Pyoverdine Receptor: a Case of Positive Darwinian Selection in Pseudomonas aeruginosa

It would seem justified to assert that, so far, no revision of the Darwinian paradigm has become necessary as a consequence of the spectacular discoveries of molecular biology. But there is something else that has indeed affected our understanding of the living world: that is its immense diversity.

Ernst Mayr (1904-2005) (24)

These phrases are taken from an essay by Ernst Mayr, the leading evolutionary biologist of the 20th century, published on the occasion of his 100th birthday. Mayr writes that evolutionary biologists had studied speciation in birds and mammals but had largely ignored the evolution of species in all other phyla, particularly in unicellular protists and prokaryotes, which will be the field for forthcoming discoveries (24). Mayr, who died on 3 February 2005, would have liked the article by Smith and colleagues in a recent issue of the Journal (32), who report on their sequence analysis of the pyoverdine locus, the most divergent region in the Pseudomonas aeruginosa core genome. The pyoverdine outer membrane receptor fvpA was found to be the most divergent alignable gene in the region. The fvpA gene showed substantial intratype variation and evidence of positive selection. Being not just another sequence paper, it provides food for thought how positive selection and adaptive evolution may work in species with very large population sizes, which is the domain of the nearly neutral theory of molecular evolution (20, 27).

In molecular evolutionary genomics, the most commonly used criterion for detecting positive selection is the rate of nonsynonymous (amino acid changing) and synonymous (silent) nucleotide substitutions in protein-coding DNA sequences. A significant excess of nonsynonymous (d_N) over synonymous (d_S) nucleotide substitutions is interpreted as evidence for the action of positive selection. Using parsimony (38) or maximum-likelihood (44) models (see references 23, 37, and 43 for the most recently developed tests and algorithms), numerous genes in viral and mammalian genomes have been identified to be evolving under the influence of positive selection. Signals of positive Darwinian selection were detected in about 5 to 6% of the genes of the human, pig, and mouse lineages (18).

Reports on positive selection in bacteria are rare. Since most bacterial species are believed to have very large population sizes (22), bacteria are expected to evolve under purifying or negative natural selection—i.e., natural selection acting to decrease the frequency of deleterious alleles. Low d_N/d_S ratios of about 0.14 that reflect the elimination of deleterious nonsynonymous substitutions were indeed observed when large data sets of phylogenetically independent pairs of genes between closely related pairs of bacterial species were compared (11, 17). On the other hand, a recent examination of DNA sequence of polymorphism in 149 data sets from 84 bacterial species indicated that bacterial populations harbor abundant slightly deleterious, nearly neutral nonsynonymous substitutions, which are subject to ongoing purifying selection (16). The major exceptions to this trend were seen among surface proteins, particularly those of bacteria parasitic on vertebrates. Evasion of host immune and defense mechanisms apparently favors amino acid sequence diversity of immunogenic bacterial proteins.

Correspondingly the few cases of positive selection that yet have been reported are components of the outer membrane or cellular appendages of pathogens that are exposed to the mammalian immune system: the outer surface protein of Borrelia burgdorferi (40), the pilus protein PilE (1) and the outer membrane protein PorB of Neisseria meningitidis (33, 41), and the flagellin and intimin proteins of pathogenic Escherichia coli (31, 39).

At first glance, the positive selection found for two genes in the pyoverdine locus of P. aeruginosa cannot be easily attributed to the evolutionary dynamics of a mammalian host-bacterial parasite relationship as it applies to all previously reported cases. First, the metabolically versatile species P. aeruginosa is ubiquitously distributed in soil and aquatic habitats and is not a classical pathogen with pronounced host and tissue tropism, albeit it may cause disease in predisposed animals, plants, and humans (5). Second, pyoverdines are not subject to immune surveillance by the infected host but are fluorescent high-affinity peptidic siderophores that are excreted by all fluorescent Pseudomonas species (30). In P. aeruginosa, pyoverdines are not only iron(III) chelators but also signal molecules for the production of virulence factors (21). Each strain of P. aeruginosa makes one of three pyoverdine types, each with a distinct peptide chain that is synthesized by nonribosomal peptide synthetases (6, 9, 25).

The pyoverdine receptor is also normally type specific, transporting only its corresponding pyoverdine. Now, the receptors for the three different pyoverdine types of P. aeruginosa have been identified (8, 32, 36). From the article by Spencer et al., it became clear that the type II pyoverdine receptors could be split into two subtypes: one corresponding to the sequence described for the rhizosphere P. aeruginosa 7NSK2 (8) and one corresponding to one sequence derived from a cystic fibrosis isolate (36).

To clarify why Smith's comparative sequence analysis of the 50-kb pyoverdine locus in nine P. aeruginosa strains is a substantial contribution to the evolutionary genomics of bacteria, some information about our current knowledge on genome organization and population biology of P. aeruginosa will be helpful for the reader to put the paper into perspective: The P. aeruginosa genome is a mosaic of a conserved core and variable accessory segments, the latter made up by genome islands and genome islets (10, 14, 19, 36). The core genome is characterized by a conserved synteny of genes and a low average nucleotide divergence of 0.5% (19, 36). The global d_N/d_S ratio of about 0.16 provides strong evidence for purifying selection in the core genome. P. aeruginosa has a nonclonal population structure (7, 19, 29). All genotypes, each of which is characterized by nonrandom association of alleles, are in linkage equilibrium to each other. In other words, the frequency of recombination is high enough to prevent linkage disequilibrium among gene loci and hitchhiking effects on phenotype.

