Molecular diversity and evolution of myticin-C antimicrobial peptide variants in the Mediterranean mussel, Mytilus galloprovincialis

peptides 29 (2008) 1094–1101

available at www.sciencedirect.com

journal homepage: www.elsevier.com/locate/peptides

Molecular diversity and evolution of myticin-C antimicrobial peptide variants in the Mediterranean mussel, Mytilus galloprovincialis Abinash Padhi *, Bindhu Verghese Department of Biological Sciences, 800 S. Tucker Drive, University of Tulsa, OK 74104, USA

article info

abstract

Article history:

Mussels have diverse groups of cysteine rich, cationic antimicrobial peptides (AMPs)

Received 2 January 2008

(defensins, mytilins, myticins, and mytimycin) that constitute an important component

Received in revised form

of their innate immune defence. Despite the identification and characterization of these

5 March 2008

AMPs in mussels, the underlying genetic mechanisms that maintain high diversity among

Accepted 6 March 2008

multiple variants of the myticin-C isoform are poorly understood. Using phylogeny-based

Published on line 15 March 2008

models of sequence evolution and several site-by-site frequency spectrum statistical tests for neutrality, herein we report that positive selection has been the major driving force in

Keywords:

maintaining high diversity among the allelic-variants of the myticin-C AMP of Mytilus

Mussel

galloprovincialis. The statistical tests rejected the hypothesis that all polymorphism within

Antimicrobial peptides

myticin-C loci is neutral. Although a majority of the codons constrained to purifying

Myticin

selection (rate of amino acid replacement to the silent substitution, v < 1), approximately

Molecular evolution

8% of the codons with v  5.5 are under positive selection (v > 1), thus indicating adaptive

Maximum-likelihood models

evolution of certain amino acids. Direct interaction of these peptides with the surrounding

Positive selection

pathogens and/or altered/new pathogens in the changing environment is the likely cause of molecular adaptation of certain amino acid sites in myticin-C variants. # 2008 Elsevier Inc. All rights reserved.

1.

Introduction

The Mediterranean mussel, Mytilus galloprovincialis, which belongs to the Phylum Mollusca and Family Mytilidae, is primarily found in estuarine and marine habitats, and often in exposed rocky substrata or sandy bottoms [9]. Like other marine invertebrates [34], this ecologically and geographically diversified organism also relies solely on its innate immune response, and antimicrobial peptides (AMPs) are the major component of its innate immune defense system [5,21,34]. Due to their direct involvement with altered/new pathogens, AMPs exhibit an extraordinary diversity in their structure and function [23,33,34]. For example, four diverse groups of AMPs

with multiple isoforms (defensins: MGD1, MGD2; mytilins: A, B, C, D, and G1; myticins: A, B and C; and mytimycin) have been identified and characterized in marine mussels, M. galloprovincialis and M. edulis [21,25]. All these AMPs in marine mussels showed similar structural features: the presence of an Nterminal signal peptide, followed by the mature peptide region that is characterized by 6–8 conserved cysteine residues, and a C-terminal extension rich in anionic residues [21]. These small cationic AMPs are active against Gram-positive and Gramnegative bacteria, and some of these peptides have also shown anti-fungal activities [4,20,21]. Among the three isoforms (A, B, and C) of myticin discovered so far in the marine mussel M. galloprovincialis [19,25], myticin-C

* Corresponding author. Present address: Center for Infectious Disease Dynamics, Department of Biology, 208 Mueller Laboratory, The Pennsylvania State University, University Park, PA 16802, USA. Tel.: +1 814 867 2122. E-mail addresses: [email protected], [email protected] (A. Padhi). 0196-9781/$ – see front matter # 2008 Elsevier Inc. All rights reserved. doi:10.1016/j.peptides.2008.03.007

peptides 29 (2008) 1094–1101

is reported to have several allelic-variants [25]. Despite the high structural similarity with other molluscan AMPs, myticin-C is the only peptide reported to be expressed in larval developmental stages of mussels, and is thus believed to be developmentally regulated as well as associated with antimicrobial activities from the larval stages [25]. Since these peptides might

1095

be directly interacting with altered/new pathogens in the surrounding environment/habitats, one would expect that some form of selection might have played a crucial role in generating high allelic diversity in myticin-C from the larval stages. Using phylogeny-based models of sequence evolution [39] and several statistical tests for neutrality [7,8,31,42], here we

Fig. 1 – Maximum likelihood-based tree showing phylogenetic relationships among different myticin-C variants. The tree is rooted with myticin-A and -B proteins. Bootstrap support I50 is given at the base of the nodes.

