Adaptive Divergence in Experimental Populations of Pseudomonas fluorescens. IV. Genetic Constraints Guide Evolutionary Trajectories in a Parallel Adaptive Radiation

The capacity for phenotypic evolution is dependent upon complex webs of functional interactions that connect genotype and phenotype. Wrinkly spreader (WS) genotypes arise repeatedly during the course of a model Pseudomonas adaptive radiation. Previous work showed that the evolution of WS variation was explained in part by spontaneous mutations in wspF, a component of the Wsp-signaling module, but also drew attention to the existence of unknown mutational causes. Here, we identify two new mutational pathways (Aws and Mws) that allow realization of the WS phenotype: in common with the Wsp module these pathways contain a di-guanylate cyclase-encoding gene subject to negative regulation. Together, mutations in the Wsp, Aws, and Mws regulatory modules account for the spectrum of WS phenotype-generating mutations found among a collection of 26 spontaneously arising WS genotypes obtained from independent adaptive radiations. Despite a large number of potential mutational pathways, the repeated discovery of mutations in a small number of loci (parallel evolution) prompted the construction of an ancestral genotype devoid of known (Wsp, Aws, and Mws) regulatory modules to see whether the types derived from this genotype could converge upon the WS phenotype via a novel route. Such types—with equivalent fitness effects—did emerge, although they took significantly longer to do so. Together our data provide an explanation for why WS evolution follows a limited number of mutational pathways and show how genetic architecture can bias the molecular variation presented to selection.UNDERSTANDING—and importantly, predicting—phenotypic evolution requires knowledge of the factors that affect the translation of mutation into phenotypic variation—the raw material of adaptive evolution. While much is known about mutation rate (e.g., Drake et al. 1998; Hudson et al. 2002), knowledge of the processes affecting the translation of DNA sequence variation into phenotypic variation is minimal.Advances in knowledge on at least two fronts suggest that progress in understanding the rules governing the generation of phenotypic variation is possible (Stern and Orgogozo 2009). The first stems from increased awareness of the genetic architecture underlying specific adaptive phenotypes and recognition of the fact that the capacity for evolutionary change is likely to be constrained by this architecture (Schlichting and Murren 2004; Hansen 2006). The second is the growing number of reports of parallel evolution (e.g., Pigeon et al. 1997; ffrench-Constant et al. 1998; Allender et al. 2003; Colosimo et al. 2004; Zhong et al. 2004; Boughman et al. 2005; Shindo et al. 2005; Kronforst et al. 2006; Woods et al. 2006; Zhang 2006; Bantinaki et al. 2007; McGregor et al. 2007; Ostrowski et al. 2008)—that is, the independent evolution of similar or identical features in two or more lineages—which suggests the possibility that evolution may follow a limited number of pathways (Schluter 1996). Indeed, giving substance to this idea are studies that show that mutations underlying parallel phenotypic evolution are nonrandomly distributed and typically clustered in homologous genes (Stern and Orgogozo 2008).While the nonrandom distribution of mutations during parallel genetic evolution may reflect constraints due to genetic architecture, some have argued that the primary cause is strong selection (e.g., Wichman et al. 1999; Woods et al. 2006). A means of disentangling the roles of population processes (selection) from genetic architecture is necessary for progress (Maynard Smith et al. 1985; Brakefield 2006); also necessary is insight into precisely how genetic architecture might bias the production of mutations presented to selection.Despite their relative simplicity, microbial populations offer opportunities to advance knowledge. The wrinkly spreader (WS) morphotype is one of many different niche specialist genotypes that emerge when experimental populations of Pseudomonas fluorescens are propagated in spatially structured microcosms (Rainey and Travisano 1998). Previous studies defined, via gene inactivation, the essential phenotypic and genetic traits that define a single WS genotype known as LSWS (Spiers et al. 2002, 2003) (Figure 1). LSWS differs from the ancestral SM genotype by a single nonsynonymous nucleotide change in wspF. Functionally (see Figure 2), WspF is a methyl esterase and negative regulator of the WspR di-guanylate cyclase (DGC) (Goymer et al. 2006) that is responsible for the biosynthesis of c-di-GMP (Malone et al. 2007), the allosteric activator of cellulose synthesis enzymes (Ross et al. 1987). The net effect of the wspF mutation is to promote physiological changes that lead to the formation of a microbial mat at the air–liquid interface of static broth microcosms (Rainey and Rainey 2003).Open in a separate windowFigure 1.—Outline of experimental strategy for elucidation of WS-generating mutations and their subsequent identity and distribution among a collection of independently evolved, spontaneously arising WS genotypes. The strategy involves, first, the genetic analysis of a specific WS genotype (e.g., LSWS) to identify the causal mutation, and second, a survey of DNA sequence variation at specific loci known to harbor causal mutations among a collection of spontaneously arising WS genotypes. For example, suppressor analysis of LSWS using a transposon to inactivate genes necessary for expression of the wrinkly morphology delivered a large number of candidate genes (top left) (Spiers et al. 2002). Genetic and functional analysis of these candidate genes (e.g., Goymer et al. 2006) led eventually to the identity of the spontaneous mutation (in wspF) responsible for the evolution of LSWS from the ancestral SM genotype (Bantinaki et al. 2007). Subsequent analysis of the wspF sequence among 26 independent WS genotypes (bottom) showed that 50% harbored spontaneous mutations (of different kinds; see Open in a separate windowFigure 2.—Network diagram of DGC-encoding pathways underpinning the evolution of the WS phenotype and their regulation. Overproduction of c-di-GMP results in overproduction of cellulose and other adhesive factors that determine the WS phenotype. The ancestral SBW25 genome contains 39 putative DGCs, each in principle capable of synthesizing the production of c-di-GMP, and yet WS genotypes arise most commonly as a consequence of mutations in just three DGC-containing pathways: Wsp, Aws, and Mws. In each instance, the causal mutations are most commonly in the negative regulatory component: wspF, awsX, and the phosphodiesterase domain of mwsR (see text).To determine whether spontaneous mutations in wspF are a common cause of the WS phenotype, the nucleotide sequence of this gene was obtained from a collection of 26 spontaneously arising WS genotypes (WS_A-Z) taken from 26 independent adaptive radiations, each founded by the same ancestral SM genotype (Figure 1): 13 contained mutations in wspF (Bantinaki et al. 2007). The existence of additional mutational pathways to WS provided the initial motivation for this study.

