Comparative analysis of the circadian rhythm genes period and timeless in Culex pipiens Linnaeus, 1758 (Diptera, Culicidae)

Abstract Nucleotide sequences of the circadian rhythm genes, period and timeless, were studied for the first time in mosquitoes Culex pipiens Linnaeus, 1758. In this work we evaluated variations of the studied genome fragments for the two forms of Culex pipiens (forma “pipiens” – mosquitoes common for aboveground habitats, forma “molestus” – underground mosquitoes). We compared Culex pipiens from Russia with transatlantic Culex pipiens and subtropical Culex quinquefasciatus Say, 1823. Our results show that intraspecies variability is higher for the gene period than for the gene timeless. The revealed substitutions in nucleotide sequences and especially in amino acid sequences grouped the individuals of the two forms into distinct clusters with high significance. The detected fixed amino acid substitutions may appear essential for functioning of the circadian rhythm proteins in Culex pipiens, and may be correlated with adaptations of the taxa within the group Culex pipiens. Our results suggest that natural selection favors fixed mutations and the decrease in diversity of the genes period and timeless in mosquitoes of the Culex pipiens f. “molestus” compared with the Culex pipiens f. “pipiens”, is probably correlated with adaptive features of Culex pipiens f. “molestus”. The studied genome regions may be considered as promising molecular-genetic markers for identification, population and phylogenetic analysis of similar species and forms of the Culex pipiens complex.


Introduction
The Culex pipiens Linnaeus, 1758 complex considered by some authors as a 'polytypic species' includes up to seven morphologically identical or very similar forms (Harbach et al. 1984, 1985, Vinogradova 2000).By the second half of the 20 th century, the taxonomic status of these forms changed several times from species to subspecies and back.At present only two species, Culex pipiens Linnaeus, 1758, and Culex quinquefasciatus Say, 1823 have been left within the Culex pipiens complex based on morphological similarity (Harbach 2012).Both species are known as bridge-vectors of West Nile and Saint Louis encephalitis flaviviruses, the etiological agents of dangerous human diseases (Vinogradova 2000).The medical significance of the Culex pipiens complex generates much interest in its studies, including taxonomy.
Only one species of the Culex pipiens complex, C. pipiens, has been found in Russia.This species includes two forms, C. pipiens f. "pipiens" and C. pipiens f. "molestus", originally described as distinct species (Harbach et al. 1984(Harbach et al. , 1985)).C. pipiens forms designation is provided in accordance with the rules of International Code of Zoological Nomenclature (http://www.iczn.org/iczn/index.jsp).The two forms are morphologically identical, but have notably distinct biological features.The mosquitoes C. pipiens f. "pipiens" are anautogenous (females require a blood meal to mature each egg raft), mate in swarms, oviposit in a wide variety of natural and manmade habitats, feed preferentially on avian hosts and enter diapause to overwinter (Vinogradova 2000).In contrast C. pipiens f. "molestus" are autogenous (females oviposit the first egg raft without bloodmeal), develop without winter diapause in urban flooded basements and tunnels, feed preferentially on mammal hosts and are able to mate in a confined space.The specific features of reproduction and development of the two forms has resulted in their spatial isolation in moderate climate areas, suggesting genetic isolation.This suggestion is confirmed by the isoenzyme analysis of autogenous and anautogenous populations of C. pipiens from England (Byrne and Nichols 1999), Russia (Lopatin 2000) and Germany (Weitzel et al. 2009) as well as by study of populations from Europe with CQ11 assay (Bahnck and Fonseca 2006).The results of these investigations showed that in these regions the forms are genetically distinct, with no or poor gene flow between populations of different forms.However, in the Mediterranean area, in N Africa and the Middle East, both autogenous and anautogenous specimens develop in the same pools.These populations display highly variable autogeny rates, from 10-90% in Egypt (Gad et al. 1995) to 4-55% in Israel (Nudelman et al. 1988), and both autogenous and anautogenous females were encountered in the progenies of autogenous or anautogenous female parents (Gad et al. 1995).Consequently, the question of divergence of the two forms in moderate climates remains still unclear.
Among the specific behavioral/physiological traits which remained up to now the important criteria for defining populations of C. pipiens f. "pipiens" and C. pipiens f. "molestus", differences in mating behavior are under the special interest.Mating ac-tivity of C. pipiens f. "pipiens" is restricted within the crepuscular period when males aggregate in swarms where they copulate with virgin females attracted to a swarm (Ivanov 1984, Fyodorova and Serbenyuk 1999, Vinogradova 2000).In contrast, males of C. pipiens f. "molestus" never swarm and have irregular locomotor and mating activity (Shinkawa et al. 1994).Such temporal differences in mating activity may represent the temporal isolation between two forms.
Clock genes, especially period and timeless, play an essential role in regulation of mating rhythms in insects.In Drosophila, null mutants of the clock gene period (per) lost the circadian rhythm in mating activity (Sakai and Ishida 2001).Similar effects have been described for gene timeless (tim) in Drosophila and for gene per in Grillus bimaculatus De Geer, 1773 (Sehgal et al. 1994, Moriyama et al. 2008).The analysis of mating activity in transformant lines carrying per transcription units derived from Drosophila melanogaster Meigen, 1930or Drosophila pseudoobscura Frolova & Astaurov, 1929, showed that per controls species-specific mating rhythms, at least in flies (Tauber et al. 2003).
It may be suggested that differences in the rhythm of mating activity in two forms of C. pipiens resulted from the variations in circadian clocks genes.To test this hypothesis, we selected the genes per and tim.The aim of our work was to study variable nucleotide sequences in these genes, and to estimate the possible evolutionary significance of the detected variations.