1096

peptides 29 (2008) 1094–1101

Table 1 – Polymorphisms and tests for neutrality in myticin-C variants of M. galloprovincialis Summary statistics S p uW DTajima DFu and H-test E-test

Li

Signal peptide 12 0.0224 0.0410 1.207 1.544 0.591 1.465

( p = 0.097) ( p = 0.084) ( p = 0.224) (p = 0.024)

Mature peptide 54 0.0583 0.0923 1.172 3.257 0.639 1.587

( p = 0.086) (p = 0.004) ( p = 0.213) (p = 0.023)

C-terminal anionic-rich region 51 0.0474 0.0872 1.447 3.330 0.546 1.754

(p = 0.034) (p = 0.005) ( p = 0.335) (p = 0.008)

All regions 117 0.0467 0.0800 1.370 3.585 0.633 1.810

(p = 0.048) (p = 0.001) ( p = 0.225) (p = 0.007)

p-Value is given in parentheses. Significant p-values (<0.05) are in bold. S, number of segregating sites; p, average number of nucleotide differences per site; uW, nucleotide diversity based on the proportion of segregating sites [36]; DTajima, Tajima’s D [31]; DFu and Li [8], H-test [7], Etest [42].

evaluate the importance of natural selection in the evolution of multiple variants of myticin-C. Identifying genes that underlie adaptation is central to understanding molecular evolution. Therefore, analyzing patterns of amino acid substitutions would provide additional insight into an understanding of protein adaptation by identifying amino acids that have evolved under positive selection (e.g., [24]). At the genomic level, although a vast majority of genes have evolved under purifying selection [12], many genes that are associated with specific functions, for example, genes associated with antimicrobial activities, are reported to have evolved under positive selection [24,33,35]. If the rate of nonsynonymous substitutions (dn) is higher than the silent/synonymous substitutions (ds), positive selection (dn/ ds = v > 1) is said to be operating. Alternatively, if v is less than one, the gene is said to be under purifying selection [11]. This approach has been widely used in detecting selection pressures in protein coding DNA sequences by comparing the ‘dn’ and ‘ds’ in many antimicrobial peptides [24,33]. In this paper, we employed the same approach to evaluate the importance of natural selection in generating high diversity in myticin-C. To our knowledge, this is the first attempt to study molecular evolution of myticin antimicrobial peptides in a marine mussel.

2.

Materials and methods

Using the M. galloprovincialis myticin-C (GenBank accession: CAM56789) as a query for a homology search, we performed PSI-

Fig. 2 – dn to ds ratios vs. ds estimated from the myticin-C antimicrobial peptide variants. Signal peptides were excluded from the analyses.

Fig. 3 – Sliding window analyses showing the divergence at synonymous and nonsynonymous sites (window length = 20, step size = 10).

BLAST [1] against the NCBI non-redundant protein database. From the PSI-BLAST query result, we retrieved a total of 74 unique coding nucleotide sequences representing myticin-C allelic-variants [25], and two sequences representing myticin-A (GenBank accession: AAD47638) and myticin-B (GenBank accession: AAD47639) (Fig. 1). For selection analyses we reconstructed a ML tree using the appropriate nucleotide substitution model selected by the hierarchical likelihood ratio test (hLRT) implemented in Modeltest ver. 3.5 [26]. The inferred tree was used in the ML-based selection analyses. Sequences were aligned using DAMBE ver. 4.5.2 [37,38], BioEdit ver. 7.0.5.3 [10], and Mesquite ver. 1.12 [16]. The Hasegawa–Kishino–Yano (HKY) model with an estimation of the proportion of invariable sites (I = 0.3835) and a gamma distribution shape parameter (G = 0.8726) was identified as the appropriate nucleotide substitution model by hLRT. The ML tree was reconstructed with a TBR tree searching method implemented in PAUP* 4.0b10 [30]. We used the Maxchi method [17] to test for recombination. DNAsp ver. 4 [27] was used for sliding window analyses. We used several ‘site-by-site frequency spectrum’-based statistical tests to test whether all polymorphism within loci is neutral. Two measures of nucleotide variability, p [22] and uW [36], were calculated. For a neutrality test, we used several statistical tests such as Tajima’s D [31], Fu and Li’s D [8], Fay and Wu’s H-test [7], and Zeng et al.’s E-test [42] implemented in the program NeutralityTest ver. 1.2 ([15]; available at http:// hgc.sph.uth.tmc.edu/neutrality_test/). To know whether there is significant heterogeneity in the distribution of v across the entire codon of myticin-C variants,