TABLE 1

Mutational causes of WS

Evolutionary Strata in a Small Mating-Type-Specific Region of the Smut Fungus Microbotryum violaceum

DNA sequence analysis and genetic mapping of loci from mating-type-specific chromosomes of the smut fungus Microbotryum violaceum demonstrated that the nonrecombining mating-type-specific region in this species comprises ∼25% (∼1 Mb) of the chromosome length. Divergence between homologous mating-type-linked genes in this region varies between 0 and 8.6%, resembling the evolutionary strata of vertebrate and plant sex chromosomes.EVOLUTION of mating types or sex-determining systems often involves the suppression of recombination around the primary sex-determining or mating-type-determining locus. In animals and plants, it is often an entire or almost entire chromosome (Y or W in male or female heterogametic species, respectively) that ceases to recombine with its homologous (X or Z) chromosome (Charlesworth and Charlesworth 2000; Charlesworth 2008). Self-incompatibility loci in plants are also thought to be located in regions of suppressed recombination (Charlesworth et al. 2005; Kamau and Charlesworth 2005; Kamau et al. 2007; Li et al. 2007; Yang et al. 2007). Regardless of the phylogenetic position of a species, such nonrecombining regions are known to follow similar evolutionary trajectories. The nonrecombining region on the sex-specific chromosome expands in several steps, forming evolutionary strata—regions of different X/Y (or Z/W) divergence (Lahn and Page 1999; Handley et al. 2004; Sandstedt and Tucker 2004; Nicolas et al. 2005)—and genes in the nonrecombining regions gradually accumulate deleterious mutations that eventually render them dysfunctional (Charlesworth and Charlesworth 2005; Charlesworth 2008).Fungal mating-type systems are very diverse, with the number of mating types varying from two to several hundred (Casselton 2002). Like sex chromosomes in several animals and plants, suppressed recombination has evolved in regions near fungal mating-type loci, including in Ustilago hordei (Lee et al. 1999), Cryptococcus neoformans (Lengeler et al. 2002), and Neurospora tetrasperma (Menkis et al. 2008). These species have two mating types, but no morphologically distinct sexes. The mating-type locus (the region of suppressed recombination) of C. neoformans is small (∼100 kb) compared with known sex chromosomes and contains only ∼20 genes that, unlike many sex chromosomes (Y or W chromosomes), show no obvious signs of genetic degeneration (Lengeler et al. 2002; Fraser et al. 2004). Judging from the divergence between the homologous genes on the two mating-type-specific chromosomes, C. neoformans started to evolve sex chromosomes a long time ago because silent divergence between the two mating types in the most ancient region exceeds 100% (Fraser et al. 2004). Genes in the younger mating-type-specific region are much less diverged between the two sex chromosomes, suggesting that the evolution of the sex locus in C. neoformans might have proceeded through several steps. The nonrecombining region around the mating-type locus of N. tetrasperma is much larger than in C. neoformans (at least 6.6 Mb), and silent divergence between homologous genes on the mating-type-specific chromosomes ranges from zero to 9%, demonstrating that these mating-type-specific chromosomes evolved recently (Menkis et al. 2008).M. violaceum, which causes anther smut disease in Silene latifolia and other species in the family Caryophyllaceae, has two mating types, A1 and A2 (reviewed by Giraud et al. 2008), which are determined by the presence of mating-type-specific chromosomes (hereafter A1 and A2 chromosomes, or sex chromosomes) in the haploid stage of the life cycle (Hood 2002; Hood et al. 2004). The A1 and A2 chromosomes are distinguishable by size in pulsed-field electrophoresis, and it is possible to isolate individual chromosomes electrophoretically (Hood et al. 2004). Random fragments of A1 and A2 chromosomes have previously been isolated from mating-type-specific bands of pulsed-field separated chromosomes of M. violaceum (Hood et al. 2004). These fragments were assumed to be linked to mating type. The same method was used to isolate fragments of non-mating-type-specific chromosomes. On the basis of the analysis of their sequences, (Hood et al. 2004) proposed that mating-type-specific chromosomes in M. violaceum might be degenerate because they contained a lower proportion of protein-coding genes than other chromosomes. However, it was not determined whether the sequences isolated from the mating-type chromosomes originated from the mating-type-specific or from the recombining regions (Hood et al. 2004), and the relative sizes of these regions are not known for these M. violaceum chromosomes. We tested the mating-type specificity of 86 of these fragments and demonstrate that fewer than a quarter of these loci are located in the mating-type-specific region, suggesting that the nonrecombining region on the A1 and A2 chromosomes is quite small, while the rest of the chromosome probably recombines (like pseudoautosomal regions of sex chromosomes) and is therefore not expected to undergo genetic degeneration. Genetic mapping confirms the presence of two