|Home | About | Journals | Submit | Contact Us | Français|
The formation of the embryonic brain requires the production, migration, and differentiation of neurons to be timely and coordinated. Coupling to the photoperiod could synchronize the development of neurons in the embryo. Here, we consider the effect of light and melatonin on the differentiation of embryonic neurons in zebrafish. We examine the formation of neurons in the habenular nuclei, a paired structure found near the dorsal surface of the brain adjacent to the pineal organ. Keeping embryos in constant darkness causes a temporary accumulation of habenular precursor cells, resulting in late differentiation and a long-lasting reduction in neuronal processes (neuropil). Because constant darkness delays the accumulation of the neurendocrine hormone melatonin in embryos, we looked for a link between melatonin signaling and habenular neurogenesis. A pharmacological block of melatonin receptors delays neurogenesis and reduces neuropil similarly to constant darkness, while addition of melatonin to embryos in constant darkness restores timely neurogenesis and neuropil. We conclude that light and melatonin schedule the differentiation of neurons and the formation of neural processes in the habenular nuclei.
The light-dark cycle synchronizes the circadian clock of organisms with their environment (Vallone et al., 2007). In zebrafish, light can be perceived not only by the eyes and pineal gland (photoreceptive organs) but, uniquely among model vertebrates, also by other organs and cultured cells (Kaneko et al., 2006; Tamai et al., 2004; Tamai et al., 2007; Whitmore et al., 2000; Whitmore et al., 1998). Light has been shown to initiate molecular oscillations in the zebrafish embryo (Dekens and Whitmore, 2008; Kazimi and Cahill, 1999; Vatine et al., 2009; Vuilleumier et al., 2006) and affect the timing of the cell cycle (Dekens et al., 2003), as well as modulate predator avoidance behavior in zebrafish larvae (Budaev and Andrew, 2009). However, the consequences of light on neurogenesis have only recently begun to be characterized (D’Autilia et al., 2010; Dulcis and Spitzer, 2008; Toyama et al., 2009).
Melatonin acts as a marker of photoperiod in vertebrates, regulating both daily and seasonal behavior in adults via receptors found in specific brain regions (Pandi-Perumal et al., 2008). In the zebrafish pineal organ, melatonin is synthesized from serotonin by a series of enzymes including arylalkylamine-N-acetyltransferase (aanat2). Transcription of aanat2 is cyclic, with peaks during the night and troughs during the day. Under conditions of alternating light:dark (L:D) periods, aanat2 is expressed by 22 hours post fertilization (hpf) in zebrafish embryos (Gothilf et al., 1999; Zilberman-Peled et al., 2007) and robust, cyclic melatonin production can be detected by 37 hpf (Kazimi and Cahill, 1999). This rhythmic expression depends on the synchronization of oscillations so that aanat2 expression is in phase in all pineal cells. The oscillators are synchronized by Period-2 (Per2), a transcriptional repressor induced by light in cells of the zebrafish pineal organ. In the absence of Per2 activity due to constant darkness, aanat2 expression and melatonin production reach a constant, intermediate level (Kazimi and Cahill, 1999; Ziv et al., 2005). Melatonin receptors are present at high levels in the embryonic brain (Rivkees and Reppert, 1991; Seron-Ferre et al., 2007), and in mammals, melatonin can be transferred to the developing fetus via the placenta (Klein, 1972) and to the newborn via milk (Reppert and Klein, 1978). Low melatonin synthesis due to mutation of the biosynthetic enzyme acetylserotonin O-methyltransferase (ASMT) has been linked to autism spectrum disorders (Melke et al., 2008). Melatonin treatment of mammalian neural stem cells induces their differentiation (Bellon et al., 2007; Kong et al., 2008; Moriya et al., 2007). Finally, melatonin stimulates increased cell division in zebrafish embryos (Danilova et al., 2004). Therefore, a link between light stimulation, gene expression and melatonin exists during early development, but its influence on neurogenesis is not well understood.
