Glucose Down-regulates Per 1 and Per 2 mRNA Levels and Induces Circadian Gene Expression in Cultured Rat-1 Fibroblasts ∗ ∗

In mammals, peripheral circadian clocks are present in most tissues, but little is known about how these clocks are synchronized with the ambient 24-h cycles. By using rat-1 fibroblasts, a model cell system of the peripheral clock, we found that an exchange of the culture medium triggered circadian gene expression that was preceded by slow down-regulation of Per1 and Per2 mRNA levels. This profile contrasts to the immediate up-regulation of these genes often observed for clock resetting. The screening of factor(s) responsible for the down-regulation revealed glucose as a key component triggering the circadian rhythm. The requirement of both glucose metabolism and RNA/protein synthesis for the down-regulation suggests the involvement of gene(s) immediately up-regulated by glucose metabolism. An analysis with high density oligonucleotide microarrays identified >100 glucose-regulated genes. We found among others immediately up-regulated genes encoding transcriptional regulators TIEG1, VDUP1, and HES1, in addition to cooperatively regulated genes that are associated with cholesterol biosynthesis and cell cycle. The immediate up-regulation of Tieg1 and Vdup1 expression was dependent on glucose metabolism but not on protein synthesis, suggesting that the transcriptional regulators mediate the glucose-induced down-regulation of Per1 and Per2 expression. These results illustrate a novel mode of peripheral clock resetting by external glucose, a major food metabolite.


INTRODUCTION
Almost all organisms on earth exhibit daily changes in a variety of physiological processes, such as gene expression, metabolism, and behavior (1)(2)(3). Many of the daily changes persist under constant conditions with intrinsic period lengths (approximately 24 hours) under the control of autonomous biological pacemakers called circadian clocks. The circadian clock can be reset by environmental time cues (such as light) to synchronize with the ambient 24-hour cycles.
In mammals, genetic and molecular analyses revealed transcription/translationbased negative feedback loop(s) that constitutes the core oscillator of the circadian clock. This loop involves a well-concerted regulation of Per, Cry, Clock, Bmal genes and their products, resulting in robust oscillations of Per and Bmal mRNA levels in antiphase and of several output genes such as Dbp (4,5). The core oscillatory mechanism seems to be common to both the central clock localized in the hypothalamic suprachiasmatic nucleus (SCN) 1 and peripheral clocks distributed in most tissues and cells (6-8). Peripheral clocks are, however, distinct from the central clock in that the circadian expression of the clock genes persists only for several days in culture (9). The cyclic gene expression in peripheral clocks is likely sustained by synchronization with some neural and/or humoral signals generated by the SCN in a circadian manner (9-11).
Recent studies showed that restricted feeding cycle synchronizes peripheral clocks independently of the central clock (12-14), suggesting that feeding may be a dominant time cue for peripheral clocks. These observations led to an interesting idea that the central clock indirectly synchronizes peripheral clocks by regulating feeding behaviors (12)(13)(14). This model predicts peripheral clocks to be reset by feeding-associated events, such as food processing, metabolite absorption, and/or change in hormone levels, but little is known about the molecular identity of the feeding signal to be sent to peripheral clocks. Cultured cells such as rat-1 fibroblasts have been used as a model for the study on the peripheral clock system, by guest on November 17, 2017 http://www.jbc.org/ Downloaded from 4 especially on the resetting mechanism (7,8,(15)(16)(17)(18)(19). This is because circadian gene expression in these cells is induced not only by serum shock that was originally found to be effective (7) but also by treatment with many chemicals that activate a variety of signal transduction pathways (8,(15)(16)(17)(18)(19). Such induction of circadian rhythm is always preceded by a stimulus-induced immediate up-regulation of Per1 and/or Per2 mRNA levels (7,8,(15)(16)(17)(18)(19), an event that plays an important role in the photic-resetting of the central clock (20-22). Here we demonstrate that in rat-1 fibroblasts, an exchange of the culture medium induced circadian expression of the clock genes, which was preceded by slow down-regulation, not by rapid up-regulation, of Per1 and

Preparation of RNA Samples and RT-PCR Analysis
The cultured cells were washed with ice-cold PBS, homogenized with 1 ml (for 35-mm dish) or 3 ml (for 10-cm dish) of TRIZOL reagent (Invitrogen), and stored at -80°C until use. The extraction of total RNA and quantitative RT-PCR analysis were performed as described previously (23) with some modifications (see Supplemental Experimental Procedures).