pseudoautosomal regions in the M. violaceum mating-type-specific chromosomes.As these chromosomes are mating type specific in the haploid stage of M. violaceum, mating-type-specific loci (or DNA fragments) can be identified by testing whether they are present exclusively in A1 or A2 haploid strains. We therefore prepared haploid A1 and A2 M. violaceum cultures from S. latifolia plants from two geographically remote locations (accessions Sl405 from Sweden and Sl127 from the French Pyrenees). Haploid sporidial cultures were isolated by a standard dilution method (Kaltz and Shykoff 1997; Oudemans and Alexander 1998). Mating types were determined by PCR amplification of each culture with primers designed for A1 and A2 pheromone receptor genes linked to A1 and A2 mating types (Yockteng et al. 2007). The primers were as follows: 5′-TGGCATCCCTCAATGTTTCC-3′ and 5′-CACCTTTTGATGAGAGGCCG-3′ for the A1 pheromone receptor (GenBank accession no. {"type":"entrez-nucleotide","attrs":{"text":"EF584742","term_id":"156254864","term_text":"EF584742"}}EF584742) and 5′-TGACGAGAGCATTCCTACCG-3′ and 5′-GAAGCGGAACTTGCCTTTCT-3′ for the A2 pheromone receptor (GenBank accession no. {"type":"entrez-nucleotide","attrs":{"text":"EF584741","term_id":"156254862","term_text":"EF584741"}}EF584741). Cultures with PCR product amplified only from an A1 or A2 pheromone receptor gene were selected for further use. The mating types of the cultures were verified by conjugating them in all combinations.The GenBank nucleotide database was searched using BLAST for sequences similar to those isolated by Hood et al. (2004). Sequences with similarity to transposable elements (TE) and other repeats were excluded. The resulting set of nonredundant sequences was used to design PCR primers for 98 fragments. Half of these were originally isolated from the A1 and half from the A2 chromosomes and are hereafter called A1-NNN or A2-NNN (where NNN is the locus number; supporting information, Table S1), which does not imply that these loci are A1 or A2 specific, but merely indicates that they were originally isolated from the A1 or A2 chromosomes. Amplification of these regions from new A1 and A2 M. violaceum cultures, independently isolated by ourselves, revealed that only 5 of the 49 loci isolated from the A1 chromosome are indeed A1 specific and only 6 of 49 isolated from the A2 chromosome are A2 specific. All other loci amplified from both A1 and A2 cultures. Figure 1 illustrates some of these results from the Swedish sample (Sl405).Open in a separate windowFigure 1.—Testing of mating-type specificity for loci isolated from A1 and A2 chromosomes. (a) PCR amplifications from haploid cultures from Sl405 using primers designed from six A1-originated loci. Loci in which a PCR product could be amplified only from A1 cultures (boxed) were classified as specific to mating type A1. (b) PCR tests of six A2-originated loci on the same set of haploids as in a. Loci in which a PCR product amplified only from A2 cultures (boxed) were classified as specific to mating type A2. Loci amplified from both A1 and A2 cultures are not mating type specific.The fragments that amplified from both A1 and A2 mating types may be in recombining regions, or they could be present in mating-type-specific regions on both A1 and A2 chromosomes. If they are in recombining regions, the A1- and A2-linked homologs should not be diverged from each other, but if they are in nonrecombining, mating-type-specific regions, the divergence of the A1- and A2-linked homologs should be roughly proportional to the time since recombination stopped in the region. We therefore sequenced and compared PCR fragments amplified from the two mating types of Sl405 or Sl127 cultures (GenBank accession nos. {"type":"entrez-nucleotide","attrs":{"text":"FI855822","term_id":"258612459","term_text":"FI855822"}}FI855822–{"type":"entrez-nucleotide","attrs":{"text":"FI856001","term_id":"258612570","term_text":"FI856001"}}FI856001). Sequencing of PCR products showed that 12 (4 A1 and 8 A2) loci have more than one copy, and they were excluded from further analysis. Sequences of 61 loci were identical between the A1 and A2 strains, and four loci demonstrated low total divergence (0.24–0.61%) between the two mating types (otintseva and D. Filatov, unpublished results). Thus, these loci might be located in the recombining part of the mating-type-specific chromosomes. Ten of 75 loci that amplified in both mating types demonstrated multiple polymorphisms fixed between the mating types rather than between the locations. Given that the strains that we used in the analysis originated from two geographically distant locations, it is highly unlikely that multiple polymorphisms distinguishing the A1 and A2 sequences arose purely by chance; thus, these loci are probably located in the nonrecombining mating-type-specific region of the M. violaceum A1 and A2 chromosomes.

TABLE 1

Loci from mating-type-specific chromosomes of M. violaceum used for PCR analysis and genetic map construction

		With nonzero A1/A2 divergenceb
Loci	Mating type specific	<1%	>1%	With zero A1/A2 divergenceb	Total
A1a	5	2 (1)	3 (3)	35 (3)	45 (7)
A2a	6	2 (0)	7 (7)	26 (3)	41 (10)
Subtotal		4 (1)	10 (10)
Total	11	14 (11)		61 (6)	86 (17)

