into these processes. Kanai et al. constructed affinity
columns composed of various regions of one kinesin
isoform, KIF5, fused to GST and then applied extracts
derived from the mouse brain to them. One particular
region of KIF5, a 59 amino acid “minimal binding se-
quence” in the tail (as opposed to the motor domain)
interacted with up to 42 proteins, which were identified
by mass spectrometry or immunoblotting. These pro-
teins include several RNA helicases (e.g., DDX1, vasa),
hnRNP proteins (e.g., hnRNP U, hnRNP A1), factors in-
volved in general protein synthesis (e.g., EF-1?, eIF2?,
lation (e.g., FMR1, FXR1), and not surprisingly, one fac-
tor already thought to be involved in RNA transport
(staufen) (Kiebler et al., 1999; Tang et al., 2001). The
authors show that the proteins reside in huge (up to
?1000 S) complexes, together with the mRNAs encod-
ing CamKII? and ARC (activity regulated cytoskeleton-
associated protein). At least based on size and their
inclusion of staufen, eIF2?, and CaMKII? mRNA, these
complexes may be analogous to those identified by Kri-
chevsky and Kosik (2001), who isolated huge RNP parti-
cles from polysome sucrose gradients of rat brain ex-
tract. Ultrastructural analysis of the Krichevsky and
Kosik particles showed that they contained ribosomes,
but because the initiation factors eIF4E and eIF4G were
not detected, they assumed that the particles were not
engaged in mRNA translation. With the exception of one
ribosomal protein (L3), Kanai et al. surprisingly did not
detect any other component of the ribosome. Perhaps
a possible association of ribosomes with KIF5 is too
indirect or weak to be retained on an affinity column.
Kanai et al. next obtained or generated antibody
against many of the proteins and performed an exten-
sive series of coimmunoprecipitation experiments and
found that RNA was necessary for most of the proteins
to reside in the complex. Thus, many of the interactions
are probably indirect and occur through RNA intermedi-
ates. More interesting, immunocytochemical analysis of
cultured mouse hippocampal neurons revealed that
many of the proteins are in dendrites. Furthermore, a
double in situ hybridization for CaMKII? mRNA and indi-
rect immunofluorescence for one of the proteins in the
complex (Pur ?, an RNA binding protein) showed them
to be colocalized. While additional experiments demon-
strated that particle movement was reduced by a domi-
nant-negative KIF5, the pie `ce de re ´sistance of the paper
clearly is found in the final figure; here the authors use
siRNA to knock down six of the factors and find that in
tract binding protein-associated factor, and staufen),
the transport of an CaMKII? 3? UTR in dendrites is sub-
whether the transport of other mRNAs to dendrites is
reduced when these or other factors are knocked down
Eom, T., Antar, L.N., Singer, R.H., and Bassell, G.J. (2003). J. Neu-
rosci. 23, 10433–10444.
Huang, Y.-S., Carson, J.H., Barbarese, E., and Richter, J.D. (2004).
Genes Dev. 17, 638–653.
Kanai, Y., Dohmae, N., and Hirokawa, N. (2004). Neuron 43, this
Kiebler, M.A., Hemraj, I., Verkade, P., Kohrmann, M., Fortes, P.,
Marion, R.M., Ortin, J., and Dotti, C.G. (1999). J. Neurosci. 19,
Krichevsky, A.M., and Kosik, K.S. (2001). Neuron 32, 683–696.
Martin, K.C., Barad, M., and Kandel, E.R. (2000). Curr. Opin. Neuro-
biol. 10, 587–592.
Mayford, M., Baranes, D., Podsypanina, K., and Kandel, E.R. (1996).
Proc. Natl. Acad. Sci. USA 93, 13250–13255.
Miller, S., Yasuda, M., Coats, J.K., Jones, Y., Martone, M.E., and
Mayford, M. (2002). Neuron 36, 507–519.
Ouyang, Y., Rosenstein, A., Kreiman, G., Schuman, E.M., and Ken-
nedy, M.B. (1999). J. Neurosci. 19, 7823–7833.
Silva, A.J., Stevens, C.F., Tonegawa, S., and Wang, Y. (1992). Sci-
ence 257, 206–211.
Tang, S.J., Meulemans, D., Vazquez, L., Colaco, N., and Schuman,
E. (2001). Neuron 32, 463–475.
A Rhythmic Ror
The circadian clock mechanism in mammals involves
two interlocking transcriptional feedback loops. Rev-
erb ?, through its role as a transcriptional repressor,
back loop that regulates Bmal1 transcription. Results
reported by Sato et al. in this issue of Neuron now
show that the transactivator Rora acts coordinately
with Rev-erb ? and that their competing activities on
the same promoter element drive the rhythm in Bmal1
transcription. This finding defines the second feed-
back loop in mammals.
