SEARCH

SEARCH BY CITATION

Keywords:

  • aphid;
  • photoperiod;
  • clock genes;
  • circadian expression

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

The molecular basis of circadian clocks is highly evolutionarily conserved and has been best characterized in Drosophila and mouse. Analysis of the Acyrthosiphon pisum genome revealed the presence of orthologs of the following genes constituting the core of the circadian clock in Drosophila: period (per), timeless (tim), Clock, cycle, vrille, and Pdp1. However, the presence in A. pisum of orthologs of a mammal-type in addition to a Drosophila-type cryptochrome places the putative aphid clockwork closer to the ancestral insect system than to the Drosophila one. Most notably, five of these putative aphid core clock genes are highly divergent and exhibit accelerated rates of change (especially per and tim orthologs) suggesting that the aphid circadian clock has evolved to adapt to (unknown) aphid-specific needs. Additionally, with the exception of jetlag (absent in the aphid) other genes included in the Drosophila circadian clock repertoire were found to be conserved in A. pisum. Expression analysis revealed circadian rhythmicity for some core genes as well as a significant effect of photoperiod in the amplitude of oscillations.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

Circadian clocks are internal endogenous oscillators governing daily cycles of activity in most organisms including their physiology and behaviour (Bell-Pedersen et al., 2005). Photoperiod is the main stimulus entraining or synchronizing the clock to match the day–night cycles and, since it is also the main cue for season change, the participation of the circadian clock to keep seasonal rhythms has also been suggested (Lincoln et al., 2003; Stoleru et al., 2007).

Basic molecular constituents of animal circadian clocks are highly evolutionarily conserved and have been best characterized in Drosophila and mouse. In Drosophila, six transcription factors, organized into two feedback loops (Fig. 1), are central to the circadian clock (Cyran et al., 2003; Hardin, 2005). The per/tim feedback loop is based on the rhythmic expression of the period (per) and timeless (tim) genes, which encode for the transcription factors PER and TIM which repress their own transcription. Expression of per and tim is promoted by CLOCK-CYCLE (CLK-CYC) heterodimers binding E-box promoter regulatory elements. Levels of per and tim transcripts peak at dusk but PER and TIM accumulate in a delayed fashion during the night forming PER-TIM complexes that can enter into the nucleus. Once in the nucleus, PER interacts with CLK-CYC heterodimers bound to E-boxes, which inhibits CLK-CYC-mediated transcriptional activation, leading to minimum levels of per and tim mRNAs at dawn (Hardin, 2005). The kinases double-time (DBT), Casein kinase 2 (CK2) and shaggy (SGG), the protein phosphatase 2a (PP2A) and the F-box degradation signalling protein supernumerary limbs (SLMB) are involved in regulating the stability, subcellular localization and eventual degradation of PER and TIM proteins. In a second feedback loop (the Clk loop), CLK-CYC heterodimers directly activate at dusk the transcription of genes vrille and Pdp1 encoding the transcription factors VRI and PDP1 (specifically, the PDP1epsilon isoform) which inhibit and activate, respectively, the expression of Clock so that peak levels occur approximately in anti-phase to that of per and the other CLK-CYC-activated genes (Cyran et al., 2003). Photoperiodic entrainment is accomplished by the light-induced degradation of TIM through the participation of the photoreceptor CRYPTOCHROME (CRY1 in Fig. 1) and the F-box E3 ubiquitin ligase JETLAG (Koh et al., 2006; Peschel et al., 2006).

image

Figure 1. Model of the circadian clock in Drosophila. Genes (represented by crooked arrows) are organized into two interconnected feedback loops (Cyran et al., 2003) including the orthologs present in Acyrthosiphon pisum and major differences found in the aphid are highlighted in black boxes. Proteins constituting the core of the two feedback loops in Drosophila are indicated by different shapes. See text for details. Proteins committed to degradation are indicated by dotted shapes. The pea aphid genome contains two copies of a mammalian-type cryptochrome, CRY2, which is absent in Drosophila. Their putative role as repressor of CLK/CYC dimers (Yuan et al., 2007) is suggested by a question mark. CLK/CYC dimers and PDP1 attached to crooked arrows indicate their roles as transcription activators of corresponding genes. Lines ending in bars indicate negative regulation. Wavy lines indicate rhythmic transcription in Drosophila.

Download figure to PowerPoint

Two transcriptional feedback loops similar to those just described for Drosophila are also at the core of the circadian clockwork in the mouse, with conserved orthologs of most of the genes involved but having several differences. First, multiple copies of several clock genes occur in mammals leading to increased complexity. Second, orthologs of both timeless and the Drosophila-type cryptochrome are absent and there is an inclusion of two paralogs of an evolutionarily distinct cryptochrome (a so-called mammalian-type cryptochrome or mCry1 and mCry2, reviewed in Looby & Loudon, 2005). In the Drosophila central pacemaker, cryptochrome synchronizes the per/tim feedback loop with the light–dark cycle, whereas mCRYs are an integral part of the clockwork, substituting for TIM as the partner/s of PER to repress CLK-BMAL heterodimer activity (Bmal is the mammalian ortholog of cycle; Ivanchenko et al., 2001; Lin & Todo, 2005; Looby & Loudon, 2005; see also Collins et al., 2006, for a repressor role of Drosophila CRY in peripheral clocks). In mammals, the second loop works similarly to the Clk loop described for Drosophila, but cyclic expression of the activator Bmal instead of Clk is controlled by the products of two genes, Rev-Erbα and Rora not related with the Drosophila vrille and Pdp1 genes (reviewed in Bell-Pedersen et al., 2005; Looby & Loudon, 2005). As a net result of cyclic activity of clock genes, oscillations in the level of expression or in the activity of clock-controlled genes finally lead to the overt biological rhythms.

Given this high degree of conservation between Drosophila and mammals, it was expected that the Drosophila clock model would be highly conserved among other insects (Rubin et al., 2006). However, current data derived from circadian specific studies and from the analysis of recently available insect genomes suggest that, contrary to expectations, the Drosophila model cannot be generalized to the rest of the insect species and that diverse clock mechanisms may exist among them (Yuan et al., 2007). Differences in localization, pattern of expression, gene structure and in the gene repertoire involved in the clockwork have been reported (Sauman & Reppert, 1996; Chang et al., 2003; Zhu et al., 2005, 2008; Rubin et al., 2006; Yuan et al., 2007; Ikeno et al., 2008). More importantly, a mammalian-type cryptochrome, designated insect CRY2, that is absent in Drosophila is present in all non-drosophilid insects studied to date. Moreover, all insect CRY2 proteins potently repress CLK-CYC transcriptional activation in vitro, a property that is shared with mCRYs (Yuan et al., 2007). The Drosophila-type cryptochrome (here designated insect CRY1 as proposed by Yuan et al., 2007) is absent from the genomes of Tribolium castaneum and Apis mellifera and the latter species also lacks a timeless ortholog in its genome. Considering all these elements together, three major types of clockwork model have been proposed in insects (Yuan et al., 2007). First, the previously described Drosophila type in which PER would be the main repressor of CLK-CYC and CRY1 would merely transmit photic information to the clockwork through TIM. Second, the butterfly model having both CRY1 and CRY2, but the latter instead of PER would act as the main repressor of CLK-CYC transcriptional activation. Third, the absence of CRY1 in Tribolium and Apis depicts a clockwork type where, as in the butterfly, the main repressing function would reside in CRY2 which could also have a role in light entrainment.