In order to investigate the effects of light and melatonin on neurogenesis, we examined the development of the habenular nuclei. These are a pair of brain nuclei that are adjacent to the pineal organ and make up part of the highly conserved dorsal diencephalic conduction system (DDCS) implicated in modulation of the dopamine and serotonin systems (Hikosaka, 2010; Sutherland, 1982). The habenular nuclei express opsin proteins in fish and amphibians (Bertolucci and Foa, 2004) and receive projections from pinealocytes in the Djungarian hamster (Korf et al., 1986). In addition, neurons of the habenular nuclei express melatonin receptors in mice (Weaver et al., 1989) and undergo seasonal changes in morphology in frogs (Kemali et al., 1990). We examined neuronal differentiation and gene expression in the zebrafish habenular nuclei and find that light and melatonin control the timing of neuronal differentiation. In particular, reduction of light and melatonin produce a delay in differentiation which ultimately alters the DDCS by reducing the extension of neuronal processes in the habenular nuclei. Our results demonstrate that light and melatonin have significant effects on vertebrate brain formation.
Zebrafish were raised at 28.5°C on a 14/10 hour light/dark cycle (LD), a 10/14 hour dark/light cycle (DL), constant light (LL) or constant darkness (DD). For DL and DD conditions, embryos were put into darkness by 5 minutes post fertilization. Embryos and larvae were staged according to hours (h) or days (d) post fertilization. The wild-type AB strain (Walker, 1999) was used. To prevent melanosome darkening, embryos were raised in water containing 0.003% phenylthiourea.
Embryos were treated by placing them in egg water containing melatonin (0.001 or 0.02 μmolar, Sigma), U0126 (100 μmolar, Sigma), or luzindole (5, 7.5 or 10 μmolar, Sigma) for the duration of the treatment. The peak concentration of melatonin used in egg water (0.02 μmolar) is equivalent to the peak concentration of endogenous melatonin in untreated LD embryos at 67 hpf (0.46 pg/embryo [from Figure 4]=0.02 μmolar; molarity caluculated using the molecular weight of melatonin as 232.28 g/mole and the volume of an embryo at 128 nl volume based on Cheung et al, 2006). For controls, embryos were placed in egg water with vehicle alone (ethanol for melatonin or DMSO for luzindole and U0126).
For cloning of melatonin receptor mtnr1aa by RT-PCR, total RNA was isolated from 24 hpf zebrafish embryos using Trizol (Invitrogen), and cDNA prepared using Superscript II reverse transcriptase (Invitrogen). cDNA was amplified using primers within the ORF of mtnr1aa and cloned into the pCRII-Topo vector (Invitrogen). For cloning of melatonin receptors mtnr1a-like and mtnr1ba, total genomic DNA was isolated from zebrafish caudal fin samples using alkaline lysis. The largest exon of each gene was amplified using primers within the exon and cloned into the pCRII-Topo vector. An EST for mtnr1bb was purchased from Open Biosystems.
Whole-mount RNA in situ hybridization was performed as described previously (Snelson et al., 2008), using reagents from Roche Applied Bioscience. RNA probes were labeled using fluorescein-UTP or digoxygenin-UTP. To synthesize antisense RNA probes, pBK-CMV-leftover(kctd12.1) (Gamse et al., 2003) was linearized with EcoRI and transcribed with T7 RNA polymerase; pBK-CMV-right on (kctd12.1) (Gamse et al., 2005) with BamHI and T7 RNA polymerase; pBS-gfi1 (Dufourcq et al., 2004) with SacII and T3 RNA polymerase. pBK-CMV-cpd2 (cadps2) (Gamse et al., 2005) with Sal I and T7 RNA polymerase, pCR4-nrp1a ((Kuan et al., 2007b) with NotI and T3 RNA polymerase, cxcr4b (Chong et al., 2001) with EcoRV and SP6 RNA polymerase, pBS-otx5 (Gamse et al., 2002) with Not1 and T7 RNA polymerase, mtnr1aa with XhoI and SP6 RNA polymerase, mtnr1bb with EcoRI and SP6 RNA polymerase, and mtnr1a-like and mtnr1ba with EcoRV and SP6 RNA polymerase. Embryos were incubated at 70°C with probe and hybridization solution containing 50% formamide. Hybridized probes were detected using alkaline phosphatase-conjugated antibodies and visualized by 4-nitro blue tetrazolium (NBT) and 5-bromo-4-chloro-3-indolyl-phosphate (BCIP) staining for single labeling, or NBT/BCIP followed by iodonitrotetrazolium (INT) and BCIP staining for double labeling. All in situ data was collected on a Leica DM6000B microscope with a 10x or 20x objective.