peptides 29 (2008) 1094–1101

we compared the model M0 that assumes uniform v for the entire codon with the model M3, which distinguishes three v site classes, one of which is allowed to exceed v = 1 [39,41]. To assess the significance of the findings, a likelihood ratio test (LRT) for rate heterogeneity was carried out. To determine whether positive selection is operating in any of the codon sites of myticin-C variants, we estimated parameters under five different codon substitution models [39] as implemented in the CODEML program of PAML ver. 3.15 [41]. Model performance was evaluated using LRTs. The LRTs were used to compare models that assume no positive selection (v < 1) with those that assume positive selection (v > 1). The five-codon substitution models are M1a (nearly neutral), M2a (positive selection), M7 (b-distribution; 0  v  1), M8 (b + v > 1; continuous) [39], and M8a (b + v = 1) [29]. The LRTs were conducted to evaluate the model that best fit the data. The null models (M1a, M7, and M8a) were compared with their corresponding alternative models (M2a and M8) [39]. Posterior probabilities of the inferred positively selected sites were estimated by the Bayes empirical Bayes (BEB) approach that takes sampling errors into account [40]. Amino acid alignments were used to calculate mean hydrophobicity profiles [13] and the flexibility index [2] using the web server: http://www.expasy.ch/tools/protscale.html. Secondary structure predictions and associated confidence values for the myticin-A, -B and -C mature peptides were made by using the PSIPRED [18] Protein Structure Prediction Server (http://bioinf.cs.ucl.ac.uk/psipred/).

3.

Results

The inferred phylogeny showed that all the myticin-C variants formed a distinct cluster with strong bootstrap support (Fig. 1), thus allowing us to perform selection analyses. The number of polymorphic sites within each region is listed in Table 1. The number of segregating sites (S), average number of nucleotide differences per site (p), as well as the nucleotide diversity based on the proportion of segregating sites (uW) for mature peptide regions are relatively higher than their corresponding values for signal peptide and C-terminal anionic-rich regions. While Tajima’s D, Fu and Li’s D as well as Zeng et al.’s E-test

1097

rejected the hypothesis that all the polymorphims within the loci are neutral, Fay and Wu’s H-test could not rejected the null hypothesis of neutral evolution (Table 1). The pairwise comparisons (Fig. 2) and the sliding window analyses (Fig. 3) suggested that although a majority of the codons are constrained to purifying selection, certain codons appear to be under positive selection (diversifying selection). However, since the sliding window analyses or pairwise comparisons are unlikely to identify codons that are under positive selection, we used phylogeny-based codon substitution models to identify these codons. There is no evidence of recombination among the variants of myticin (Maxchi test, p > 0.1), thus allowing us to perform a phylogeny-based test for positive selection. To find whether there is any evidence of rate heterogeneity among the codons, we compared the M0 model with the model M3 that has three classes of v. The M0– M3 comparison revealed that M3 is a better fit to the data (x2 = 138.75, d.f. = 4, p < 0.00001), thus suggesting significant variation in v among the codons (Fig. 4). The M3 model showed that approximately 72% of the codons have v0 = 0.257, whereas 24% of codons have v1 = 2.23, and only 4% of the codons have v2 = 7.48. Under the M3 model, while sites: 23, 24, 26, 28, 29, 41, 46, 59 that are within the mature peptide regions are under positive selection, sites: 76, 77, 82, 86, 90, 91, 92, 97, 98 that are within the C-terminal anionic region are under positive selection with naı¨ve posterior probability 0.95. The comparison of null models (M1a, M7, and M8) with their corresponding alternative models (M2a and M8) rejected the null models (Table 2), thus indicating the variants of myticin-C have evolved under positive selection. All the selection models (M2a and M8) are consistent with the fact that approximately 8% of the codons with overall v  5.5 are positively selected (Table 2). To avoid the possibility of false positives we considered a site under positive selection if the BEB posterior probability is 0.95. Of the six positively selected sites each with BEB posterior probability 0.95, only one site (site-59) is within the mature peptide region, and remaining sites are within the C-terminal anionic region (Fig. 5a). A secondary structure prediction of the mytcin-C mature peptide domain suggests that the positively selected residue (site-59) selected by both M2a and M8 selection models falls exclusively in predicted coils (Fig. 5a). However, under the M3

