Extreme Conservation and Non-Neutral Evolution of the cpmA Circadian Locus in a Globally Distributed Chroococcidiopsis sp. from Naturally Stressful Habitats

Abstract

Cyanobacteria are among the most ancient organisms known to have circadian rhythms. The cpmA gene is involved in controlling the circadian output signal. We studied polymorphism and divergence of this gene in six populations of a stress-tolerant cyanobacterium, Chroococcidiopsis sp., sampled in extreme habitats across the globe. Despite high haplotype diversity (0.774), nucleotide diversity of cpmA is very low (π = 0.0034): the gene appears to be even more conserved than housekeeping genes. Even though the populations were sampled thousands kilometers apart, they manifested virtually no genetic differentiation at this locus (F_ST = 0.0228). Using various tests for neutrality, we determined that evolution of cpmA significantly departures from the neutral model and is governed by episodic positive selection.

Introduction

The circadian system controls the expression of many genes in a genome and thus is essential for maintaining cellular homeostasis. Circadian rhythms have been observed in cyanobacteria since the 1990s, making them one of the simplest known organisms to possess circadian rhythmicity (Kondo et al. 1993). This circadian programming can enhance the ability of an organism to anticipate important cyclic changes in the environment and generate appropriate responses (Woelfle et al. 2004). Cyanobacteria are a major model system in prokaryotes for analyzing and testing predictions about circadian rhythms and possible effects of evolutionary forces on them (Mackey et al. 2011).

Having an internal clock that can match the external light/dark cycle is thought to be advantageous for cyanobacteria as photosynthetic organisms (Johnson et al. 1998; Ouyang et al. 1998). Cyanobacteria are quite versatile and have a high adaptive potential that allows them to occupy the most extreme habitats on our planet (Whitton 1987). Given its significance for controlling essential physiological processes, such as cell division, regulation of nitrogen fixation, and photosynthesis (Johnson and Golden 1999), the circadian system is important for adaptability of cyanobacteria.

Macroevolutionary studies of various circadian genes showed that their evolution was shaped by many factors, including duplications, lateral transfers, and various types of selection (Dvornyk et al. 2003; Dvornyk 2006, 2009; Baca et al. 2010). However, very little is known about the factors, which govern evolution of circadian genes at the population level. Previously, we reported multiple duplications and an elevated neutral mutation rate in the core circadian genes, kaiB and kaiC, from a filamentous cyanobacterium Nostoc linckia under acute environmental stress (Dvornyk et al. 2002). Apart from this, no other reports are available.

The cpmA (circadian phase modifier) gene is an element of the output pathway of the circadian clock (Katayama et al. 1999). Mutations in the cpmA gene alter the phasing of the circadian rhythm of a restricted subset of genes, which results in severe growth defects (Katayama et al. 1999). The gene was reported to consist of two domains of similar length (N- and C-terminal domains), which harbor several highly conserved regions of potential circadian function (Dvornyk 2006). The C-terminal domain shares partial homology with the N-terminus of the AIR carboxylase (AIRC) (pfam00731) and the NCAIR mutase (COG0041) (Dvornyk 2006).

Chroococcidiopsis is a coccoid unicellular cyanobacterium first reported from hot deserts (Fewer et al. 2002). Chroococcidiopsis is not only capable of surviving extreme desiccation but also shows a remarkable resistance to high doses of ultraviolet radiation (Cordoba-Jabonero et al. 2005). Resemblance of morphological characteristics of Chroococcidiopsis to proterozoic microfossils suggests it as one of the evolutionary oldest cyanobacteria (Friedmann 1980; Friedmann and Ocampo-Friedmann 1995).

This study was designed to examine polymorphism and diversity of the cpmA gene in the Chroococcidiopsis strains sampled from various extremely stressful habitats around the Earth and to determine factors that govern microevolution of this gene.