Open in a separate window^aA1, loci originated from the A1 sex chromosome; A2, loci originated from the A2 sex chromosome.^bThe number of loci used for genetic map construction is in parentheses.To confirm the mating-type-specific or pseudoautosomal locations of the loci with and without A1/A2 divergence, we conducted genetic mapping in a family of 99 individuals, 50 of which were of mating type A1 and 49 of mating type A2. The family was generated by a cross between A1 and A2 M. violaceum strains from S. latifolia accessions Sl405 (Sweden) and Sl127 (France), respectively. The choice of strains from geographically distant locations was motivated by the hope of maximizing the number of DNA sequence differences between them that can be used as molecular genetic markers in segregation analysis. We inoculated S. latifolia seedlings with sporidial cultures of both mating types. For inoculation, petri dishes with 12-day-old seedlings of S. latifolia were flooded with 2.5 ml of inoculum suspension. Inoculum suspension consisted of equal volumes of the A1 and A2 sporidial cultures that were mixed and conjugated overnight at 14° under rotation (Biere and Honders 1996; Van Putten et al. 2003). Seedlings were potted 3 days after inoculation. Two months later, teliospores were collected from the flowers of the infected plant and grown in petri dishes on 3.6% potato dextrose agar medium. Haploid sporidia formed after meiosis were isolated and grown as separate cultures for DNA extraction. The mating types of single sporidia cultures were identified as described above. The loci analyzed in the segregation analysis were sequenced in the two parental haploid strains and in 99 (50 A1 and 49 A2) haploid strains that were generated in the cross. Single nucleotide differences between the parental strains were used as molecular genetic markers for segregation analysis in the progeny. The genetic map was constructed using MAPMAKER/EXP v3.0 (Lincoln et al. 1992) and MapDisto v1.7 (http://mapdisto.free.fr/).The resulting genetic map is shown in Figure 2. As expected, no recombination was observed between the 10 loci with diverged A1- and A2-linked copies. In addition, one marker with no A1/A2 divergence, A2-397, was also completely linked to the loci with significant A1/A2 divergence. This locus either may be very tightly linked to the nonrecombining mating-type-specific region or may have been added to that region more recently than the loci that had already accumulated some divergence between the alleles in the two mating types. The mating-type-specific pheromone receptor locus (Devier et al. 2009) and 11 mating-type-specific loci are also located in this nonrecombining region (Figure 2). Interestingly, the cluster of nonrecombining markers is flanked on both sides with markers that recombine in meiosis, demonstrating that there are pseudoautosomal regions on both ends of the mating-type-specific chromosomes.Open in a separate windowFigure 2.—Genetic map of the mating-type-determining chromosome in M. violaceum. Genetic distance (in centimorgans) and the relative positions of the markers are shown to the left and the right of the chromosome, respectively. The position of the nonrecombining region corresponds to the cluster of linked markers shown on the right of the figure. Total A1/A2 divergence is shown in parentheses. Eleven mating-type-specific markers (for which sequences are available from only one mating type), located in the nonrecombining mating-type-specific region, are not shown.Our results demonstrate that although the loci reported by Hood et al. (2004) were isolated from the A1 and A2 chromosomes, most of these loci are not located in the nonrecombining mating-type-specific regions. In fact, the nonrecombining region might be relatively small: of 86 tested fragments, only 21 appeared to be either mating type specific or linked to the mating-type locus. Assuming that these loci represent a random set of DNA fragments isolated from the A1 and A2 chromosomes, it is possible to estimate the size of the nonrecombining region using the binomial distribution: the nonrecombining region is expected to be 24.4% (95% CI: 16.7–33.6%) of the chromosome length. As the sizes of the A1 and A2 chromosomes are ∼3.4 and 4.2 Mb long (Hood 2002; Hood et al. 2004), the nonrecombining region might be ∼1 Mb long.Interestingly, total A1/A2 divergence for the 11 loci with A1- and A2-linked copies mapped to the nonrecombining region varied from 0% to 8.6% (Figure 2). In addition, 11 loci amplified from only one mating type. These genes could represent degenerated genes, some of which degenerated in A1 strains, and some in A2 strains. Alternatively, they might be highly diverged genes, such that the PCR primers amplify only one allele, and not the other. Variation in divergence may be the result of the stepwise cessation of recombination between the A1 and A2 chromosomes in M. violaceum, resembling the evolutionary strata reported for human, chicken, and white campion sex chromosomes (Lahn and Page 1999; Handley et al. 2004; Bergero et al. 2007). However, only the differences between the most and the least diverged loci are statistically significant (Devier et al. 2009), the M. violaceum mating-type region has at least three strata: one oldest stratum, including the pheromone receptor locus; a younger stratum with ∼5–9% A1/A2 divergence; and the youngest stratum with 1–4% divergence between the two mating types. There may also be an additional very recently evolved stratum containing the locus named A2-397, which is also present in all A1 strains tested, with no fixed differences between the A1 and A2 strains (

One- and Two-Locus Population Models With Differential Viability Between Sexes: Parallels Between Haploid Parental Selection and Genomic Imprinting

A model of genomic imprinting with complete inactivation of the imprinted allele is shown to be formally equivalent to the haploid model of parental selection. When single-locus dynamics are considered, an internal equilibrium is possible only if selection acts in the opposite directions in males and females. I study a two-locus version of the latter model, in which maternal and paternal effects are attributed to the single alleles at two different loci. A necessary condition for the allele frequency equilibria to remain on the linkage equilibrium surface is the multiplicative interaction between maternal and paternal fitness parameters. In this case the equilibrium dynamics are independent at both loci and results from the single-locus model apply. When fitness parameters are additive, analytic treatment was not possible but numerical simulations revealed that stable polymorphism characterized by association between loci is possible only in several special cases in which maternal and paternal fitness contributions are precisely balanced. As in the single-locus case, antagonistic selection in males and females is a necessary condition for the maintenance of polymorphism. I also show that the above two-locus results of the parental selection model are very sensitive to the inclusion of weak directional selection on the individual''s own genotypes.PARENTAL genetic effects refer to the influence of the mother''s and father''s genotypes on the phenotypes of their offspring, not attributable just to the transfer of genes. Examples have been documented across a wide range of areas of organism biology; see, for example, Wade (1998) and ​and22 in Rasanen and Kruuk (2007). Parental selection is a more formal concept used in theoretical modeling and concerns situations where the fitness of the offspring depends, besides other factors, on the genotypes of its parent(s) (generalizing from Kirkpatrick and Lande 1989).