Aphids, like most organisms, exhibit diurnal changes in their physiology and behaviour which are likely to be controlled by an endogenous circadian clock. Whilst limited, there are examples reporting the daily rhythm of release of sex pheromones (Eisenbach & Mittler, 1980; Thieme & Dixon, 1996), the rhythm in fresh weight-gain and larviposition time (Hodgson & Lane, 1981) or the rhythm in host-finding behaviour (Narayandas & Alyokhin, 2006). However, there is one aspect of aphid biology, related to the circadian clock, that has received much attention and that is their mode of reproduction by cyclical parthenogenesis. The switch from a parthenogenetic to a sexual mode of reproduction (or its inhibition by manipulations of the light–dark cycles) constitutes the first case of photoperiodism studied in animals (Markovitch, 1924) and has since been one of the most thoroughly studied examples of photoperiodism in insects. Briefly, shortening of day-length in autumn is the main cause of the switch from viviparous parthenogenesis to oviparous sexual reproduction in aphids living in nature. Moreover, this process can be mimicked or prevented in the laboratory by convenient manipulations in the photoperiod experienced by the reared aphids (Hardie & Vaz Nunes, 2001). Aphids had long been considered to measure photoperiodic time by a non-oscillatory (non-circadian) or hourglass mechanism (Hillman, 1973; Lees, 1973). Current evidence, however, suggests that photoperiodic responses in insects, including aphids, as in other major taxa are a function of the circadian system (Vaz Nunes & Hardie, 1993; Hardie & Vaz Nunes, 2001; Saunders et al., 2004; but see Veerman, 2001; Bradshaw et al., 2003; Emerson et al., 2009). Determining the existence and analysing the expression of known clock genes in species with robust photoperiodic responses is an essential step towards elucidation of this important issue. The recent availability of the genome of the aphid Acyrthosiphon pisum by the International Aphid Genomics Consortium offers the opportunity not only to characterize the first clockwork system in a hemimetabolous insect but also in a robustly photoperiodic system. This will allow a better understanding of the evolution of insect circadian clocks (current models are built based only on holometabolous species) and will elucidate whether the circadian clockwork (or some of its elements) also participates in the photoperiodic response.

In this report we present results of a search through the A. pisum genome for homologs of genes known to participate in the circadian clocks of Drosophila and other insects in order to establish the gene repertoire constituting the aphid clockwork. Using genome information, we cloned and sequenced aphid cDNAs from the genes constituting the core of the two feedback loops described above including per and tim homologs, for which predictions were rather poor, but also Clk, cyc, Cry1, Cry2, vri and Pdp1. For genes involved in stability/degradation of clock proteins present in the aphid genome, models available from the gene prediction programs, were either validated or modified by comparison with sequences from other insect species. Phylogenetic analysis was also carried out both to ensure orthology of the sequences and to better understand the evolution of clock genes. Finally, we investigated the expression of the eight core genes along the day–night cycle under two different photoperiod conditions.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

Identification of clock gene homologs in Acyrthosiphon pisum

Homologs for most of the Drosophila clock genes (Fig. 1) were identified in the A. pisum genome (Table 1). Most of these genes were found among the set of NCBI's RefSeq predictions. However, two key genes in the Drosophila clock (period and timeless) were absent from this set. For these two genes no expressed sequence tags (ESTs) were present in the databases (Table 1) and only partial predictions were available among non-RefSeq gene models included in the GLEAN set (Elsik et al., 2007) (Table 1). For AcpPer two poorly predicted non-overlapping models were available on two different scaffolds, adding up to 1069 translated amino acids. In the case of AcpTim a single model that could be translated into 477 amino acids was available (Table 1). A timeless paralog (a true ortholog of the Drosophila Timeout gene, whose involvement in the clockwork system is unknown), was, however, found among RefSeq predictions (see below). Both Drosophila-type (AcpCry1) and mammalian-type (AcpCry2) cryptochrome genes were identified in the aphid genome. Interestingly, the AcpCry2 gene was also duplicated. Seven of the eight genes putatively involved in the stability/degradation of core clock proteins were also searched for, and found to be present in the aphid genome (Table 1). However, our searches failed to identify a true ortholog of Drosophila E3 ubiquitin ligase jetlag. Sequence analysis, including alignments of the predicted genes with available sequences from different species and phylogenetic analysis, allowed us to confirm the true orthology of the sequences indicated in Table 1 and to validate the RefSeq models for the most part. All the identified genes were manually annotated in the Apollo GMOD-Chado database accessible at the AphidBase web portal (http://www.aphidbase.com). Except for period and timeless, only a few gene models needed minor modifications. Additionally, we designed primers based on the genome sequence (Table S1), and cDNAs from the core clock gene set were amplified by PCR and sequenced. This allowed experimental validation of the corresponding gene models and was essential for obtaining the true coding sequence of A. pisum genes period and timeless (Table 1 and below). Finally, the numbers of ESTs available for putative core clock genes were very low, ranging from none to four, while those available for genes involved in the stability/degradation of core clock genes were more abundant (Table 1).

Table 1.  Summary of data obtained for clock gene sequences identified in the Acyrthosiphon pisum genome
Genes1Symbol2Prediction3Aphidbase4ESTs5NCBI6Length7Comp.8Rate9
  • 1

    Drosophila genes searched for in the A. pisum genome.

  • 2

    Symbols given to the identified genes.

  • 3

    If available, NCBI RefSeq accession number of the predicted orthologous protein is given. Otherwise available GLEAN models are given.

  • 4

    Aphidbase Gbrowse accession numbers (http://www.aphidbase.com/aphidbase/).

  • 5

    Number of ESTs found among the approximately 170.000 A. pisum ESTs available at Aphidbase.

  • 6

    NCBI accession numbers of cDNAs sequenced in the present study ($, partial sequences).

  • 7

    Length of the A. pisum gene sequences (amino acids)/Length of the alignments of conserved regions used for phylogenetic reconstructions. For the core genes (except AcpVri), the length derived from our sequenced cDNAs is given.

  • 8

    p-values obtained for A. pisum sequences in the chi-square tests performed using TREE-PUZZLE (Schmidt et al., 2002) to test for homogeneity of amino acid composition in insect sequences. **, highly significant (P < 0.01).

  • 9

    Summary of the results obtained using RRTree (Robinson-Rechavi & Huchon, 2000) to test for homogeneity in rates of amino acid sequence evolution among insect sequences. ++, A. pisum sequences showing highly accelerated rates in all the comparisons; +, A. pisum sequences showing accelerated rates in some of the comparisons (the number of significant comparisons over a total of 5 is indicated in parenthesis); =, A. pisum sequences not showing accelerated rates in any comparison. Insect species included in the composition and rate analyses are as in Fig. 2.

