Mapping and quantification of cryptochrome expression in the brain of the pea aphid Acyrthosiphon pisum

Aphids are paradigmatic photoperiodic animals often used to study the role of the circadian clock in the seasonal response. Previously, we described some elements of the circadian clock core (genes period and timeless) and output (melatonin, AANATs and PTTH) that could have a role in the regulation of the aphid seasonal response. More recently, we identified two opsins (C‐ops and SWO4) as candidate input photoperiodic receptors. In the present report, we focus on the study of cryptochromes (cry) as photoreceptors of the circadian clock and discuss their involvement in the seasonal response. We analyse the expression of cry1 and cry2 genes in a circadian and seasonal context, and map their expression sites in the brain. We observe a robust rhythmic expression of cry2 peaking at dusk in phase with core clock genes period and timeless, while cry1 shows a weaker rhythm. Changes in cry1 and cry2 expression correlate with activation of the seasonal response, suggesting a possible link. Finally, we map the expression of cry1 and cry2 genes to clock neurons in the pars lateralis, a region essential for the photoperiodic response. Our results support a role for cry as elements of the aphid circadian clock and suggest a role in photoreception for cry1 and in clock repression for cry2.


INTRODUCTION
Life has adapted to daily and seasonal rhythmic changes in the environment caused by Earth's movements. Organisms adopt different strategies to cope with cycles of resource availability, temperature variations, light exposure, etc. Biological rhythms adapt organisms to those cyclic variations. The best known, endogenously generated rhythms that adapt organisms to the day-night cycles are known as circadian rhythms, and those evolved to deal with yearly-based changes (specially seasons in temperate regions of the planet) are referred to as annual (mainly seasonal) rhythms. In order to keep track of time, and anticipate to daily changes, living beings have developed an internal molecular machinery called 'circadian clock'. The clock consists of a series of interlocked feedback loops whose relation helps organizing the physiological needs at every time of the day (Doherty and Kay, 2010). The clock is temperature compensated and synchronized with the environment mainly by light, but also by other cues in some organisms. Similarly, annual activities need to be coordinated with external conditions to produce adequate seasonal responses. The main input that controls seasonal cycles is the length of the day or photoperiod (Nijhout, 2010). In many organisms, changes in day length that anticipate the arrival of a particular season induce physiological and behavioural responses required to overcome the corresponding adverse conditions. This capability, shown by many organisms to respond to photoperiodic changes is called 'photoperiodism'. While the genes involved in the circadian clock have been extensively studied (especially in Drosophila melanogaster) (Özkaya & Rosato, 2012;Tataroglu & Emery, 2015), it is less clear how the photoperiodic or annual calendar works at the molecular level (Dolezel, 2015; Miquel Barberà and Jorge Mariano Collantes-Allegre contributed equally to this study. Koštál, 2011;Nijhout, 2010). However, it has been proposed that photoperiodic clocks should share many elements with the circadian clock (Abrieux et al., 2020;Iiams et al., 2019;Meuti & Denlinger, 2013;Nagy et al., 2019;Pegoraro et al., 2014;Saunders, 2010).
Whether independent or integrated, both circadian and calendar (photoperiodic) systems need an input system based on light detection to retrieve information on the moment of the day or of the year, a central rhythmic oscillator and an output pathway to communicate time information to the rest of the body. In addition, a molecular calendar would need a counter to keep track of the number of accumulated photoperiodic cycles of particular length (Saunders, 1981).
Aphids are paradigmatic photoperiodic insects (Markovich, 1924) whose complex life cycles, mostly characterized by their cyclical parthenogenesis, are greatly controlled by photoperiod. In particular, the pea aphid, Acyrthosiphon pisum that exhibits a rather simplified life cycle when compared to other aphid species, consists of several generations of viviparous parthenogenetic females that succeed one another along the favourable seasons (spring and summer). However, when pea aphids sense the shortening of day length at the end of the summer, which announces the arrival of the winter, they give birth to a single sexual generation made of sexual (oviparous) females and males that will mate and produce overwintering (diapausic) eggs. This behaviour can be easily reproduced in the laboratory. Indeed, A. pisum can be reared indefinitely reproducing through viviparous parthenogenesis under long day (LD) conditions (i.e., >16 h of light), but if switched to short day (SD) conditions (e.g., 12 h of light), it responds to the change in photoperiod by producing the sexual morphs in the next generation. Aphids exhibiting both sexual and asexual phases of the life-cycle are called 'holocyclic', in contrast with 'anholocyclic' strains, which are natural mutants who fail to produce a photoperiodic response (i.e., they do not switch to sexual reproduction under SD conditions) thus continuously reproducing through parthenogenesis.