Organisms time their physiology and behavior to cope
with the predictable daily alterations in the environment
that result from the Earth’s rotation. Such biological
timing is controlled by genetically determined timers
oscillate for months in constant conditions, while main-
taining a remarkably stable period length of about 24
hr. A key question in circadian biology is how a stable
molecular oscillator with such a long cycle length is
generated. In this issue of Neuron, Sato et al. (2004)
take us a step closer to answering this question by
identifying a new element important for adding stability
to the mammalian clock mechanism.
The circadian clocks of animals were initially envi-
sioned to be comprised of a single intracellular negative
transcriptional feedback loop (Figure 1, core loops). In
Drosophila (the organism in which the loop was first
described), the model posited that the heterodimeric
transcription factor complex of dCLOCK/dCYCLE (dCLK/
and timeless (tim) by binding to E box elements in their
Joel D. Richter
Program in Molecular Medicine
University of Massachusetts Medical School
Worcester, Massachusetts 01605
Figure 1. The Mammalian and Drosophila
Circadian Clockworks Are Each Comprised
of Two Interlocked Feedback Loops
The core loop determines the period and the
amplitude of circadian oscillations, while the
second (stabilizing loop) is important to fine-
tune and stabilize these oscillations. PDP1
but the possibility that they heterodimerize
with an unknown partner to bind their target
sequence has not been excluded. Abbrevia-
tions: VP box, VRI/PDP box; RRE, Rev-erb/
promoters (Stanewsky, 2002). The resultant PER and
TIM proteins repress their own transcription by forming
dCLK/dCYC activity. A circadian rhythm with a stable
24 hr period would be achieved by delaying PER and
TIM nuclear accumulation via the activity of a set of
kinases regulating PER and TIM stability (Stanewsky,
The negative transcriptional feedback loop of mam-
mals is similar (Reppert and Weaver, 2002), with the
orthologs of dCLK and dCYC, CLOCK and BMAL1, driv-
ing, through E box enhancers, the rhythmic expression
of three Period genes (Per1–Per3, with Per1 and Per2
tochrome genes (Cry1 and Cry2). The resultant PER and
back into the nucleus to inhibit CLOCK/BMAL1-medi-
ated transcription. Posttranslational mechanisms are
also felt to contribute to the time delays needed for a
24 hr clock in mammals (Reppert and Weaver, 2002).
The singlefeedback loopmodel waschallenged afew
years ago when a second, interlocking transcriptional
feedback loop was discovered that involves the circa-
dian regulation of dClk transcription in flies and Bmal1
in mammals (Figure 1; Glossop et al., 1999; Shearman
et al., 2000). In Drosophila, the search for regulators of
dClk transcription led to the identification of a negative
regulator, VRILLE (VRI), and a positive regulator, PDP1
(Blau and Young, 1999; Cyran et al., 2003; Glossop et
al., 2003). These proteins belong to the same subfamily
of basic leucine zipper transcription factors and com-
pete for the same binding site in the dClk promoter. The
vri and pdp1 genes are themselves regulated positively
by dCLK through E box enhancer elements. The discov-
closed the second feedback loop in flies.
Inmammals, searchfor regulatorsof Bmal1transcrip-
tion initially focused on Rev-erb ?, which was identified
of Bmal1 expression (Preitner et al., 2002). Rev-erb ? is
a transcription factor that belongs to the retinoic acid-
related orphan receptor family. Proteins of this nuclear
receptor family bind as monomers to Rev-erb/Ror ele-
ments in responsive genes, with Rev-erb ? and Rev-erb
? acting as transcriptional repressors, while Rora, Rorb,
and Rorc act as transactivators. Indeed, Rev-erb ? was
found to bind to Rev-erb/Ror elements in the Bmal1
promoter. Moreover, there is a circadian rhythm in Rev-
erb ? mRNA levels that is antiphase to the Bmal1 mRNA
rhythm in both the master brain clock, the suprachias-
matic nuclei (SCN), and in an SCN-entrained peripheral
clock in liver, as would be expected for a Bmal1 repres-
sor (Preitner et al., 2002; Ueda et al., 2002). In Rev-erb
? knockout mice, Bmal1 mRNA levels in both tissues
are constantly high, with little daily variation (Preitner
et al., 2002). Surprisingly robust circadian behavioral
rhythms persist in the knockout animals, with a modest
(0.38 hr) reduction in period length. However, Rev-erb
? knockout mice do show a greater interindividual vari-
ability of period length, compared to wild-type mice,
indicating that the precision of the circadian pacemaker
is affected in the knockout animals. Interestingly, the
Rev-erb ? mRNA rhythm is probably directly regulated
by CLOCK/BMAL1 heterodimers, through E box ele-
ments in the Rev-erb ? promotor, interconnecting the
two feedback loops. It was thus proposed that Rev-erb
?, through its rhythmic repressive action, is the primary
determinant of the circadian oscillation of Bmal1 tran-
scription (Preitner et al., 2002).