Feedback loop core genes        
 periodAcpPerGLEAN_27669ACYPI476690FM9986461018 / 4620,009**++
GLEAN_36135ACYPI48687
 timelessAcpTimGLEAN_16439ACYPI364390FM998651$606 / 3820,006**++
 cycleAcpCycXP_001943647ACYPI0046862FM998648648 / 2730,519+ (4)
 clockAcpClkXP_001944549ACYPI0048123FM998647613 / 3230,983+ (2)
 vrilleAcpVriXP_001949829ACYPI0088510FM998650$485 / 1481,000=
 PAR-domain protein 1AcpPdp1XP_001943257ACYPI0007024FM998649293 / 1471,000=
 cryptochromeAcpCry1XP_001944402ACYPI0057572FN377569557 / 4660,447=
AcpCry2-1XP_001950693ACYPI0065840FN377570501 / 4910,830+ (4)
AcpCry2-2XP_001950192ACYPI0041970FN377571491 / 4910,664+ (4)
Stability/degradation        
 double-timeAcpDbtXP_001951697ACYPI00816210417 / 2611,000=
 supernumerary limbsAcpSlmbXP_001949585ACYPI00687435525 / 4181,000=
 Casein kinase II α subunitAcpCk2αXP_001942962ACYPI00200633361 / 3221,000=
 Casein kinase II β subunitAcpCk2βNP_001119697ACYPI00008926220 / 2191,000=
 protein phosphatase 2a (twins)AcpTwsXP_001949195ACYPI00974114448 / 4401,000=
 protein phosphatase 2a (widerborst)AcpWdbXP_001944239ACYPI00066620579 / 4281,000=
 shaggyAcpSggXP_001951519ACYPI00925711425 / 3491,000=
 jetlagNot found      

Evolutionary distinctiveness of clock genes in Acyrthosiphon pisum

During the course of sequence analysis it soon became evident that, although some aphid clock gene sequences were readily aligned with orthologous sequences from other insect species, the alignments contained a high number of mismatches. Sequencing errors were excluded after examining traces available at NCBI for some of these genes and after cDNA sequencing of others (see Experimental procedures). Homogeneity tests performed with TREEPUZZLE (Schmidt et al., 2002, see Experimental procedures) on available insect sequences, representative of different insect orders, including the hemipteroid Pediculus humanus, revealed significant differences in amino acid composition in A. pisum AcpPER and AcpTIM (Table 1). All other insect species showed homogeneous composition for all sequences analysed. Moreover, relative rate tests performed on amino acid sequences using RRTREE (Robinson-Rechavi & Huchon, 2000 see Experimental procedures), showed that AcpPER, AcpTIM, and other genes orthologous to those participating in the Drosophila per/tim feedback loop (Fig. 1) were evolving at accelerated rates with respect to their respective insect orthologs (Table 1). Additionally, by computing the percentage of amino acid differences among available clock gene sequences representative of different insect orders and comparing with the average distance between A. pisum and the rest of insect sequences, it was evident that some aphid clock genes were particularly divergent (Fig. 2). This unusual divergence was not observed, however, between equivalent sequences from a second hemipteroid (P. humanus) or in any other insect sequences. As a result, some aphid sequences, although clearly orthologous of corresponding insect genes, occupied aberrant positions in the phylogenetic trees (some of them are shown in the next section). These results support the hypothesis that genes involved in the putative per/tim feedback loop (Fig. 1), especially period and timeless, are evolving faster in A. pisum than their orthologs in other insects. However, putative genes from the Clk feedback loop and genes involved in the regulation of the stability and degradation of core clock proteins (Fig. 1) appear to have evolved at rates comparable to other insect species (Table 1; Fig. 2).

image

Figure 2. Average amino acid distances ± SE for several clock genes. Striped bars represent average distances between Acyrthosiphon pisum and other insects belonging to different orders including a second hemipteroid, Pediculus humanus (Anoplura), from the Hemimetabola and Drosophila melanogaster (Diptera), Bombyx mori (Lepidoptera), Apis mellifera (Hymenoptera) and Tribolium castaneum (Coleoptera) from the Holometabola. Black bars show the average distances for each gene among all the above-mentioned species (A. pisum included). Grey bars represent average distances between the hemipteroid P. humanus and the rest of insect species, including A. pisum. The orthologous sequences of PER and CRY2 from Riptortus pedestris (Hemiptera) and of PER from Blattela germanica (Orthoptera) were also included in the corresponding analysis. Due to problems with some sequences retrieved from databases, or to the lack of genes in some species, Anopheles gambiae (Diptera) replaced D. melanogaster in CRY2 and CYC analyses, Dianemobius nigrofasciatus (Orthoptera) replaced P. humanus in CRY1 analysis and Antheraea pernyi and Danaus plexippus (Lepidoptera) replaced B. mori in CLK and SGG analyses, respectively. Significant difference (P < 0.01), as revealed by one-way anova comparing bar heights, were due to extreme aphid divergence (indicated by asterisks).

Download figure to PowerPoint

Characterization of Acyrthosiphon pisum core-clock gene sequences period and timeless

Cloning and sequence analysis of AcpPer cDNA yielded a 3275 bp sequence that contained both an initial AUG and a stop codon and a predicted protein of 1018 amino acids. The predicted protein clearly grouped with other insect PER sequences after a phylogenetic analysis that revealed, however, a high divergence of the A. pisum sequence (Figs. 2, 3A; Table 1). As a PAS protein (PER-ARNT-SIM; Hirayama & Sassone-Corsi, 2005), AcpPER was shown, after a SMART search, to contain two tandemly organized PAS (PAS-A and PAS-B) domains and a PAC (PAS associated C-terminal) domain highly conserved among all animal PER proteins (Fig. 3B). While these three domains are recognized in A. pisum, they are rather divergent. For instance, two regions corresponding to proposed TIM binding sites in Drosophila were found to be highly divergent, as well as a region orthologous to the CLK-CYC interaction domain or CCID (Fig. 3B; Saez & Young, 1996; Chang & Reppert, 2003). Similarly, two Nuclear Localization Signal (NLS) motifs necessary for nuclear import of PER that are well conserved across insect species are either missing or highly modified in AcpPER (Fig. 3B).

image

Figure 3. Phylogenetic analysis and structure comparisons of PER and TIM proteins in different species. (A and C) Phylogenetic relationships of insect PER and TIM proteins, respectively, inferred by neighbor joining on Poisson corrected distances. For PER, the Daphnia sequence was used as outgroup. In TIM analysis, the Drosophila melanogaster and Acyrthosiphon pisum sequences for the TIMEOUT paralog (here abbreviated as TIM2) were included as outgoups. Bootstrap values are shown for nodes only when higher than 50%. (B and D) Schematic representation of PER and TIM protein features from D. melanogaster and A. pisum. For Drosophila, isoform D of the TIM protein was used as reference. The numbers at the end of each diagram indicate protein size. For AcpTIM the filled box represents the sequenced portion of the protein while empty boxes with a question mark represent the hypothetical missing parts. Abbreviations: NLS, nuclear localization signal; CLD, cytoplasmic localization domain; PAS, PER-ARNT-SIM; TG, threonine-glycine region. Two PAS domains (PAS-A and PAS-B) followed by a PAC (PAS associated C-terminal) domain are found in PER protein. Black lines indicate the TIM and PER binding sites present in PER and TIM proteins, respectively, in D. melanogaster (Saez & Young, 1996). The dotted line indicates the CLK/CYC inhibition domain described in Drosophila (Chang & Reppert, 2003). Numbers below these lines represent percentage of identity between D. melanogaster and A. pisum and between D. melanogaster and Pediculus humanus, respectively. The alignments of different insect sequences corresponding to two PER and one TIM regions containing conserved NLS motifs are shown. Numbers on the left and right sides of each sequence in the alignment blocks indicate the relative positions of the first and last residues shown with respect to the N-terminus (position 1) for each sequence.