Methods
The larvae of mosquitoes of both intraspecific forms were collected mostly in August 2006 in Volgograd City and nearby areas.The sampling sites, methods of larvae collection and rearing in lab, and methods of evaluating autogenity have been described earlier (Fedorova and Shaikevich 2013).The DNA of mosquitoes collected in the underground sampling sites in Nizhny Novgorod, Moscow and St Petersburg, as well as in aboveground sampling site Iksha, Moscow region, was used to analyze the diversity of the first exon of the gene tim.The methods of mosquito sampling at these sites have been described earlier (Vinogradova and Shaikevich 2007).

DNA isolation and analysis
The DNA was isolated using the kit DIAtom™ DNA Prep (Isogen Russia).Each of the amplification reactions used 0.1 μg of the total DNA.The polymerase chain reaction (PCR) was run on the thermocycler GeneAmpR PCR System 2700 (Applied Biosystems USA), with amplification Encyclo PCR kit (Evrogen Russia), following the manufacturer's instructions.For PCR and sequencing of amplification products, specific primers were constructed which were complementary to the conserved sequences of exons in the published sequences of the genes period and timeless from the total genome of a similar species C. quinquefasciatus (Vector Base Gene ID CPIJ007193 and CPIJ007082, respectively) (Arensburger et al. 2010).When the first sequences were obtained, the primers were constructed basing on DNA sequences of C. pipiens.The PCR conditions were adjusted using the program Oligo6 (http://www.oligo.net/): primary denaturing 95°C -5 min; 35 cycles at 95°C -30 s, Tm (for each primer pair) -1 min, 72°C -1,5 min; final synthesis at 72°C for 7 min.Primer sequences and annealing temperatures for the PCR are shown in Table 1.Higher temperature was used if two primers in the pair had different annealing temperatures.Negative control was run for all amplification reactions.The DNA of introns was analysed by direct sequencing of amplicons without cloning.Amplified fragments of the genes per and tim were purified from the gel using QIAquick Gel Extraction kit (Qiagen USA).The fragments were cloned using the kit pGEM-T Easy Vector Systems (Promega USA); the DNA of the three clones for each individual mosquito was sequenced using the equipment ABI PRISM 310 and the BigDye Termination kit (Applied Biosystems USA), according to the manufacturer's instructions and deposited to GenBank under accession numbers: KU133680-KU133745.The sequences of separate exons of each clone were combined into a single sequence.Nine combined sequences from individual C. pipiens f. "pipiens" and nine combined sequences from individual C. pipiens f. "molestus" were investigated for each of the two genes studied, per and tim.Extended study of the coding sequences of exon 1 of the gene tim in two forms of C. pipiens was performed using the DNA from the 26 individual mosquitoes C. pipiens f. "molestus" and 17 individual mosquitoes C. pipiens f. "pipiens".21 new different haplotypes are submitted to GenBank (KU997646 -KU997666).