Materials and Methods

Environmental Samples and Cyanobacterial Strains

Environmental samples were collected from four locations representative of extreme arid climate regimes (Bahl et al. 2011; Caruso et al. 2011). Major environmental parameters and geographical coordinates are given in table 1. To standardize the nature of samples collected, hypoliths were targeted. These comprise biofilms that colonize the ventral surface of quartz, a relatively inert, and ubiquitous substrate in deserts worldwide. Hypoliths develop independently of the surrounding soil and are dominated by cyanobacteria (Pointing et al. 2009). The sampling sites were Antarctic Dry Valleys (McKelvey Valley), Canadian Arctic (Devon Island), cold desert (Taklimakan, China), and hot desert (Kalahari, Southern Africa). Climate was delineated according to the standards based on the long-term mean annual precipitation and temperature (Peel et al. 2007).

Table 1.

Characteristics of the Four Sampling Sites Used in This Study.

Sampling Sites	Population Name	Coordinates	Environmental Conditions at the Site
McKelvey Valley, Antarctica	AC (chasmolith)	S77°24.595′ E161°11.747′	Strong winds
	AE (endolith)		Ice cap deposits
	AH (hypolith)		Mean annual rainfall <100 mm
			Average yearly temperature −20°C to −35°C
Canadian Arctic Circle, Devon Island	CH (hypolith)	N75°23.224′ W89°40.335′	High levels of solar radiation
			Variation in salinity
			Mean annual rainfall <500 mm
			Average yearly temperature less than −5°C
Taklimakan Desert, China	TH (hypolith)	N38°24.233′ E88°53.806′	Large fluctuation in temperature
			Shifting sand
			Salt accumulation
			Mean annual rainfall <25 mm
			Average temperature less than −20°C (winter) and <40°C (summer)
Kalahari Desert, Botswana	SH (hypolith)	S 22°11.0′ E29°7.0′	Dry air
			Shifting sands
			Large fluctuation in temperature
			Mean annual rainfall <200 mm
			Average yearly temperature more than −11°C (winter) and >40°C (summer)

The collected environmental samples were subcultured to establish cyanobacterial cultures containing predominantly Chroococcidiopsis sp. The cultures were maintained on BG-11 medium in conical flasks at 25°C in sterilized incubators with constant light. In addition to the environmental cultures, a strain of Chroococcidiopsis ATCC 27 900 was used in the analyses for identification purposes and comparisons.

DNA Extraction, Amplification, Cloning, and Sequencing

Genomic DNA was isolated by the cetyltrimethylammonium bromide DNA extraction protocol with minor modifications (Stewart and Via 1993).

The polymerase chain reaction (PCR) primers were designed using Net Primer (http://www.premierbiosoft.com/netprimer) and the cpmA nucleotide sequences of the following cyanobacterial species: Acaryochloris marina MBIC11017 (GenBank accession no. NC_009925, locus tag: AM1_4350), Trichodesmium erythraeum IMS101 (NC_008312, locus tag: Tery_4722), Microcystis aeruginosa NIES-843 (NC_010296, locus tag: MAE_62660), Thermosynechococcus elongatus BP-1 (NC_004113, locus tag: tll1189), and Nostoc sp. PCC 7120 (BA000019, locus tag: alr3885).

The PCR was conducted using forward primer 5 F (5′-ACCGGATTTCCCGAAGTGATTTGG-3′) and reverse primer 4 R (5′-GCCGCACCAAATCCATTATC-3′). The master mix of 50 µl contained 38.5 µl of water, 5 µl of 10× RED Taq buffer (Sigma-Aldrich), 2 µl of dNTPs (Sigma-Aldrich), 1 µl of each primer, 0.3 µl of Red Taq polymerase (Sigma-Aldrich), 2 µl of MgCl₂, and 1.2 µl of Tween-20 were used. The PCR was conducted using the following profile: denaturation at 92°C for 1 min, annealing at 50°C, and extension at 72°C both for 2 min with 40 cycles. The PCR products were then purified using a Qiagen purification kit and cloned into a pDrive cloning vector (Qiagen, Inc.) following the manufacturer’s protocol.

The cloned gene fragments were isolated and sequenced using an automated sequencer (3730xl DNA Analyzer, Applied Biosystems) in Genome Research Center (University of Hong Kong) using the dye primer method (Applied Biosystems).

Data Analyses

The sequence chromatograms were proofread and trimmed for unresolved bases. The obtained nucleotide sequences were aligned using ClustalW (Thompson et al. 1994) as implemented in the BioEdit software v. 7.0.9 (Hall 1999).