To retrieve information about the duration of light availability, aphids (as most organisms) need a photosensitive system. The aphid counts with several putative photosensitive proteins, such as opsins and cryptochromes (Collantes-Alegre et al., 2018;Cortés et al., 2010;Michael et al., 2017;Terakita, 2005). Although classically considered the main light sensors responsible for colour sensitivity and image formation, opsins have been shown to play diverse roles in insects also including entrainment of the circadian clock and other nonvisual functions (Leung & Montell, 2017;Ni et al., 2017;Senthilan et al., 2019). Cryptochromes, on the other hand, are known to play crucial roles in insect's circadian clocks (Emery et al., 1998;Michael et al., 2017;Yuan et al., 2007).
Cryptochromes are blue light-sensitive proteins noncovalently bound to a flavin adenine dinucleotide molecule that is related to the photolyase family (a group of proteins in charge of repairing light-induced DNA damage). Despite their similarities with photolyases, cryptochromes have lost the ability to repair DNA but have evolved to perform other functions such as photoreception, control of transcription, or magnetoreception (reviewed in Michael et al., 2017). Two types of cryptochromes can be found in insects: the Drosophila-like cryptochrome (dCRY) and the mammalian-type cryptochrome (mCRY). While D. melanogaster lacks mCRY, bees and beetles lack the dCRY. However, butterflies and aphids have both types of cryptochrome, which is considered the ancestral condition in insects (Yuan et al., 2007). dCRY works as a photoreceptor (Emery et al., 2000) and, when present, is in charge of synchronizing the circadian clock with the light-dark cycle (H. Zhu et al., 2008). After hit by the morning light, dCRY binds to TIMELESS (TIM, a central circadian clock protein) and targets it for the ubiquitinase protein JETLAG (JET) marking it for degradation (Koh et al., 2006). This process synchronizes the clock with the outer light conditions in a series of feedback loops together with the product of gene period (per) (Hardin, 2005). While dCRY protein and mRNA levels fluctuate daily in D. melanogaster (Emery et al., 1998), transcript levels remain constant in some butterflies, rotifers, annelids and mosquitoes (Gentile et al., 2009;Meuti et al., 2015;Rund et al., 2011Rund et al., , 2016Zantke et al., 2013;H. Zhu et al., 2008). The mCRY in those insects where it is present seems not to be photosensitive (Merlin et al., 2013;Zhang et al., 2017;H. Zhu et al., 2005) but instead acts as a transcription factor along with other clock elements. Indeed, it represses (probably coordinated with PER) transcription mediated by the products of genes Clock and Cycle (CLK and CYC, respectively). Aphids, like butterflies, have both dCRY and mCRY types of cryptochrome (referred to as CRY1 and CRY2 respectively). Gene duplications are frequent in aphids, and this is also true for CRY2, which is present in two copies, cry2-1 and cry2-2 in the pea aphid genome . In contrast to Drosophila and other insects, aphids lack a JET homologue , which puts into question a putative role of CRY1 as synchronizer of the circadian clock in aphids similar to its role in Drosophila. Similar to other transcription factors at the core of the aphid clock (such as PER and TIM), both copies of CRY2 in A. pisum are highly divergent and are evolving faster than corresponding homologue proteins in other insects pointing to a role as a transcription factor at the core of the circadian clock .
Apart from a putative role in the aphid circadian clock, a role for either CRY1 or CRY2 (or both) as photoreceptors in the photoperiodic system remains to be elucidated. In this respect, old reports already established that photoreceptors involved in aphid photoperiodism reside in the aphid brain and not in the eyes (Lees, 1964) showing the expression in the aphid brain of particular opsins that could thus have a role in photoperiodism (Collantes-Alegre et al., 2018;Gao et al., 1999). Cryptochromes are also good candidates to act as photoperiodic photoreceptors and thus deserve studying their expression in aphids. In a previous report , we characterized the dCRY and mCRY aphid homologues (cry1 and cry2) present in the pea aphid genome. In the present report, we have quantified their expression in some detail measuring transcript levels along the day-night cycle and under different photoperiod regimes. In addition, through in situ hybridization, we have located cry1 and cry2 transcripts in regions of the aphid brain related to the circadian clock and photoperiodism.