Download figure to PowerPoint

PCR amplification of AcpTim cDNA usually needed two rounds of nested PCR reactions and, in many instances, fragments apparently containing unspliced introns were obtained. We PCR amplified, cloned and sequenced 1818 bp of cDNA corresponding to a partial sequence of the A. pisum timeless homolog encoding 606 amino acids (about half the size of the protein in other insects) with the 5′- and 3′- ends missing (about 70 and 350–400 amino acids, respectively, when compared with other insect sequences) (Fig. 3D). Attempts to amplify further upstream or downstream coding sequences using different strategies always produced fragments containing stop codons in both directions. Orthology of the predicted protein with other insect TIM sequences was evident after phylogenetic analysis that included the paralog AcpTIM2 (Fig. 3C). However, as already described for AcpPER, although the sequenced fragment could be reliably aligned with other TIM sequences, the high divergence throughout the A. pisum sequence (Fig. 2) was reflected in its anomalous position in the phylogenetic tree and in the length of the branch leading to the A. pisum sequence (Fig. 3C). The presence of a somewhat modified NLS motif, that is highly conserved in other TIM sequences, was observed within a region relatively well conserved in other insect species but poorly conserved in A. pisum (Fig. 3D). In Drosophila, this motif has been shown to participate in the interaction with PER (Saez & Young, 1996).

Clock and cycle

Apart from minor point changes, predicted RefSeq models for Clk and cyc were coincidental with our cDNA derived sequences. Predicted proteins from both gene sequences contained the domains characteristic of bHLH-PAS transcription factors conserved in other insects and mammals (Hirayama & Sassone-Corsi, 2005). These included a ‘basic helix-loop-helix’ (bHLH) domain containing a well conserved NLS motif and PAS (A and B) and PAC domains (Fig. 4). The predicted AcpCLK protein was about 400 amino acids shorter than the corresponding Drosophila homolog having a size comparable to the homologous sequence in other insects (Fig. 4A; Rubin et al., 2006). AcpCLK contained a C-terminal domain 60% similar to the Δ19 region found conserved in mouse and several insects (Chang et al., 2003; Rubin et al., 2006) that is apparently involved in the stabilization of the CLK-CYC heterodimer (Takahata et al., 2000; Chang et al., 2003). The polyglutamine-rich regions found in Drosophila are absent in AcpCLK and, hence, similar to the observations in Antheraea pernyi (Chang et al., 2003).

image

Figure 4. Schematic representation comparing the structure and some protein features in AcpCLK (A) and AcpCYC (B) with corresponding Drosophila proteins. The numbers at the end of each diagram indicate protein size. Abbreviations: NLS, nuclear localization signal; bHLH, basic helix-loop-helix domain; Q, polyglutamine-rich region; T, transactivation domain. PAS and PAC domains are as in Fig. 3. Triangle indicates a so-called Δ19 region (Chang et al., 2003). Sequence alignments showing the degree of conservation of NLS and the transactivation domain in CLK and CYC, respectively, are shown. Numbers on the left and right sides of each sequence in the alignment blocks indicate the relative positions of the first and last residues shown with respect to the N-terminus (position 1) for each sequence.

Download figure to PowerPoint

The predicted AcpCYC protein was about 200 amino acids longer than the Drosophila homolog, as observed in other insects, and contained a C-terminal domain highly conserved in all insects studied except Drosophila which lacks this domain. This domain is highly similar to the transactivation domain described in the mouse CYC homolog BMAL (Takahata et al., 2000; Chang et al., 2003) (Fig. 4B). Although less dramatic than was observed for AcpPER and AcpTIM, both AcpCLK and AcpCYC seemed to accumulate numerous amino acid differences when comparing sequences from conserved aligned blocks with other insect sequences. These differences were reflected in the long branches leading to A. pisum sequences in the phylogenetic trees obtained, and in the unexpected groupings of some sequences, especially for AcpCYC (Fig. S1). Although amino acid composition of AcpCLK and AcpCYC was not significantly different from other insects, relative rate tests showed that AcpCYC was evolving at accelerated rates in most comparisons while AcpCLK was accelerated in some of them (Table 1).

Vrille and Pdp1

SMART analysis of the A. pisum orthologs of these two basic zipper transcription factors revealed the presence of typical BRLZ (Basic-Region-Leucine-Zipper) domains in both of them with highly conserved DNA-binding domains (Fig. 5). As observed for other insects, both proteins were much shorter in A. pisum than in Drosophila. AcpVRI contained a glycine-serine rich stretch similar to Drosophila and a glutamine-rich domain which is absent from Drosophila VRI (Fig. 5A). AcpPDP1 contained a proline and acidic rich region (PAR) similar to Drosophila and other insects, but lacked the glutamine- and alanine-rich domains present in Drosophila (Cyran et al., 2003). Interestingly, our cDNA sequence data, in agreement with the RefSeq prediction, revealed that AcpPDP1 appeared to lack seven amino acids from an extended basic domain including the highly conserved KKSRK amino acids that, in Drosophila, are necessary for proper DNA binding (Reddy et al., 2000). However, a thorough analysis of sequencing traces derived from our directly sequenced PCRs and analysing additional recombinant plasmids containing cloned fragments from this gene showed a minority of transcripts including the missing seven amino acids (Fig. 5B). By looking at the genome sequence, it became clear that the seven amino acids corresponded to a 21 bp DNA sequence starting with a canonical GT donor site that was affected by alternative splicing – either incorporating this 21 bp or splicing it out as part of the adjoining intron. Finally, both AcpVRI and AcpPDP1 proteins had a homogeneous composition with other insect sequences and did not seem to be evolving at accelerated rates (Table 1).

image

Figure 5. Schematic representation comparing the structure and some protein features in AcpVRI (A) and AcpPDP1 (B) with corresponding Drosophila proteins (isoforms A and D, respectively). The numbers at the end of each diagram indicate protein size. Abbreviations: BRLZ, basic region leucine zipper; G/S, glycine-serine rich stretch; Q, glutamine-rich domain; A, alanine-rich domain; PAR, proline and acidic rich domain; E, extended DNA-binding domain. Alignments containing partial BRLZ domains in different species for both genes are shown with boxes indicating well conserved DNA-binding domains. Numbers on the left and right sides of each sequence in the alignment blocks indicate the relative positions of the first and last residues shown with respect to the N-terminus (position 1) for each sequence. A ‘VKKSRKQ’ motif between arrows indicates a second AcpPDP1 splicing variant found containing this motif apart from the most abundant variant lacking these seven amino acids (included in the alignment).