Data analysis
The DNA sequences were translated into amino acids sequences using ExPASy software (Swiss Institute of Bioinformatics), and compared with amino acids sequences of C. quinquefasciatus (Arensburger et al. 2010) and C. pipiens from the USA (Meuti et al. 2015) using programs MAFFT (http://mafft.cbrc.jp/alignment/server/)and MEGA6 (Tamura et al. 2013).Evolutionary analysis was run using MEGA6.Maximum Composite Likelihood model (Tamura et al. 2004) and Kimura 2-parameter model (Kimura 1980) were used to describe the nucleotide substitution pattern.Tables below show the data obtained using the Maximum Composite Likelihood model.The Kimura 2-parameter model produced somewhat higher estimates.The optimal model describing evolutionary patterns was found using the option 'Find best DNA/Protein substitution model' in MEGA6.For our data, the Jones-Taylor-Thornton (JTT) model (Jones et al. 1992) showed the lowest BIC (Bayesian Information Criterion) scores for amino acid sequences and was selected to describe the amino acids substitution pattern.For estimating polymorphism within each group and evolutionary divergence between each group the number of base substitutions per site from averaging over all sequence pairs was calculated, all positions containing gaps and missing data were eliminated.Codon positions included were 1st+2nd+3rd+Noncoding.For the estimation of Maximum Likelihood Estimate of Transition/Transversion Bias (R) substitution pattern and rates were estimated under the Kimura 2-parameter model.Phylogeny analysis was run in MEGA6.The evolutionary history was inferred using the Neighbor-Joining method.The percentage of replicate trees in which the associated taxa clustered together in the bootstrap test (1000 replicates) is shown next to the branches.All ambiguous positions were removed for each sequence pair.
Natural selection and the probability of rejecting the null hypothesis of strictneutrality (dN = dS) was evaluated using MEGA6.For these purposes was used a codon-based Z-test (MEGA6).For a pair of sequences, this is done by first estimating the number of synonymous substitutions per synonymous site (dS) and the number of nonsynonymous substitutions per nonsynonymous site (dN), and their variances: Var(dS) and Var(dN), respectively.With this information, we tested the null hypothesis that there is no impact of selection (dN = dS) and the probability (P) of rejecting the null hypothesis of strict-neutrality.Also was tested an alternative hypothesis of pu-rifying selection (dN < dS) and the probability of rejecting the null hypothesis of strictneutrality in favor of the alternative hypothesis using a codon-based Z-test (MEGA6).Values of P determine statistical significance in a hypothesis test.A low P value suggests that sample provides enough evidence for the rejecting of the null hypothesis for the entire population.Values of P less than 0.05 are considered significant at the 5% level.The variance of the difference was computed using the analytical method (Kimura 1980).All ambiguous positions were removed for each sequence pair.

