Analysis of in vitro glucose utilization in a circadian pacemaker model

An in vitro glucose utilization method, based upon 14C-2-deoxyglucose kinetics in brain slices, has been used to study circadian rhythms in hypothalamic slices containing the suprachiasmatic nucleus (SCN). Spontaneous SCN metabolic activity in vitro is similar to that observed in vivo with higher metabolic rates in subjective daytime and lower rates during subjective night. However, in vitro SCN metabolic activity during late subjective day is above that seen when glucose utilization is measured in vivo, suggesting that an inhibitory influence normally active in vivo is lost during slice isolation. Incubation of slices containing SCN in the presence of TTX exposes a TTX-insensitive component of metabolic activity in early subjective day, supporting prior suggestions that glucose utilization by the circadian oscillator continues in the absence of Na(+)-dependent action potentials. Studies with high Mg2+ concentrations are consistent with the hypothesis that most metabolic activity above the basal level observed with the glucose utilization method is related to synaptic activity. Pharmacological studies of the SCN brain slice model with radiotracers offer potential for analysis of both circadian rhythmicity and neural regulation.

cleoxyglucose kinetics in brain slices, has been used to study circadian rhythms in hypothalamic slices containing the suprachiasmatic nucleus (SCN). Spontaneous SCN metabolic activity in vitro is similar to that observed in vivo with higher metabolic rates in subjective daytime and lower rates during subjective night. However, in vitro SCN metabolic activity during late subjective day is above that seen when glucose utilization is measured in vivo, suggesting that an inhibitory influence normally active in vivo is lost during slice isolation. Incubation of slices containing SCN in the presence of TTX exposes a TTX-insensitive component of metabolic 'activity in early subjective day, supporting prior suggestions that glucose utilization by the circadian oscillator continues in the absence of Na+-dependent action potentials. Studies with high Mg*+ concentrations are consistent with the hypothesis that most metabolic activity above the basal level observed with the glucose utilization method is related to synaptic activity. Pharmacological studies of the SCN brain slice model with radiotracers offer potential for analysis of both circadian rhythmicity and neural regulation.
The suprachiasmatic nucleus (SCN) of hypothalamus is the principal circadian pacemaker in mammals, driving a wide variety of behavioral and physiological rhythms (Moore and Eichler, 1972;Stephan and Zucker, 1972). Despite detailed knowledge about SCN anatomy and physiology (Meijer and Rietveld, 1989), the cellular mechanisms involved in circadian pacemaker function remain largely unknown. Intrinsic circadian rhythms demonstrated in SCN in vivo thus far include a rhythm of neuronal action potential firing rate, which persists even when the nucleus is isolated as a surgical "island" (Inouye and Kawamura, 1979), a rhythm of glucose utilization demonstrated with the 14C-2deoxyglucose (2DG) method , and a rhythm of vasopressin (VP) synthesis (Schwartz and Reppert, 1985;Reppert and Uhl, 1987;Robinson et al., 1988). SCN glucose utilization is high in daytime and low at night. Recently it has been shown that hypothalamic infusion of TTX in vivo appears to block SCN input and output pathways without affecting the circadian oscillator itself (Schwartz et al., 1987a), suggesting that neural firing is a pacemaker output rather than an intrinsic component of oscillator function. This also implies that there may be components of the rhythm of glucose utilization related to intrinsic oscillator function that may be exposed by eliminating the large metabolic costs of neuronal electrical activity with TTX. In order to address this and other aspects of SCN neurochemistry, we have developed an in vitro model of SCN utilizing hypothalamic brain slices and 2DG autoradiography. The unique circadian properties of the SCN, combined with the advantages of in vitro pharmacology, can also help to define the cellular processes that contribute to glucose utilization measurements. Circadian rhythms of neural firing have been repeatedly observed in brain slices (Green and Gillette, 1982;Groos and Hendriks, 1982;Shibata et al., 1982;Wheal and Thomson, 1984), indicating that oscillator function continues in vitro. Initial hypothalamic slice experiments demonstrated that SCN uptake of 2DG is high in subjective daytime and low in subjective night, even after 8 hr in vitro (Newman and Hospod, 1986). Subsequently, we have developed a detailed kinetic model to permit quantification of in vitro glucose utilization so that the glucose consumption related to individual cellular activities might be separated (Newman et al., 1990). In this article, we describe the circadian cycle of SCN glucose utilization in brain slices. We also show, for the first time, that a rhythm of glucose utilization persists in the presence of TTX and describe the effects of Mg2+ on daytime SCN glucose utilization in vitro.