TABLE 1

Frequencies of genotypes and fitness parameterizations in model 1

Retrograde Intraflagellar Transport Mutants Identify Complex A Proteins With Multiple Genetic Interactions in Chlamydomonas reinhardtii

The intraflagellar transport machinery is required for the assembly of cilia. It has been investigated by biochemical, genetic, and computational methods that have identified at least 21 proteins that assemble into two subcomplexes. It has been hypothesized that complex A is required for retrograde transport. Temperature-sensitive mutations in FLA15 and FLA17 show defects in retrograde intraflagellar transport (IFT) in Chlamydomonas. We show that IFT144 and IFT139, two complex A proteins, are encoded by FLA15 and FLA17, respectively. The fla15 allele is a missense mutation in a conserved cysteine and the fla17 allele is an in-frame deletion of three exons. The flagellar assembly defect of each mutant is rescued by the respective transgenes. In fla15 and fla17 mutants, bulges form in the distal one-third of the flagella at the permissive temperature and this phenotype is also rescued by the transgenes. These bulges contain the complex B component IFT74/72, but not α-tubulin or p28, a component of an inner dynein arm, which suggests specificity with respect to the proteins that accumulate in these bulges. IFT144 and IFT139 are likely to interact with each other and other proteins on the basis of three distinct genetic tests: (1) Double mutants display synthetic flagellar assembly defects at the permissive temperature, (2) heterozygous diploid strains exhibit second-site noncomplemention, and (3) transgenes confer two-copy suppression. Since these tests show different levels of phenotypic sensitivity, we propose they illustrate different gradations of gene interaction between complex A proteins themselves and with a complex B protein (IFT172).CILIA and flagella are microtubule-based organelles that are found on most mammalian cells. They provide motility to cells and participate in many sensory processes. Defects in or loss of cilia/flagella cause a variety of human diseases that include polycystic kidney disease, retinal degeneration, infertility, obesity, respiratory defects, left–right axis determination, and polydactyly (Fliegauf et al. 2007). Mouse mutants demonstrate that cilia are essential for viability, neural tube closure, and bone development (Eggenschwiler and Anderson 2007; Fliegauf et al. 2007). Cilia and flagella are also present in protists, algae, moss, and some fungi.The assembly and maintenance of cilia and flagella require intraflagellar transport (IFT) (Kozminski et al. 1995). IFT involves the movement of 100- to 200-nm-long protein particles from the basal body located in the cell body to the tip of the flagella using the heterotrimeric kinesin-2 (anterograde movement) (Kozminski et al. 1995) and movement back to the cell body (retrograde movement) using the cytoplasmic dynein complex (Pazour et al. 1999; Porter et al. 1999). IFT particles change their direction of movement as well as their size, speed, and frequency at the ends of the flagella as they switch from anterograde to retrograde movement (Iomini et al. 2001). Biochemical isolation of IFT particles reveals that they are composed of at least 16 proteins and that these particles can be dissociated into two complexes in vitro by changing the salt concentration (Cole et al. 1998; Piperno et al. 1998). Recent genetic and bioinformatics analysis adds at least 7 more proteins to the IFT particle (Follit et al. 2009) (Eggenschwiler and Anderson 2007).

TABLE 1

Proteins and gene names for the intraflagellar transport particles in Chlamydomonas, C. elegans, and mouse

A Two-Pathway Analysis of Meiotic Crossing Over and Gene Conversion in Saccharomyces cerevisiae

Several apparently paradoxical observations regarding meiotic crossing over and gene conversion are readily resolved in a framework that recognizes the existence of two recombination pathways that differ in mismatch repair, structures of intermediates, crossover interference, and the generation of noncrossovers. One manifestation of these differences is that simultaneous gene conversion on both sides of a recombination-initiating DNA double-strand break (“two-sidedness”) characterizes only one of the two pathways and is promoted by mismatch repair. Data from previous work are analyzed quantitatively within this framework, and a molecular model for meiotic double-strand break repair based on the concept of sliding D-loops is offered as an efficient scheme for visualizing the salient results from studies of crossing over and gene conversion, the molecular structures of recombination intermediates, and the biochemical competencies of the proteins involved.EUKARYOTES transit from the diplophase to the haplophase via meiosis, which is associated with a number of interrelated processes, including crossing over and gene conversion. These processes involve meiosis-specific, programmed DNA double-strand breaks (DSBs) and their repair (DSBr). DSBr, in turn, is associated with mismatched base pairs and their rectification, referred to as “mismatch repair” or MMR (Bishop et al. 1987). Current efforts to accommodate both the genetic and molecular phenomena associated with meiotic DSBr in yeast (Saccharomyces cerevisiae) have been thoroughly reviewed (e.g., Hollingsworth and Brill 2004; Hoffmann and Borts 2004; Surtees et al. 2004; Hunter 2007; Berchowitz and Copenhaver 2010), but none of the reviews commits to an overall picture with quantitative predictions. This work aims to remedy that lack. Specifically, we have made use of salient published studies to develop, step-by-step, a comprehensive model of meiotic DSBr and MMR. The main features of this model are summarized in FeaturesPairing pathwayDisjunction pathwayProductsCrossovers and noncrossoversCrossovers onlyCrossover InterferenceNo positive interferencePositive interferenceMsh4–Msh5 dependenceNoneTotalBimolecular intermediateLong with junctions not fully ligatedShort with fully ligated Holliday junctionsInvasion heteroduplexPartly ephemeralEphemeralMMR at invasion and annealingDependent on Msh2 and Mlh1NoneMMR near the DSB siteDirected by 3′ invading and annealing endsMlh1 dependent; directed by junction resolutionRole of Msh2 in MMRRecognizes mismatches and attracts Mlh1NoneRole of Msh4–Msh5 in MMRNoneAttracts Mlh1Open in a separate window     