The gene period (per) in two forms of C. pipiens
The structure of the gene per was studied in three individual C. pipiens f. "molestus" and in three individual C. pipiens f. "pipiens".Coding sequences of the three exons of the gene per were analysed: exon 2, 333 bp, exon 3, 738 bp, and exon 4, 1229 bp.In total, the 18 compared sequences spanned each 2300 bp (Suppl.material 1).
In the exon 2 of the gene per (333 bp) 11 variable sites were found; six of these substitutions resulted in amino acid substitutions in both forms of C. pipiens (Fig. 1).The exon 3 (738 bp) had 12 variable nucleotide sites, resulting in three amino acid substitutions (Fig. 1).The exon 4 (1229 bp) had 27 variable nucleotide sites, resulting in four amino acid substitutions (Fig. 1).In total, the nucleotide sequence of the three exons of the gene per for the both intraspecific forms had 50 (2.2%)variable nucleotide sites and 13 (1.7%)polymorphic amino acid sites, 48 nucleotide sites being parsimony-informative.The estimated Transition/Transversion bias (R) is 2.83.The DNA polymorphism of the gene per among individuals of the C. pipiens f. "pipiens" (0.003) and of the C. pipiens f. "molestus" (0.002) were both low, variability of the amino acid sequences also was low (Table 2).The genetic distances between two forms of C. pipiens from Volgograd were 0.010 based on nucleotide sequences and 0.011 based on amino acid sequences of the gene per (Table 2).

Comparison of the gene per for transatlantic C. pipiens
The obtained sequences of the gene per of C. pipiens from Volgograd and C. pipiens f. "pipiens" from the USA (GenBank acc.number KM355980) using BLAST software were compared.The identity of nucleotide sequences of C. pipiens f. "pipiens" mosquitoes from different continents is 98-99%; 4-16 amino acid substitutions were detected.Pairwise comparison showed that C. pipiens f. "pipiens" from the USA is slightly different from the Volgograd C. pipiens f. "pipiens" (0.008) and from C. pipiens f. "molestus" (0.013); these values are comparable with the differences between the studied C. pipiens f. "pipiens" and C. pipiens f. "molestus" (Table 2).In upper right section in bold: the number of nucleotide base substitutions (NA) per site from averaging over all sequence pairs between groups is shown.All results are based on the pairwise analysis of 20 sequences.There were a total of 2300 positions of per gene and 1560 positions of tim gene in the final dataset.In lower left section: the number of amino acid substitutions (AA) per site from averaging over all sequence pairs between groups are shown.The analysis involved 20 amino acid sequences.A total of 766 positions of the gene per and 520 positions of the gene tim were analysed as the final dataset.

Comparison of the gene per in C. pipiens and C. quinquefasciatus
The identity of the nucleotide sequences of the gene per for the two species was 97%.
Comparison of the DNA from both forms of C. pipiens and C. quinquefasciatus (CPIJ007193) revealed 113-116 (4.9-5.5%)variable nucleotide sites (Suppl.material 1): 37 nucleotide substitutions were non-synonymous, resulting in amino acid substi-tutions (Fig. 1).64 nucleotide substitutions and 24 amino acid substitutions are specific for C. quinquefasciatus, with nine substitutions in each of the exons 2 and 3, and six in exon 4 (Fig. 1).The mean genetic divergence between C. pipiens and C. quinquefasciatus is 0.03 by both DNA and amino acid sequences.The difference between C. pipiens and C. quinquefasciatus is three times higher than the difference between the two forms of C. pipiens (Table 2).
In exon 1 of the gene tim (1037 bp) 15 variable nucleotide sites and five variable amino acid sites were found, four of them showing variations only for the C. pipiens f. "pipiens" and were not found in C. pipiens f. "molestus" (Fig. 2).In exon 5 (379 bp) five DNA substitutions were found, and in exon 6 (145 bp) there were three variable nucleotide sites; all substitutions in the exons 5 and 6 were synonymous, resulting in similar amino acid sequences for C. pipiens f. "pipiens" and C. pipiens f. "molestus" (Suppl.material 2, Fig. 2).The estimated Transition/Transversion bias (R) is 2.79.
Comparing the nucleotide sequences of the gene tim for the specimens of C. pipiens f. "molestus" one variable DNA site was found, the detected nucleotide substitution does not result in amino acid substitution.The aligned DNA sequences of the C. pipiens f. "pipiens" had 11 variable sites, one mutation resulting in amino acid substitution (Fig. 2).The DNA polymorphism of the gene tim among specimens of the C. pipiens f. "pipiens" (0.0038) was higher than for C. pipiens f. "molestus" (0.0004), and variability of the amino acid sequences was 0.001 and 0.000, respectively.Comparing the total sequence of the three exons of the gene tim, between C. pipiens f. "pipiens" and C. pipiens f. "molestus" 23 (1.5%) variable nucleotide sites were found (all 23 sites were parsimonyinformative) and six (0.4%) polymorphic amino acid sites.Genetic distance between the two forms was 0.012 for DNA sequences and 0.008 for amino acid sequences (Table 2).