RESULTS
We designed an experiment to test for rhythmic expression of both cryptochrome genes, cry1 and cry2 and to compare their expression in two aphid strains reared under two photoperiod regimes (i.e., LD and SD photoperiods). Since both copies of aphid cry2 (cry2-1 and cry2-2) are 98% identical at the nucleotide level, our primers did not discriminate between them and thus, we quantified their expression jointly.
This first experiment was complemented with a second one intended to confirm the characteristic self-sustainability of the rhythmic expression of circadian clock genes under DD (constant darkness).

Expression of cryptochrome genes in holo-and anholocyclic aphids
We studied the expression of cryptochrome genes in heads of holocyclic YR2 and anholocyclic GR aphids ( Figure 1). For each strain, we quantified the expression at six different time points and compared aphids reared in LD and SD photoperiods (see "Materials and Methods" section). The expression patterns of cry1 ( Figure 1a) were similar in groups of aphids that reproduce by parthenogenesis (i.e., holocyclic YR2 aphids kept under LD and anholocyclic GR aphids in both LD and SD light regimes). However, a COSINOR analysis rejected a rhythmic expression for cry1 in those groups (Table S2).
Differently, aphids in which the seasonal response had been activated (holocyclic strain YR2 reared in SD) had a particular pattern of cry1 expression with a significant rhythmic expression peaking at the end of the night (Figure 1a and Table S2).
Analysis of the expression of cry2 revealed two main patterns ( Figure 1b). On one hand, we found a significant rhythmic expression in aphids with active parthenogenesis (i.e., holocyclic YR2 aphids under LD and anholocyclic GR aphids under LD and SD) with a peak around ZT15 (Figure 1b). On the other hand, aphids with an active seasonal response (i.e., YR2 aphids reared under SD) showed a similar pattern (though with increased expression) but COSINOR analysis did not support significant rhythmic expression probably because the high variation observed at some ZTs, especially at ZT21 (Table S2).
F I G U R E 1 Relative expression of cry1 (a) and cry2 (b) genes in holocyclic YR2 (left) and anholocyclic GR (right) aphid strains reared under long day (LD) (16L:8D) and short day (SD) (12L:12D) photoperiods. Mean relative expression of three biological replicates AESEM is plotted at 6 h intervals over 1.5 days. Individual values for each biological replicate are included as crosses. Scotophases of LD and SD photoperiods are indicated by dark and light grey backgrounds, respectively. Fitted cosine curves were included as red dotted lines in cases with a significant rhythmic expression after COSINOR tests (Table S2) It is noteworthy that expression of both cry1 and cry2 genes was always higher in aphids reared under SD than in aphids reared under LD photoperiods and this was observed in both strains at any time point ( Figure 1).

Self-sustainability of cry2 expression
The observation of a significant rhythmic expression of cry2 in aphids reared under LD photoperiod with a pattern that coincided with the pattern previously reported for core clock genes period and timeless , prompted us to explore if this gene exhibits the characteristic self-sustainability in the expression of genes at the core of the circadian clock (Özkaya & Rosato, 2020). To do that, we followed the expression of this gene in holocyclic aphids YR2 for 3 days both at LD and under continuous darkness (DD). As seen in Figure 2, aphids reared under LD exhibited a significant rhythmic expression of cry2, peaking at late day around ZT15 (Table S2). Notably, the observed pattern was identical to that reported previously for period and timeless (indicated in Figure 2 for comparison) . However, in DD cry2 expression rapidly dampens after the first day and loses its synchronicity with per and tim expression Although the signal was weaker than that in cry1 experiments, the localization of cry2 transcripts allowed us to identify at least two cell clusters similar to those previously described for cry1, thus the same nomenclature was used ( Figure 4). Again, specificity of the signal F I G U R E 2 Relative expression of cry2 gene in holocyclic YR2 aphid strain reared under long day (LD) (16L:8D) and constant darkness (DD) (0L:24D) photoperiods. Mean relative expression of three biological replicates AESEM is plotted at 6 h intervals over 3 days. Individual values for each biological replicate are included as crosses. Scotophases of LD and subjective scotophase of DD photoperiods are indicated by dark and light grey backgrounds, respectively. Expression of per and tim genes was included for comparison purposes (data from . Fitted cosine curves were included as red dotted lines in cases with a significant rhythmic expression after COSINOR tests (Table S2)