Materials and Methods
Materials. All buffers and inorganic salts were obtained from Sigma Chemical (St. Louis, MO) and were cell culture grade. The 95% O,, 5% CO, mixture and liquid nitrogen were purchased from General Welding (Long Island City, NY). All 14C-2-deoxyglucose (specific activity, 59 mCi/mmol) and 14C standards were from Amersham (Arlington Heights, IL). Isopentane, class IA, was obtained from Fisher Scientific (Fair Lawn, NJ), and scintillation cocktail, from National Diagnostics (Manville, NJ). OM-1 film for autoradiography was obtained from Kodak (Rochester, NY). All water was deionized and purified to a resistance of 17.5 MQ with a Bamstead NANOpure system (Boston, MA). Sprague-Dawley rats were obtained from Taconic Farms (Germantown, NY).
Tissue preparation and incubations. Brain slice incubations are conducted in a chamber specifically designed for biochemistry and morphology. Details of our methods for tissue preparation, 2DG incubation, freezing, cryostat sectioning, image analysis, and calculation of glucose  (Newman et al., 1989(Newman et al., , 1990 groups using the following limits (with circadian time, CTOO, defined and therefore our methods are discussed only briefly here. Male as the time of lights on): CT0 1, CTOO-O 1.5; CT03, CT0 1.5-04; CT06, Sprague-Dawley rats weighing 175-225 gm are caged unrestrained in ~04-07.5;CT09,CT07.5-10;CT12,CT1~13;CT15,~13-16;CT18, groups of three at 22°C with free access to food and water in a 12 hr: 12 hr light/dark (L/D) cycle for at least 3 weeks. The three cagemates are killed on a single day and divided among control and experimental groups so that a time-matched control accompanies most tetrodotoxintreated preparations. All animals are killed under dim red lighting and the eyes are covered with black electrical tape prior to turning on room lights for brain removal. For animals killed during subjective daytime, the lights are not turned on the day of death so that they receive no light stimulus for at least 12 hr prior to death. Animals killed during the dark phase of the L/D cycle are handled similarly but without alteration of their lighting schedule prior to death. The brain is removed by a posterior approach to avoid traction on the optic nerve. The hypothalamus is block dissected, and slices are chopped coronally at 540 hrn on a Smith-Farquhar tissue chopper (Sorvall, Newtown, CT). Two slices containing the entire SCN are placed in a chamber within 4 min of animal death for preincubation in Krebs-Ringer (K-R) with an atmosphere of prehumidified 95% O,, 5% CO, and ld rni glucose, 1.5 mM Ca*+, pH 7.37, 305 mOsm and PO, of 715 mm Hg. Slices are preincubated for 75 min prior to isotope exposure. Slices incubated with CT16-19; CT21, CTl9-22; CT23, CT22-24. The same limits were applied for control or TTX-treated slices and were used throughout the text and figures.
Analysis of rostrocaudal profile. In order to study the rostrocaudal profiles of SCN glucose utilization, three control slices, one each from CT06, CT09, and CT12, were chosen for serial section analysis based upon uniform sectioning and tissue chopping that isolated most of the SCN within a single brain slice. Values for right and left SCN were averaged and values for sections within 80 pm of the slice surface were corrected for edge artifacts (Newman, 1991). Cross-sectional areas of each SCN half-nucleus was determined using the MCID image analyzer. The three slices were aligned for comparison by using the most rostra1 section with a unilateral cross-sectional SCN area of 0.1 mm* as a reference. Differences in dorsomedial and ventrolateral glucose utilization were also sought using SCN regions defined with reference to slices that were prepared and incubated in the usual manner and then immunostained with antibodies to VP or vasoactive intestinal peptide.
TTX or the various M$+ concentrations are exposed to the altered buffer for 15 min prior to incubation with 2DG, during the 2DG in-Results cubation, and throughout the rinse period.
