Abstract

   Most organisms possess circadian clocks that are able to anticipate the
   day/night cycle and are reset or “entrained” by the ambient light. In
   the zebrafish, many organs and even cultured cell lines are directly
   light responsive, allowing for direct entrainment of the clock by
   light. Here, we have characterized light induced gene transcription in
   the zebrafish at several organizational levels. Larvae, heart organ
   cultures and cell cultures were exposed to 1- or 3-hour light pulses,
   and changes in gene expression were compared with controls kept in the
   dark. We identified 117 light regulated genes, with the majority being
   induced and some repressed by light. Cluster analysis groups the genes
   into five major classes that show regulation at all levels of
   organization or in different subset combinations. The regulated genes
   cover a variety of functions, and the analysis of gene ontology
   categories reveals an enrichment of genes involved in circadian
   rhythms, stress response and DNA repair, consistent with the exposure
   to visible wavelengths of light priming cells for UV-induced damage
   repair. Promoter analysis of the induced genes shows an enrichment of
   various short sequence motifs, including E- and D-box enhancers that
   have previously been implicated in light regulation of the zebrafish
   period2 gene. Heterologous reporter constructs with sequences matching
   these motifs reveal light regulation of D-box elements in both cells
   and larvae. Morpholino-mediated knock-down studies of two homologues of
   the D-box binding factor Tef indicate that these are differentially
   involved in the cell autonomous light induction in a gene-specific
   manner. These findings suggest that the mechanisms involved in period2
   regulation might represent a more general pathway leading to light
   induced gene expression.

Introduction

   The ability to perceive sunlight provides animals with many adaptive
   advantages. Light perception can be used for orientation in the
   environment, the location of prey and the escape from predators as well
   as for communication via visual signals. The daily changes in lighting
   conditions also represent an important time cue for the optimal
   temporal distribution of activities and physiological processes of the
   organism, which in turn enhances survival. In this case, environmental
   lighting signals cooperate with endogenous timers, the most important
   of which is the circadian clock.

   Circadian clocks regulate daily changes in physiology in most
   organisms, ranging from cyanobacteria to humans [45][1]. These clocks
   consist of an oscillator mechanism that generates rhythms with a period
   of roughly 24 hours even in the absence of external cues. Molecularly,
   the oscillator is composed of proteins participating in a
   transcription-translation feedback loop (reviewed in [46][2], [47][3],
   [48][4]). In vertebrates for example, the transcription factors Clock
   and Bmal1 activate transcription of the period (per) and cryptochrome
   (cry) genes via so-called E-box elements in their promoters. The Period
   and Cryptochrome proteins accumulate in the cytoplasm, re-enter the
   nucleus and act there to inhibit the transcriptional activity of the
   Clock-Bmal1 complex. This reduces the transcription of the cry and per
   genes, hence less protein is made, and the repression is released, so
   that the cycle can start again. Additional feedback loops and
   posttranslational modifications are thought to confer the 24 h period
   of this mechanism and to render it more robust. This molecular clock
   mechanism is encountered in the neural circadian pacemaker of the
   brain, the paired suprachiasmatic nucleus (SCN) situated in the
   hypothalamus above the optic chiasm, which drives rhythms of behavior
   and other “centrally” regulated aspects of physiology. However, this
   clock mechanism is also present in most other tissues of the body,
   constituting so-called “peripheral” clocks. These clocks govern many
   aspects of cell and tissue physiology, acting either autonomously or in
   concert with systemic cues.

   The circadian oscillator can be synchronized with (or “entrained” to)
   the environment via a number of different cues, including temperature,
   food, various chemical compounds and, perhaps most pervasively, light.
   In mammals, a subset of retinal ganglion cells which are intrinsically
   photosensitive (so-called ipRGCs, [49][5]) projects to the SCN. The
   opsin photopigment, melanopsin, is expressed in the ipRGCs and is
   sufficient for circadian light responses in the absence of functional
   rods and cones [50][6], [51][7], [52][8]. However, melanopsin knock-out
   mice still entrain to light-dark (LD) cycles when rods and cones are
   present, therefore the ipRGCs can also transmit light information
   received through the rods and cones [53][9]. Light induced changes in
   ipRGC activity are signalled via glutamate and PACAP containing
   projections to the neurons of the SCN, leading to acute changes in
   neuronal activity and also to gene expression changes mediated by the
   transcription factor CREB [54][10]. Currently poorly defined signals,
   which might include hormones such as glucocorticoids, as well as body
   temperature changes and the activity of the autonomic nervous system,
   then transmit timing information from the SCN to the peripheral tissue
   clocks [55][11], [56][12]. However, under some conditions, e.g.
   unnaturally timed feeding schedules [57][13], peripheral clocks can be
   “decoupled” from the SCN clock and run even in antiphase to it,
   revealing the principal autonomy of peripheral tissue clocks from the
   central pacemaker. Nevertheless, to perceive changes in environmental
   lighting conditions, mammalian peripheral clocks rely upon systemic
   signals from the SCN.