Phylogenetic Analysis and Taxonomical Identification of Sequences

The likelihood ratio test as implemented in the ProtTest 3.0 software (Darriba et al. 2011) was used to determine the most appropriate substitution model for the data set. According to the Bayesian information criterion, the LG model (Le and Gascuel 2008) with gamma distribution (α = 0.55) turned out to fit our data best.

Because the cpmA gene was amplified from enriched but nonaxenic cultures, it was necessary to verify which of the obtained cpmA gene sequences belong to Chroococcidiopsis sp. This was done by constructing a maximum likelihood phylogenetic tree using the CpmA sequences of various cyanobacteria (supplementary table S1, Supplementary Material online) and that of Chroococcidiopsis thermalis PCC 7203 (GenBank accession no. HQ113459.1) as a control for comparison. The statistical significance of the tree was evaluated by a bootstrap resampling with 100 replications. The sequences, which appeared on the tree outside the 100% bootstrap support clade with the control sequence, were assumed not to belong to Chroococcidiopsis sp. According to the sequencing of ∼4 kbp of the SSU/ITS/LSU region, the used strains are all Chroococcidiopsis, which currently is considered as a monospecific genus (Bahl et al. 2011). Finally, a total of 50 sequences were retained for the analyses. From these sequences, 26 were unique haplotypes. This analysis was conducted using PhyML 3.0 (Guindon and Gascuel 2003).

Analysis of Sequence Polymorphism and Population Genetic Diversity

The sequences were grouped into six populations according to their location and type of microbial community (table 1). Intra- and interpopulation nucleotide diversity was computed assuming the Kimura two-parameter model (Kimura 1980). The analyses were carried out using MEGA5 (Tamura et al. 2011).

For each population, the following parameters were calculated: average number of nucleotide differences (K), number of haplotypes (H), haplotype diversity (Hd), and θ per site (Watterson 1975). These analyses were performed using DnaSP v5.10 (Librado and Rozas 2009). Interpopulation differentiation was estimated by F_ST computed with the analysis of molecular variance (AMOVA) (Excoffier et al. 1992) as implemented in the Arlequin 3.5 software (Excoffier et al. 2005) and with Jost’s D (Jost 2008) as implemented in SPADE (Chao and Shen 2010).

The recombination parameter R and the minimum number of recombination events in the sample, R_m, were estimated according to Hudson (Hudson 1987). The genetic association between polymorphic sites in the whole sample was measured by the Z_nS statistic (Kelly 1997), and the effect of intragenic recombination on the observed DNA variation was estimated by the ZZ statistics (Rozas et al. 2001). The confidence intervals for the above estimates were obtained by coalescent simulations with 10,000 replicates under an assumption of no recombination. DnaSP v5.10 (Librado and Rozas 2009) was used for these computations.

Tests for Neutrality and Positive Selection

Correspondence of the obtained data to the neutral expectations was examined using several estimates: Tajima’s D (Tajima 1989), Fay and Wu’s H (Fay and Wu 2000), Fu’s F_s test of selective neutrality (Fu 1997), Fu and Li’s D* and F* (Fu and Li 1993), and Achaz’ Y/Y* (Achaz 2008) as implemented online at http://wwwabi.snv.jussieu.fr/achaz/neutralitytest.html. To detect positive selection, we applied a compound DHEW test (Zeng et al. 2007), which combines Tajima’s D, Fay and Wu’s normalized H, and Ewens–Watterson estimates of neutrality. The synonymous (dS) and nonsynonymous diversity (dN), and the dN/dS ratio were calculated using the modified Nei–Gojobori method (Nei and Gojobori 1986). The analyses were conducted using DnaSP v5.10 (Librado and Rozas 2009).

Analysis of Population History

We analyzed the data for signs of historical population size changes using two estimates: Fu’s F_s (Fu 1997) and R₂ statistics (Ramos-Onsins and Rozas 2002). Extensive computer simulations suggested that these tests are most robust for detecting population growth/decline (Ramos-Onsins and Rozas 2002). We also used a Bayesian analysis as implemented in LAMARC 2.1.6 to compute the exponential population growth rate, g (Kuhner 2006). The growth rate relates to the scaled time-dependent mutation parameter θ as follows: θ_t = θ_{present time} exp(−gt), where t is time before present (Kuhner 2006).

