Report

A Screen for Genes Induced in the Suprachiasmatic Nucleus by Light

See allHide authors and affiliations

Science  06 Mar 1998:
Vol. 279, Issue 5356, pp. 1544-1547
DOI: 10.1126/science.279.5356.1544

Abstract

The mechanism by which mammalian circadian clocks are entrained to light-dark cycles is unknown. The clock that drives behavioral rhythms is located in the suprachiasmatic nucleus (SCN) of the brain, and entrainment is thought to require induction of genes in the SCN by light. A complementary DNA subtraction method based on genomic representational difference analysis was developed to identify such genes without making assumptions about their nature. Four clones corresponded to genes induced specifically in the SCN by light, all of which showed gating of induction by the circadian clock. Among these genes are c-fos and nur77, two of the five early-response genes known to be induced in the SCN by light, andegr-3, a zinc finger transcription factor not previously identified in the SCN. In contrast to known examples, egr-3induction by light is restricted to the ventral SCN, a structure implicated in entrainment.

Daily rhythms of biological activity, manifested by forms as diverse as cyanobacteria, fungi, plants, and animals, are driven by self-sustaining, endogenous oscillators called circadian clocks (1), which typically run with an intrinsic period that is close to, but not exactly, 24 hours. Under natural conditions, circadian clocks become precisely entrained to the 24-hour light-dark cycle because exposure to light at certain times induces a phase shift of the clock. Entrainment to light-dark cycles ensures that the clock adopts a specific and stable phase relation to the natural day, setting the clock to local time and enabling the organism to anticipate daily environmental events (2).

In mammals, the circadian clock that drives daily rhythms of behavioral activity is located within the SCN of the hypothalamus (3). Entrainment of the clock to light-dark cycles is mediated by photoreceptors in the retina (4), and light information is conveyed directly from the retina to the SCN by the retinohypothalamic tract (5). Although the molecular basis of entrainment to light-dark cycles in mammals is unknown, the process likely involves light- and clock-dependent transcriptional regulation within the SCN (6). When a rodent kept in constant darkness is exposed to a brief light pulse during the subjective night, a time when the clock responds to light with a phase shift, five known early-response genes—c-fos, fos-B, jun-B,zif268 (NGFI-A), and nur77 (NGFI-B)—are specifically induced within the SCN (7). The genes are not induced by exposure to light during the subjective day, a time when the clock is not phase-shifted by light. Together, these and related experiments (8) strongly suggest that induction of genes in the SCN by light is an intermediate step in a pathway mediating entrainment of the clock to light-dark cycles. They further suggest that gating of this induction by the clock contributes to the restriction of phase-shifting by light to certain times, a feature that is essential for achieving stable entrainment (9).

To discover potential components of the entrainment pathway, we sought to identify genes induced in the SCN by light without making assumptions about their nature. We developed a cDNA subtraction method based on genomic representational difference analysis (RDA) (10) and carried out subtractions as follows: Syrian hamsters were entrained to a light-dark cycle for 3 weeks and then placed in constant dim light (<1 lux) for 1 week (11). Animals were then assigned to either of two equal groups for light treatment (30 min, 250 lux) or sham treatment (similar handling, <1 lux) at circadian time (CT) 19, a time during subjective night that is optimal for a phase advance by light. At the end of the treatment, SCNs were removed by micropunch from 40 animals in each group. The remaining animals served as controls (Fig. 1). As expected, light treatment resulted in a phase advance (Fig. 1A), whereas sham treatment resulted in little or no phase shift (Fig. 1B).

Figure 1

Examples of controls confirming that the light or sham treatments had the expected effect on the phase of the circadian clock. Shown are spontaneous wheel-running activity records of two hamsters in constant dim light. Successive days are represented by horizontal lines, and the 24 hours within each day are represented on the x axis. Tick marks on horizontal lines represent bouts of spontaneous wheel-running activity. (A) Record from a hamster receiving a 30-min light pulse at the indicated time (diamond), corresponding to CT 19, as determined from the animal's spontaneous activity rhythm (not from the scale on thex axis). On the days after the light pulse, the shift to an earlier daily onset of wheel-running activity marks a phase advance, calculated here to be +1.10 hours (2). The mean from two light-treated hamsters was +1.23 hours. (B) Record from a hamster receiving a 30-min sham treatment at the indicated time (circle), corresponding to CT 19, as determined from the animal's spontaneous activity rhythm. The calculated phase shift was –0.04 hours; the mean from two sham-treated hamsters was –0.07 hours.