Comparison of the gene tim for the transatlantic C. pipiens
The obtained sequences of the gene tim for C. pipiens from Volgograd and C. pipiens f. "pipiens" from the USA (KM355979) were compared using BLAST software.Identity of nucleotide sequences for mosquitoes of C. pipiens f. "pipiens" from different continents is 96-97%.We found 7-12 amino acid substitutions.Unexpectedly, we found a 60-bp deletion within the coding sequence of exon 1 in C. pipiens f. "pipiens" from the USA (KM355979), positions 263-282 in Fig. 2. No similar deletion was found in either of the studied C. pipiens forms from Volgograd and no similar deletions were found in C. quinquefasciatus (CPIJ007082).The genetic distance between C. pipiens f. "pipiens" from Volgograd and C. pipiens f. "pipiens" from the USA (KM355979) is 0.025, two times higher than the distance between both forms of C. pipiens in Volgograd 0.012 (Table 2).

Comparison of the gene tim from C. pipiens and C. quinquefasciatus
Comparison of DNA sequences of exons 1, 5 and 6 of the gene tim between representatives of the two species, C. pipiens and C. quinquefasciatus, revealed 50 variable sites (see Suppl.material 2), which result in 8-14 amino acid substitutions, eight of which are found only in C. quinquefasciatus (Fig. 2).The genetic distance between the species is 0.029 in the DNA sequences and 0.019 in amino acid sequences (Table 2).A striking similarity should be noted for the gene tim from C. quinquefasciatus and C. pipiens f. "pipiens" from the USA (KM355979).Their amino acid sequences have only three variable sites: 136, 280, and 375 (Fig. 2), their DNA sequences differ in 13 single-nucleotide substitutions and one deletion.The genetic distance for the gene tim between C. quinquefasciatus and C. pipiens f. "pipiens" from the USA (KM355979) is 0.009 based on DNA sequences and 0.004 based on amino acid sequences, lower that the distance between the two forms of C. pipiens in Volgograd (Table 2).

Extended study of exon 1 of the gene tim in two forms of C. pipiens
Our results showed that exon 1 of the gene tim in C. pipiens f. "molestus" differ from that of in C. pipiens f. "pipiens" (Fig. 2).Contrary to the gene per, no shared polymorphisms were found in amino acid sequences of gene tim between two forms (Figs 1, 2).To confirm these findings we studied the structure of exon 1 (1037 bp) of the gene tim in 23 specimens of C. pipiens f. "molestus" and 14 specimens of C. pipiens f. "pipiens" in addition to 6 samples of gene tim described above.In total, 43 samples were examined.
The obtained nucleotide sequences showed overlapping peaks in one or more sites for 6 individuals.Four of them were identified as C. pipiens f. "molestus" and two as C. pipiens f. "pipiens".Exon 1 of the gene tim of these six samples was studied by cloning and the DNA of the five clones for each specimen was sequenced.In total 79 sequences were obtained for comparative analysis (Suppl.material 3).In five specimens one allele was identical to C. pipiens f. "pipiens" and other one was identical to C. pipiens f. "molestus", i. e. these mosquitoes represented hybrids.In one C. pipiens f. "molestus" (NN23) the alleles differed by two nucleotide substitutions in 3'end.All hybrids were collected in Volgograd, where both forms develop in the same pools in summer.
The DNA polymorphism of the exon 1 of gene tim among specimens of the C. pipiens f. "pipiens" (0.007) was ten times higher than for the C. pipiens f. "molestus" (0.0006), and variability of the amino acid sequences was 0.0053 and 0.0001, respectively.Genetic distance between the two forms was 0.011 for DNA sequences and 0.009 for amino acid sequences.Genetic distance between C. pipiens of both forms and C. quinquefasciatus was 0.029 for DNA and 0.02 for amino acid sequences (Suppl.material 3).The DNA polymorphism, as well as genetic distances between the two forms in extended study of the exon 1 are very close to the results obtained for the three exons of the gene tim (see above) (Table 2).