A Problem With the Correlation Coefficient as a Measure of Gene Expression Divergence

The correlation coefficient is commonly used as a measure of the divergence of gene expression profiles between different species. Here we point out a potential problem with this statistic: if measurement error is large relative to the differences in expression, the correlation coefficient will tend to show high divergence for genes that have relatively uniform levels of expression across tissues or time points. We show that genes with a conserved uniform pattern of expression have significantly higher levels of expression divergence, when measured using the correlation coefficient, than other genes, in a data set from mouse, rat, and human. We also show that the Euclidean distance yields low estimates of expression divergence for genes with a conserved uniform pattern of expression.IT is now possible to measure the expression levels of thousands of genes in multiple tissues at multiple times. This has led to investigations into the evolution of gene expression and how the pattern of expression changes on a genomic scale. In some analyses, the evolution of expression is considered only within one tissue, but in many studies the evolution across multiple tissues is investigated. In this latter case, the evolution of an expression profile—a vector of expression levels of a gene across several tissues—is considered.Several different statistics have been proposed to measure the divergence between gene expression profiles. The two most popular measures are the Euclidean distance (Jordan et al. 2005; Kim et al. 2006; Yanai et al. 2006; Urrutia et al. 2008) and Pearson''s correlation coefficient (Makova and Li 2003; Huminiecki and Wolfe 2004; Yang et al. 2005; Kim et al. 2006; Liao and Zhang 2006a,b; Xing et al. 2007; Urrutia et al. 2008). The correlation coefficient is often subtracted from one, so that the statistic varies from zero, when there has been no expression divergence, to a maximum of two; we refer to this statistic as the Pearson distance. Here we describe a significant shortcoming of the Pearson distance that is not shared by the Euclidean distance.To investigate properties of these two measures of expression divergence, we compiled a data set of 2859 orthologous genes from human, mouse, and rat for which we had microarray expression data from nine homologous tissues: bone marrow, heart, kidney, large intestine, pituitary, skeletal muscle, small intestine, spleen, and thymus). The expression data for rat came from Walker et al. (2004), the mouse data from Su et al. (2004), and the human data from Ge et al. (2005). Each tissue experiment had two replicates in mouse, a varying number of replicates in rat, and one in humans; some genes were also matched by multiple probe sets. To obtain an average across experiments and probe sets we processed the data as follows:

1. Raw CEL files of gene expression levels were obtained from the NCBI Gene Expression Omnibus database (http://www.ncbi.nlm.nih.gov/projects/geo/).
2. The results from the mouse, rat, and human arrays were normalized separately using both the MAS5 (Affymetrix 2001) and the RMA algorithms (Irizarry et al. 2003) as implemented in Bioconductor (Gentleman et al. 2004). The results are qualitatively similar for the two normalization procedures, although recent analyses suggest that MAS5 normalization is generally better (Ploner et al. 2005; Lim et al. 2007).
3. The expression of each gene within a tissue was averaged across experiments and probe sets.