DISCUSSION
During the last decades, there has been a debate on the involvement of the circadian clock in the control of photoperiodic responses in insects (Bradshaw & Holzapfel, 2010;Koštál, 2011). However, as indicated by recent studies in different insect species (Ikeno et al., 2013;Ikeno, Katagiri, et al., 2011;Ikeno, Numata, & Goto, 2011;Meuti et al., 2015;Omura et al., 2016;Pegoraro et al., 2014), there is growing evidence that the circadian clock is involved in the regulation of the seasonal response. In this regard, aphids are paradigmatic photoperiodic organisms that have been used as a model to study the regulation of photoperiodism, being a suitable model to shed light on the debate. Both the circadian and the (putative) photoperiodic clock need input elements that provide the information on light availability either to synchronize the circadian clock or to provide day-length information to the photoperiodic clock. The identity of these input elements has not yet been determined in aphids. In a recent report, we analysed the expression of the opsin repertoire in the aphid A. pisum and determined that two particular opsins could be good candidates to have a role as photoperiodic photoreceptors (Collantes-Alegre et al., 2018). In the present report, we focus on cryptochromes, which are members of a different family of photoreceptors, and study the expression of the two types of cryptochrome present in the aphid genome (i.e., a Drosophila-type and a mammal-type, or cry1 and cry2, respectively) to gain some insights in their possible involvement either as input elements of the circadian clock or in the aphid seasonal response.

Cry1 expression
The expression of the Drosophila-type cryptochrome encoding gene (cry1) was not significantly rhythmic in aphids committed to the parthenogenetic mode of reproduction (i.e., YR2 holocyclic aphids reared under LD conditions or GR anholocyclic aphids irrespective of the photoperiod). Our results partially agree with previous quantifications made in aphids in a rather limited survey (Cortés, 2010;Cortés et al., 2010), in which a different holocyclic strain did not show rhythmic expression of cry1. Interestingly, we did find rhythmic expression, with a peak at ZT21, in holocyclic aphids of the G1 generation reared under SD conditions. Our findings in the pea aphid differ from Drosophila, in which cry1 transcripts oscillate in a daily manner with a maximum at the beginning of the day (Egan et al., 1999;Emery et al., 1998). This difference can derive from the specific regulation of the fly circadian clock, which only encodes one type of photosensitive cryptochrome (Emery et al., 1998) that synchronizes the central oscillator with the day-night cycle (Özkaya and Rosato, 2012). However, most other insect species encode for two F I G U R E 5 Integrative graphical summary of the putative elements of the aphid photoperiodic system, including potential photoperiodic photoreceptors. Our current hypothesis proposes CRY1 expressed in the dorsal neurons (DN) as the site of light photoperiodic input. C-ops and SWO4 opsins may also participate in the photoreception. The input would convey the light information to the photoperiodic clock core located in the DN that would, in turn, regulate the production or release of insulins by the insulin producing cells (IPCs). Genes expressed in each group of clock neurons are indicated in parentheses. See discussion for further details. ol, optic lobe; Pc, protocerebrum types of cryptochromes: a cry1 (or Drosophila-type cry) that plays a photosensitive role probably acting in the synchronization of the circadian clock with the day-night cycles, and a cry2 (or mammalian-type cry) that would act as a repressor in the negative feedback loop at the core of the circadian clock along with the clock gene period (Merlin et al., 2013;Yuan et al., 2007;Zhang et al., 2017). In insects in which both cry1 and cry2 are present, as in aphids, it has been reported an absence of rhythmicity in the expression of cry1 (Gentile et al., 2009;Lugena et al., 2019;Rund et al., 2011;Tokuoka et al., 2017;Ueda et al., 2018;Xu et al., 2016).
Recent work in A. pisum by Colizzi et al. (2021) shows rhythmic CRY protein levels with a peak at ZT16 under 16L:8D photoperiod, in contrast to the lack of significant rhythmicity at the transcript level we describe in Figure 1a. However, although not significant a weak oscillating pattern, with a shallow peak during the day, is present in all aphid groups with active parthenogenesis (Figure 1a). This pattern would be compatible with the delayed peak at the protein level observed at dusk (Colizzi et al., 2021). In addition, it cannot be ruled out that CRY2 could also be contributing to the rhythmicity observed for CRY by Colizzi et al. (2021), as the authors discuss.