to be a negative regulator of Bmal1 transcription, Ueda
et al. (2002) discovered that the Bmal1 Rev-erb/Ror ele-
ments alone can drive rhythmic expression of a minimal
promoter in cell culture, suggesting that the Rev-erb/
role.As aconsequence,it seemedlikelythat atransacti-
vator, able to bind to the Rev-erb/Ror elements, was
also involved in the regulation of Bmal1 and other Rev-
erb/Ror-dependent target genes. The Ror proteins thus
became viable candidates as positive rhythmic regula-
tors of Bmal1 transcription. In support of this proposed
role in Bmal1 transactivation is the fact that Rora mRNA
levels are under circadian regulation in the SCN, while
Rorc mRNA levels show circadian oscillation in liver; the
peak of the oscillation in each tissue roughly coincides
with the peak in Bmal1 mRNA levels (Preitner et al.,
2002; Ueda et al., 2002).
Now Sato and colleagues (2004) have moved the Ror
part of the circadian clock story to a new level. They
regulators by functionally characterizing the genes that
were found by DNA microarrays to oscillate in multiple
tissues, with the idea that “cross-tissue cycling genes”
would include key components of the circadian clock-
work. These rhythmically regulated genes were trans-
fected into HeLa cells, and their effect on the Bmal1
promoter were tested. Lo and behold, of the genes that
activated Bmal1 transcription were two orphan nuclear
receptor family members, Rora and Rorc. Sato and co-
workers (2004) went on to demonstrate that this Bmal1
transactivation was dependent on Rev-erb/Ror binding
sites and that the Ror proteins physically compete with
Rev-erb proteins for the same binding site, similar to
the PDP1/VRI competition in Drosophila.
The authors focused their behavioral studies on Rora,
because it is expressed in the SCN, while Rorc is not,
and because Rora null mutant mice were already avail-
aframe-shift deletionmutationin theRora gene,leading
to a nonfunctional protein (see references in Sato et al.,
2004). They found Bmal1 mRNA levels were moderately
reduced in the homozygous mutant mice. The loss-of-
function mutation shortened circadian period, as in the
Rev-erb ? knockout mice. Sato et al. (2004) also pro-
vided evidence that CLOCK regulates Rora expression,
probably through E box elements in its promoter.
The modest molecular and behavioral phenotypes in
staggerer mutant mice could be explained by redun-
dancy of the transactivation function between Rora and
Rorb within the SCN; Rorb mRNA is robustly expressed
in the SCN (Andre et al., 1998). In fact, 6 years ago,
Rorb knockout mice were shown to have a circadian
phenotype, as the circadian period expressed by the
mutants is significantly lengthened (Andre et al., 1998).
pret because mammalian circadian genes were just be-
ginning tobe identified(Reppert andWeaver, 2002).The
current identification of Rora as a positive regulator of
Bmal1 expression may help shed new light on the pre-
viously described circadian phenotype of the Rorb
knockout. Future experiments need to examine the mo-
lecular clock in the SCN of Rorb mutants and test the
An additional finding by Sato and coworkers (2004)
is that staggerer mutant mice have normal clock gene
rhythms in peripheral tissues, including a normal Bmal1
mRNA rhythm. Thus, it is likely that Rorc is the major
Bmal1 transactivator in peripheral clocks like the liver.
Taken together, it is entirely possible that each of the
Ror proteins (a, b, and c) contributes to rhythmic Bmal1
activation and that the specific contribution of each de-
pends on the tissue examined. A similar situation may
apply to Rev-erb ? and Rev-erb ? activities as Bmal1
repressors, because Rev-erb ? is rhythmically ex-
pressed in the SCN and liver and may partially compen-
sate for the circadian defects in Rev-erb ? mutant mice.
Two interacting transcriptional feedback loops are a
central featureof the circadianclock mechanismin both
mammals and Drosophila (Figure 1). The negative tran-
scriptional feedback loop is the core loop, because it is
the primary determinant of period length and amplitude.
