
BMAL1:CLOCK complex, as the expression of these performed according to protocols described by Akashi and
genes peaks 20–24 h after serum shock (Akashi & Nishida Nishida (Akashi & Nishida 2000). Flag-mBMAL1 was over-
2000). Moreover, the nuclear accumulation of CRY1/ expressed by transfecting COS7 cells with mBMAL1b cDNA

© Blackwell Publishing Limited Genes to Cells (2003) 8, 973–983 981


T Tamaru et al.

ligated into pCMV-Tag2B (Stratagene Cloning Systems, La Jolla,


CA, USA) using Lipofectamine 2000 (Invitrogen, Carlsbad, Quantitative PCR
USA). The mouse brain and liver samples were prepared from At the indicated time points, total RNA was extracted using an
male 8-week-old C57/BL6 mice (Sankyo Laboratory Service, RNAwith™ and DNA-free kit (Ambion, Austin, TX, USA).
Tokyo, Japan). Nuclear and cytosolic fractions at the indicated RNA quantification was performed using fluorescence-based
time points were prepared as previously described (Tamaru et al. real-time PCR technology (TaqMan Real-Time PCR, Applied
2000). For immunoprecipitation analysis, the samples were Biosystems, Tokyo, Japan) as previously described (Tamaru et al.
incubated at 4 °C for 16 h with 1 µg of antibody in TTBS 2003) using the following probes (5′-labelled with FAM and 3′-
(Tris-buffered saline; TBS with 0.1% Tween-20) containing Ser/ labelled with TAMRA) and primers: mBMAL1-For, GAC CTT
Thr phosphatase inhibitor cocktail (Sigma-Aldrich, Tokyo, Japan) ATT GGC CAG AGC TTG TT; mBMAL1-Rev, CGC AGT
and complete™ protease inhibitor cocktail (Roche Diagnostics, GTC CGA GGA AGA TAG; mBMAL1-Probe, CCT GCA TCC
Tokyo, Japan). After incubating with Protein G-Sepharose AAA AGA TAT TGC CAA AGT TAA GGA; the probes and
(Amersham Pharmacia Biotech, Tokyo, Japan), the beads were primers for mPER1 were previously described (Tamaru et al.
washed with TTBS. The immunoprecipitates were treated with 2003). As a control, GAPDH mRNA was amplified from the
l protein phosphatase (500 units) (Calbiochem-Novabiochem, same RNA samples. The relative mRNA abundance was calcu-
Darmstadt, Germany) at 30 °C for 2 h. Immunoblot analyses lated using the comparative delta-Ct method according to the
were performed as previously described (Tamaru et al. 2000). manufacturer’s protocol.
The following antibodies were used: anti-active MAPK antibody
(Promega, Tokyo, Japan), anti-CLOCK antibody (Calbiochem-
Novabiochem), anti-CKIe antibody (Takano et al. 2000), anti- Acknowledgements
ERK1/2 and anti-phospho CREB antibodies (Upstate, Lakeplacid,
The authors thank Drs Hitoshi Okamura, Masaaki Ikeda and
USA), anti-PER1 antibody (kindly donated by Dr Okamura;
Steven M. Reppert for kindly donating the anti-PER1 antibody,
Yagita et al. 2001a,b) and anti-CRY1 antibody (Alpha Diagnostic,
the mBMAL1 cDNA, and the mCRY1/mCRY2 cDNA, respec-
San Antonio, TX, USA).
tively. This work was supported by a grant-in-aid for Scientific
Research from the Ministry of Education, Culture, Sports, Science
Immunocytochemistry and Technology of Japan (to T. Tamaru and K. Takamatsu), the