Microarray Analysis
Poly(A) + RNA was isolated from total RNA by Oligotex-dT30 (Roche) and used to prepare cRNA sample as described in the GeneChip Expression Analysis The raw data for each probe set were calculated from the scanned array image by using GeneChip Analysis Suite software (Affymetrix) and subsequently analyzed by using GeneSpring software (Silicon Genetics). To compare the data from different arrays, the signal intensity value for each probe set on each array was normalized as follows: negative control value (median intensity value of negative control genes included in the array) was subtracted from the raw values, and the calculated values were divided by their median value. The values below 0 were set to 0. Then, the normalized values from 4 independent arrays were averaged for each probe set at each time point, and they were compared between two time points (time 0 with 1-hour or 4-hour) to select probe sets for up-or down-regulated genes with the following criteria: (i) the intensity exhibits 3-fold or greater change, (ii) the change is significant (mean ± SE at each time point do not overlap with each other), and (iii) the signal is marked as 'present' by the GeneChip software in at least 2 out of 4 arrays at the time point for the higher average intensity. Expressed sequence tags (ESTs) in the selected probe sets were characterized by searching GenBank database using BLAST program. Nomenclature of the selected genes was updated by using OMIM and GenBank databases. Overlapping probe sets for the same gene were unified by leaving one that exhibited the strongest signal, when their expression profiles were similar to each other. Hierarchical clustering of the selected probe sets was then carried out using Cluster program (24) in uncentered correlation/average linkage clustering mode, and the result was visualized with the aid of TreeView program (24). Functional assignment of the selected genes was performed by searching OMIM and PubMed databases.

Induces Circadian Gene Expression in Rat-1 Fibroblasts
In a search for a stimulus that induces rhythmic expression of some clock genes by guest on November 17, 2017 http://www.jbc.org/ Downloaded from 7 in rat-1 fibroblasts, we unexpectedly found that an exchange of the culture medium to serum-free medium triggered circadian changes in mRNA levels of Per2, Dbp, Bmal1, and Cry1 genes ( Fig. 1 and Supplemental Fig. 1, medium change). The phase of the expression rhythm of each gene induced by the medium exchange was advanced by about 4 hours relative to that observed after a pulse treatment with 50% serum and Per2 expression after the serum shock (Fig. 1B). We further analyzed immediate changes of Per1, Per2, and Bmal1 mRNA levels after the medium exchange, and found that these mRNA levels began to decrease 45-60 min after the treatment (Fig. 2). These results indicate that the circadian rhythm is induced by the medium exchange in a manner quite different from that triggered by the serum shock, and predict a novel mechanism of rhythm-induction that is preceded by slow downregulation of Per1 and Per2 expression.

Glucose Down-regulates Per1 and Per2 mRNA Levels and Induces Circadian Gene Expression
The exchange of the culture medium might have multiple effects on the cultured cells, and hence we searched for factor(s) that causes the down-regulation of Per1 and Per2 mRNA levels (Fig. 3). An exchange of the culture medium with medium recovered from another culture dish had a minimal effect on Per1 and Per2 expression (Fig. 3A), eliminating the possibility that the down-regulation was caused by any physical stimuli such as a transient change in temperature and/or pH of the medium or exposure of the cells to the air. Next, we screened the components of DMEM (salts, glucose, pyruvate, amino acids, and vitamins) by adding them separately to the culture medium, and found that glucose remarkably reduced not only Per1 but also Per2 expression (Fig. 3B) in a manner similar to that observed after the medium exchange (Fig. 3A). These suggest glucose as a key molecule    (Fig. 4). After the addition of glucose solution (5.6 mM final concentration) to the culture medium, Per2, Dbp, and Bmal1 genes exhibited robust circadian expression with profiles nearly identical to those observed after the medium exchange (Fig. 4B, compare solid lines with dashed lines), and the profiles were obviously different from those after the serum shock (Fig. 1B, dashed lines). These results demonstrate that glucose can trigger circadian rhythm, and suggest that the induction of circadian rhythm by the medium exchange is mainly due to the supply of glucose.

Metabolism and RNA/Protein Synthesis
We were interested in the molecular mechanism underlying the induction of circadian rhythm that is preceded by the down-regulation of Per1 and Per2 expression, and we first investigated whether glucose itself or its metabolism is important for the down-regulation (Fig. 5A). Among several glucose-related compounds examined, metabolizable carbohydrates such as galactose, fructose, and mannose reduced Per1 and Per2 expression to the level as was caused by glucose   We searched for genes which respond immediately or slowly to the glucose addition by using high-density oligonucleotide array technology. The cells were harvested just before, 1 hour after, or 4 hours after the glucose addition, and the cRNA sample from each preparation was hybridized to microarray containing 8,800 probe sets, some of which recognize same genes. The array contained probe sets for Per2, Dbp, and Bmal1 genes, and their signal intensities exhibited temporal profiles similar to those measured by RT-PCR analysis (Supplemental Fig. 2), ensuring the reliability of the microarray data. We found 176 probe sets exhibiting 3-fold or greater changes in their signal intensities after the glucose addition (Supplemental Table I; representing 130 known genes with 3 overlapping probe sets and 43 ESTs without sequence homology to any known gene). Based on the temporal expression pattern, the 176 probe sets were grouped into 8 clusters by hierarchical clustering method (Fig. 7A, clusters a-h). Per2 and Dbp genes belonged to cluster e (containing delayed down-regulated genes), and Bmal1 gene belonged to cluster f (containing steadily down-regulated genes; Fig. 7A, boxed).
Among 24 probe sets exhibiting immediate up-regulation after the glucose addition