Melatonin was isolated from zebrafish embryos as previously described (Kazimi and Cahill, 1999) with the following modifications: Methylene chloride was evaporated under vacuum using a rotary evaporator with the collection vial semi-submerged in a room-temperature water bath. Dried extracts were eluted in 0.2 mL 0.1% porcine gelatin (type a) in PBS. This volume was used in full to generate duplicate samples that were subsequently analyzed using a Direct Saliva Melatonin ELISA (Alpco) following manufacturer’s instructions, beginning with acid/base pretreatment.
In order to validate the use of the ELISA assay for detecting melatonin from zebrafish embryos, we quantified the amount of melatonin in 43 hpf embryos raised in LD conditions, with a sample of 5 versus 15 embryos. The amount of melatonin that was reported by the ELISA increased by 2.7 times when the number of embryos was increased 3-fold, indicating that the assay is valid for use with zebrafish embryos.
For LL experiments, the concentration of melatonin per embryo is extremely low (<0.01 pg/embryo). In order to ensure that the ELISA was able to detect the small amount of melatonin in these samples, a large number of embryos were pooled for each time point. A circadian variation in melatonin concentration was detected in these samples, indicating that the ELISA was working properly.
For whole-mount immunohistochemistry with rabbit or mouse-derived antibodies, larvae were fixed overnight in 4% paraformaldehyde or Prefer fixative (Anatech). Paraformaldehyde-fixed samples were permeabilized by treatment with 10 μg/ml Proteinase K (Roche Applied Bioscience) and refixed in 4% paraformaldehyde. Prefer-fixed samples were not permeabilized. All samples were blocked in PBS with 0.1%TritonX100, 10% sheep serum, 1% DMSO, and 1% BSA (PBSTrS). For antibody labeling, rabbit anti-Lov (Kctd12.1) or rabbit anti-Ron (Kctd12.2) (1:500; Gamse et al, 2005), rabbit anti-GFP (1:1000, Torrey Pines Biolabs), HuC-D (1:200, Invitrogen), SV2 (1:500, Developmental Studies Hybridoma Bank), acetylated alpha-tubulin (1:1000, Sigma) were used. Larvae were incubated overnight in primary antibody diluted in PBSTrS. Primary antibody was detected using goat-anti-rabbit or goat-anti-mouse secondary antibodies conjugated to the Alexa 568 or Alexa 488 fluorophores (1:350, Invitrogen). Samples were counterstained with TOPRO3 (1:10,000, Invitrogen).
For quantitation of neuropil, confocal data was imported into Volocity (Improvision), and the lasso tool was used to select all anti-acetylated tubulin fluorescence within the left or right habenular nucleus, excluding the habenular commissure. The volume of this region was calculated using Quantitation module of Volocity. We recorded the volume of the largest contiguous labeled region as the volume of neuropil in the habenula (in order to exclude the large amount of small speckle artifacts).
All immunofluorescence data were collected on a Zeiss LSM510 confocal microscope with a 40x oil-immersion objective and analyzed with Volocity software (Improvision).
To test the effects of photoperiod on neuronal differentiation, we examined the development of the habenular nuclei under different light/dark conditions (Fig. 1A). In a 14 hour light:10 hour dark (LD) photoperiod, habenular neurons express the potassium-channel-tetramerization-domain (KCTD) containing genes kctd12.1 and kctd12.2. kctd12.1 is expressed in the lateral subnucleus, which is larger in the left habenula, while kctd12.2 is expressed in the medial subnucleus, which is larger in the right habenula (Gamse et al., 2005). Neurons of both subnuclei express the synaptic vesicle priming protein calcium dependent activator protein for secretion 2 (cadps2) (Gamse et al., 2005). Under LD conditions, transcription of kctd12.1, kctd12.2 and cadps2 transcription is first detectable at 38, 45, and 44 hpf respectively in the habenular nuclei (Fig. 1B, F, J). However, when embryos are exposed to constant darkness (DD) conditions, neuronal development is significantly postponed. Expression of kctd12.1, kctd12.2, and cpd2 is delayed until 48, 49, and 52 hpf (delay of 10, 4, and 8 hours) respectively (Fig. 1C-E, G-I, K-M).