Scientific Research Promotion Fund from the Japan Private School
At the indicated time points, the cells cultured on slide glasses were Promotion Foundation (to K.Takamatsu), and the Project Research
fixed with 4% paraformaldehyde in TBS at room temperature for Grant from Toho University School of Medicine (to T.Tamaru).
30 min. After a brief incubation with 0.3% Triton X-100 in TBS, This work was performed within the Cooperative Research Pro-
blocking was performed with BLK buffer (TBS with 0.5% BSA gram of the Institute for Protein Research, Osaka University.
and 0.1% Triton X-100) at room temperature for 1 h. The speci-
mens were incubated with the anti-mBMAL1 antibody (20 µg/
mL in BLK) at 4 °C for 48 h. The antibody binding was detected References
by using Alexa Fluor 488-labelled goat anti-rabbit IgG as a sec- Abe, H., Honma, S., Namihira, M., Tanahashi, Y., Ikeda, M. &
ondary antibody, while nuclear staining was performed by using Honma, K. (1998) Circadian rhythm and light responsiveness of
DAPI (Molecular Probes, Eugene, USA). The slides were then BMAL1 expression, a partner of mammalian clock gene Clock,
visualized by laser confocal microscopy (Bio-Rad Laboratory, in the suprachiasmatic nucleus of rats. Neurosci. Lett. 258, 93 –96.
Tokyo, Japan). Akashi, M. & Nishida, E. (2000) Involvement of the MAP kinase
cascade in resetting of the mammalian circadian clock. Genes
Dev. 14, 645–649.
Immunohistochemistry
Balsalobre, A., Damiola, F. & Schibler, U. (1998) A serum shock
Mice were deeply anaesthetized under a dim red light and per- induces circadian gene expression in mammalian tissue culture
fused intracardially with ice cold saline and then with 4% parafor- cells. Cell 93, 929–937.
maldehyde (Sigma, St Louis, MO, USA) in phosphate-buffered Bunger, M.K., Wilsbacher, L.D., Moran, S.M., et al. (2000) Mop3
saline (PBS). Livers were removed, post-fixed in 4% PFA in PBS is an essential component of the master circadian pacemaker in
for 2 h at room temperature, and cryoprotected in 20% sucrose for mammals. Cell 103, 1009–1017.
two nights. The brains were sliced using a cryomicrotome Crosthwaite, S.K., Dunlap, J.C. & Loros, J.J. (1997) Neurospora
(CM1900, Leica, Germany) into 10 µm-thick sections. Immuno- wc-1 and wc-2: transcription, photoresponses, and the origins
histochemical analyses of BMAL1 were performed by using anti- of circadian rhythmicity. Science 276, 763–769.
mBMAL1 antibody (20 µg/mL) as the primary antibody. The Darlington, T.K., Wager-Smith, K., Ceriani, M.F., et al. (1998)
antibody binding was detected by using Alexa Fluor 488-labelled Closing the circadian loop: clock-induced transcription of its
goat anti-rabbit IgG as a secondary antibody, while nuclear stain- own inhibitors per and tim. Science 280, 1599–1603.
ing was performed by using PI (Molecular Probes) according to Edery, I., Zwiebel, L.J., Dembiska, M.E. & Rosbash, M. (1994)
the manufacturer’s instruction.This was followed by laser confocal Temporal phosphorylation of the Drosophila period protein.
microscopy (Bio-Rad). Proc. Natl. Acad. Sci. USA 91, 2260–2264.