Results

Nucleotide Polymorphism and Intrapopulation Diversity of the cpmA Locus in Chroococcidiopsis sp.

The phylogenetic analysis of the obtained cpmA sequences of Chroococcidiopsis sp. yielded a completely unresolved tree (supplementary fig. S1, Supplementary Material online). For the determined 26 unique haplotypes (alleles), there were 35 segregating sites out of 513 sequenced. Among the polymorphisms, 11 were synonymous and 23 nonsynonymous. One polymorphism, a singleton at position 283, was either one depending on the evolutionary path. Figure 1 shows distribution of the polymorphism along the sequenced region of the gene.

Fig. 1.

The distribution of nucleotide polymorphism along the 50 partial sequences of the Chroococcidiopsis sp. cpmA gene. Sliding window of 100 bp with increments of 10 bp. The putative functional domains are shaded with tints of gray. The two hydrophobic motifs (positions 376–408 and 433–513) are depicted by black boxes .

The average total nucleotide diversity of the gene for the whole species (π ± SE) was 0.0034 ± 0.0005; the synonymous diversity was 0.0036, almost the same as the nonsynonymous diversity (0.0033). The total DNA diversity of the two gene domains was slightly different: 0.0030 and 0.0036 for N- and C-terminal domains, respectively. The dN values were a bit higher in the N-terminal domain (0.0037 ± 0.0013 and 0.0032 ± 0.0008, respectively), whereas the rate of synonymous substitutions was significantly higher at the C-terminal domain (0.0049 ± 0.0014) than at the N-terminal domain (0.0010 ± 0.0010). Interestingly, six replacement substitutions versus only one synonymous substitution occurred in one of the two hydrophobic motifs at the C-terminal domain (positions 126–146 and 155–171 in the translated CpmA protein sequence of Chroococcidiopsis sp.). Some of these replacements are radical. For example, highly hydrophobic cysteine at position 138 is replaced by hydrophilic positively charged arginine, and hydrophobic nonpolar leucine at position 159 is replaced by polar hydrophilic serine.

All studied populations and the whole species manifested very low genetic diversity at the cpmA locus (table 2). The highest value of total nucleotide diversity (π = 0.0049) was determined in populations AH and SH, and the lowest in population Canadian Arctic Circle (CH; π = 0.0021). Populations AH and SH also showed the highest rate of nonsynonymous substitutions (0.0056 and 0.0053, respectively), whereas the highest rate at synonymous sites (0.0070) was recorded for population AE. Overall, the populations from Canadian Arctic Circle (CH) and Taklimakan Desert (TH) were the least polymorphic.

Table 2.

Genetic Diversity Parameters of the Studied Populations.

Population	H	Hd	K	dS	dN	θη	π
AC	5	0.933	1.67	0.0033	0.0035	0.0043	0.0033
AE	6	0.893	1.93	0.0070	0.0026	0.0053	0.0038
AH	6	0.889	2.56	0.0034	0.0056	0.0072	0.0049
CH	6	0.641	1.08	0.0023	0.0020	0.0044	0.0021
SH	3	0.833	2.50	0.0038	0.0053	0.0053	0.0049
TH	5	0.647	1.40	0.0030	0.0026	0.0048	0.0028
Average for the species	4.3	0.774	1.74	0.0036	0.0033	0.0152	0.0034

Note.—H, number of haplotypes; Hd, haplotype (gene) diversity; K, average no. of differences; θη, theta (per site) from the total number of mutations; π, intrapopulation diversity; dS, synonymous substitutions; dN, non-synonymous substitutions.

On the other hand, all populations manifested high haplotype diversity, which ranged from 0.641 to 0.993, averaging at 0.774 for the whole species.

Recombination and Linkage Disequilibrium

The analysis yielded the overall R value of 61.2 and R_m value of 2. The values of Z_nS and ZZ were 0.0247 and −0.0092, respectively. The coalescent simulations showed that R_m, Z_nS, and ZZ are not significant. These results suggest that intragenic recombination is not a significant factor for the observed nucleotide variation at the cpmA gene.