Variation in non-coding regions of the gene tim
The sequences of some non-coding regions were analysed, expecting to find differences not only in coding DNA structure but also in intron size between C. pipiens f. "pipiens" and C. pipiens f. "molestus".The primers were constructed for the conserved sites of the exons using the obtained sequences, and by homology with the gene tim from C. quinquefasciatus (CPIJ007082).The sequences of the introns 1-2 (7158 bp in length) and 10-11 (5189 bp), being too long for efficient PCR and sequencing and containing numerous repeats were not analysed.As for the other introns, sequencing of the PCR products showed no variability between two intraspecific forms in intron between exons 5 and 6 (59 bp).In intron 6-7 (61 bp) three variable sites and in intron 7-8 (160 bp) six variable sites were found.Studied introns showed no mutations common with either of the two C. pipiens forms.In the intron 9-10 (167 bp) seven variable sites were found, six of which differed between the two forms (Suppl.material 4).The length of all amplified intron sequences was identical for C. pipiens f. "pipiens" and C. pipiens f. "molestus".

Phylogenetic analysis
Phylogenetic dendrograms were constructed applying the Neighbor-Joining method to amino acid sequences of the three coding regions of genes per and tim, C. quinquefasciatus was used as an out-group.C. quinquefasciatus and C. pipiens form two well differentiated clusters.Basing on similarity of the gene per the individuals of the C. pipiens f. "pipiens" group together and form a joint cluster with a bootstrap coefficient of 97.The studied specimens of the C. pipiens f. "molestus" had more polymorphic amino acid sequences, but also are grouped into one cluster with bootstrap coefficient of 63 (Fig. 4).Based on the similarity of the gene tim, C. pipiens f. "pipiens" and C. pipiens f. "molestus" group into separate clusters with a bootstrap coefficient of 88 (Fig. 4B).
On the dendrogram for the exon 1 of the gene tim, constructed using the results of our extended study, most specimens of the C. pipiens f. "molestus" form separate clusters with a bootstrap coefficient of 96.A separate subcluster is formed by sequences of the hybrid V219 clones with haplotype H4.The studied specimens of the C. pipiens f. "pipiens" have polymorphic DNA sequences (Suppl.material 3).The dendrogram basing on amino acid sequences shows similar configuration.

Evolutionary analysis
One way to test whether natural selection is operating on a gene is to compare the relative abundance of synonymous and nonsynonymous substitutions within the gene sequences (Tamura et al. 2013).Analysing evolution of the nucleotide sequences, the Codon-based Test of Neutrality rejected the null hypothesis of strict-neutrality with strong statistical support in both genes (Table 4).Though comparison of some haplotypes within the C. pipiens f. "pipiens" also shows deviation from neutrality, difference between the forms is considerably higher (Suppl.materials 5, 6).Analysis of dN-dS between the per nucleotide sequences of both intraspecific forms indicates that the probability of rejecting the null hypothesis of strict-neutrality ranges from 0 to 0.015 across the sequences with an overall average of 0.003.Between tim nucleotide sequences of the three exons of C. pipiens f. "pipiens" and C. pipiens f. "molestus", the probability of rejecting the null hypothesis of strict-neutrality ranges from 0.003 to 0.017 across the specimens with an overall average of 0.006.Table 4 shows average mean dN-dS and of the probability of rejecting the null hypothesis of strict-neutrality for each individual.
Analysis of dN-dS between the 79 nucleotide sequences of exon 1 of the gene tim indicates that the probability of rejecting the null hypothesis of strict-neutrality between intraspecific forms ranges from 0.002 to 0.15 across the sequences with an overall average of 0.05.The number of synonymous substitutions per site (dS) was higher that the number of non-synonymous substitutions per site (dN), indicating Purifying Selection.The probability of rejecting the null hypothesis of strict-neutrality (dN = dS) in favor of the alternative hypothesis of Purifying Selection (dN < dS) ranges for the tim nucleotide sequences of C. pipiens f. "pipiens" and C. pipiens f. "molestus" from 0.001 to 0.10 with an overall average of 0.025 (Suppl.material 7).