The effect of DD conditions on neuronal differentiation in the epithalamus is not a generalized delay of brain development. Expression of neuropilin 1a (nrp1a), a semaphorin receptor required for habenular axon targeting (Kuan et al., 2007b), is unaffected by DD treatment (Fig. 1N-O). Accordingly, habenular axons innervate their appropriate targets in the interpeduncular nucleus of the midbrain (Supplemental Figure 1A-B). Furthermore, the formation of the pineal complex occurs on time, marked by expression of the genes otx5 (Gamse et al., 2002) and gfi-1 (Dufourcq et al., 2004) (Fig. 1P-S) and by HuC/D in projection neurons (Fig. 2I-J, white arrowheads). In addition, kctd12.1 expression in the pituitary and kctd12.2 expression throughout other regions of the brain is unchanged (Fig. 1B-I).
The initial expression of kctd12.1 at 38 hpf in the habenular nuclei coincides with the start of the second dark phase of the photoperiod, while kctd12.2 and cadps2 initiates in the middle of the second dark phase. By reversing the phase of the photoperiod to 10 hours of darkness followed by 14 hours of light (dark-light [DL]), the second dark phase would occur 14 hours earlier (Figure 2A). To determine if expression of genes in the habenular nuclei is correlated with the second dark phase, we incubated embryos in DL conditions, and examined whether gene expression was advanced by 5 hours. We harvested LD and DL embryos 5 hours and 0 hours before the first time point that we can detect expression in LD embryos (33 and 38 hpf for kctd12.1, 40 and 45 hpf for kctd12.2, 39 and 44 hpf for cadps2, respectively). Expression of kctd12.1, kctd12.2, and cadps2 was absent in DL embryos at the earlier time point, and present at the later time point, similar to LD control embryos (Figure 2B-M). Therefore, gene expression is not advanced by 5 hours in DL embryos relative to LD controls. However, a slight advance in the timing of kctd12.1 expression may be present. Expression of kctd12.1 at its onset is low, and increases gradually over time (compare Figure 1B to 1D). In the habenular nuclei, the number of embryos with high expression of kctd12.1 was greater in DL than in LD embryos at 38 hpf (compare Figure 2D to E; 100% of LD embryos (n=30) had expression equal or less than the example shown in Figure 2D, whereas 83% of DL embryos (n=18) had expression equal or greater to the example shown in Figure 2E, and the remainder resembled Figure 2D; p<0.0001, two-tailed Fisher’s exact test). We find evidence for slightly premature kctd12.1 expression in the habenular nuclei of DL embryos compared to LD siblings, and no change in the timing of expression for kctd12.2 and cadps2.
The late development of habenular neurons in DD conditions could result from delayed specification of progenitor cells, or delayed differentiation of progenitors into post-mitotic habenular neurons. We examined expression of cxcr4b, a marker of habenular progenitor cells and newly born neurons (Roussigne et al., 2009). Initially, similar numbers of cxcr4b+ cells are detected in the epithalamus of LD and DD embryos, but as development progresses, excess cxcr4b+ cells accumulate in DD embryos relative to LD embryos (Fig. 3A-F; 100% of DD embryos had excess cxcr4b+ cells). Similar to Roussigne et al, we note a left-biased initial appearance of cxcr4b+ cells in both LD and DD conditions. By 72 hpf, the number of cxcr4b+cells in DD embryos is similar to LD embryos (Fig. 3G-H). The RNA binding proteins HuC/D are expressed in post-mitotic habenular neurons (Kim et al., 1996; Roussigne et al., 2009). Under DD conditions, many fewer HuC/D-expressing precursors are detected in the habenular nuclei at 38 hpf relative to LD siblings (average of 31 total HuC/D-expressing cells for LD versus 9 for DD, p<0.001 in two-tailed T-test; Fig. 3I-J). Therefore, it appears that in constant darkness, an appropriate number of habenular progenitor cells are specified, but they exit the progenitor state late.