Between-Population Diversity

The between-population differentiation at the cpmA locus was very low (table 3). The genetic distances between the populations ranged from 0.0024 to 0.0049 with the mean interpopulation distance value of 0.0034 ± 0.0006, that is, the same as the mean intrapopulation nucleotide diversity. The matrix of pairwise F_ST and Jost’s D values (table 3) provides further support for the low between-population differentiation. The average interpopulation F_ST was only 0.0228 and Jost’s D was only 0.011. Overall, the results of AMOVA indicated that 97.72% of the variability resided within populations.

Table 3.

Genetic Differentiation between the Studied Populations.

Population	AC	AE	AH	CH	SH	TH
AC		−0.0064/0	0.0127/0.103	0.0018/0.153	0.0201/0.125	0.0065/0.141
AE	0.0035		0.0370/0.068	0.0266/0.086	0.0355/0.066	0.0109/0.075
AH	0.0042	0.0046		0.0466*/0.185	−0.0237/0.167	0.0254/0.175
CH	0.0026	0.0030	0.0037		0.0889/0.033	0.0034/0
SH	0.0041	0.0044	0.0049	0.0035		0.0515/0.029
TH	0.0030	0.0033	0.0040	0.0024	0.0038

Note.—Below diagonal, pairwise between-population genetic distance; above diagonal, pairwise F_ST values/Jost’s D.

*P < 0.01

Tests for Neutrality

The results of all tests for neutrality suggested that the cpmA locus did not follow neutral expectations at the species level (table 4). The compound DHEW test, which analyzes data for presence of positive selection (Zeng et al. 2007), yielded a highly significant P value of 0.0034. There were signs of non-neutral evolution and positive selection in particular populations too. The dN/dS ratio was above 1 in three populations: AC, AH, and SH (table 4).

Table 4.

Results for Tests of Neutrality.

Population	D	H	Fs	D*	F*	Y	Y*	E	DHEW	dN/dS
AC	−1.3369, P = 0.06	0.9361, P = 0.783	−2.5180, P = 0.012	−1.3683, P > 0.10	−1.4501, P > 0.10	NA	NA	0.9044	0.7331	1.06
AE	−1.3593, P = 0.100	−0.2882, P = 0.2306	−2.7262, P = 0.012	−1.3604, P > 0.10	−1.5029, P > 0.10	−2.4510, P = 0.0419	−1.0615, P = 0.1746	0.7167	0.5792	0.37
AH	−1.4219, P = 0.076	0.9685, P = 0.8899	−1.5490, P = 0.115	−1.3748, P > 0.10	−1.5486, P > 0.10	−0.7952, P = 0.2173	−1.3463, P = 0.0910	0.0628	0.3723	1.64
CH	−1.9819, P = 0.006	0.6275, P = 0.6698	−2.9045, P = 0.005	−2.4988, P < 0.02	−2.6914, P < 0.02	NA	NA	0.4386	0.3416	0.87
SH	−0.7968, P = 0.179	0.2531, P = 0.3782	0.4611, P = 0.501	− 0.7968, P > 0.10	− 0.7529, P > 0.10	−1.2248, P = 0.1922	NA	0.1525	1	1.39
TH	−1.8391, P = 0.015	0.7643, P = 0.7324	−1.4301, P = 0.083	−2.1369, P < 0.05	−2.3170, P < 0.05	NA	NA	0.5489	0.4369	0.87
Average for the species	−2.6443, P < 0.001	−1.7418, P = 0.049	−30.0980, P < 0.001	−4.7781, P < 0.02	−4.7860, P < 0.02	−2.4699, P = 0.001	−2.3260, P = 0.002	0.0143	0.0034	0.92

Note.—D, Tajima’s D (Tajima 1989); H, Fay and Wu’s H (Fay and Wu 2000); F_s, Fu’s F_s test of selective neutrality (Fu 1997); D* and F*, Fu and Li’s D* and F* test statistics (Fu and Li 1993); Y and Y*, Achaz’ Y and Y* test statistics (Achaz 2008); E, test for directional selection (P values) (Zeng et al. 2006); DHEW, compound test for positive selection (P values) (Zeng et al. 2007); dN/dS, ratio of nonsynonymous to synonymous substitutions; NA, not available.