Measurement of glucose utilization. Measurement of in vitro glucose utilization with 2DG is analogous to in vivo glucose utilization (Sokoloff et al., 1977) except that it is possible to deliver a square wave pulse of radioactivity to the tissue. Incubation with 2DG is initiated by moving

Spontaneous in vitro metabolic activity
Analysis of the means of right and left SCN glucose utilization values for hundreds of slices in this and other experiments has revealed no consistent difference between the two nuclei at anv the slices toa second chamber preequilibrated with K-R containing 0.2 &i/ml of isotope. The slices are incubated for 45 min, removed from the incubation chamber, briefly rinsed in warm K-R, and returned to the preincubation chamber for 30 min of rinse. Following rinse, the slices are rapidly frozen in isopentane cooled to -80°C and cryostat sectioned to 20 pm. Sections are collected on a glass slide, exposed to x-ray film along with standards for 1 week, and then stained with cresyl violet for image analysis. Optical density is measured with an image analvzer (MCID. St. Catherine, Ontario. Canada) that nermits imaae overlay so that the region of interest is chosen while viewing only the Nissl section. Alternate sections are analyzed throughout the rostrocau-da1 extent of the nucleus except for sections within 80 pm of the slice surface that are excluded (Newman,199 1). Radioactivity is quantified in nCi/gm tissue using the standard curve. Glucose utilization is then calculated in right and left SCN and the average ofright and left anterior hypothalamic areas (AHA) using the measured radioactivity and Equation 1, which has been derived using a five-compartment, eight-parameter kinetic model (Newman et al., 1990). The mean ofall sections from each animal is then calculated for each right and left SCN and the combined AHA: time of day; therefore, we present all SCN data as the average of the right and left nuclei. A robust rhythm of SCN glucose utilization persists in vitro (Figs. 1,2A). SCN glucose utilization is low from CT18 to CT23, rises in early subjective day, peaks in late day at CT09, and gradually returns to low levels between CT 12 and CT 15. Although inspection of the AHA data suggests that there is no diurnal rhythm in that region (Fig. 3A), comparison between all daytime (CTOO-CT12) and dark-phase (CT1 2-CT24) control slices reaches significance at the p < 0.05 level. Most of the in vitro SCN and AHA glucose utilization values are about 20% higher than in vivo values at the corresponding time of day. SCN values in late subjective day and early dark phase are much higher, however. Individual slice glucose utilization values for nighttime SCN, and for AHA at all times of day, are about 50 pmol/ 100 gm/min, while daytime SCN values are considerably higher with several particularly active slices in late day. Variation of SCN glucose utilization In Equation 1, R, is rate of glucose utilization; C, is the concentration R of elucose in the nerifusate: C.* is the total radioactivitv in the tissue: -.l= C,,;is the perifusate radioactivity; LC is the lumped constant; K,*, k,*; K,*k,* and k,* are rate constants; E(i), B(i), M(i), G(i), and X, are kinetic c,*/c,* parameters derived from the solution of the differential equations for the model; To is the time of incubation (45 min); and A is the duration C" of rinse (30 min). Although this equation appears complex, in practice, once the rate constants have been determined for the tissue and C,, r,, and A are set, the entire equation reduces to a single constant that is multiplied by C,*/C,* to obtain glucose utilization. For hypothalamic slices-incubated .undkr the conditions stated above, Equation 1 reduces to R. = 104.19 x C,*/C.*. The circadian time assianed to each slice incubation corresponds to the midpoint of the 45 mm incubation with 2DG.
values during subjective daytime is considerably greater than variations of SCN values at night or of AHA values at any time of day, similar to the situation observed for neural firing rates Rostrocaudal profile Rostrocaudal analyses of slices incubated with 2DG at CT06, CT09, and CT 12 reveal that in vitro metabolic activity is significantly higher in caudal SCN than in rostra1 SCN (Fig. 4). Glucose utilization correlates closely with cross-sectional area in the rostra1 half of the SCN but not at all in the caudal half where values of over 200 /Imol/lOO gm/min are found. It is thus apparent that the high levels of glucose utilization seen during the later half of subjective daytime reflect primarily increased metabolic activity in the caudal half of the nucleus. No consistent differences were observed between the dorsomedial and ventrolateral SCN subdivisions.  does not alter glucose utilization of slices incubated at CT03 or between CT18 and CT23 (Fig. 2B). For all daytime groups except CT03, comparison of control and TTX slices reveals highly significant differences; however, at CT03 there is no significant decrease produced by TTX. Although there is also no difference between control and TTX-treated slice SCN glucose utilization during the dark phase of the L/D cycle, control SCN glucose utilization between CT 18 and CT2 1 is low, similar to AHA glucose utilization, whereas SCN glucose utilization of control slices at CT03 is considerably elevated compared to nighttime SCN values or AHA glucose utilization. Glucose utilization of slices incubated with TTX at CT03 also differs significantly from that of slices exposed to TTX between CT06 and CT12 (p < O.OOl), while slices incubated between CT06 and CT 12 do not differ significantly from slices incubated with TTX at CT 18 and CT2 1. Thus, there is persistent spontaneous metabolic activity in SCN at CT03 that continues even in the presence of TTX, consistent with the hypothesis that voltageregulated Na+-dependent action potentials are not essential for circadian oscillator function (Schwartz et al., 1987a).
