Circadian Control of Dendrite Morphology in the Visual System of Drosophila melanogaster

Background In the first optic neuropil (lamina) of the fly's visual system, monopolar cells L1 and L2 and glia show circadian rhythms in morphological plasticity. They change their size and shape during the day and night. The most pronounced changes have been detected in circadian size of the L2 axons. Looking for a functional significance of the circadian plasticity observed in axons, we examined the morphological plasticity of the L2 dendrites. They extend from axons and harbor postsynaptic sites of tetrad synaptic contacts from the photoreceptor terminals. Methodology/Principal Findings The plasticity of L2 dendrites was evaluated by measuring an outline of the L2 dendritic trees. These were from confocal images of cross sections of L2 cells labeled with GFP. They were in wild-type and clock mutant flies held under different light conditions and sacrified at different time points. We found that the L2 dendrites are longest at the beginning of the day in both males and females. This rhythm observed under a day/night regime (LD) was maintained in constant darkness (DD) but not in continuous light (LL). This rhythm was not present in the arrhythmic per01 mutant in LD or in DD. In the clock photoreceptor cryb mutant the rhythm was maintained but its pattern was different than that observed in wild-type flies. Conclusions/Significance The results obtained showed that the L2 dendrites exhibit circadian structural plasticity. Their morphology is controlled by the per gene-dependent circadian clock. The L2 dendrites are longest at the beginning of the day when the daytime tetrad presynaptic sites are most numerous and L2 axons are swollen. The presence of the rhythm, but with a different pattern in cryb mutants in LD and DD indicates a new role of cry in the visual system. The new role is in maintaining the circadian pattern of changes of the L2 dendrite length and shape.


Introduction
The visual system of different fly species show circadian rhythms in many physiological and structural processes. They have been detected in the retina, in changes of the electroretinogram amplitude [1], and in the first optic neuropil (lamina). In the lamina they include rhythms in the number of presynaptic profiles [2], in changes of interneuron axon sizes [3,4] and in migration of screening pigment granules in the photoreceptor terminals [5]. Those circadian rhythms existing in the retina photoreceptors are probably generated by circadian oscillators, called peripheral oscillators, located in the photoreceptors themselves [6]. The photoreceptors, like the clock neurons in the brain, the so-called lateral (LNs) and dorsal (DNs) neurons, show expression of clock genes. In the clock neurons the main clock gene period (per) and its protein (PER), oscillate in a circadian pattern [7][8][9][10]. In contrast to the retina photoreceptors, an expression of clock genes has not been observed in the lamina neurons that show circadian structural plasticity. PER staining, however, has been detected in the lamina glial cells [10]. In the retinal photoreceptors in D. melanogaster PER staining is the highest at the end of the night and begins to decrease at the beginning of the day [9]. While intensity of PER staining during the night is the same in both the LNs and the photoreceptors, during the day it decays faster in photoreceptors and glia than in the LNs [9]. The molecular clock in most clock cells is based on circadian transcription and translation feedback loops with two main core genes, per and timeless (tim). It is entrained by light and other environmental cues to drive daily rhythms in output processes [11]. The circadian clock in the LNs is entrained by light inputs mainly from the extraretinal photoreceptors of the Hofbauer-Buchner eyelet. It is entrained by light inputs also from the compound eyes. In addition it possesses its own photoreceptor cryptochrome (CRY), the blue-light absorbing protein, encoded by the cryptochrome (cry) gene [12]. CRY is responsible for light-dependent degradation of tim protein TIM and resetting the circadian rhythms. In turn TIM is required for stabilization of PER. CRY might also be an element of the molecular circadian clock, as it has been suggested in peripheral oscillators [13,14]. In the current model of the molecular clock in the pacemaker neurons in the brain of Drosophila, the transcription factors CLOCK (CLK) and CYCLE (CYC) activate expression of per and tim, and PER and TIM accumulate and form PER/TIM heterodimers in the cytoplasm in darkness. About 6 h after lightsoff they enter the nucleus, dissociate and PER represses transcription by preventing the CLK/CYC heterodimers from binding to per and tim promotores [15]. The timing of PER, TIM and other clock protein functions, their stability and subcellular localization depend on phosphorylation [16,17]. In the nucleus PER and TIM do not accumulate simultaneously in time suggesting that those proteins do not act only as heterodimers and have distinct functions in the nucleus [18].