Population History

Both results of the Fu’s F_s (Fu 1997) and R₂ (Ramos-Onsins and Rozas 2002) tests showed significant departure from the constant population size. The F’_s and R₂ values (−30.0980 and 0.0244, respectively) were far outside their 95% confidence intervals as determined by coalescent simulations: (−4.2967 to 4.7621) and (0.0501 to 0.1966), respectively. Along with the highly significant Tajima’s D (table 4), these results point out to the population expansion of Chroococcidiopsis sp. Further support for this conclusion comes from the Bayesian estimates of the exponential population growth parameter, g, which ranged from 851 to 933 for all six studied populations.

Discussion

Low Level of Intra- and Interpopulation Nucleotide Diversity at the cpmA Locus of Chroococcidiopsis sp.

Our results are in further support of very low nucleotide diversity at circadian genes of cyanobacteria. Despite the extreme environments and the very large geographical distances between the sampling locations, both the intrapopulation (table 2) and interpopulation diversity (table 3) of Chroococcidiopsis sp. at the cpmA locus were quite low. Previously, we studied microevolution of the two core circadian genes, kaiB and kaiC, in a filamentous cyanobacterium Nostoc linckia from the environmentally contrasting slopes of the ecological model microsites, Evolution Canyons I and II (Israel) (Dvornyk et al. 2002). These genes manifested approximately 1,000-fold higher mutation rate in the cyanobacterial strains from the environmentally stressful south-facing slopes when compared with the strains from the temperate north-facing slopes.

Data on intra- and interpopulation DNA diversity of protein-coding bacterial genes are very limited and fragmentary; some available information of several housekeeping genes of some bacterial species in comparison with the respective values for cpmA is provided in table 5. The cpmA gene appears to be even less polymorphic than housekeeping genes, which are thought to be extremely conserved due to their significance for basic functions of an organism (Jordan et al. 2002). For example, the total nucleotide diversity of the cpmA gene is at least 2-fold lower than that of the most conserved housekeeping genes among those presented in table 5, rplL and gyrA of Clostridium perfringens.

Table 5.

Polymorphism of Some Housekeeping and Symbiotic Genes of Bacteria in Comparison with cpmA of Chroococcidiopsis sp.

Species	Genes	Hd	dS	dN	θ/θη	π	References
Clostridium perfringens	pfoS	—	0.042	0.004	9.349/—	0.014	Rooney et al. (2006)
	rplL	—	0.019	0.004	4.415/—	0.007
	gyrA	—	0.029	0.001	9.868/—	0.007
Bradyrhizobium elkanii	recA	0.263	—	—	—	0.0350	Perrineau et al. (2011)
	dnaK	0.238	—	—	—	0.0257
	glnII	0.238	—	—	—	0.0386
Bradyrhizobium canariense	recA	0.928	0.1003	0.0016	—/0.0246	0.0271	Vinuesa et al. (2005)
	atpD	0.902	0.0473	0.0012	—/0.0103	0.0127
	glnII	0.863	0.0455	0.0014	—/0.0108	0.0117
Rhizobium gallicum	glnII	0.939	—	—	—/0.0248	0.0230	Silva et al. (2005)
	atpD	0.923	—	—	—/0.0256	0.0209
	nifH	0.903	—	—	—/0.0442	0.0458
	nodB	0.855	—	—	−/0.0876	0.1243
Microcystis aeruginosa	ftsZ	0.939	—	—	—	0.026	Tanabe et al. (2007)
	glnA	0.948	—	—	—	0.025
	gltX	0.969	—	—	—	0.023
	gyrB	0.938	—	—	—	0.017
	pgi	0.956	—	—	—	0.043
	recA	0.957	—	—	—	0.013
	tpi	0.951	—	—	—	0.019
Chroococcidiopsis sp.	cpmA	0.774	0.0036	0.0033	7.814/0.0152	0.0034

Note.—atpD, beta subunit of ATPase; ftsZ, cell division protein FtsZ; glnA, glutamine synthetase; gltX, glutamyl-tRNA synthetase; gyrB, DNA gyrase subunit B; pgi, glucose-6-phosphate isomerase; recA, recombination protein RecA; tpi, triosephosphate isomerase; gyrA, DNA gyrase subunit A; pfoS, regulatory protein PfoS; rplL, 50 S ribosomal protein; dnaK, heat shock chaperone protein; glnII, glutamine synthetase II; nifH, dinitrogenase reductase; nodB, N-acetylglucosmine deacetylase; θ, theta (per gene) from the total number of mutations; θη, theta (per site) from the total number of mutations; —, not provided.