Twenty-five regions of significantly elevated sequence variation were uncovered in pairwise comparisons between PAO1 and three other partially sequenced strains (36), whereby the flagellar regulon, the O-antigen biosynthesis locus, the pilin gene pilA, and the pyoverdine locus exhibited the largest sequence diversity in the core genome. The maintenance of multiple alleles indicates diversifying selection for these gene loci. Particularly the pilin and pyoverdine loci share numerous features. First, they exhibit P. aeruginosa atypical codon usage (19, 32) and atypical tetranucleotide frequency distribution (32, 42), indicating acquisition by horizontal gene transfer. Second, the homologous proteins show fewer similarities between types than with other species. In the case of the pyoverdine receptor FpvA, the highest similarity is with a nonpseudomonad: type II FpvA is similar to a receptor from Agrobacterium tumefaciens, and type I is similar to a receptor from Azotobacter vinelandii (32). Similarly, the five groups of type IV pili of P. aeruginosa are no more closely related to each other than they are to the pili of other β-, γ-, or δ-Proteobacteria (34). Although pilins are highly divergent by amino acid alignment, they share function, protein domains, and a similar arrangement of secondary structure elements and primary structure sequence motifs. The same applies to the highly divergent genes in the central pyoverdine region. The proteins encoded by colinear genes share protein domains, secondary structure elements, and—probably—function (32).

Mosaic genes are a major source of genetic diversity. Known mosaic genes in P. aeruginosa are ampC (35), fleP (2), fliC (19, 35), mucABCD (4), and oprD (28). In all these cases, intragenic recombination did not change the sequence of the recombined blocks. The situation is different for the pyoverdine locus. Eleven regions of about 100 bp in length that are located within about 400 bp of a region of intertype divergence show pronounced nucleotide divergence within a type (32). By shuffling sequences between types at the edges of divergent regions, recombination created intratype divergence and simultaneously eroded the boundary regions of high intertype divergence.

Recombination with other pyoverdine types accounts for most intratype divergence at the pyoverdine locus. The exceptions are the two genes PA2403 and PA2398 that flank the nonribosomal peptide synthetase complex. PA2403 encodes a protein involved in the regulation of pyoverdine production, and PA2398 encodes the outer membrane pyoverdine receptor FpvA. Both genes exhibit high d_N/d_S ratios as evidence of positive selection. Hydropathy profiles and known tertiary structures of homologs suggest that the positively selected amino acid residues cluster in spatially defined regions.

In their article, Smith and colleagues (32) propose that FpvA drives the diversity of the pyoverdine locus. Receptor and pyoverdine peptide should coevolve to maintain mutual specificity. Consistent with this hypothesis, fpvA and the nonribosomal peptide synthetase genes are adjacent to each other, which facilitates coevolution of the siderophore peptide chain and its receptor en bloc by counterselection of recombinations that separate the genes. Coevolution makes sense, but what could be the sources for this diversifying selection? First of all, the coordinate synthesis of siderophore and receptor and the regulation thereof combine the acquisition of the essential, but scarcely bioavailable, nutrient iron(III) from the environment with the control by the cognate receptor to keep intracellular iron below the hazardous level. Second, this strategy may defend against the stealing of the ferric siderophore by other strains because ferric pyoverdine is supposed to be taken up only by strains of the same type (15). Within one type, siderophore production is an altruistic cooperative trait that is costly for the individual but provides a local (group) benefit because other individuals can take up the siderophore-iron complex (13). However, this cooperative behavior can be undermined by cheaters that do not synthesize the siderophore but express a compatible receptor. Such a situation has been studied in an experimental model of the growth of P. aeruginosa PAO1 (cooperator) and an isogenic pyoverdine-negative mutant (cheater) under iron limitation (13). Cooperation was favored by higher relatedness and more global competition, whereas the cheaters dominated the population in the case of local competition. P. aeruginosa mutants that do not produce siderophores have been seen to evolve in the lungs of cystic fibrosis patients (9). Pyoverdine-negative isolates accumulated with lung colonization time, and numerous patients were cocolonized with pyoverdine-positive and pyoverdine-negative strains of the same genotype. The pyoverdine-negative mutants retained the capacity to take up pyoverdine, and numerous type II strains (and type III) could also utilize type I pyoverdine, due to the production of a ubiquitous second receptor for type I pyoverdine receptor, FpvB (12). Also, the type III receptor was found to lack specificity since it can recognize the type II pyoverdine as well (12). This scenario of defense and sharing of and cheating over resources makes it plausible why positive selection was detected for the pyoverdine receptor FpvA. It may be lucky coincidence that four of the nine sequenced P. aeruginosa strains were isolated from patients with cystic fibrosis (32).

There is an additional major point which may account for positive selection and substantial intratype sequence variation of the type II ferripyoverdine receptor: it is used by pyocin S3 to gain entry into the cell (3). Pyocins are bacteriocins produced by P. aeruginosa that kill strains of the same species. Different types have been described: among them are the S-type pyocins, which have two components, a protein with DNase activity and an immunity protein which confers protection to the producing strain (26). Strains producing type II pyoverdine are susceptible to killing by pyocin S3, whereas strains producing type I or type III pyoverdines are pyocin S3 resistant. Type II pyoverdine decreases the killing activity of the pyocin for a type II cell, whereas the presence of type I or type III pyoverdines increases the killing activity (3). Such an antagonism that the ferripyoverdine receptor confers the supply of a vital nutrient, but also mediates the entry of a lethal weapon (the latter situation aggravated by the presence of P. aeruginosa strains of other pyoverdine types) makes it plausible why, despite global purification selection of the core genome, diversification and positive selection dominate at the pyoverdine locus. In summary, the pyoverdine region is the most divergent locus of the core genome because it is subject to speciation and coevolution; encodes a trait of altruistic cooperation (the production of siderophores), and encodes a receptor that is both a major fitness allele and a major deleterious allele. Ernst Mayr probably would have loved this unusual story of speciation in a prokaryote.