Download figure to PowerPoint

Cryptochromes

Four different sequences that were highly similar to Drosophila CRY were found in the A. pisum genome by BLAST search. In order to ascertain orthology relationships, we built up a phylogenetic tree including representative sequences of the different insect and vertebrate cryptochromes including evolutionary related 6-4 DNA photolyases (Yuan et al., 2007) (Fig. S2). The phylogenetic analysis revealed that both mammal-type and Drosophila-type cryptochrome genes were present in the A. pisum genome as well as a 6-4 DNA photolyase. Following Yuan et al. (2007) we designated the aphid cryptochromes as AcpCry1 (Drosophila-type) and AcpCry2 (mammal-type). We found AcpCry2 is duplicated, with both versions differing by 3.5% of their total (18 out of 512 amino acids). Moreover, both versions of AcpCRY2 cDNAs were cloned and sequenced in our A. pisum strain (see Experimental procedures), revealing several polymorphisms when compared with the NCBI sequences (data not shown), which supported their paralogous character. SMART analysis showed that AcpCRYs contained typical flavin adenine dinucleotide (FAD) binding and photolyase domains (Fig. 6). Additionally, both AcpCRY2 paralogs, but not AcpCRY1, contained conserved domains that, in the mouse and zebrafish, are necessary for the CLK-CYC repressing function of mCRYs (Hirayama et al., 2003; Rubin et al., 2006). NLS and Coiled-coil motifs described in mammal-type CRYs (Hirayama et al., 2003; Chaves et al., 2006) were also conserved in AcpCRYs (Fig. 6). Interestingly, both AcpCRY2 genes showed significant accelerated rates of change in most comparisons (Table 1; Fig. 2).

image

Figure 6. Schematic representation comparing the structure and some protein features in the mouse mCRY1 with its aphid ortholog AcpCRY2. The numbers at the end of each diagram indicate protein size. Dark grey and white backgrounds demarcate FAD- binding and photolyase domains, respectively. Three regions (RD-2a, RD-1 and RD-2b) necessary for the CLK-BMAL repressing function are shown as 2a, 1 and 2b, respectively (Hirayama et al., 2003). Percentage identity between Mus musculus and A. pisum and between M. musculus and Riptortus pedestris (Ikeno et al., 2008), respectively, are indicated for these three regions and for a CLD domain. The C-terminal region (C-T) is indicated. All mouse features conserved in the aphid are indicated on the aphid molecule by similar boxes. Two multiple sequence alignments corresponding to the mouse RD-2b region containing an NLS motif and to a Coiled-coil domain (C-C), both necessary for nuclear entry (Chaves et al., 2006) are shown. The alignments also include AcpCRY1 and other insect sequences representative of the Drosophila-type cryptochrome (Insect CRY1) showing the lack of the above domains in this paralog. Numbers on the left and right sides of each sequence in the alignment blocks indicate the relative positions of the first and last residues shown with respect to the N-terminus (position 1) for each sequence.

Download figure to PowerPoint

Expression analysis of core-clock genes in Acyrthosiphon pisum

Temporal expression patterns of putative core-clock genes analysed through quantitative real-time PCR (see Experimental procedures) revealed weak to moderate oscillations in mRNA abundance for some genes in aphid heads along the day–night cycle, especially in aphids reared under long-day (LD) conditions (Fig. 7). In general, the oscillating patterns observed in mRNA levels in heads of aphids reared under short-day conditions (SD) were similar to those observed in aphids reared under long-day conditions (LD). However, for some genes, a significantly elevated expression in SD aphids was observed (Fig. 7). This was the case for AcpPer, AcpPdp1, and AcpCry2 (Fig. 7) and also AcpCry1 (not shown). No significant oscillation along the day–night cycle was observed for AcpClk and AcpCry1. Conversely, a significant oscillation was observed in AcpCyc under LD, with a peak of expression at the beginning of the light phase (Fig. 7, P= 0.04). AcpPer also significantly oscillated in LD aphids with the maximum expression at the end of the light phase (Fig. 7, P= 0.0002). As for other genes, oscillation of AcpPer under SD conditions was not significant, likely attributable to high variability at some time points. AcpCry2 oscillated both in LD and SD and, interestingly, showed the same temporal pattern of expression as AcpPer under both photoperiod conditions (Fig. 7, P= 0.05 and P= 0.04, respectively). AcpVri and AcpPdp1 both oscillated in LD aphids with a pattern similar to that observed for AcpCyc, having maximum expression at the beginning of the light phase and the minimum at the end of the light phase (Fig. 7, P= 0.05 and P= 0.005, respectively). AcpVri also significantly oscillated in SD aphids (P= 0.008). Finally, expression of AcpTim was also rhythmic in LD, with a peak at the beginning of the dark phase (as observed for AcpPer). However, difficulties experienced in the amplification of AcpTim cDNAs (including the frequent presence of dimers and the occasional amplification of fragments of unexpected size) suggest our results for AcpTim should be treated with caution until additional data becomes available. These problems led us to exclude from the analysis AcpTim results from SD aphids.

image

Figure 7. Temporal expression patterns of core clock genes in Acyrthosiphon pisum. The plots show the average (± SE) relative head mRNA levels for each time point, represented as filled circles and filled triangles for LD and SD regimes, respectively (see Experimental procedures). Continuous lines and dotted lines represent sine wave curves fitted to 4 h interval data (see Experimental procedures). Open bars at the bottom of the plot indicate light and black bars indicate dark during the two illumination regimes used. For each gene P-values obtained after the zero-amplitude test for cosinor analysis are shown when significant. One way anova was performed in order to check for significant differences in relative head mRNA levels between the two photoperiods investigated. P-values are indicated for those genes with significant differences between LD and SD photoperiods. NS, non significant; NA, not applicable.

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

In the present study we have described the gene repertoire of the putative circadian clock of the aphid Acyrthosiphon pisum based on the analysis of its recently available genome. We have also analysed the structure, evolution and the expression of genes at the core of the clockwork. Although some individual clock genes have been described in other species, this is the first hemimetabolous insect in which the full core clockwork gene set has been investigated and it will therefore contribute to a better understanding of the functioning and evolution of the circadian clock in insects.

We have shown the presence in A. pisum of orthologous sequences of the full gene set constituting the core of the two feedback loops described in the Drosophila circadian clock, as well as most of the ancillary genes that control the stability/degradation of core clock proteins (Fig. 1; Table 1). Notably, the A. pisum genome contains a mammalian-type cryptochrome (AcpCry2), confirming its presence in hemimetabola (see Ikeno et al., 2008) and suggesting it has been lost in drosophilids (Rubin et al., 2006; Yuan et al., 2007; Zhu et al., 2008). Moreover, the presence of two copies of this gene in the aphid is similar to the situation in many vertebrates, but this is the first time it is reported in an insect, which points to an additional relevance for this aphid gene. In addition, contrary to the situation in T. castaneum and A. mellifera but similar to Lepidoptera, A. pisum also contains a copy of a Drosophila-type cryptochrome (AcpCry1), which would place the aphid clockwork closer to the hypothetical ancestral clock model proposed by Yuan et al. (2007) as in the butterfly (see Introduction). Under this model, AcpCRY1 would work primarily as a circadian photoreceptor whereas AcpCRY2 would be an essential component of the core clockwork acting as a major transcriptional repressor of the clock alone, or in conjunction with AcpPER (see below). The presence of almost the entire clockwork gene set in the aphid genome (with the exception of a jetlag ortholog, see below) might suggest a high conservation of the clock mechanism. However, a detailed analysis of individual genes shows that the aphid system is somewhat unique. First, the A. pisum orthologs of the four transcription factors at the core of the Drosophila per/tim feedback loop (AcpPER, AcpTIM, AcpCYC and AcpCLK) are present in the aphid but are highly divergent and are evolving faster than corresponding proteins in other insects. This is especially true for AcpPER and AcpTIM (Table 1; Fig. 2). Similarly, the two copies of AcpCRY2 are also evolving at accelerated rates. If, as has been proposed for other insects (Rubin et al., 2006; Yuan et al., 2007; Zhu et al., 2008), AcpCRY2 does have a central role in the aphid clock, then all genes at the core of the putative per/tim loop in the aphid oscillator are changing rapidly with the exception of AcpCRY1.