The rhythms detected in the lamina of the fly's optic lobe seem to be controlled by a circadian clock located in the brain [19]. An involvement of circadian oscillators in the photoreceptors and in the optic lobe glia is possible, however. The oscillation in size of two lamina large monopolar cells L1 and L2 have been detected in three flies species; Musca domestica, Drosophila melanogaster and Calliphora vicina [3,4,20]. In the housefly the axons of L1 and L2 cells change their girth during the day and night. This rhythm is maintained in constant darkness (DD) and in continuous light (LL). They are endogenously generated by a circadian clock. The daily pattern of these changes is species-specific and correlated with the locomotor activity pattern. Both cells are larger when locomotor activity of a fly increases [21]. In D. melanogaster L1 and L2 cells swell at the beginning of both the day and night. In the case of L2, in addition to axons, circadian oscillations have also been detected in size of nuclei but not in cell bodies [22]. As it has been found in M. domestica, structural changes in L1 and L2 monopolar cells are offset by changes in the lamina epithelial glial cells [23].
The function of rhythmic remodeling of L1 and L2 axons is still unknown. The aim of the present study was to examine the L2 dendrite morphology which may affect the size of the L2 axon. The dendrites extend from the L2 axons and surround the photoreceptor terminals in the lamina. They receive photic and visual information as postsynaptic sites through tetrad synapses from the photoreceptors. We hypothesized that changes in the size of L1 and L2 axons are correlated with changes in the size of dendrites. The length of the dendrites, in turn, is correlated with the number of tetrad synapses. In the housefly it has been found that tetrad presynaptic structures, called T-bars, show daily changes in frequency. The highest number of tetrad presynaptic profiles has been found when axons of both cells L1 and L2 were the largest [2].
Here we examined the length of the dendrites of the L2 monopolar cell. We did this by measuring the cross-sectional outline (size) of the L2 dendritic tree under different light conditions and at different times during the 24 h period. It was done on transgenic lines with targeted expression of green fluorescence protein (GFP) in the L2 cells in wild-type background D. melanogaster (Fig. 1,2) and in the arrhythmic per 01 null mutant and cry b mutant with an amino-acid substitution in CRY (Fig. 3). We found that the size of L2 dendritic tree changes during the day and night and that this structural plasticity is controlled by a circadian clock. Using cry b mutants we also detected that CRY is not only the circadian photoreceptor. It also controls the phase of the circadian rhythm in changes of the L2 dendrite morphology.

Results
Morphological plasticity of the L2 dendrites in wild-type Drosophila LD12:12. The daily statistically significant differences in the cross-sectional size of the L2 dendritic tree were found in both males and females (p,0.05) ( Fig. 2A). There were no statistically significant differences in the L2 dendritic tree size between males and females, however. The morphology of the L2 dendritic tree was also similar in males and females, usually we observed 8 dendritic spines per cell (Fig. 3). In females the size of the dendritic tree was the largest at ZT1 (25.95 mm63.43 SD) and the smallest in the middle of the night (ZT16: 19.05 mm64.05 SD) (Fig. 4). This means it was smaller by 27%. This difference was statistically significant (F = 41.33; p,0.001). At another time point studied (ZT4: 20.53 mm64.35 SD and ZT13: 19.59 mm63.92 SD) the L2 dendritic tree size was not statistically different from that at ZT1.
In addition to changes in the size of the dendritic tree, there were also changes in morphology of the L2 dendritic spines which were long and thin during the day but short and thick during the night ( Fig. 2A). These changes were observed in both sexes.
DD condition. In DD the pattern of size changes of the L2 dendritic tree was similar to that one detected in LD 12:12 (Figs 2B,5). This indicated a circadian rhythm in changes of morphology of the L2 dendrites. No significant differences were found between males and females.
LL condition. There was not significant changes in the size of the L2 dendritic tree during the subjective day and night in LL in males and females as well as between these groups (Fig. 6).
Morphology of the L2 dendrites in Drosophila arrhythmic mutant per 01 LD 12:12. There were not statistically significant differences in the cross-sectional sizes of the L2 dendritic tree during the day and night in males or in females (Fig. 7).     In addition to disruption of the circadian rhythm in the L2 dendrite plasticity in per 01 mutants, there were also observed changes in morphology of dendritic spines that were shorter in per 01 mutants than in wild-type flies at ZT1 (Fig. 3).
DD condition. No significant differences were found in the cross-sectional sizes of the L2 dendritic tree in per 01 females or males held in DD (Fig. 8).
LL condition. There were no statistically significant differences in the cross-sectional sizes of the L2 dendritic tree in per 01 males or females held in LL (Fig. 9).
Morphology of the L2 dendrites in Drosophila clock photoreception mutant cry b LD 12:12. In cry b flies the pattern of the L2 dendritic tree size changes was different in males than in females (Fig. 10) and dendritic spines were less regular in shape than in wild-types flies (Fig. 3). Some dendritic spines of one cell we longer than others in cry b while in wild-type flies all spines had mostly similar size.