Cry2 expression
In contrast with results for cry1, we found a strong daily rhythmic expression of the mammalian-type cryptochrome (cry2) in aphids committed to the parthenogenetic mode of development (i.e., YR2 holocyclic aphids reared under LD conditions or GR anholocyclic aphids irrespective of the photoperiod). However, no significant rhythmicity was detected in aphids with active seasonal response (i.e., holocyclic aphids reared under SD conditions). As before, this result partially agrees with previous, more limited, quantifications of cry2  that showed significant daily rhythms in holocyclic aphids both reared in LD and SD, with higher amplitude in SD. However, the lack of significance in our COSINOR test for aphids of the YR2 strain reared under SD conditions probably results from a nonstable MESOR that decreases over time (Figure 1b). Interestingly, the peak of cry2 expression at dusk observed in parthenogenetic aphids of the YR2 and GR strains coincides in phase with the peak of expression of period and timeless for the same aphid morphs Cortés et al., 2010). A peak at the beginning of the night in the expression of cry2, in many cases in phase with peaks in the expression of per and tim genes, has also been reported in other insects (Bertossa et al., 2014;Gentile et al., 2009;Ingram et al., 2012;Jiang et al., 2018;Rubin et al., 2006;Ueda et al., 2018;Werckenthin et al., 2012;Xu et al., 2016) but rhythmic expression has not been found in some insects (Ikeno et al., 2008;Kotwica-Rolinska et al., 2017;L. Zhu et al., 2017). Furthermore, we found that the expression of cry2 rapidly dampens under DD conditions, similar to our results previously reported for genes period and timeless, which are at the core of the circadian clock . This quick dampening of oscillator-related genes might be related to the fact that daily rhythms in parthenogenetic aphids are weak and shallow (Beer et al., 2017;Joschinski et al., 2016). Thus, our results showing similar patterns of expression and behaviour under continuous darkness of cry2 and core clock genes, support a putative role, yet to be elucidated, of cry2 in the negative feedback loop of the circadian clock core in conjunction with period and/or timeless.

Differences between strains and photoperiods
Previous studies on the pea aphid show that the activation of the photoperiodic response in holocyclic aphids triggers changes in the expression of some circadian clock genes Cortés et al., 2010).
However, these changes could be originated by the switch of the photoperiod itself, the activation of the seasonal response, or a combination of both. That prompted us to analyse the expression of circadian clock genes including in the analysis anholocyclic aphids, in which the switch from LD to SD photoperiod does not activate the seasonal response. In this way, we could separate the effect inherent to the photoperiod from the effect of the seasonal response. In our previous work, we found that the expression of period and timeless genes correlated with the activation of the seasonal response and not to the photoperiodic conditions . In the present report, the quantification of expression of cryptochrome genes clearly shows two expression patterns for cry1 and cry2: a basal level of expression in aphids committed to the parthenogenetic mode of reproduction (holocyclic aphids under LD and anholocyclic aphids at both LD and SD photoperiods) that was altered (increased expression) in aphids that had activated the seasonal response (holocyclic aphids of the G1 generation reared under SD). Higher expression was also previously reported for core clock genes period and timeless in holocyclic aphids maintained under SD conditions . In this respect, changes in the expression of clock genes have been associated with activation of the photoperiodic response, mainly diapause, in other insect species (Abrieux et al., 2020;Bajgar, Dolezel, & Hodkova, 2013;Bajgar, Jindra, & Dolezel, 2013;Iiams et al., 2019;Meuti et al., 2015;Nagy et al., 2019;Werckenthin et al., 2012), although in some other cases there was no differential expression between photoperiodic conditions (Ikeno et al., 2008).