Tieg1 and Vdup1 are Glucose-responsive Immediate-early Genes
To characterize properties of the glucose-induced up-regulation of Tieg1, Vdup1, and Hes1 expression, we investigated detailed temporal changes of their mRNA levels (Fig. 9). The addition of glucose transiently up-regulated Tieg1 and Hes1 expression with a peak at about 1 hour after the treatment, whereas it persistently up-regulated Vdup1 expression ( Fig. 9, solid lines). On the other hand, the addition of nonmetabolizable 3-OMG had little or no effect on Tieg1 and Vdup1 mRNA levels, but increased Hes1 mRNA levels in a manner similar to that after the glucose addition (Fig. 9, compare solid lines with dashed lines). It is most likely that We then tested the effects of protein and RNA synthesis inhibitors on the up- indicating that the up-regulation is independent of new protein synthesis. On the other hand, the effect of glucose addition was blunted by the actinomycin D treatment, suggesting that the glucose action centers at the level of transcription. These results indicate that Tieg1 and Vdup1 are glucose-responsive immediate-early genes, which may act as a direct mediator of the glucose effect.

Glucose as a Direct Resetting Signal for Peripheral Clocks
This study stemmed from our fortuitous observation that an exchange of the culture medium induced circadian gene expression in rat-1 fibroblasts ( Fig. 1 and Supplemental Fig. 1). Screening of factors associated with the medium exchange led to identification of glucose as the key molecule (Fig. 3), and glucose addition was shown to induce the circadian rhythm in culture (Fig. 4). Glucose is one of the major food metabolites, and in rodents, plasma glucose level exhibits diurnal rhythm (30).
In accordance with this rhythm, 3 out of 4 genes known to contain the glucoseresponse element (31) exhibit diurnal expression patterns in the liver (32,33). In addition, glucose, but not fat or non-nutritive bulk, causes a phase-shift of foodanticipatory activity rhythm, which occurs after a restriction of feeding time (34,35).
Thus, our present results predict an important role of glucose as a direct resetting signal for peripheral clocks in vivo, and this idea would explain how the peripheral clock in the liver responds to a change in feeding time faster than the clocks in other tissues (12,13).
Besides glucose, the levels of glucose-regulated hormones such as insulin and glucagon exhibit diurnal rhythms in plasma (30), and insulin immediately up-regulates Per1 and Per2 expression in rat-1 fibroblasts (18). In addition, recent studies affected by the food intake. Paradoxically, due to such a network, it is not feasible to evaluate the effect of glucose as a direct resetting signal in vivo. It is the model cell system such as rat-1 fibroblasts that enables this study.

Down-regulation of Per1 and Per2 mRNA Levels by Glucose
The induction of circadian rhythm by glucose seems quite different in mechanism from either the rhythm-induction by serum shock in cultured fibroblasts or expression, and it is less likely that a change in NADH/NAD + ratio plays an essential role in it. The glucose metabolism would initiate new RNA and protein synthesis that is required for the down-regulation (Fig. 6). As the candidates for such glucoseresponsive genes, we found two genes for transcriptional regulators TIEG1 and VDUP1 (Figs. 7 and 8A), whose mRNA levels were immediately up-regulated in a glucose metabolism-dependent manner (Fig. 9). Identification of Tieg1 and Vdup1 as glucose-responsive immediate-early genes (Fig. 10) may implicate these gene

Other Cellular Responses to Glucose
The microarray analysis revealed glucose-induced up-or down-regulation of genes associated with a variety of metabolic pathways (Fig. 7). Among them, 13 genes were related to cholesterol biosynthesis ( In conclusion, the present study identified glucose as a key molecule that can directly reset the peripheral clock, and illustrates a novel mode of the clock resetting.
Finding of the genes regulated by the glucose addition suggested activation of several transcriptional regulators and cellular pathways that have not been known to respond to glucose. Further analyses of these glucose-responsive genes including their regulatory cis-elements and trans-factors would lead to elucidation of not only a novel intracellular pathway mediating the glucose response but also an unidentified mechanism by which peripheral clocks are synchronized with feeding cycle.       Hirota et al.  Hirota et al.