Severe mutations in most genes of this loop result in a
Stanewsky, 2002). On the other hand, no mutation in
the second loop has yet been described that leads to
that PDP1 and VRI null mutations are lethal, and there-
fore cannot be properly tested for their effect on circa-
not yet been tested for Rev-erb or Ror family members,
and, as mentioned above, redundancy is quite possible.
However, almost complete disruption of dClk mRNA
cycling in flies (transgenic dClk overexpression) and
out mice) does not result in severe behavioral pheno-
types (Kim et al., 2002; Preitner et al., 2002). Therefore,
the role of the Bmal1 or dClk feedback loop is probably
to fine-tune the clock mechanism, assuring its precision
and stability even when environmental conditions vary.
In this regard, it is interesting that Rev-erb ? mutants
excessively phase shift their clock after a light pulse
(Preitner et al., 2002).
Other control mechanisms augment the interacting
feedbackloops toprovideacoherent clockmechanism.
For example, there is apparently a shunt between the
two loops in mammals, as both E box and Rev-erb/Ror
elements contribute to the circadian regulation of Cry1
mRNA levels (Etchegaray et al., 2003; Sato et al., 2004).
Also, there is evidence suggesting that Rora regulates
Rev-erb ?, and Rev-erb ? can negatively regulate its
own expression (see Adelmant et al., 1996; Delerive et
al., 2002). This would add extra levels of regulation that
critical importance of the posttranslational regulation of
core clock proteins to clock function, including phos-
phorylation, nuclear entry, ubiquitination, and proteoly-
sis, which has been largely elucidated in Drosophila
(Stanewsky, 2002), needs to be more fully delineated in
mammals. And finally, the fundamental role of rhythmic
changes in chromatin structure to core clock function,
which has been initiated in mammals (Etchegaray et al.,
2003), needs to be expanded and extended to Drosoph-
ila. It is clear that the parallel, complementary nature of
circadian clock discoveries between flies and mammals
is what keeps circadian biologists Roring.
Patrick Emery and Steven M. Reppert
Department of Neurobiology
University of Massachusetts Medical School
364 Plantation Street
Worcester, Massachusetts 01605
Adelmant, G., Begue, A., Stehelin, D., and Laudet, V. (1996). Proc.
Natl. Acad. Sci. USA 93, 3553–3558.
Andre, E., Conquet, F., Steinmayr, M., Stratton, S.C., Porciatti, V.,
and Becker-Andre, M. (1998). EMBO J. 17, 3867–3877.
Blau, J., and Young, M.W. (1999). Cell 99, 661–671.
Hardin, P.E., Young, M.W., Storti, R.V., and Blau, J. (2003). Cell
Delerive, P., Chin, W.W., and Suen, C.S. (2002). J. Biol. Chem.
Etchegaray, J.-P., Lee, C., Wade, P.A., and Reppert, S.M. (2003).
Nature 421, 177–182.
Glossop, N.R., Lyons, L.C., and Hardin, P.E. (1999). Science 286,
Glossop, N.R., Houl, J.H., Zheng, H., Ng, F.S., Dudek, S.M., and
Hardin, P.E. (2003). Neuron 37, 249–261.
Kim, E.Y., Bae, K., Ng, F.S., Glossop, N.R., Hardin, P.E., and Edery,
I. (2002). Neuron 34, 69–81.
Preitner, N., Damiola, F., Lopez-Molina, L., Zakany, J., Duboule, D.,
Albrecht, U., and Schibler, U. (2002). Cell 110, 251–260.
Reppert, S.M., and Weaver, D.R. (2002). Nature 418, 935–941.
Sato, T.K., Panda, S., Miraglia, L.J., Reyes, T.M., Rudic, R.D., McNa-
mara, P., Naik, K.A., FitzGerald, G.A., Kay, S.A., and Hogenesch,
J.B. (2004). Neuron 43, this issue, 527–537.
Shearman, L.P., Sriram, S., Weaver, D.R., Maywood, E.S., Chaves,
I., Zheng, B., Kume, K., Lee, C.C., van der Horst, G.T., Hastings,
M.H., and Reppert, S.M. (2000). Science 288, 1013–1019.
Stanewsky, R. (2002). Cell Tissue Res. 309, 11–26.
Ueda, H.R., Chen, W., Adachi, A., Wakamatsu, H., Hayashi, S., Taka-
sugi, T., Nagano, M., Nakahama, K., Suzuki, Y., Sugano, S., et al.
(2002). Nature 418, 534–539.