Localization of cry1 and cry2
If either CRY1 and/or CRY2 do have a role in the circadian clock or/and participate in the photoperiodic response in aphids, localizing their site(s) of expression in the aphid brain is highly relevant. Thus, making use of in situ hybridization, we mapped the neurons expressing cry1 and cry2 genes in the aphid brain. We believe that we have identified most of the cells expressing cry1 as the intensity of the signal revealed by the cry1 probe was rather strong. However, in the case of cry2, it is likely that other cry2-expressing neurons remain to be discovered, since the intensity revealed by the cry2 probe was rather faint, probably reflecting low levels of expression of this gene.
Although no co-localization experiment of both cryptochromes has been performed, our results suggest that both cryptochromes are expressed in the same cells. Most relevant, the position and number of cryptochrome expressing neurons in the pars lateralis and the optic lamina and lobula are very similar to that of the clock neurons expressing the circadian clock genes period and timeless that we previously described . Indeed, cry1 and cry2 positive signal was detected in cells in the pars lateralis, probably in the large dorsal neurons (l-DN) but also, at least cry1, in other DN (Figures 3   and 4). Both cry1 and cry2 signal was also detected in the so-called LN. Finally, cry1 was detected in the LaN and, although not detected, we cannot discard that these LaN also express cry2, but at low levels thus not yielding a visible signal. We thus propose a co-expression of cryptochromes together with period and timeless at least in the l-DN and the LN that should be confirmed in future experiments. In support of this, CRY-immunoreactivity (ir) has been recently found in the aphid brain co-localizing with PER-ir in DN, LN and LaN (Colizzi et al., 2021). In addition, cryptochromes have been co-localized together with per and tim in the clock neurons of other insects (Bazalova et al., 2016;Benito et al., 2008).
Taking into account the pattern of expression of cry2 (similar to that of clock genes period and timeless, see above) as well as its role in other insects (Ikeno, Katagiri, et al., 2011;Ikeno, Numata, & Goto, 2011;Merlin et al., 2013;Zhang et al., 2017;H. Zhu et al., 2008), we can assume that the aphid CRY2 is not photosensitive, but it rather plays a role as transcriptional factor similar to PER and TIM. That leaves CRY1 as a good candidate for the photoreception in the seasonal response, and probably also for synchronization of the circadian clock. Classic studies in aphids using localized illumination narrowed the photoperiodic sensitive region down to the dorsal medial region of the head (Lees, 1964). So far, expression of C-opsin and SWO4 opsins have been found in the underlying brain in the vicinity of this photoperiodic sensitive head region (Collantes-Alegre et al., 2018). Furthermore, classical ablation experiments (C. G. H. G. Steel, 1977;C. G. Steel & Lees, 1977) that removed the region in the pars lateris where clock cells  and cryptochrome expressing cells (this report) have been identified, dramatically affected the photoperiod response, which led the authors to propose the localization of both the clock and the photoperiodic photoreceptors in that region (Lees, 1964;C. G. Steel & Lees, 1977). In Figure 5, we propose a scenario with the putative actors that would have a role in the aphid photoperiodic response integrating our current and past results. In this scenario, we describe a photoperiodic system in which CRY1 plays a light input role in the DN, that is, complemented by C-opsin and SWO4  (Barberà et al., 2019). In agreement with this hypothesis, a neuronal link has been found between DN and IPC in aphids (Colizzi et al., 2021;Cuti et al., 2021). Moreover, a circadian controlled photoperiodic response has been described in other insects (Abrieux et al., 2020;Denlinger & Armbruster, 2016;Iiams et al., 2019).
Although the aphid photoperiodic response was classically thought to be independent of a circadian clock, our results support the subsequent interpretations that modelled it as a damping oscillator (Hardie & Vaz Nunes, 2001). Functional analysis including RNAi and CRISPR on cryptochrome genes (and other targets shown in Figure 5) would be highly valuable to show the involvement of these genes in the photoperiodic response in aphids and, potentially, in other insects.

EXPERIMENTAL PROCEDURES Aphids
A. pisum strains YR2 (York Red 2) and GR (Gallur Rojo) were used for the different analysis. YR2 strain was collected originally near York (UK) and strain GR, from Gallur (Spain) are our holocyclic and anholocyclic reference strains, respectively. Stocks of both aphid strains have been kept for years in our laboratory as viviparous parthenogenetic clones on Vicia faba seedlings under LD conditions (16 h of light and 8 h of darkness or 16L:8D) at 18 C. To study differences in mRNA expression potentially related to photoperiodism, third nymphal stage (N3) aphids from both strains were transferred to SD conditions (12L:12D). These were considered the generation 0 (SD-G0). Sexuparae aphids born from SD-G0 are referred to as SD-G1. Adult aphids of the SD-G1 were collected, quickly decapitated and the heads were immediately frozen in TRI Reagent (Ambion, Waltham, MA, USA) with liquid nitrogen and stored at À80 C. Sister SD-G1 insects of both strains were allowed to produce the next generation (SD-G2) to confirm the activation of the seasonal response (production of sexual morphs) in holocyclic YR2 aphids and the maintenance of parthenogenesis in the anholocyclic GR ones. Parallel groups of aphids were kept under LD conditions.