More comparable evidence comes from the study of another cyanobacterium, toxic Microcystis aeruginosa (Tanabe et al. 2007). The average nucleotide diversity reported for seven housekeeping genes of this prokaryote was 0.023, ranging from 0.013 (recA) to 0.043 (pgi), which is up to 10-fold higher than that for the cpmA gene (table 5).

Recent multilocus study using the 16S rRNA-5.8S ITS-23S rRNA gene region sequences of globally distributed Chroococcidiopsis sp. reported differentiation between the ecotypes of the cyanobacterium from cold and hot deserts (Bahl et al. 2011). It was suggested that this differentiation was driven by adaptation to specific environmental conditions rather than by geographic isolation. We did not find any such differentiation at the cpmA locus. This inconsistency may be due to the much higher conservation of cpmA even when compared with the rRNA genes.

The observed extreme conservation of the circadian gene, cpmA, may result from two main factors. First, cpmA belongs to the AIRC superfamily. Members of this superfamily are of critical importance, as they are involved into de novo biosynthesis of purines, an essential component of DNA (Luckens and Buchanan 1959). Taking account of this, cpmA may actually be considered as a housekeeping gene. Furthermore, the circadian system as a whole is critically important for maintenance of intracellular homeostasis and adaptation to environment (Johnson 2005). According to the estimates by different methods, a proportion of genes that are expressed rhythmically in the cyanobacterial genome varies from 2% in Synechocystis sp. PCC 6803 (Kucho et al. 2005) to up to 30% in Synechococcus elongatus PCC 7942 (Liu et al. 1995). The importance of the circadian system assumes the importance of all its elements. Indeed, mutations in cpmA significantly reduce a growth rate of the cyanobacterium (Katayama et al. 1999)

The high haplotype diversity of the cpmA gene is not quite surprising. It has been frequently reported for many genes in bacteria (table 5 and references therein). However, high haplotype diversity usually results in significant interpopulation differentiation of bacteria (Silva et al. 2005; Vinuesa et al. 2005), which is not a case for Chroococcidiopsis sp. Even the Jost’s D test (Jost 2008), which is particularly sensitive to “private” alleles, failed to detect significant differentiation between the populations of Chroococcidiopsis sp. (table 3).

One of the possible reasons for the high gene diversity of bacteria and, particularly, Chroococcidiopsis sp. may be the existence of multiple ecotypes in the same habitat, which are adapted to ecological microniches (Tanabe et al. 2007). This assumption is plausible, given the extreme environmental conditions and quite pronounced fluctuations in environmental factors (temperature, light, humidity, etc.) at the sampling sites.

Non-Neutral Evolution of the cpmA Locus in Chroococcidiopsis sp.

CpmA is one of the genes, which control the circadian output in cyanobacteria (Katayama et al. 1999). There is limited information about its physiological function. It was reported that inactivation of cpmA results in a significantly earlier phase and lower amplitude of rhythmic expression of a core circadian gene, kaiA, and severe decrease in growth of a cyanobacterium (Katayama et al. 1999). Based on these data, it was suggested that cpmA might contribute to basic cell metabolism (Katayama et al. 1999). Although there is no direct evidence of whether cpmA itself has adaptive significance or does not, but signals for positive selection in the whole gene and, specifically, the character of some replacement substitutions in its hydrophobic motifs suggest that this possibility should not be excluded. Moreover, this seems likely considering the adaptive significance of the circadian system as a whole (Johnson et al. 1998; Ouyang et al. 1998).