Circadian rhythm in morphology of dendrites
The results obtained in the present study showed that the L2 cell dendrites, postsynaptic sites of tetrad synapses in the first optic neuropil of D. melanogaster, show circadian plasticity in morphology. Under the day/night regime LD 12:12, the L2 dendrites, measured as the cross-sectional outline of the dendritic tree are the longest at the beginning of day and shorter later during the day and at night. This daily rhythm in changes of size of the dendritic tree was maintained in DD. This indicates that the circadian plasticity of the L2 dendrites is controlled endogenously by a circadian clock. This rhythm was detected in both males and females. It had the similar pattern in both sexes under both conditions. The rhythm was not observed, however, in LL. Similarly to circadian rhythms in behavior, continuous light abolished the circadian plasticity of dendrites. It confirms other studies that the molecular mechanism of the clock in D. melanogaster  is disrrupted in LL and PER protein is absent in the retina photoreceptors and in the LNs [9]. In contrast to the present observation, the results obtained in our earlier studies showed that daily oscillations in L1 and L2 axon sizes of the housefly are maintained in both DD and LL. Moreover, L1 and L2 mean sizes are larger in LL than in LD 12:12 and are smallest in DD [3]. The differences in the size of L2 dendritic tree in LL and under different light conditions were not observed in D. melanogaster.
In LD 12:12, L1 and L2 cells of M. domestica increase their axons during the day and show a decrease during the night. In turn, our study of L1 and L2 interneuron plasticity of D. melanogaster at electron microscope level has shown that both monopolar cells change their cross-sectional area of axons twice during the 24 h period. This change takes place at the beginning of the day and at the beginning of the night [4]. The observed differences between M. domestica and D. melanogaster indicate species specificity of visual and circadian systems. Circadian plasticity of L1 and L2 interneurons, however, have been found in both species. Moreover, in both species we have also found a correlation between the size of L1 and L2 interneurons and the level of locomotor activity [3,4,21]. Both cells swell when flies are highly active. The pattern of circadian L1 and L2 plasticity is synchronized with their species-specific pattern of the circadian rhythm in locomotor activity. In the present study we did not observe two peaks in the size of the L2 dendritic tree during the 24 h period but only one at the beginning of the day. It means that the largest L2 dendritic tree forms when the girth of the L2 axon increases at the beginning of the day. The second peak in changes of L1 and L2 axon size may be connected to other processes going on in monopolar cells at the beginning of the night or indicate that changes of axon sizes are not the same along the axons. In D. melanogaster the L2 axons change shape from an inverted conical form during the day to a cylindrical one at night [4]. The previous measurements of L1 and L2 axons were carried out at the proximal depth of the lamina while the present ones at more distal depth. In addition to axons, the L2 nuclei and cell bodies have also been studied in Drosophila transgenic line with GFP targeted to the L2 cytoplasm. This study showed that the nuclei are also larger  during the day than during the night. In contrast, L2 cell bodies do not change their sizes during the day and night [22].
The role of per and cry genes in control of circadian plasticity and morphology of the L2 dendrites Using the transgenic flies in per 01 background we found that the observed circadian changes in size of the L2 cell dendritic tree were abolished in the null per 01 mutant flies. It confirms that the circadian plasticity of the L2 dendrites is controlled by the circadian clock for which per is crucial in its molecular mechanism. Unlike to the circadian rhythm in locomotor activity, the rhythm in the L2 dendrite plasticity in the per 01 mutants was not entrained by external daily changes of light and darkness. It means that the circadian input is necessary for maintaining circadian plasticity of dendrites also in LD conditions. The rhythm in circadian plasticity of L1 and L2 axons have also not been observed in the per 01 mutant in LD 12:12 (Pyza E. and Meinertzhagen I.A., unpublished results).
In the present study we also examined involvement of cry gene in the circadian plasticity of dendrites. Cryptochrome has been suggested as an element of molecular mechanism of the circadian clock in peripheral clocks [13,14,25,26]. This is in addition to its role as a circadian photoreceptor in the central pacemaker neurons [27,28]. Since CRY has been detected in photoreceptors of the compound eyes [25,26], CRY may play both functions in these peripheral oscillators. In the lamina, however, circadian rhythms may be regulated by the central pacemaker, at least in the housefly. This is because the severance of the optic lobe in this species abolishes the rhythm in swelling and shrinking of L1 and L2 axons [19]. In D. melanogaster, however, we have found that per expression in the LNs is not sufficient to generate the circadian plasticity of L1 and L2 monopolar cells (Pyza E, Hardin PE, Meinertzhagen IA, unpublished results). This rhythm was also not completely rescued after per expression in both the LNs and the retina photoreceptors. It indicates that per expression in additional cells, probably glial ones is also important for this rhythm. In case of dendrites their rhythmic morphological changes are correlated with circadian rhythms in the photoreceptor terminals, in the abundance of a presynaptic active zone protein Bruchpilot (Górska-Andrzejak J, Niańko E, Pyza E, unpublished results).