We hypothesized that delayed melatonin production by the pineal organ may contribute to the delay in habenular neurogenesis, so we examined melatonin production in DD embryos. A previous report demonstrated that raising zebrafish embryos in continuous darkness beginning at 14 hpf resulted in near-basal production of melatonin until 55 hpf (Kazimi and Cahill, 1999). We confirmed these results by beginning the dark period within 5 minutes post fertilization, and harvesting embryos at 12-hour intervals for analysis by ELISA (Figure 4A). We find that in LD embryos, melatonin is first detectable at 43 hpf, at a concentration of 0.10 pg/embryo (Figure 4B). By contrast, DD embryos do not produce a similar concentration of melatonin until 55 hpf. The 12 hour delay in melatonin synthesis is similar to the 10-hour delay in kctd12.1 gene expression, the earliest marker of habenular neuron differentiation that we have tested.
We examined the expression of melatonin receptors in the embryonic zebrafish. A previous report had shown that the melatonin receptors mtnr1aa (previously Z1.7), mtnr1ba (previously Mel1b) and mtnr1bb (previously Z2.6-4), are expressed in zebrafish embryos between 18 and 36 hpf (Danilova et al., 2004). To examine which of these receptors is expressed in habenular precursor cells, we performed in situ hybridization at 24, 36, and 48 hpf. We find that mtnr1ba and mtnr1bb are expressed throughout the central nervous system at all time points examined (Figure 4C-H), including in habenular cells, which are marked by expression of cxcr4b (Roussigne et al., 2009) (Figure 4I-K). Expression of mtnr1aa was found in the ventral hindbrain but was not detected in habenular cells (data not shown).
Next we assessed the role of melatonin signaling in the development of the habenular nuclei. To test whether reduced melatonin levels contributed to the delay in neuronal development in DD conditions, we first tested the ability of melatonin to rescue habenular development in DD conditions. DD embryos were treated with exogenous melatonin, either in imitation of LD conditions (14 hours (h) low melatonin:10 h high) or continuously (24 h high melatonin) (Fig. 5A), starting at 14 hpf. Either treatment rescues habenular development in DD embryos to resemble LD embryos (Fig. 5B, C, F, G, H). Next, we tested if blocking melatonin receptor activity in LD embryos could replicate the delayed differentiation phenotype of DD embryos. We found that treatment of embryos in LD conditions with the transmembrane melatonin receptor MT1/2 antagonist luzindole (Dubocovich, 1988) can delay habenular development similar to DD conditions (Fig. 5D, E, F, I, J). Therefore, melatonin signaling can contribute to the timely differentiation of habenular neurons.
The MT1/2 melatonin receptors are 7-pass G-protein coupled proteins that can signal intracellularly via a number of pathways, including the MEK/ERK MAP kinase cascade (Jockers et al., 2008). We treated LD embryos with 2x 1-hour pulses of the MEK–specific inhibitor U0126, at 24 and 36 hpf (Fig. 5K). Following this treatment, habenular development was delayed similar to DD or luzindole treatment (Fig. 5L,M).
Adult fish or pineal organs kept in constant light (LL) conditions exhibit constitutively low levels of melatonin production (Bolliet et al 1995, Oliveira et al 2007, Amano et al 2006). We find a similar result in zebrafish embryos incubated under LL conditions (Figure 4B). The concentration of melatonin in LL embryos remains below 0.01 pg/embryo at all time points assayed.
Since addition of exogenous melatonin to embryos in DD conditions was sufficient to rescue the timely appearance of gene expression in the habenular nuclei, and because pharmacological inhibition of melatonin receptors was sufficient to inhibit timely gene expression in LD embryos, we examined LL embryos to determine if melatonin was necessary for the timing of gene expression. However, we detected no delay in gene expression in the habenular nuclei or the pineal complex of LL embryos (Figure 6 B-S). In fact, a slight advance in the timing of gene expression may be present in the expression of kctd12.1 and cadps2. Expression of both of these genes at the onset is low, and increases gradually over time (compare Figure 6B to D, Figure 6J to M). In the habenular nuclei of LL embryos, 63% of embryos exhibited moderate to high expression of kctd12.1 at 38 hpf (moderate expression is shown in Figure 6C), while 66% of LD embryos exhibited low or no expression of kctd12.1 at 38 hpf (low expression is shown in figure 6B). A similar although less striking increase in expression is detected for cadps2 (LL: 35% with moderate expression [Figure 6K], 63% with low expression, n=42. LD: 0% with high expression, 61% with low expression [Figure 6J], 39% with no expression, n=44). By 48 and 52 hpf, the number of embryos with moderate or high expression of kctd12.1 and cadps2 is nearly equal for LL and LD embryos (LL: 78% for kctd12.1, n=83; 98% for cadps2, n=48. LD: 70% for kctd12.1, n=70; 98% for cadps2, n=51. Expression is low for both genes in the remainder of embryos). We conclude that under LL conditions, where melatonin production is greatly diminished, development of the habenular nuclei occurs on time, and may be slightly advance.