Starting with 1 μg of polyadenylated RNA from each of the two groups of 40 SCN tissue punches, we generated cDNA representations (12). Each RDA experiment (13) was performed so as to identify genes induced by light (“forward”) and genes suppressed by light (“reverse”) (13), the latter used here solely as a control. After three rounds of RDA, polymerase chain reaction (PCR) products were cloned and recombinant plasmids were chosen at random for screening by differential hybridization (14). Duplicate Southern (DNA) blots were prepared for sets of plasmids, and one blot from each duplicate was hybridized to32P-labeled cDNA derived from SCNs of light-treated animals and the other to a comparable probe derived from SCNs of sham-treated animals; these are denoted +light and –light probes, respectively (Fig. 2A). Most inserts showed equivalent hybridization to the two probes, some showed no detectable hybridization to either probe, and some showed stronger hybridization to the +light probe than to the –light probe (asterisks in Fig. 2A). Of 792 RDA clones tested, 101 showed differential hybridization to the +light probe. After culling likely artifacts (15), sequencing and cross-hybridization indicated that the remaining 60 differentially hybridizing inserts corresponded to seven different clones. Notably, we obtained c-fos and nur77, two of the five early-response genes known to be induced in the SCN by light.

Figure 2

(A) Example of initial characterization of RDA clones by differential hybridization. Duplicate Southern blots show hybridization of RDA clones to32P-labeled cDNA derived from SCNs of light-treated animals (+light probe) or to a comparable probe from SCNs of sham-treated animals (–light probe). Asterisks mark two inserts showing stronger hybridization to the +light probe; these are two copies of rda-7 (see text). (B) Progressive enrichment of differentially hybridizing RDA clones through the RDA procedure. Southern blots of cDNA (150 ng) from each stage of RDA were probed with full-length c-fos cDNA or rda-7 cDNA, as indicated. P, initial cDNA representation derived from SCNs of hamsters receiving light treatment; S, initial cDNA representation derived from SCNs of hamsters receiving sham treatment; DP1, DP2, and DP3, difference products obtained after one, two, and three rounds of RDA subtraction, respectively (+, forward subtraction; –, reverse subtraction) (13).

To exclude a fluke of sampling as the explanation for this result, we examined whether differentially hybridizing fragments had become progressively enriched through the RDA procedure (Fig. 2B). With a full-length c-fos probe, several PCR products showed enrichment at each stage and were observed only in the forward subtractions (Fig. 2B, DP1, DP2, and DP3, + lanes). For a probe corresponding to one of our differentially hybridizing inserts (designated rda-7), a PCR product corresponding in size to the rda-7 insert showed progressive enrichment and was observed only in the forward subtractions. Phosphorimaging indicated that c-foswas enriched by a factor of ∼250 and rda-7 was enriched by a factor of >3000.

Next, we tested the in vivo regulation of genes corresponding to the differentially hybridizing RDA clones (Fig.3) (16). A c-fosantisense riboprobe gave the expected results, showing specific c-fos induction in the SCN by light at CT 19 and CT 14 but not at CT 6; this reflects gating of induction by the circadian clock. Of the five RDA candidate clones, two, rda-7 and rda-65, detected transcripts showing specific induction in the SCN by light and gating by the circadian clock, much like that of c-fos, but with a cellular distribution within the SCN that appeared to differ from that of c-fos. The anatomical patterns and time courses of induction at CT 19 detected by the rda-7 and rda-65 riboprobes appeared identical; induction peaked during the first 30 min after the onset of exposure to light, became moderate at 2 hours, and returned to baseline by 4 hours (17). Little or no hybridization to the rda-7 and rda-65 riboprobes was detected in SCN sections from hamsters kept in constant dim light and killed at CT 2, 8, 14, or 20 (17); this result suggested that the transcripts do not contribute to the mechanism of the clock beyond a possible role in entrainment. As with c-fos, we detected no other prominent sites of induction in the brain by light, and sense riboprobe controls showed only background hybridization that did not differ between light- or sham-treated animals (17). Riboprobes from the remaining three candidate RDA clones did not reveal light-induced or clock-regulated genes.