The possibility that the observed accelerated rates of change in the genes of the per/tim loop is reflecting a different clockwork system in hemimetabola can be ruled out as sequence comparisons always included at least a second hemimetabolous insect (usually the human lice P. humanus, whose genome is also currently being sequenced) not showing these high rates of sequence change. This was most evident after the analysis of AcpPER which, in addition to the P. humanus sequence, also included sequences from distant hemimetabolous representatives (Blatella and Gryllus) and from the hemipteroid Riptortus pedestris (Fig. 3A; Ikeno et al., 2008) which are more similar to Drosophila and other holometabola sequences than the A. pisum sequence. Similarly, the sequenced portion of AcpTIM was found to be more divergent than the Daphnia ortholog when compared to other insects (Fig. 3C). Therefore the high divergence observed for core clock genes would be characteristic of aphid sequences and not the result of a holo- vs. hemi-metabolous differential clock evolution. Although it is possible that the genes at the core of the per/tim feedback loop in the aphid are each evolving independently to adapt to new aphid-specific functions, it seems more reasonable to think that the observed high rates of change in all these genes are reflecting a co-ordinated evolution to adapt the clock to the (as yet unknown) aphid-specific needs. It is worth to mention in this context the unexpected phylogenetic position reported for the PAS dimerization domain in the Musca domestica PER protein (Piccin et al., 2000) which, along with a functional assay, led the authors to hypothesize an intermolecular coevolution between PER and TIM. Contrary to observations in core clock genes, the other (ancillary) genes identified in the aphid's genome, putatively involved in the stability/degradation of the aphid core clock proteins, all seemed to be evolving according to the pace in other insect species, which agrees with these genes having additional non-clock related functions (Vallone et al., 2007). Moreover, two of the three genes participating at the putative aphid second feedback loop (genes AcpVri and AcpPdp1 controlling gene Clock expression, see Fig. 1) did not show any evidence of accelerated rates of change (Table 1; Fig. 2), which agrees with their orthologs being, at least in Drosophila, transcription factors controlling the expression of other genes in addition to Clock (see reports for both genes at http://flybase.org/). Therefore the distinctive pattern observed for genes at the core of the aphid circadian clock reflects a process of adaptation of the aphid clockwork rather than a global process affecting all aphid genes.

More studies are obviously needed to know the force(s) behind the fast evolution of aphid core clock genes. However, our present results on sequence analysis and expression data on particular genes may help to shed some light on the issue and constitute the starting point for further research. For example, in the case of the aphid PER protein, the high divergence detected at the amino acid level also affects two NLS motifs (highly conserved in all other insects) that have been proposed to be involved in the nuclear import of the protein, which are absent or highly modified (Fig. 3B). As nuclear import is essential for the transcription inhibition role attributed to PER, it is likely that AcpPER on its own lacks this function which, along with the high substitution rate detected, might be indicative of a modified functionality of this gene. However, it is also possible that the second NLS, by conserving some basic residues in A. pisum (Fig. 3B), is still functional, allowing independent entry of AcpPER into the nucleus to exert its repressing role on the putative AcpCLK-AcpCYC dimers. Finally, we cannot disregard that AcpPER may enter the nucleus with the help of another protein. Both AcpTIM and AcpCRY2 would be the candidates for this role as their orthologs in Drosophila and the mouse, respectively, have been shown to stabilize and help PER nuclear entry (Saez & Young, 1996; Hirayama et al., 2003; Sakakida et al., 2005). Our data do not show if these interactions take place among these proteins in the aphid. However, the high divergence of putative domains that, in Drosophila, are involved in the binding of PER and TIM, and in the interaction with the CLK-CYC dimer (Fig. 3B, D), suggests that these interactions may not occur in the aphid. Since AcpCRY2 domains, that participate in nuclear entry and repression of CLK-CYC activity in mammalian-type Cryptochrome (Hirayama et al., 2003), are highly conserved (Fig. 6), this strongly supports a conserved repressing function of this aphid gene. Whether such repression requires AcpPER remains to be elucidated. With respect to the aphid orthologous of Drosophila PER's partner (AcpTIM), the fact that it is even more divergent than AcpPER (its definitive size pending future research) supports the thesis that it has evolved a modified functionality, either within or outside the clock machinery. In this context, the absence of a jetlag homolog in the A. pisum genome provides support, albeit indirect, for a modified functionality (or even a loss of function) of AcpTIM. JETLAG (JET) is an F-box protein of the SCF E3 ubiquitin ligase family that, in Drosophila, has been shown to be essential for synchronising the clock with the day–night cycles by promoting the degradation of TIM in response to light (Koh et al., 2006). It has also been demonstrated that CRY1 is necessary in the process (as the light photoreceptor transmitting photic information) and that both proteins must physically interact with TIM (Koh et al., 2006). We investigated the presence of jet in different insect species with full genome sequence available and found that it was only absent in species in which either cry1 or tim orthologs were missing (e.g. T. castaneum, A. mellifera and P. humanus; own unpublished results). Therefore, the absence of JET in A. pisum may indicate that, if aphid clock cells have an intrinsic light entrainment mechanism, it may not rely on AcpTIM. We do not discount the possibility that the accelerated evolution detected in A. pisum per and tim genes may be the result of selection pressure on functions of these genes not related with the circadian clock. In this respect, both per and tim genes in Drosophila have been involved in different species-specific functions directly related with reproductive success, including courtship and mating behaviour (Kyriacou & Hall, 1980, 1986; Sakai & Ishida, 2001; Tauber et al., 2003; Beaver & Giebultowicz, 2004) and gamete production (Beaver et al., 2002, 2003; Kotwica et al., 2009).