In addition to delaying neurogenesis, raising embryos in DD conditions resulted in reduced neuropil in the habenular nuclei. This neuropil consists of defasciculated axons from the forebrain and dendrites from habenular neurons (Concha et al., 2000; Hendricks and Jesuthasan, 2007; Moutsaki et al., 2003). We used confocal imaging and volumetric analysis of neuropil to assay changes in LD versus DD larvae. At 48 hpf, DD embryos form an average of 28.5% less neuropil volume in the left habenula than LD siblings (Fig. 7 A-C). By 72 hpf, a 21% average reduction in total neuropil volume is seen (Fig. 7 D-F). We find that decreased neuropil under DD conditions is due to reduced melatonin receptor signaling. Treatment of LD embryos with luzindole causes decreased neuropil relative to untreated LD embryos (Fig. 7 G-I). Conversely, DD embryos treated with melatonin exhibit an increase in neuropil relative to DD alone (Fig. 7 J-L).
In DD embryos, the volume of presynaptic densities in forebrain axons terminating on habenular dendrites was unchanged relative to LD (Supplemental Figure 2A-C). In addition, at 72 hpf the number of cells in the L/R asymmetric lateral subnucleus was unaffected by LD versus DD (Supplemental Figure 2D-F). Because cell number and inputs appear unchanged, the reduced neuropil is best explained as decreased dendritogenesis by habenular neurons.
Light plays a crucial role in starting the circadian oscillator in the pineal organ as well as synchronizing the oscillations of individual pineal cells so as to generate nighttime peaks of melatonin output (Dekens and Whitmore, 2008; Kazimi and Cahill, 1999; Tamai et al., 2007; Vuilleumier et al., 2006; Whitmore et al., 2000; Ziv et al., 2005). We find that both light and melatonin are important for the timing of neuronal differentiation in the habenular nuclei and ultimately for the appropriate elaboration of dendrites from these neurons.
We were able to rescue habenular neuron differentiation in 100% of DD embryos with melatonin, and recapitulate delayed differentiation in 100% of LD embryos with the melatonin receptor inhibitor luzindole, indicating an important role for melatonin in the development of habenular neurons. However, we noted changes in the size of the precursor pool (cxcr4b-expressing cells) in only a fraction (~30%) of the melatonin- or luzindole-treated embryos. In addition, LL embryos receiving a constant light signal produce no detectable melatonin, yet they showed no delay of habenular differentiation, and may exhibit a modest advancement in the timing of some genes’ expression in the habenular nuclei. Therefore, although melatonin is sufficient to promote the differentiation of habenular neurons under DD conditions, it is not necessary under LL conditions. One explanation for the sufficiency but not necessity of melatonin is that light may act in a parallel pathway independent of melatonin to stimulate differentiation of habenular neurons. Many tissues of the zebrafish have been demonstrated to be light responsive (Cahill, 1996; Dekens et al., 2003; Tamai et al., 2005; Whitmore et al., 2000), express photosensitive pigments including cryptochromes and teleost multiple tissue (tmt) opsin (Moutsaki et al., 2003; Tamai et al., 2007) and photosensitive enzymes such as acetyl-CoA oxidases (Hirayama et al., 2007; Hockberger et al., 1999; Thisse and Thisse, 2004). Habenular cells receiving both light and melatonin input could integrate the intensity of signaling over time. Once the total light- and melatonin-mediated signaling reaches a threshold amount, habenular progenitors would differentiate. In this model, light signaling during the day plus melatonin signaling during the night in LD conditions would reach the threshold for activation of kctd12.1 expression at 38 hpf, at 45 hpf for kctd12.2 expression, and so on. Under DD conditions, in the absence of light input, the threshold would not be reached until 5-10 hours later, when melatonin production had become great enough for a long enough period of time. Conversely, under LL conditions, constant signaling by light input could reach the threshold more quickly, even in the absence of melatonin, since the amount of time that the embryos are exposed to light is almost doubled. Finally, under DL conditions, in which the second dark period occurs 14 hours earlier than in LD conditions, it is conceivable that a peak of melatonin production occurs earlier than in LD, and accordingly we do detect a slight advance in gene expression under DL conditions. Knocking down each of the photoreceptive proteins in the context of melatonin receptor inhibition should reveal if habenular precursor cells integrate light and melatonin signals in order to time their differentiation.
