Rhythmic glucose utilization precedes the development of the molecular clockworks in embryonic stem cells; however clock gene expression becomes rhythmic upon short-term differentiation. Previous investigations of the ontogeny of circadian rhythms concluded that undifferentiated mouse embryonic stem cells do not contain a functional clock based upon the lack of molecular rhythmicity in the expression of known components of the molecular clock 
, both in synchronized cultures as well as at the single cell level 
. These studies, however, limited their analyses to some, but not all, putative clock genes. Here we show – in agreement with those studies as well as in a more comprehensive manner – that undifferentiated cells indeed do not possess a functioning canonical molecular clock, based upon expression of mRNA of most of these genes as well and protein expression of PER2. Undifferentiated cells were not rhythmically expressed with respect to the clock genes tested, which is also in accord with previous studies. Upon non-specific differentiation, however, all of the tested clock genes were expressed rhythmically, with the exception of mClock
. Furthermore, the rhythm of luciferase bioluminescence in mPER2::LUC dESCs confirms the expression data. Although previous studies also looked at differentiated cells, the fate of those cultures was directed towards that of neural tissues. These data show the earliest developmental point at which clock genes exhibit circadian rhythms. This study investigated gene expression rhythms in primary cell cultures without the use of chemical synchronization as well as a physiological output of the clock, glucose uptake, which is a measure directly indicative of glucose utilization 
. In addition to investigating whether undifferentiated cells exhibit uninduced rhythmicity, undirected differentiation was included in this set of experiments in order to observe any potential reorganization of clock elements in a manner that recapitulates the development of the embryo in utero.
Remarkably, rhythmic glucose utilization in undifferentiated stem cells does not necessarily coincide with rhythmic canonical clock gene expression; these processes are developmentally and experimentally separable. Previous studies in juvenile chicks showed that enucleation abolishes 2-DG uptake in the brain while clock genes remained rhythmic 
. Similarly, rhythmic administration of melatonin to embryonic astrocytes was sufficient to drive rhythms of 2-DG uptake, but not of all the canonical clock genes 
. Along with the previously mentioned rhythm of 2-DG uptake in neonatal rats 
as well as the recently discovered transcription-independent rhythm of redox cycles in human red blood cells 
the data presented here provide compelling evidence that metabolic rhythms are not regulated solely by the canonical molecular clockworks.
The glucose utilization rhythms in both ESCs and dESCs were corroborated by rhythmic glucose transporter expression, mGlut8
in both ESCs and dESCs, indicating that the rhythms are driven by a transcriptional mechanism separate from the rhythmic expression of the canonical clock genes. Previously, Tonack, et al. 
showed that ESCs expressed several mGlut
transcripts throughout embryoid body differentiation, including glut1 and glut8 in undifferentiated cultures. Although neither mGlut1
have been implicated in circadian rhythms, mGlut1
is necessary for ESC viability 
and upregulation of mGlut8
in embryoid bodies suggests an increased need for glucose in differentiating cells 
. Interestingly, the amplitude of mGlut8
in both cultures remained the same, while the average glucose utilization was markedly increased in differentiated cells.
The rhythmic clock gene expression upon differentiation is quite remarkable for a number of reasons. The relative profiles of those genes that were rhythmic are consistent with the canonical molecular mechanism of circadian transcription, the negative elements mPer1
were identical in their phasing, and the positive element mBmal1
was expressed in anti-phase. Also, the cultures were rhythmic in the absence of any chemical synchronization. Previous studies had used dexamethasone 
or forskolin 
to synchronize cultures, as is common practice. In this study, the only conceivable synchronizing factors could have been centrifugation during passage or the absence of LIF, however there is no evidence of either phenomenon occurring in ESCs. Despite this, rhythms in 2-DG and clock genes were synchronized differentially between ESCs and dESCs. Secondly, the signal for differentiation away from the pluripotent state immediately synchronized the cultures in this study, as seen in the bioluminescence data. In contrast, Yagita et al 
showed that clock genes remained arrhythmic throughout an induced differentiation process that required several days of incubation with retinoic acid. In combination with the data presented here, it would appear that clock gene rhythmicity is a dynamic phenomenon that can change depending on developmental stage. Indeed, previous studies have shown that the synchrony of clock genes changes throughout development in both the SCN 
and the liver 
. The nature of this synchronization is unknown but, considering the aggregate nature of the cultures, functional gap junctions between cells may facilitate communication and synchrony 
. Finally, the relative phases of the clock genes, once synchronized, align in a manner that would suggest functional molecular clockworks, with the positive and negative elements expressed anti-phase to each other.
Unlike previous studies examining rhythms in fate-specific differentiated cells, the media in this study differed with respect to one ingredient; the differentiation inhibitor LIF. LIF acts as a cytokine binding to a heterodimeric receptor complex of its own receptor, LIFR, and gp130. The pathway ultimately leads to activation and translocation of STAT3 to the nucleus where it binds and activates various genes, presumably those involved in maintaining pluripotency 
. However, there is no known, direct link between the LIF pathway and glucose transporter/uptake. The absence of LIF in differentiated cultures may account for the rhythmic expression of clock genes, but it does not explain the persistence of the 2-DG uptake rhythm in undifferentiated cells.
These data cannot be explained by possible effects of 2-DG on metabolism itself. While high mM concentrations 2-DG inhibits cellular metabolism in vitro
, and high dosages of 2-DG can affect circadian clock light sensitivity in vivo
, the concentrations of 14
C-2DG employed here (8 µM) and elsewhere in vitro
, and dosages employed for metabolic markers in vivo
have no effect on metabolism 
or circadian clock function 
. Further, the 2-DG concentrations employed here are 1/250 of the 2 mM Kondoh et al. 
have shown to have no effect on ES cell proliferation. Much higher concentrations than 2 mM are required to affect differentiation. Moreover, these data cannot be explained by cell cycle effects on clock gene expression and/or metabolism, since cell-cycle period for these cells is 11–12 hours rather than the 24 hours observed here 
. The most parsimonious explanation of the present results is that a circadian clockworks that does not entail the rhythmic expression of clock genes is present in mouse embryonic stem cells before differentiation into germ lines.
Thus, although undifferentiated ESCs do not possess a functioning canonical molecular clock, a circadian rhythm of glucose utilization persists in these cells. This rhythm is coincident with rhythmic expression of one glucose transporter gene, suggesting rhythmic transcriptional control of glucose utilization. Whether this rhythm is initiated intrinsically or in response to glucose availability is unknown. In addition, acute differentiation by withdrawal of LIF increases the amplitude of glucose utilization rhythms and initiates rhythms of the expression of canonical clock genes, suggesting a potential role for the cytokine in initiating the transcriptional-translational feedback loop.