We computed expression distances (ED) between orthologous gene expression profiles, for each of the three species comparisons, rat–mouse, rat–human, and mouse–human, according to the two different distance metrics, the Euclidean distance and the Pearson distance:(1)Here x_ij is the expression level of the gene under consideration in species i in tissue j, and is the average expression level of the gene in species i across tissues. Expression levels are known in a total of k tissues.Because expression levels are measured on different microarray platforms in the three species, we compute relative abundance (RA) values, before calculating the Euclidean distance (Liao and Zhang 2006a). The RA is the expression of a gene in a particular tissue divided by the sum of the expression values of that gene across all tissues. We calculated RA values to remove “probe” effects (the tendency for a gene to bind its probe set on one platform more efficiently than on another platform). Because of probe effects it is not easy to distinguish absolute changes in expression and differences in binding efficiency. Calculating RA values removes this problem from the Euclidean distance. Pearson''s distance does not change under such a rescaling and so this is unnecessary.In some analyses the logarithm of the expression or RA values are used (e.g., Makova and Li 2003; Kim et al. 2006; Xing et al. 2007), and in others the expression values are used without this transformation (e.g., Huminiecki and Wolfe 2004; Jordan et al. 2005; Yang et al. 2005; Liao and Zhang 2006a,b; Yanai et al. 2006; Urrutia et al. 2008). We calculated both the Pearson and the Euclidean distances on log-transformed and untransformed expression values. The results are qualitatively similar so here we present only the results obtained using the logarithm of the expression or RA values.It is natural to expect the two measures of expression divergence to be positively correlated with one another; however, the Euclidean and Pearson distances are almost completely uncorrelated (MAS5 normalization, mouse–rat correlation coefficient = 0.06, human–rat r = 0.13, human–mouse r = 0.10; RMA normalization, mouse–rat correlation coefficient = −0.12, human–rat r = −0.00, human–mouse r = −0.08; Figure 1). This could, plausibly, be because the two statistics measure different aspects of divergence. However, irrespective of this, there is a potential problem associated with the Pearson distance. Imagine that we have a gene that is expressed at identical levels in all tissues in two species (i.e., expression levels are uniform between tissues and also between species). We quite reasonably assume that measured expression levels contain noise. Thus each measured expression level (x_ij) is the sum of the (assumed) uniform expression level and an independent random number representing noise. In this case there is no real divergence in the expression profile between the species. However, the two measures of divergence may differ greatly in this case. The Euclidean distance reflects only the noise present in the data and hence will be small if the noise is small. By contrast, the Pearson distance will have a value close to 1 since the second term in PeaD in Equation 1 will be close to zero, reflecting the fact that the noise components of different expression levels are independent. Thus the Pearson distance will give the impression that expression divergence is great, but all this apparent divergence is noise. This will be a problem with Pearson''s distance whenever measurement error is of the same magnitude as the differences in expression between tissues. This will therefore tend to be a problem for lowly expressed genes, where measurement error can be large relative to the true value.Open in a separate windowFigure 1.—The correlation between the Euclidean and Pearson distances for (a) mouse–rat, (b) human–rat, and (c) human–mouse. Only the results from MAS5 normalization are shown; qualitatively similar results were obtained with RMA.The above example is unrealistic because real gene expression profiles are rarely perfectly uniform. To investigate whether this shortcoming of the Pearson distance is a problem in real data sets, we determined genes with a relatively uniform pattern of expression in all three species considered above. To do this we computed the entropy of a gene''s expression, which is a measure of uniformity in expression across tissues (Schug et al. 2005): the higher the value of the entropy, the more uniform is the expression. We calculated the entropy for each gene in each of the three species, averaged these across species, and then took those genes in the upper quartile of mean entropy values as a data set of genes with a relatively conserved pattern of uniform expression.It is natural to expect those genes with a conserved uniform pattern of expression to have relatively low expression divergence; however, on average these genes have significantly higher Pearson distances than other genes (Figure 2; supporting information, Figure S1 and Figure S2). By contrast, the Euclidean distance shows the pattern one would anticipate; all of the conserved uniform genes have low expression divergence. It therefore seems likely that the Pearson distance is sensitive to measurement error and hence may not be a good measure of expression divergence.Open in a separate windowFigure 2.—The distribution of expression divergence values for those genes with a uniform pattern of expression that is conserved across species vs. the distribution for all genes for (a) Pearson and (b) Euclidean distances for mouse–rat. We present similar values for human–mouse and human–rat in Figure S1 and Figure S2. Only the results from MAS5 normalization are shown; qualitatively similar results were obtained with RMA.

TABLE 1

The median expression divergence for genes that have a conserved uniform pattern of expression (upper quartile of mean entropy values) vs. all other genes

Characterization of a Thermostable Short-Chain Alcohol Dehydrogenase from the Hyperthermophilic Archaeon Thermococcus sibiricus

Short-chain alcohol dehydrogenase, encoded by the gene Tsib_0319 from the hyperthermophilic archaeon Thermococcus sibiricus, was expressed in Escherichia coli, purified and characterized as an NADPH-dependent enantioselective oxidoreductase with broad substrate specificity. The enzyme exhibits extremely high thermophilicity, thermostability, and tolerance to organic solvents and salts.Alcohol dehydrogenases (ADHs; EC 1.1.1.1.) catalyze the interconversion of alcohols to their corresponding aldehydes or ketones by using different redox-mediating cofactors. NAD(P)-dependent ADHs, due to their broad substrate specificity and enantioselectivity, have attracted particular attention as catalysts in industrial processes (5). However, mesophilic ADHs are unstable at high temperatures, sensitive to organic solvents, and often lose activity during immobilization. In this relation, there is a considerable interest in ADHs from extremophilic microorganisms; among them, Archaea are of great interest. The representatives of all groups of NAD(P)-dependent ADHs have been detected in genomes of Archaea (11, 12); however, only a few enzymes have been characterized, and the great majority of them belong to medium-chain (3, 4, 14, 16, 19) or long-chain iron-activated ADHs (1, 8, 9). Up to now, a single short-chain archaeal ADH from Pyrococcus furiosus (10, 18) and only one archaeal aldo-keto reductase also from P. furiosus (11) have been characterized.Thermococcus sibiricus is a hyperthermophilic anaerobic archaeon isolated from a high-temperature oil reservoir capable of growth on complex organic substrates (15). The complete genome sequence of T. sibiricus has been recently determined and annotated (13). Several ADHs are encoded by the T. sibiricus genome, including three short-chain ADHs (Tsib_0319, Tsib_0703, and Tsib_1998) (13). In this report, we describe the cloning and expression of the Tsib_0319 gene from T. sibiricus and the purification and the biochemical characterization of its product, the thermostable short-chain ADH (TsAdh319).The Tsib_0319 gene encodes a protein with a size of 234 amino acids and the calculated molecular mass of 26.2 kDa. TsAdh319 has an 85% degree of sequence identity with short-chain ADH from P. furiosus (AdhA; PF_0074) (18). Besides AdhA, close homologs of TsAdh319 were found among different bacterial ADHs, but not archaeal ADHs. The gene flanked by the XhoI and BamHI sites was PCR amplified using two primers (sense primer, 5′-GTTCTCGAGATGAAGGTTGCTGTGATAACAGGG-3′, and antisense primer, 5′-GCTGGATCCTCAGTATTCTGGTCTCTGGTAGACGG-3′) and cloned into the pET-15b vector. TsAdh319 was overexpressed, with an N-terminal His₆ tag in Escherichia coli Rosetta-gami (DE3) and purified to homogeneity by metallochelating chromatography (Hi-Trap chelating HP column; GE Healthcare) followed by gel filtration on Superdex 200 10/300 GL column (GE Healthcare) equilibrated in 50 mM Tris-HCl (pH 7.5) with 200 mM NaCl. The homogeneity and the correspondence to the calculated molecular mass of 28.7 kDa were verified by SDS-PAGE (7). The molecular mass of native TsAdh319 was 56 to 60 kDa, which confirmed the dimeric structure in solution.The standard ADH activity measurement was made spectrophotometrically at the optimal pH by following either the reduction of NADP (in 50 mM Gly-NaOH buffer; pH 10.5) or the oxidation of NADPH (in 0.1 M sodium phosphate buffer; pH 7.5) at 340 nm at 60°C. The enzyme exhibited a strong preference for NADP(H) and broad substrate specificity (Table ​(Table1).1). The highest oxidation rates were found with pentoses d-arabinose (2.0 U mg⁻¹) and d-xylose (2.46 U mg⁻¹), and the highest reduction rates were found with dimethylglyoxal (5.9 U mg⁻¹) and pyruvaldehyde (2.2 U mg⁻¹). The enzyme did not reduce sugars which were good substrates for the oxidation reaction. The kinetic parameters of TsAdh319 determined for the preferred substrates are shown in Table ​Table2.2. The enantioselectivity of the enzyme was estimated by measuring the conversion rates of 2-butanol enantiomers. TsAdh319 showed an evident preference, >2-fold, for (S)-2-butanol over (RS)-2-butanol. The enzyme stereoselectivity is confirmed by the preferred oxidation of d-arabinose over l-arabinose (Table ​(Table1).1). The fact that TsAdh319 is metal independent was supported by the absence of a significant effect of TsAdh319 preincubation with 10 mM Me²⁺ for 30 min before measuring the activity in the presence of 1 mM Me²⁺ or EDTA (Table ​(Table3).3). TsAdh319 also exhibited a halophilic property, so the enzyme activity increased in the presence of NaCl and KCl and the activation was maintained even at concentration of 4 M and 3 M, respectively (Table ​(Table33).