   This is not the case in many other organisms. In Drosophila, peripheral
   tissue clocks respond directly to light [58][14]. Cryptochrome, which
   functions as part of the core clock feedback loop in mammals, serves as
   a photopigment in the fruit fly [59][15]. Strikingly, direct light
   responsiveness of peripheral clocks is also encountered in a
   vertebrate, the zebrafish. Clock gene expression in zebrafish organ
   cultures can be entrained to LD cycles [60][16], and even single
   zebrafish cell culture cells are able to entrain their clocks in
   response to light pulses [61][17], [62][18]. In the absence of light,
   rhythms in single cells continue, but drift out of phase with respect
   to the other cells, leading to an apparent dampening of the rhythm of
   the entire culture. The nature of the photoreceptor mediating this
   peripheral light responsiveness is still elusive, although various
   candidates have been proposed, including Teleost Multiple Tissue (TMT-)
   Opsin and Cryptochromes [63][19], [64][20]. A more recent study has
   implicated specifically the cry1a gene in mediating the effects of
   light on the clock. However, Cry1a is hypothesized to act as an element
   of the signalling pathway, rather than serving as a light receptor
   itself [65][21]. The expression of several clock genes acutely responds
   to light in zebrafish cells, e.g. per1, per2 and cry1a [66][17],
   [67][22], [68][23], [69][24]. Light induced expression of the per2 gene
   has very recently been shown to depend on D- and E-box enhancer
   elements within its promoter [70][25]. Strikingly, evidence points to
   the signalling pathways mediating light induced clock gene expression
   in fish being conserved in mammals, despite the lack of peripheral
   photoreception. Fibroblasts transfected with melanopsin become light
   sensitive and can entrain their clocks in response to exposure to LD
   cycles [71][26], [72][27]. Furthermore, the light responsive D- and
   E-box enhancer elements of the per2 gene are conserved in chicken and
   mammals, including humans, and a human promoter fragment containing
   these elements mediates light induction of a reporter gene when
   transfected into zebrafish cells. One way to interpret these findings
   would be that, in mammals, SCN-controlled signals for the entrainment
   of peripheral clocks might have co-opted pathways that formerly acted
   downstream of the now lost peripheral light receptor [73][25].

   Interestingly, the direct light reception in peripheral tissues might
   affect physiology not only indirectly, via regulating the clock, but
   also more directly. For example, one gene that has been shown to be
   light inducible already in the early embryo, even before the clock is
   running properly, is the DNA repair enzyme 6,4-photolyase/cry5
   [74][28]. This regulation seems likely to have physiological relevance,
   since mortality caused by UV treatment was reduced when zebrafish
   embryos had been exposed to light prior to treatment [75][28]. Similar
   findings have recently been reported in the zebrafish z3 cell line
   [76][29].

   Here, we set out to characterize light induced transcription in the
   zebrafish more globally by identifying light responsive genes at three
   levels of organization: whole larvae, heart organ cultures and cell
   culture cells. We identified a relatively restricted set of 117
   regulated genes, the majority (90) being upregulated by light. Induced
   genes fell into a variety of functional categories, including genes
   involved in circadian clock function, DNA repair, retinal light
   reception and metabolism, highlighting the pervasive effects of light
   exposure on physiology. Strikingly, examination of the promoters of the
   upregulated genes revealed an enrichment for E- and D-box binding
   sites, indicating that the role E- and D-box binding factors play in
   light induction of the per2 gene might also extend to many other light
   responsive genes.