Analysis of the expression of putative aphid clock genes should provide clues on their roles (if any) in the clockwork. In this respect, the difference in the number of ESTs present in public databases between core clock genes and genes involved in stability/degradation of core products (Table 1) is most notable. Attending to those data, the expression of core clock genes in the aphid must be rather limited compared with the expression of ancillary genes which are involved in various other processes apart from their putative roles in the circadian clock. Our experimental results on the expression of core clock genes draws a preliminary picture of how the circadian clock works in A. pisum. First, significant rhythm in transcript levels for some genes shown in Fig. 7 shows their regulation in response to the zeitgeber (the light–dark cycles) and would suggest these genes are also part of the A. pisum endogenous circadian oscillator. Second, the coincident phasing of peak AcpPer, AcpTim and AcpCry2 transcripts (our assay did not distinguish between the two versions of the latter gene), which all peak at the end of the day, points to the coordinate regulation of these genes. Whether their expression is activated by AcpCLK-AcpCYC dimers binding to common elements in their promoter regions will need further investigation, including the search for E-boxes in the promoter regions of these genes. In Drosophila both per and tim mRNAs peak synchronously at dusk (within the dark phase; Williams & Sehgal, 2001). The same pattern of expression has been observed in butterflies for orthologs of these genes (Zhu et al., 2008) and in A. mellifera for amPer and amCry2 genes (a timeless ortholog is absent in the honey bee) (Rubin et al., 2006). The pattern observed for the orthologs of these three genes in A. pisum was similar to the pattern described in these organisms but somewhat advanced: in the aphid these genes were shown to peak at the end of the day rather than at the beginning of the dark phase (Fig. 7). Third, similar to Apis mellifera (Rubin et al., 2006) AcpCyc, but not AcpClk, showed a significant oscillation with maximum expression at dawn, in anti-phase with the expression of AcpPer, AcpTim and AcpCry2 (Fig. 7). This, along with the high conservation in A. pisum of a transactivation domain present at the C-terminal region of CYC proteins in all non-drosophilid insect species but also in the mammalian ortholog BMAL (Fig. 4B), places the aphid clock closer to the mammal than to the Drosophila system. It is worth remembering that in the mouse mBmal, and not mClk, oscillates which is contrary to the situation in Drosophila (Stanewsky, 2003). Fourth, the synchronous oscillation of both AcpPdp1 and AcpVri in antiphase with AcpPer, AcpTim and AcpCry2 is different from Drosophila where their expression is also controlled by the CLK-CYC dimer (Cyran et al., 2003). This fact and the previous discussion on AcpCyc and AcpClk expression suggest that the second loop works differently in the aphid circadian clock and probably in other insects (Fig. 1). Unfortunately, only two studies on the expression of vrille (Ikeno et al., 2008) and Pdp1 (Dolezel et al., 2008) orthologs in other insects are available, which prevents us from making comparisons in order to make a general statement on the role/s, if any, of these two genes in the circadian clock in non-drosophilid insects. Fifth, the lack of any evident oscillation of AcpCry1 is similar to observations in A. pernyi (Zhu et al., 2008) compatible with a mere role in photoreception rather than as a core pacemaker. Sixth, according to our results modifications of the photoperiod regime do not appear to substantially alter the oscillating pattern of expression of core clock genes. However, a significant increase in the expression of some of them was observed in aphids reared under short days suggesting a link between the control of circadian rhythm and photoperiod controlled processes. It would be tempting to propose that the aphid clockwork has become biased towards the control of seasonal (i.e. photoperiodic), rather than daily, responses. Most relevant in this context, several reports have shown the involvement of orthologs of clock genes timeless and period in photoperiodic responses (Goto & Denlinger, 2002; Pavelka et al., 2003; Shimada, 2005; Goto et al., 2006; Mathias et al., 2007; Stehlik et al., 2008), supporting the view that photoperiodism in insects relies on the circadian clock machinery. It has also been shown that natural selection has shaped genetic variation at the period and timeless loci in Drosophila producing latitudinal clines which are usually interpreted as evidence of adaptation to seasonal conditions at different latitudes (Costa et al., 1992; Sandrelli et al., 2007; Tauber et al., 2007; reviewed in Kyriacou et al., 2007). However, as discussed by Bradshaw & Holzapfel (2007a,b), it is possible that circadian genes are only incidentally involved in photoperiodism and both circadian and seasonal rhythms would rely on independent gene circuits.

Finally, we have shown that A. pisum has the set of genes necessary for a circadian clock to function. Moreover, the expression observed for some genes suggest a clockwork system closer to the hypothetical ancestral system in insects rather than to the Drosophila system (Ikeno et al., 2008). In this system, AcpCRY2 (alone or together with AcpPER) would have an essential role as repressor of a putative AcpCLK-AcpCYC-induced transcription. It is likely that AcpCRY1 works in this system as a photoreceptor synchronizing in some unknown way the clock with the day–night cycles. It is also likely that the second feedback loop described in Drosophila (Fig. 1) is different in A. pisum and that a loop centered on AcpCyc instead of AcpClk works in aphids similar to vertebrates. However, despite having all the necessary elements and having shown that expression of some genes are compatible with a circadian clock working in the aphid, in order to demonstrate that the studied genes are bona fide components of the circadian mechanism in the pea aphid, it is necessary to show that their mRNAs continue to oscillate in constant darkness (DD, Bell-Pedersen et al., 2005). Although in the present report we have put forward some hypotheses, there are some additional questions that need immediate clarification. For instance, why are some clock genes, especially AcpPER and AcpTIM, evolving so fast? Has AcpTIM (or AcpPER) been recruited for a different function? Do any of the aphid genes participate in the photoperiodic (seasonal) rhythmicity? We believe that full length sequencing of AcpTim and more thorough analysis of the expression of core clock genes (including in situ hybridizations) are urgently needed to address these questions. Although much work remains, our analysis provides the grounds for subsequent studies aimed at understanding how the clock works in aphids and at elucidating the possible role of circadian clock elements in photoperiodism.

Experimental procedures

  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

Identification of clock gene homologs in Acyrthosiphon pisum

The pea aphid genome assembly Acyr 1.0 (http://www.ncbi.nlm.nih.gov/projects/genome/guide/aphid/) was used to search for homologs of Drosophila melanogaster genes known to have a role in the circadian clock in this organism (Table 1; Fig. 1 for a comprehensive list of the genes). Different BLAST searches (Altschul et al., 1997) were done on Drosophila amino acid sequences, using both NCBI and AphidBase (http://www.aphidbase.com/aphidbase/) resources. When alternative protein products were available for a Drosophila gene, the longest one was chosen. Initially, the searches were carried out against A. pisum RNA (TBLASTN) and protein (BLASTP) NCBI's RefSeq databases, followed by queries against the genome scaffolds available in AphidBase. Additionally we conducted our searches across ab initio and other non-RefSeq prediction databases included within NCBI and AphidBase blasting options. TBLASTN searches were also carried out against unassembled genome sequences. Queries were also implemented against pea aphid EST databases to look for transcript support of gene models. Finally, searches against the WGS traces database available at NCBI were also performed in some instances.

Sequencing of cDNAs from putative Acyrthosiphon pisum clock genes

Genomic sequences and mRNA predictions of A. pisum homologs of Drosophila per, tim, Clk, cyc, Pdp1, vri and cry (including both Drosophila and mammal cryptochrome types) were aligned, which allowed us to design primers for PCR amplification based on exon sequences (Table S1). In addition to primers based on translated portions along the genes, we also designed several primers based on sequences located at various distances upstream of the initiation methionine and downstream of the stop codon. Our primer design allowed us to amplify each cDNA in two or more overlapping fragments that, once assembled using the Staden package (Staden et al., 1998), spanned the whole coding sequence along with partial 5′ and 3′ untranslated regions. Additionally, 3′ RACE, using forward gene specific primers and oligo-dT-T7 primers, were carried out to amplify the 3′UTRs. cDNA templates for PCR amplification were reverse transcribed (High Capacity cDNA reverse transcription kit, Applied Biosystems, Foster City, CA, USA) from total RNA obtained from whole aphids of our YR2 strain (Ultraspec-II RNA isolation system, Biotecx). PCR amplified products (5 PRIME Taq DNA polymerase) were usually sequenced directly after purification (High Pure PCR Product Purification Kit, Roche, Indianapolis, IN, USA) using previously designed primers. In some instances cloning of the PCR amplified fragments was necessary. The pGEM-T easy Vector System (Promega, Madison, WI, USA) was used for this purpose. Because of recurrent amplification problems with the different PCR strategies used, only partial sequences of timeless and vrille aphid orthologs have been characterized (about 50% of the expected coding sequence for both genes).

Our novel cDNA sequences allowed us to check the gene models or predictions available for the 8 sequenced genes. For the rest of the putative clock genes identified (Table 1) the gene models were checked by comparing the predicted amino acid sequences with the orthologous sequences in other insects (see below) and with the ESTs available for some of the genes.