Quantification of cryptochrome gene expression by RT-qPCR
Excised aphid heads from YR2 and GR strains were used to compare the expression of the cryptochrome genes under different photoperiodic conditions and at different times of the day. We followed the procedure described in Cortés et al. (2010). Two experiments were designed to perform the quantification. In the first, the material here analysed is the same that had been previously used to study the expression of genes at the core of the aphid circadian clock . From each strain, aphids were sampled at six different time points along the day-night cycle for 1.5 days starting 3 h after the lights went on (zeitgeber time 3 or ZT3) and thereon taken at 6-h intervals (i.e., ZT9, ZT15, ZT21, ZT24+3, ZT24+9). Synchronized adult aphids reared under both LD and under short day (SD-G1) conditions (see above and see Cortés et al., 2010) were sampled, starting the following day after the final nymphal to adult moult. Three biological samples were obtained from each ZT, each sample consisting of 10 aphids. A second experiment that compared holocyclic YR2 adult aphids reared under LD and constants darkness (DD, 24L:0D) conditions was also performed. In this case, aphids reared on LD from birth were collected at ZT3, 9, 15 and 21 for three consecutive days (12 sampling points in total). A parallel group of aphids that were transferred to DD before the first sampling day was also sampled during the first consecutive days in DD. The source material of this second experiment is the same analysed in   Table S1. For each sample, three technical replicates were analysed. The RpL7 gene was used as an endogenous control of constitutive expression (Nakabachi et al., 2005). Relative expression for each sample was calculated using the ΔΔCt (threshold cycle) method (Livak & Schmittgen, 2001). All relative expression values were normalized to an inter-run calibrator sample (IRC). The presence of daily rhythmicity in the expression of cryptochrome genes was tested with COSINOR analysis (Refinetti et al., 2007). The parameters defining the fitted cosine function (MESOR, amplitude and acrophase) were indicated in significantly rhythmic profiles.

Fixation and dissection of the aphid central nervous system
Aphids were fixed in PFAT (4% paraformaldehyde in 1Â phosphatebuffered saline [PBS], 0.1% TritonX-100) overnight at 4 C. After three washes of 10 min in PBST (0.1% TritonX-100 in PBS 1Â), the central nervous systems were dissected out of the head capsule in cold PBST under a stereomicroscope and pooled in cold 100% methanol. Pooled tissues were washed twice in 100% methanol for 5 min and stored at À20 C until beginning of hybridization protocol.

Synthesis of digoxigenin-labelled riboprobes and FISH
To physically localize the place of transcription of the aphid cryptochrome genes, whole-mount in situ hybridizations were performed on dissected aphid central nervous systems using digoxigenin (DIG)-labelled RNA probes obtained as indicated in . First, total RNA was extracted and used for cDNA synthesis as indicated above.
We obtained DNA templates for probe synthesis by performing PCRs on this cDNA using the primers indicated in Table S1. The riboprobe for cry1 was synthesized from an amplicon 746 bp long with primers cry1-F4/R2. Two riboprobes were obtained for cry2 from amplicons cry2-2F3/R2 and cry2-1F5/R8 663 and 470 bp long, respectively. To avoid cross-hybridization, probes for cry1 and cry2 genes were designed from non-conserved regions between them. Alkaline phosphatase-conjugated with Anti-DIG (Roche) was used to detect hybridized probes using SIGMAFAST™ Fast Red TR/Naphthol AS-MX Tablets (ref. F4523, Sigma-Aldrich). In situ hybridization and detection of probes were performed as previously described .

Microscopy
Preparations of cry1 hybridizations of central nervous systems were imaged by confocal laser scanning microscopy (FV1000, Olympus, Hamburg, Germany) with a 559 nm excitation laser with UPLSAPO 20Â NA 0.75 and UPLFLN 40Â oil NA 1.30 objectives. Noise signal was reduced with Kalman filtering. Confocal images were processed with Fiji (Schindelin et al., 2012). Preparations of cry2 hybridizations were obtained by bright field microscopy with a DS-Ri1 CCD camera (Nikon) mounted on an Eclipse E800 microscope (Nikon).