TABLE 1.

Substrate specificity of TsAdh319

Widespread Distribution of Cell Defense against d-Aminoacyl-tRNAs

Several l-aminoacyl-tRNA synthetases can transfer a d-amino acid onto their cognate tRNA(s). This harmful reaction is counteracted by the enzyme d-aminoacyl-tRNA deacylase. Two distinct deacylases were already identified in bacteria (DTD1) and in archaea (DTD2), respectively. Evidence was given that DTD1 homologs also exist in nearly all eukaryotes, whereas DTD2 homologs occur in plants. On the other hand, several bacteria, including most cyanobacteria, lack genes encoding a DTD1 homolog. Here we show that Synechocystis sp. PCC6803 produces a third type of deacylase (DTD3). Inactivation of the corresponding gene (dtd3) renders the growth of Synechocystis sp. hypersensitive to the presence of d-tyrosine. Based on the available genomes, DTD3-like proteins are predicted to occur in all cyanobacteria. Moreover, one or several dtd3-like genes can be recognized in all cellular types, arguing in favor of the nearubiquity of an enzymatic function involved in the defense of translational systems against invasion by d-amino acids.Although they are detected in various living organisms (reviewed in Ref. 1), d-amino acids are thought not to be incorporated into proteins, because of the stereospecificity of aminoacyl-tRNA synthetases and of the translational machinery, including EF-Tu and the ribosome (2). However, the discrimination between l- and d-amino acids by aminoacyl-tRNA synthetases is not equal to 100%. Significant d-aminoacylation of their cognate tRNAs by Escherichia coli tyrosyl-, tryptophanyl-, aspartyl-, lysyl-, and histidyl-tRNA synthetases has been characterized in vitro (3–9). Recently, using a bacterium, transfer of d-tyrosine onto tRNA^Tyr was shown to occur in vivo (10).With such misacylation reactions, the resulting d-aminoacyl-tRNAs form a pool of metabolically inactive molecules, at best. At worst, d-aminoacylated tRNAs infiltrate the protein synthesis machinery. Although the latter harmful possibility has not yet been firmly established, several cells were shown to possess a d-tyrosyl-tRNA deacylase, or DTD, that should help them counteract the accumulation of d-aminoacyl-tRNAs. This enzyme shows a broad specificity, being able to remove various d-aminoacyl moieties from the 3′-end of a tRNA (4–6, 11). Such a function makes the deacylase a member of the family of enzymes capable of editing in trans mis-aminoacylated tRNAs. This family includes several homologs of aminoacyl-tRNA synthetase editing domains (12), as well as peptidyl-tRNA hydrolase (13, 14).Two distinct deacylases have already been discovered. The first one, called DTD1, is predicted to occur in most bacteria and eukaryotes (see d-amino acids, including d-tyrosine (6). In fact, in an E. coli Δdtd strain grown in the presence of 2.4 mm d-tyrosine, as much as 40% of the cellular tRNA^Tyr pool becomes esterified with d-tyrosine (10).

TABLE 1

Distribution of DTD1 and DTD2 homologs in various phylogenetic groupsHomologs of DTD1 and DTD2 were searched for using a genomic Blast analysis against complete genomes in the NCBI Database (www.ncbi.nlm.nih.gov). Values in the table are number of species. For instance, E. coli is counted only once in γ-proteobacteria despite the fact that several E. coli strains have been sequenced.