Although data about primarily non-neutral evolution and positive selection in various genes of bacteria are abundant, they are quite limited with respect to the circadian genes. As to prokaryotes, positive selection at the population level was documented for two core circadian genes, kaiB and kaiC, in a filamentous cyanobacterium, Nostoc linckia, from two model ecological microsites known as Evolution Canyons I and II (Dvornyk et al. 2002). Results of other studies (Dvornyk et al. 2004; Dvornyk 2005) suggested positive selection at the above-species level for two other circadian genes, ldpA and sasA, which are elements of the circadian input and output, respectively. On the other hand, the previous macroevolutionary study of cpmA in prokaryotes did not determine any positive selection on the gene (Dvornyk 2006). The discrepancies between the results of the cpmA studies at micro- and macrolevels may be due to fairly low intensity and an episodic character of selection pressure on the gene. As such, the existing methods may not be powerful enough to detect positive selection at the above-species level, unless this selection is sufficiently strong and operates during reasonably long evolutionary period. On the other hand, selection may be population specific, as was suggested for the human period 2 gene (Cruciani et al. 2008).

Data about selection on circadian genes in eukaryotes are more abundant. In this respect, gene period of Drosophila has received much attention due to the report about a latitudinal cline in its threonine–glycine repeat polymorphism in European Drosophila melanogaster, which was implicated in selection for clock temperature compensation (Costa et al. 1992). A majority of the studies, which subsequently examined this and other polymorphisms of period in different populations and species of Drosophila, reported various types of selection, which might shape the observed polymorphism (Kliman and Hey 1993; Rosato et al. 1994; Sawyer et al. 1997, 2006). Natural selection was also reported to operate on another Drosophila circadian gene, timeless (Tauber et al. 2007). Recently, six clock-associated genes of a plant, Populus tremula, were found to experience positive selection (Hall et al. 2011).

Altogether, the above data suggest that selection is among the most important factors operating on circadian genes of prokaryotes and eukaryotes. Evidence for positive selection further supports the previous reports about the adaptive significance of the circadian clock system (Johnson et al. 1998; Ouyang et al. 1998).

Small Effective Population Size and Mutation-Drift Equilibrium

The obtained results are in favor of a relatively small effective population size of Chroococcidiopsis sp. We found high linkage disequilibrium and very low recombination rate in our data set. Given that linkage disequilibrium at neutral sites of a haploid organism depends on effective population size N_e and a recombination rate c and is determined by 1/(2N_ec). Since c is very small, N_e should also be small to explain the high disequilibrium. Using the obtained values for Z_nS (0.0247) and c (0.0004), the N_e was estimated to be only ∼50,000 individuals. Populations with small N_e should experience significant genetic drift, which reduces their diversity. Indeed, the observed nucleotide diversity of the studied populations was very low. However, despite the very large distances between the sampled populations, they show almost no genetic differentiation. This suggests virtually unlimited gene flow between them.

We also examined population history of the data, that is, whether there is any departure from demographic equilibrium. One of the signs for a recent expansion is an excess of singleton mutations, which can be measured by Tajima’s D or Fu’s F_s. Both these statistics yielded statistically significant negative values (table 4). Two other tests, R₂ (Ramos-Onsins and Rozas 2002) and g (Kuhner 2006), also supported recent expansion of the Chroococcidiopsis sp. populations. In addition, the obtained results suggest that Chroococcidiopsis sp. probably consists of many relatively small populations occupying various environmental microniches within a particular habitat.

Such population structure of Chroococcidiopsis sp. is not something unusual. Many bacteria, particularly endosymbiotic and parasitic ones, may have extremely large population size at the global scale and small effective size at the population level due to bottlenecks experienced by bacterial populations during transmission between hosts (Sharp et al. 2005). Given the harsh environmental conditions of their habitats, micropopulations of Chroococcidiopsis sp. may experience a bottleneck during their recovery after various climatic extremes, such as low temperature or high UV.

Concluding Remarks

This study provides compelling evidence for extremely high conservation and non-neutral evolution of a circadian locus, cpmA, in a stress-tolerant cyanobacterium Chroococcidiopsis sp. from extreme environments around the globe. The signs for positive selection detected in some of the studied populations and the whole species suggest that cpmA evolution is likely of adaptive nature.

Supplementary Material

Supplementary figure S1 and table S1 are available at Molecular Biology and Evolution online http://www.mbe.oxfordjournals.org.