982 Genes to Cells (2003) 8, 973–983 © Blackwell Publishing Limited


Circadian BMAL1 nucleocytoplasmic shuttling

Eide, E.J., Vielhaber, E.L., Hinz, W.A. & Virshup, D.M. (2002) Reppert, S.M. & Weaver, D.R. (2002) Coordination of circadian
The circadian regulatory proteins BMAL1 and Cryptochromes timing in mammals. Nature 418, 935–941.
are substrates of Casein kinase Iε. J. Biol. Chem. 277, 17248– Sanada, K., Okano, T. & Fukada, Y. (2002) Mitogen-activated
17254. protein kinase phosphorylates and negatively regulates basic
Gau, D., Lemberger, T., von Gall, C., et al. (1996) Phosphoryla- helix-loop-helix-PAS transcription factor BMAL1. J. Biol.
tion of CREB Ser142 regulates light-induced phase shifts of the Chem. 277, 267–271.
circadian clock. Neuron 34, 245 – 253. Scheper, T.O., Klinkenberg, D., Pennartz, C. & Pelt, J.V. (1999)
Gekakis, N., Staknis, D., Nguyen, H.B., et al. (1998) Role of the A mathematical model for the intracellular circadian rhythm
clock protein in the mammalian circadian mechanism. Science generator. J. Neurosci. 19, 40–47.
280, 1564–1569. Seaz, L. & Young, M. (1996) Regulation of the nuclear entry of
Hastings, M.H., Field, M.D., Maywood, E.S., Weaver, D.R. & the Drosophila clock proteins Period and Timeless. Neuron 17,
Reppert, S.M. (1999) Differential regulation of mPER1 911–920.
and mTIM proteins in the mouse suprachiasmatic nuclei: Shearman, L.P., Sriram, S., Weaver, D.R., Maywood, E.S., et al.
new insights into a core clock mechanism. J. Neurosci. 19, (2000) Interacting molecular loops in the mammalian circadian
RC11. clock. Science 288, 1013–1019.
Hogenesch, J.B., Gu, Y.-Z., Moran, S.M., et al. (2000) The basic Takano, A., Shimizu, K., Kani, S., Buijs, R.M., Okada, M. &
Helix-Loop-Helix-PAS protein MOP9 Is a brain-specific het- Nagai, K. (2000) Cloning and characterization of rat casein
erodimeric partner of circadian and hypoxia Factors. J. Neurosci. kinase 1epsilon. FEBS Lett. 477, 106–112.
20, RC83. Tamaru, T., Isojima, Y., Nagai, K. & Takamatsu, K. (2003)
Ishiura, M., Kutsuna, S., Aoki, S., et al. (1998) Expression of a Circadian expression of hnRNP U, a nuclear multi-potent
gene cluster kaiABC as a circadian feedback process in cyano- regulatory protein, in the murine suprachiasmatic nucleus.
bacteria. Science 281, 1519 –1523. Neurosci. Lett. 341, 111–114.
Kondratov, R.V., Chernov, M.V., Kondratova, A.A., Gorbacheva, Tamaru, T., Isojima, Y., Yamada, T., Okada, M., Nagai, K. &
V.Y., Gudkov, A.V. & Antoch, M.P. (2003) BMAL1-dependent Takamatsu, K. (2000) Light and glutamate-induced degrada-
circadian oscillation of nuclear CLOCK: posttranslational tion of the circadian oscillating protein BMAL1 during the
events induced by dimerization of transcriptional activators mammalian clock resetting. J. Neurosci. 20, 7525–7530.
of the mammalian clock system. Genes Dev. 17, 1921– Yagita, K., Tamanini, F., van der Horst, G.T.J. & Okamura, H.
1932. (2001a) Molecular mechanisms of the biological clock in
Lee, C., Etchegaray, J.P., Cagampang, F.R.A., Loudon, S.I. & cultured fibroblasts. Science 292, 278–281.
Reppert, S.M. (2001) Posttranslational mechanism regulate the Yagita, K., Tamanini, F., Yasuda, M., van der Horst, G.T.J. &
mammalian circadian clock. Cell 107, 855 – 867. Okamura, H. (2001b) Nucleocytoplasmic shuttling and mCRY-
Lema, M.A., Golombek, D.A. & Echave, J. (2000) Delay dependent inhibition of ubiquitylation of the mPER2 clock
model of the circadian pacemaker. J. Theor. Biol. 204, 565– protein. EMBO J. 21, 1301–1314.
573. Yagita, K., Yamaguchi, S., Tamanini, F., et al. (2000) Dimerization
Moore, R.Y. (1997) Circadian rhythms: basic neurobiology and and nuclear entry of mPER proteins in mammalian cells. Genes
clinical applications. Annu. Rev. Med. 48, 253 –266. Dev. 14, 1353–1363.
Morse, D. & Sassone-Corsi, P. (2002) Time after time: inputs to Yu, W., Nomura, M. & Ikeda, M. (2002) Interactivating feedback
and outputs from the mammalian circadian oscillators. Trend loops within the mammalian clock: BMAL1 is negatively
Neurosci. 25, 632– 637. autoregulated and upregulated by CRY1. CRY2 and PER2.
Oishi, K., Fukui, H. & Ishida, N. (2000) Rhythmic expression of Biochem. Biophy. Res. Commun. 290, 933–941.
BMAL1 mRNA is altered in Clock mutant mice: differential
regulation in the suprachiasmatic nucleus and peripheral tissues. Received: 22 July 2003
Biochem. Biophy. Res. Commun. 268, 164 –171. Accepted: 18 September 2003

© Blackwell Publishing Limited Genes to Cells (2003) 8, 973–983 983

You might also like