TTX uniformly reduces daytime AHA glucose utilization by 30% from a mean of 51.2 f 12.8 to 35.5 f 6.2 ( Fig. 3B   Effects of A@+ on glucose utilization at CT06 The effect of Mg*+ on SCN in vitro glucose utilization was studied in the presence of physiological concentrations of Ca*+ at a single circadian time (CT06) over a wide range of Mg2+ concentrations (Table 1). SCN glucose utilization is maximal at physiological concentrations of Mg*+ and dramatically reduced at 10 and 20 mM, while AHA glucose utilization is virtually independent of Mg2+ concentration.

Discussion
The SCN offers a unique opportunity to study cerebra1 metabolism in vitro because it expresses intrinsic rhythms of neuronal and metabolic activity. Three components of in vitro SCN glucose utilization are identified by our results. The largest component is expressed during subjective daytime, correlates closely with a high neural firing rate (Green and Gillette, 1982;Groos and Hendriks, 1982;Shibata et al., 1982;Wheal and Thomson, 1984), and is suppressed by TTX. These features are consistent with the hypothesis that this daytime component of glucose utilization reflects primarily the metabolic costs of reestablishing ionic gradients in activated neural tissue (Mata et al., 1980;Sokoloff, 1982;Yarowsky et al., 1983). The second component consists of the SCN glucose utilization observed during the sub-1988). The rostrocaudal pattern of this oxidative enzyme is thus jective dark phase. This low level ofglucose utilization correlates the inverse of in vitro glucose utilization, suggesting that inwith a low neural firing rate and is not significantly affected by creased glucose utilization in caudal SCN may reflect higher TTX. Thus, it appears to represent basal metabolic activity glycolytic capacity in that region. Since the higher in vitro glucose related to cell maintenance and low levels of neural activity and utilization was observed in caudal SCN even when that portion neurotransmission.
However, even at night, SCN glucose uti-of the nucleus was adjacent to the slice surface, diffusion of 0, lization remains above that of AHA, indicating a higher level into the slice is unlikely to explain this pattern. Furthermore, of metabolic activity in these pacemaker neurons. The third component of SCN glucose utilization is observed at CT03 as metabolic activity that is above the basal level of nighttime SCN, or AHA, but that is not suppressible with TTX. Other studies have shown that 1 PM TTX abolishes all sodium-dependent action potentials as well as optic nerve-evoked responses in SCN in vitro (Shibata et al., 1984a;Sugimori et al., 1984; S. Shibata and R. Y. Moore, unpublished observations). As might be expected from these data, TTX suppresses the in vitro rhythm of SCN VP release (Earnest et al.,199 1). TTX, at concentrations of 0.3 PM, suppresses VP release and disrupts the subsequent circadian variation in VP release if given for 6 hr in late subjective day but not if given during early subjective day or for 12 hr during subjective night. However, drinking rhythms appear to require at least three cycles for return of that by itself, this tendency toward increased glycolysis cannot explain why in vitro SCN metabolic rates in late subjective day are higher than at other times of day relative to in vivo rates. Our present working hypothesis is that the increased metabolic rate in vitro reflects the loss of an inhibitor of SCN that is active in late daytime in vivo and that is lost when the SCN is isolated as a brain slice. The proposed inhibitory feedback could occur through either neural connections or humoral influences in cerebrospinal fluid or blood. For example, interruption of the ascending serotonergic pathway from dorsal raphe in vivo has been shown to increase glucose utilization significantly in SCN (Hery et al., 1982;Maxwell and Fink, 1988). The absence of inhibitory neuropeptides or other messengers, such as melatonin (Cassone et al., 1988) may also contribute to the observed high levels of glucose utilization observed in vitro. This explanation rhythm after infusion of TTX into the region of SCN in vivo is analogous to prior suggestions that the reduced metabolic (Schwartz et al., 1987a), and there is some evidence of an un-rates of hippocampus observed in vitro are due to the loss of derlying rhythm of VP release during the last days of explant excitatory pathways during isolation (McIlwain and Bachelard, culture following TTX exposure (Earnest et al., 199 1). Although 1971;Lipton and Whittingham, 1984;Jurgensen and Wright, the nature of the TTX-resistant energy consuming processes at 1988; Newman et al., 1989). However, because the SCN in-CT03 and whether they directly relate to oscillator function are trinsically generates a reliable pattern of neural firing rate even presently unknown, the apparent rate of glucose consumption, as a surgical "island" in vivo (Inouye and Kawamura, 1979) at 30 Fmol/ 100 gm/min above baseline, is considerable. the