Discussion
For the first time the genetic structure of the circadian rhythm genes (per and tim) were analysed for mosquitoes C. pipiens f. "molestus".Our results have shown that DNA variation in individuals of C. pipiens f. "molestus" is smaller than in individuals of C. pipiens f. "pipiens".Extended study of exon 1 of the gene tim revealed 4 DNA haplotypes in C. pipiens f. "molestus" and 19 haplotypes in C. pipiens f. "pipiens".Decrease in DNA variability for the underground mosquitoes of C. pipiens f. "molestus" was also reported earlier in our study of mitochondrial DNA (Shaikevich and Zakharov 2010).In coding sequences of both genes per and tim, variations between physiologically different forms of C. pipiens were found (Table 3).In the gene per we found nine polymorphisms shared between the two forms and four fixed differences between the two forms, taking into account C. pipiens f. "pipiens" from N America (Fig. 1).The gene tim had one shared amino acid polymorphisms and one fixed difference between the forms (Fig. 3).Higher variation of the gene per is also revealed by comparison of C. pipiens and C. quinquefasciatus: basing on the amino acid sequences, the genetic distances between the species are higher for the gene per (0.036) that for the gene of tim (0.02).
C. pipiens f. "pipiens" from N America clusters with C. pipiens f. "pipiens" from Volgograd basing on comparison of the gene per and with C. quinquefasciatus based on comparison of the gene tim.It remains unknown whether this is common for all American C. pipiens f. "pipiens", shown using microsatellite analysis to differ from the European C. pipiens f. "pipiens" (Fonseca et al. 2004), or if it is a specific feature of the laboratory line, used to study the genes on circadian rhythm (Meuti et al. 2015).
Genetic structure of the studied genes is polymorphic.However, the revealed substitutions in nucleotide sequences and especially in protein sequences grouped the individuals of the two forms into distinct clusters with high significance, a longer genetic distance separating the cluster of C. pipiens from C. quinquefasciatus.Although the two studied genes differed in variability, the results of analysis of the gene per, as well as the gene tim, show that the difference between C. pipiens and C. quinquefasciatus are 2.5-3 times higher than the difference between the forms of C. pipiens.The genetic distances again confirm the order of evolutionary events in the C. pipiens complex: the divergence of the form C. pipiens f. "molestus" from C. pipiens occurred considerably later than the divergence of C. pipiens and C. quinquefasciatus (Barr 1967, Fonseca et al. 2004, Shaikevich and Zakharov 2014).
The non-coding genome sequences are considered to be highly variable.These sequences are often used to search for the markers to differentiate closely related organisms by size of the PCR products.For example, variation in spacers of the ribosomal genes cluster is a base for identification of some mosquito species of the genus Anopheles (Nicolescu et al. 2004, Gordeev et al. 2004).In sequences of three tim introns no significant difference was found between the forms.For Aedes albopictus Skuse, 1894, also no significant difference in the introns of the gene tim was reported (Summa et al. 2012).
The Test of Neutrality rejects the null hypothesis of strict-neutrality at P < 5% level and imply that both per and tim loci evolve under strong selective constraint during the divergence of intraspecific forms.Our results suggest that natural selection favored the fixed mutations and the decreased diversity of the genes per and tim in mosquitoes C. pipiens f. "molestus" compared with the C. pipiens f. "pipiens", probably preserving adaptive features of the form "molestus". Well-documented data have been reported showing that new native mutations sometimes are rapidly spreading in a population and that polymorphism in one locus may provide adaptive variations in behavioral and morphological phenotypes of the insects in nature (Tauber et al. 2007).The genes involved in circadian rhythms are proved to coordinate seasonal responses, e.g. they initiate the reproductive diapause; malfunctioning of the genes per and tim was shown to interrupt diapausing of the C. pipiens females (Meuti et al. 2015).We can assume that mutations found in per and especially in tim genes are related with functioning of the circadian rhythm proteins and contributed to divergence of the forms of C. pipiens.
The studied genes are promising candidates to evaluate the genetic basis of different behaviors of the two ecological forms within one subspecies.Further studies of the circadian rhythm genes in mosquitoes of the Culex pipiens complex would help to test this assumption.