Figure 3

In vivo regulation of transcripts corresponding to RDA clones; induction in the SCN by light and gating by the circadian clock. Shown are the results of in situ hybridization of antisense riboprobes (rda-7, rda-65, or c-fos) to coronal brain sections from hamsters subjected to a light (+) or sham (–) treatment at CT 19, 14, or 6. Variations in apparent size of the third ventricle are artifacts of tissue preparation.

Database searches revealed that rda-7 and rda-65 made significant sequence matches (18), respectively, to different parts of the 3′ untranslated region of human egr-3, an early-response gene encoding a zinc finger transcription factor (19). Further analysis of hamster cDNA and genomic clones indicated that rda-7 and rda-65 were both derived from the 3′ untranslated region ofegr-3. Although not previously known to be induced in the SCN, egr-3 is induced in various brain regions in response to stress or after focal brain injury (20). In the SCN, it is likely that egr-3 participates in the transcriptional regulation of genes in response to retinal input, as has been proposed for c-fos (6).

Light-induced c-fos and egr-3 transcripts consistently appeared to exhibit different anatomical distributions within the SCN (Fig. 3). To exclude animal-to-animal variation as the source of this difference, we examined c-fos andegr-3 induction (Fig. 4A) or c-fos, jun-B, and egr-3 induction (Fig. 4B) within SCNs of individual light-treated hamsters. As expected, c-fos induction was observed in both the dorsal and ventral regions of the SCN, as was jun-B induction. In contrast, egr-3 induction was restricted to the ventral core of the SCNs. In addition, c-fos and jun-Binduction were observed throughout the rostrocaudal extent of the SCN, whereas egr-3 induction was observed only in the central part of this rostrocaudal extent. The ventral SCN structure in whichegr-3 is induced by light resembles that stained by antisera to calbindin-D28K (21) or substance P (22).

Figure 4

Different cellular distribution within the SCN of light-induced egr-3 transcripts as compared with light-induced c-fos and jun-B transcripts. Images show central sections (at the approximate center of the rostrocaudal extent of the SCN) or rostral and caudal margins of the SCN, as indicated at the right. (A) Neighboring coronal brain sections from a light-treated hamster hybridized, respectively, to egr-3 (rda-7) or c-fos antisense riboprobes, as indicated. The right SCN in the right panel is distorted by an artifact of tissue preparation. (B) Sets of three neighboring coronal brain sections from a different light-treated hamster, each set taken from a different rostrocaudal level. Within each set, the three sections were hybridized, respectively, to c-fos, jun-B, or egr-3(rda-7) antisense riboprobes, as indicated.

The distinct cellular distribution of light-induced egr-3transcripts, as compared with light-induced c-fos andjun-B transcripts, indicates that the SCN responds to light in a complex fashion with overlapping but distinct regions activated in parallel. This divided response of the SCN to light could arise from an anatomically divided projection from the retina to the SCN, as suggested by electrical stimulation experiments (23), or from functional specialization of retinorecipient SCN cells, as suggested by the complex expression patterns of neurochemical markers (22). Pharmacological experiments have shown that retinal excitatory transmission to the ventral SCN is specifically required for circadian phase-shifting by light (24). Induction ofegr-3 itself, or activation of the ventral SCN structure that its induction by light reveals, could account for this requirement.

Note added in proof: Like egr-3, induction of the mouse per1 gene by light shows a restriction to the ventral SCN (25).

REFERENCES AND NOTES

View Abstract

Stay Connected to Science


Editor's Blog

Navigate This Article