absence of neural input need not lead to reduced metabolic The primary difference between in vitro and in vivo SCN activity. Indeed, we are suggesting that the absence of inhibitory glucose utilization is the increased metabolic activity found in neural input to SCN in slices may result in increased metabolic vitro during late subjective day. Examination of the in vitro rate. rostrocaudal profile of SCN glucose utilization reveals that, un-High regional glucose utilization has frequently been correlike in vivo glucose utilization, which correlates with SCN cross-lated with high regional synaptic content (Kennedy et al., 1976; sectional area throughout its rostrocaudal extent (Schwartz et Schwartz et al., 1979;Kadekaro et al., 1987Kadekaro et al., ) yet direct confiral., 1987b, in vitro SCN glucose utilization correlates with cross-mation of this association is lacking. We have studied SCN sectional area only in the rostra1 half of the nucleus and is metabolic activity at various Mg2+ concentrations in an effort significantly increased in the caudal half relative to the rostra1 to address such issues. High levels of Mg2+ have been observed half and in vivo values. We suggest that two factors, anaerobic to block synaptic transmission by both presynaptic and postglycolysis and neural disinhibition, are sufficient to explain this synaptic mechanisms (Crunelli and Mayer, 1984; Garthwaite discrepancy. It seems unlikely that the high SCN glucose uti-and Garthwaite, 1987;Mayer and Westbrook, 1987; Cotman lization is related solely to in vitro radiotracer kinetics or meth-et al., 1988). Despite this, mean spontaneous SCN neural firing odology. That is, it should not represent some form of calcu-rates appear to be only minimally affected by high Mg2+ when lational error, since the values from rostra1 SCN and AHA are Ca2+ is maintained in the normal range (Shibata et al., 1984b; similar to those observed for SCN and hypothalamus in vivo Thomson, 1984). The marked suppression of SCN glucose uti- Schmidt et al., 1989) in vitro kinetic lization in vitro by concentrations of Mgz+ known to block synparameters for hypothalamus are similar to the analogous in aptic transmission, combined with the relative insensitivity of vivo parameters, and the form of the rate equation employed SCN neural firing to high Mg2+ with normal Ca*+, provides direct for these studies is particularly insensitive to errors in the kinetic support for the hypothesis that physiologically activated glucose rate constants (Newman et al., 1990;Newman, 199 1). Further-utilization correlates with synaptic activity rather than with acmore, glucose utilization measurements with slices from other tion potentials in perikarya or axons. This interpretation is not regions of brain do not show values as high as those found in necessarily in conflict with the observed circadian rhythm of caudal SCN during subjective daytime, even after anoxia or 2DG uptake in fetal rats prior to synapse formation since we upon K+ stimulation (Newman et al., 1989(Newman et al., , 1991. Obviously, observe enhanced glucose utilization in SCN brain slices at CT03 the 2DG method alone cannot distinguish glycolysis associated even in the presence of TTX and, in the absence of fetal rat with oxidative metabolism from strictly anaerobic glycolysis kinetic parameters, the amplitude of the fetal rhythm is uncerthat would require much higher levels of glucose utilization for tain Schwartz, 1983, 1984; Moore and Bernstein, generating ATP. It is therefore interesting to note that SCN 1989). In fact, developmental studies of brain glucose utilization staining for cytochrome oxidase demonstrates high levels of in rats indicate that rates of glucose utilization at birth are well cytochrome oxidase in the rostra1 and middle portions of SCN below those of the adult (Nehlig et al., 1988). but much lower levels in the caudal third (Murakami and Fuller, These studies using the new quantitative method of brain The spontaneous properties of the SCN provide a unique model system for studying neural physiology. Brain slice glucose utilization is complementary to electrophysiology because infor-Meijer JH, Rietveld WJ (1989) Neurophysiology of the suprachiasmatic circadian pacemaker in rodents. Physiol Rev 69:67 l-707. Moore RY, Bernstein ME (1989) Synaptogenesis in the rat suprachiasmatic nucleus demonstrated by electron microscopy and synapsin I immunoreactivity. J Neurosci 9:2 15 l-2 162. Moore RY, Eichler VB (1972) Loss of a circadian adrenal corticostemation can be obtained even in the absence of neural action potentials. Thus, we have used TTX to expose energy consuming cellular processes in SCN apparently unrelated to Na+-dependent action potentials in early day. The quantitative nature of in vitro glucose utilization also permits direct comparison to in vivo studies. Future studies with 2DG and a variety of other radiotracers should increase our understanding of circadian rhythmicity as well as of neuronal regulation.