Conclusions
Nucleotide sequences of the circadian rhythm genes were studied for the first time in mosquitoes C. pipiens f. "molestus" and compared with those for C. pipiens f. "pipiens" and C. quinquefasciatus.These results show that intraspecies variability is higher for the gene per than for the gene tim.Revealed substitutions in nucleotide sequences and especially in protein sequences grouped the individuals of the two ecological forms of C. pipiens into distinct clusters with high significance.The results suggest that natural selection favored the fixed mutations and the decreased diversity of the genes per and tim in mosquitoes of the C. pipiens f. "molestus" compared with the C. pipiens f. "pipiens".The detected fixed amino acid substitutions may appear essential for functioning of the circadian rhythm proteins in C. pipiens, and may be related with adaptations of the taxa within the group C. pipiens.Moreover, under natural selection mutations in the key genes of circadian pattern may provide some advantage to the underground C. pipiens f. "molestus".The studied genome regions may be considered as promising molecular-genetic markers for identification, population and phylogenetic analysis of similar species and forms of the C. pipiens complex.

Figure 1 .
Figure 1.Variable amino acid sites of the gene period in C. pipiens.C. pipiens from the USA (KM355980) and C. quinquefasciatus (CPIJ007193) are taken for the comparison.Exons 2 (sites 1-111), 3 (113-358), and 4 (360-768) are separated with blank columns.Positions of the variable sites relative to combined sequences as presented in Suppl.material 1 shown on the top.

Figure 2 .
Figure 2. Variations of amino acid sites in the gene tim from C. pipiens.C.pipiens from the USA (KM355979), and C.quinquefasciatus (CPIJ007082) are taken for the comparison.Dashes show deletion in exon 1 in C. pipiens from the USA (KM355979).Exon 1 (sites 1-345) and exon 5 (sites 347-472) are separated by blank columns.Positions of the variable sites in combined tim sequences shown on the top.

Figure 3 .
Figure 3.DNA haplotypes and variable amino acid positions in the exon 1 of the gene tim from C. pipiens f. "pipiens" and C. pipiens f. "molestus".Haplotypes numbers and variable nucleotide sites are shown on the left.Variable amino acid sites are shown on the right.Only variable haplotypes are shown, all 79 sequences are presented in Suppl.material 3. Positions of the variable sites shown on the top.Dash show deletion of 12 nucleotides (sites 831-842) in exon 1 in C.pipiens f. "pipiens" from Moscow region.

Figure 4 .
Figure 4. Evolutionary relationships of the studied taxa.Neighbor-joining trees of C. pipiens based on A period and B timeless inferred amino acid sequences with the C. quinquefasciatus (CPIJ007193) as the outgroup.Percent bootstrap support based on 1000 replicates.Seven amino acid sequences were analysed with a total of 766 positions of PERIOD (A) and 519 positions of TIMELESS (B) in the final datasets.

Table 1 .
Primers constructed to study the genes per and tim.

Table 2 .
Estimates of Evolutionary Divergence over per and tim sequence pairs between Culex pipiens complex members.

Table 4 .
Codon-based Test of Neutrality between C. pipiens f. "pipiens" and C. pipiens f. "molestus".The test statistic (dN -dS) is shown above the diagonal.The probability of rejecting the null hypothesis of strict-neutrality (dN = dS) is shown.Values of P less than 0.01 are considered significant at the 1% level.There was a total of 766 positions of gene per and of 519 positions of gene tim in the final dataset.Evolutionary analyses were conducted in MEGA6.