Materials and Methods

Ethics Statement

   All zebrafish husbandry and experimental procedures were performed in
   accordance with the German animal protection standards and were
   approved by the Government of Baden-Württemberg, Regierungspräsidium
   Karlsruhe, Germany (Aktenzeichen 35-9185.64).

Raising adult and larval zebrafish

   Adult zebrafish (Tübingen strain) were raised according to standard
   methods [77][30]. Fertilized eggs were collected within 2 h of laying,
   and aliquots of 20 eggs were transferred into 20 ml of E3 buffer in
   25-cm^2 tissue culture flasks [78][31]. The flasks were incubated in a
   large-volume thermostat-controlled water bath equipped with an Osram
   L15W/41-827 light source for 5 days in constant darkness (DD) at 25°C,
   then illuminated with an approximate intensity of 1,200 lux for 1 or 3
   h or maintained in DD before RNA extraction. The intensity of
   illumination was measured with a POCKET LUX Illuminance Meter (LMT,
   Berlin).

In vitro heart culture

   In vitro heart cultures were carried out essentially as described
   before [79][32]. Briefly, freshly dissected tissue was placed in L15
   medium supplemented with 15% fetal calf serum and with gentamycin (50
   µg/ml) and Pen/Strep (100 units/ml; 100 µg/ml). Organs were dissected
   from fish and washed 4 times with medium (10 ml/5 hearts). 5 hearts
   each were placed into cell culture flasks containing 5 ml of medium and
   submerged in the water bath in constant darkness for 4 days, then
   subjected to the light pulse as performed for the larvae and finally
   directly processed for RNA extraction.

Cell culture

   PAC2 cell culture was carried out essentially as described before
   [80][16], [81][17]. Cells were seeded into 75 cm^2 flasks and submerged
   in the waterbath in constant darkness for a week before light pulse
   treatment and RNA isolation.

RNA isolation and microarray hybridization

   Total RNA was extracted from at least three biological repeat samples
   per experimental condition using Trizol RNA isolation reagent
   (GIBCO-BRL) according to the manufacturer's instructions. Synthesis and
   labeling of antisense RNA was performed as recommended by the array
   manufacturer, using kits for double-stranded cDNA synthesis
   (Invitrogen), for transcription and labeling of antisense RNA (Enzo
   Life Sciences) and for probe purification and hybridization controls
   (Affymetrix). Samples were hybridized to the Affymetrix Zebrafish
   GeneChip, representing 15,617 probes.

Microarray analysis

   Microarray hybridization data were analyzed using scripts written in
   the statistical programming language R [82][33] supported by packages
   provided by the Bioconductor project [83][34]. Background correction,
   normalization and probe set summarization were performed using the
   robust multi array algorithm with background adjustment (gcrma,
   [84][35]). Two methods were employed to detect genes differentially
   expressed in response to light exposure. First we used linear models
   and a moderated t-statistic from the package limma [85][36]. Multiple
   testing correction was performed using Benjamini and Hochberg false
   discovery rate (FDR) [86][37] and genes with an adjusted p-value of
   ≤0.05 were considered differentially expressed. Secondly, we used the
   meta-analysis technique, Rank Product [87][38], to generate a
   non-parametric statistic that detects genes consistently highly ranked
   in the comparisons between light exposed and control samples. All data
   is MIAME compliant, and raw and normalized data were stored in the
   ArrayExpress data base ([88]http://www.ebi.ac.uk/microarray-as/ae/,
   accession-no. E-MTAB-381).

   Annotations for the differentially expressed genes were retrieved with
   the Affymetrix probe set IDs using BioMart version 0.7 [89][39],
   querying the dataset of zebrafish genes based on the genome release Zv8
   in the Ensembl database (release 56). In the cases where no Ensembl
   Gene IDs were assigned, we examined the location of the probe set by
   BLAST to check whether they were located 5′ or 3′ to an annotated gene.
   If there was a unique BLAST hit close to a gene annotation, the
   corresponding Ensembl Gene ID was assigned to the probe set and marked
   as a 5′ or 3′-hit. The Ensembl Gene IDs were subsequently used to
   assign the corresponding ZFIN IDs and Entrez Gene IDs. For some probes,
   including the two most highly downregulated ones, no annotated sequence
   could be identified in the Zv8 release with either strategy, and they
   were therefore excluded from the subsequent analyses. Where no
   particular references are given, the gene description is based on