In the housefly, the pacemaker neurons transmitting circadian information to the lamina are probably the large PDF-immunoreactive neurons in the accessory medulla, similar to Drosophila's large ventral LNs. Their terminals in the lamina and medulla show circadian structural changes [4] and cyclically release PDF [29] that affects target neurons and/or glial cells. In D. melanogaster circadian information from the central pacemaker neurons to the visual system are probably transmitted to per-expressing glial cell in the medulla. They may intermediate between the central clock and target cells in the visual system [30]. In addition circadian rhythms in the lamina may be controlled by circadian oscillators in glia and in the retina photoreceptors.
The results obtained in the present study indicate that CRY is important for setting the phase of the circadian rhythm in changes of the L2 dendrite morphology. In both LD and DD conditions in the transgenic flies in cry b background, the peak in size of the dendritic tree was shifted and the daily pattern of dendrite size changes was different than in wild-type flies.
In addition to the involvement of per and cry genes in regulating circadian plasticity of the L2 interneuron structure, mutations of both genes also affect morphology of the L2 dendrites. It means that both genes may play several, at least two, roles in the nervous system. They maintain structure and cyclical structural changes of neurons in the fly's visual system. This role they may also play in other parts of the nervous system [31].

Function of circadian plasticity of dendrites
The circadian rhythms in the lamina are probably generated by several circadian oscillators. They receive circadian inputs originating from the central pacemaker neurons and clock gene expressing cells in the visual system. If they receive the circadian input from clock neurons in the central pacemaker, these neurons are different from those generating circadian rhythms in behavior. This is because they are not driven in per 01 mutants by light. It is also because cry is not only the circadian photoreceptor but also involved in the clockwork. Since the number of the tetrad presynaptic T-bars changes during the day and night, dendritic spines of the L2 interneurons, postsynaptic sites of these synaptic contacts, have to enlarge in synchrony. They have to enlarge in synchrony to the presynaptic active zone changes in the presynaptic cells. Later during the day and in the night they shorten. A space inside the cartridge is filled with glial cells which during the day decrease their sizes [23]. Because the optic lobes intermediate between the compound eyes and the central brain, rhythmic processes in the optic lobes are probably affected by circadian oscillators located in the brain and in the compound eye photoreceptors. In addition, they also synchronize rhythmic events occurring in both sites. Rhythmic remodeling of postsynaptic neurons controlled by circadian oscillators in the brain and in the visual system itself, may provide efficient transmission of photic and visual information. This is information which takes place during high behavioral activity of an animal.

Materials and Methods
We used Drosophila melanogaster strains with the bipartite GAL4/ UAS expression system [24] and generated flies that express GFP in the lamina L2 monopolar cells (Fig. 1). GAL4 enhancer trap line 21D-GAL4 which shows expression of GAL4 in L2 cells was kindly provided by Dr. Thomas Raabe (Würzburg Univ., Germany). The 21D-GAL4 strains with null mutation of gene period (per 01 ;21D/TM2) or altered cryptochrome (cry b ;21D/TM6B) backgrounds were generated in Dr. Michael Rosbash' laboratory (HHMI, Brandeis Univ., USA) (Fig. 2). The UAS construct carrying line was UAS-mCD8-GFP, encoding a GFP-variant that is targeted to the cell membrane. The line was obtained from the Bloomington Stock Center.
All experiments were performed on adult flies, males and females, of Drosophila melanogaster transgenic lines mentioned above. Flies were raised on a standard cornmeal-yeast-sugar medium at 25uC in broad-spectrum light of 2,000 lx intensity during the light phase of the light/dark cycle (LD 12:12 -12 h of light and 12 h of darkness) and in continuous light (LL). During the dark phase of LD 12:12 and in constant darkness (DD) experiments, flies were kept in a complete darkness. For experiments under DD or LL conditions, flies were held for 3 days in LD 12:12 first for entrainment and next they were transferred to DD or LL.
Six days-old flies were decapitated under CO 2 in a drop of fixative at four different time points in LD 12:12: ZT1, ZT4, ZT13, ZT16 (ZT0 -the beginning of the day and ZT12beginning of the night) and in DD or LL: CT1, CT4, CT13, CT16 (CT0 -beginning of the subjective day and CT12beginning of the subjective night). For experiments conducted