Phylogenetic analyses

To confirm the orthology of the putative clock genes identified in the pea aphid genome, we carried out phylogenetic analyses that included orthologous sequences from other organisms (Table S2). Sequences from insect species with sequenced genome available, representative of different insect orders, were used besides sequences of A. pisum and D. melanogaster (e.g. Bombyx mori, A. mellifera and T. castaneum). We also included in the analysis the sequences from P. humanus, whose genome is currently being sequenced (http://phumanus.vectorbase.org/). Other insect sequences were also used for the phylogenetic analysis of some of the genes when available (Table S2). The orthologous sequences of the water flea Daphnia pulex were generally used as outgroup sequences. The orthologs of the clock genes present in representative vertebrate genomes were also included in most of the phylogenetic analyses. For some genes, the sequences of certain paralogous genes were also included in the phylogenetic analyses (e.g. Timeout and Tango, paralogs of timeless and Clock and cycle, respectively). Other paralogs included in the analysis but not shown in Table S2 are B56, 6-4 photolyase, Casein kinase I α and Gilgamesh, and Archipelago, paralogs of widerborst, cryptochromes, double-time and supernumerary limbs, respectively. To ease the display, most trees shown in the different figures only include insect and Daphnia orthologs.

Amino acid sequences were initially aligned using CLUSTALX v1.81 (Thompson et al., 1997) in combination with MEGA v4.0 (Tamura et al., 2007), and subsequently revised manually. The alignment of the pea aphid protein predictions to other insect orthologs was used in combination with A. pisum gene models and ESTs (when available) to check the predicted gene structure and to identify exon and intron boundaries in the genomic sequence. Usually, the presence of amino acid stretches in A. pisum protein predictions not present in the rest of insect proteins or the lack of regions conserved in the other species were used as indications for likely wrong predictions in the gene models. Phylogenetic analyses were always carried out using only those reliably aligned regions of the genes (Table 1). Overall and pairwise distances among the sequences were calculated using MEGA. The homogeneity of aminoacid composition of sequences was evaluated using TREE-PUZZLE v5.2 (Schmidt et al., 2002). The program RRTree (Robinson-Rechavi & Huchon, 2000) was chosen to test for homogeneity in rates of sequence evolution. Phylogenetic trees were constructed with the neighbor-joining algorithm as implemented in MEGA, using Poisson corrected distances. Finally, node support was evaluated using 1000 bootstrap replicates (Felsenstein, 1985).

Putative functional domains and motifs on the pea aphid clock genes

Several functional domains and motifs are well described for the clock proteins in Drosophila, and other insects as well as in some vertebrates (Saez & Young, 1996; Chang & Reppert, 2003; Chang et al., 2003; Chaves et al., 2006; Hirayama et al., 2003; Hirayama & Sassone-Corsi, 2005; Sakakida et al., 2005; Rubin et al., 2006, among others). The descriptions of relevant protein regions in the literature were used to demarcate these sequences on the putative clock genes in A. pisum. The SMART server (Schultz et al., 1998) was also used to find some motifs and domains. Partial alignments shown in different figures were obtained from full alignment clustal files and manipulated using Jalview (Clamp et al., 2004).

Analysis of expression

For expression analysis, the A. pisum strain LSR1 used for genome sequencing was used. Aphids were kept on broad bean plants as a parthenogenetic reproducing clone under long day conditions (16L:8D, 18 °C). Aphids were maintained at low density conditions (up to 6 aphids per plant) over three generations in order to avoid the formation of winged forms. Aphids from two photoperiod conditions were used for expression analysis. LD aphids were obtained from the usual long day conditions (16L:8D). SD aphids were obtained from the second generation (G1) of aphids reared under short day conditions (12L:12D) that usually produce the sexual response (Cortes et al., 2008). Adult aphids from both conditions were collected and frozen in liquid nitrogen at 4 h intervals (starting at ZT0, when lights went on) along the day–night cycle. Aphids were kept at −80 °C until RNA extraction. Visual inspection of the SD descendants (G2) confirmed that induction of sexuality had occurred. Three different groups of 10 aphids were pooled from each time point and photoperiod condition. Heads were cut and separated from the aphid bodies on a Petri dish on a dry ice layer. Total RNA was extracted from heads and bodies separately using the Ultraspec-II RNA isolation system (Biotecx Laboratories, Houston, TX). RNA was treated for DNase contamination using the DNA-freeTM Kit (Applied Biosystems). One microgram of total RNA was reverse transcribed with random primers using the High Capacity cDNA reverse transcription kit (Applied Biosystems).

Temporal expression of putative clock genes AcpPer, AcpTim, AcpClk, AcpCyc, AcpPdp1, AcpVri, AcpCry1 and AcpCry2 in aphid heads and bodies was analyzed by real-time quantitative PCR using the StepOnePlus™ Real-Time PCR System with the Power SYBR® Green PCR Master Mix (Applied Biosystems). Specific primers were designed in order to ensure amplification of discrete bands with no primer-dimers. (Table S1). Expression along the day–night cycle for both photoperiod regimes was analysed in duplicate. The RpL7 gene was used as endogenous control of constitutive expression (Nakabachi et al., 2005). Relative expression was calculated using the Pfaffl method (Pfaffl, 2001) on values normalized to the arithmetic mean of Ct values across all samples for each gene within a run. MATLAB (The MathWorks, Inc) was used to perform Cosinor analysis (Nelson et al., 1979) to test for rhythmicity on relative expression values obtained as described above. Rhythm parameters were estimated by a least-squares fitting of a cosine function and statistical significance was evaluated by performing zero amplitude tests. Sine wave curves were fitted to 4-h interval data by using XL-fit software (IDBS).

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

We thank BCM-HGCS for making the Acyrthosiphon pisum genome publicly available. We thank Dr Hugh M. Robertson for his help in identifying aphid per and tim homologs. This work was supported by project CGL2007-67392/BOS from the Spanish MEC. TC received a PhD grant from the FPI program, MEC, Spain. The facilities at SCSIE (Universitat de València) were used for sequencing and RT-Q-PCR and for plant growth for aphid rearing.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

Figure S1. Phylogenetic relationships among insect paralogous genes Clock, Cycle and Arnt/Tango. A neighbor-joining topology resulting from the analysis of Poisson-corrected distances is shown with bootstrap values indicated when higher than 50%. For each gene, the Daphnia sequence was also included. The tree obtained shows the true orthology for each of sequences found in the Acyrthosiphon pisum genome. The anomalous positions of the pea aphid sequences, basal to Daphnia and the rest of insects, specially in the Cycle clade, further reflects the accelerated rates for these genes in A. pisum shown in Table 1.

Figure S2. Phylogenetic relationships among animal cryptochromes and the 6-4 Photolyase, including homolog sequences found in Acyrthosiphon pisum. Both Insect CRY1 (Drosophila-type) and Insect CRY2 (mammal-type) are present in the A. pisum genome, with two copies found for the latter (named AcpCRY2-1 and AcpCRY2-2). The phylogenetic inference was done using the neighbour-joining algorithm on Poisson-corrected distances. Bootstrap values are shown for nodes when higher than 50%. The 6-4 Photolyase of Arabidopsis thaliana was used to root the tree.

Table S1. Primers used for PCR amplification, sequencing and Real-time quantitative PCR of the genes characterized in the present study.

Table S2. Accession numbers for sequences included in the phylogenetic analysis.

Please note: Wiley-Blackwell are not responsible for the content or functionality of any supporting materials supplied by the authors. Any queries (other than missing material) should be directed to the corresponding author for the article.

FilenameFormatSizeDescription
IMB_931_sm_suppl_info.doc188KSupporting info item

Please note: Neither the Editors nor Wiley Blackwell are responsible for the content or functionality of any supporting materials supplied by the authors. Any queries (other than missing material) should be directed to the corresponding author for the article.