Integration of light and melatonin signaling might occur via a shared signal transduction cassette, the ERK MAP kinase pathway. Oxidative species, such as those generated by light-sensitive flavin-containing oxidases, induce gene expression via the ERK MAP kinase pathway (Hirayama et al., 2007). Melatonin receptors can also activate ERK MAP kinase signaling (Daulat et al., 2007; Witt-Enderby et al., 2000). In support of integration occurring at the level of intracellular signal transduction, we find that inhibition of ERK MAP kinase signaling by U0126 is capable of delaying kctd12.1 expression in the habenular nuclei. More targeted manipulation of the ERK MAP kinase pathway, such as conditional inactivation of ERK1 and 2 in the nervous system, will be necessary to test this hypothesis.
Decreased avoidance of a simulated predator is reported for zebrafish larvae raised in constant darkness (Budaev and Andrew, 2009). Budaev and Andrew have hypothesized that light input influences predator response by affecting habenular output (Budaev and Andrew, 2009) via changes in Nrp1a expression and thus axon targeting (Kuan et al., 2007a). However, we do not detect a change in nrp1a expression or axonal targeting to the IPN in LD versus DD conditions. We do find that DD conditions result in decreased neuropil density in the habenular nuclei, perhaps because habenular neurons are exposed for a shorter time to intrinsic or extrinsic signals for dendrite formation (Parrish et al., 2007). Habenular neuron function has been recently implicated in zebrafish learning whether it is best to flee or freeze in place in response to a negative stimulus (Agetsuma et al., 2010; Lee et al., 2010), a behavior that is relevant in reacting to predators. It is therefore possible that decreased predator avoidance behavior in DD-raised larvae is a consequence of reduced habenular dendritogenesis.
We show that the timing of neuronal differentiation and subsequently the appropriate outgrowth of dendrites during habenular development is an event that requires light and the hormone melatonin. Intriguingly, alteration of melatonin production is a symptom of some neurological diseases, including autism and Smith-Magenis syndrome (Elsea and Girirajan, 2008; Kulman et al., 2000; Nir, 1995; Tordjman et al., 2005). In addition, mutations of the melatonin biosynthetic enzyme ASMT are linked to increased autism risk (Melke et al., 2008). It has been proposed that altered melatonin during early postnatal development may be causative rather than simply symptomatic of these diseases, by altering formation of brain circuits (Bourgeron, 2007). The zebrafish embryo, with its easily manipulated pathway for melatonin signaling, now provides a platform to explore how melatonin influences brain development.
Supplemental Figure 1 Axonal targeting to the midbrain is unaffected by DD conditions. (A-B) Targeting of kctd12.1-expressing axons from the habenular nuclei to the interpeduncular nucleus of the midbrain (white circle) is similar in LD and DD larvae. Dorsal views. Scale bar = 20 μm.
Supplemental Figure 2 Presynaptic densities and kctd12.1-positive cell number in the habenular nuclei is unaffected by DD conditions. (A-C) At 72 hpf, the volume of SV2 signal, representing presynaptic vesicles in axons synapsing on the habenula, is similar in LD and DD larvae. (D-F) The total number of kctd12.1-expressing cells in the habenular nuclei is similar in LD and DD larvae.
We thank Qiang Guan, Heidi Beck, and Gena Gustin for expert fish care, Erin Booton for excellent in situ hybridization support, Hugo Borsetti for discussion of experiments, Jeff Johnston for help with melatonin extraction from embryos, and Marnie Halpern for in situ clones. This work was funded by NIH grant HD054534 to J.T.G. B.J.D. was supported by the Vanderbilt Medical-Scientist Training Program (T32 GM07347 from the NIH) and J.A.C. was supported by the Vanderbilt Training Program in Developmental Biology (T32 HD007402 from the NIH).
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.