Fig. 4 – Posterior probability of v for each codon site is determined by the M3 model implemented in the CODEML program [41]. Three distribution classes of v (v0, v1, and v2) and their corresponding posterior probability values for each codon site are mapped.

1098

peptides 29 (2008) 1094–1101

Table 2 – Comparison among different nested models to test for positive selection among codons Model comparison

x2-value

d.f.

p-Value for best model

v > 1 parametersa

Positively selected sitesb (posterior probability)

M1a vs. M2a

59.75847

2

0.0000

v = 5.377 p = 0.08

59-S (0.991) 77-K (0.937) 86-V (0.985) 90-V (1.000) 91-E (0.776) 92-H (0.969) 98-N (1.000)

5.394  0.983 5.134  1.372 5.359  1.039 5.433  0.898 4.320  1.928 5.274  1.156 5.433  0.898

M7 vs. M8

66.00014

2

0.0000

v = 5.5164 p = 0.082

59-S (0.997) 77-K (0.976) 86-V (0.996) 90-V (1.000) 91-E (0.912) 92-H (0.992) 98-N (1.000)

5.102  0.976 5.003  1.130 5.095  0.984 5.116  0.948 4.667  1.452 5.074  1.011 5.116  0.948

M8 vs. M8a

58.36007

1

0.0000

v = 5.5164 p = 0.082

59-S (0.997) 77-K (0.976) 86-V (0.996) 90-V (1.000) 91-E (0.912) 92-H (0.992) 98-N (1.000)

5.102  0.976 5.003  1.130 5.095  0.984 5.116  0.948 4.667  1.452 5.074  1.011 5.116  0.948

v  S.E.

a

Parameter estimates, likelihood scores, positively selected sites under different selection models with their corresponding Bayes empirical Bayes posterior probabilities, and the v value of each positively selected site with standard error are listed. b p is the proportion of codons that have v > 1. Amino acid sites are corresponding to the amino acid sites in GenBank accession number CAM56789.

model, seven additional sites were positively selected each with naı¨ve posterior probability 0.95: 2 sites were located on a coil, 3 sites on the a-helix, and the two remaining sites were located on the b-sheet (Fig. 5a). Our results have also demonstrated that although the predicted secondary structures of myticin-A, -B, and -C are remarkably similar (Fig. 5b), the mean hydrophobicity and flexibility indices among these three proteins showed subtle differences (Fig. 6a and b). While myticin-B is more hydrophilic within the mid-region (mature peptide region: residue 40–50 in Fig. 6a), myticin-A and -C are hydrophobic. Similarly, towards the C-terminal anionic region, while both A and B showed similar properties, myticin-C tends to be more hydrophilic (Fig. 6a). Many of these sites in the C-terminal anionic regions are also positively selected (Fig. 5a), thus indicating adaptive molecular evolution of this region driven by positive Darwinian selection.

4.

Discussion

The phylogenetic placement of each variant is consistent with previously reported results that are based on the distancebased neighbor joining tree [25]. If a genomic region is likely to have been subject to recent selective sweeps, p is expected to be lower than uW, and Tajima’s D [31], Fu and Li’s D [8], and Fay and Wu’s H [7] are expected to be significantly different from the neutral expectation of zero. For the entire myticin-C sequences, although Tajima’s D and Fu and Li’s D, which test whether there are too few or many rare variants than the common ones, rejected the hypothesis that all polymorphism within the loci is neutral, Fay and Wu’s H-test [7], which takes into consideration the abundance of very high-frequency

variants relative to the intermediate frequency ones, could not reject the null hypothesis of neutral evolution (Table 1). However, Zeng et al.’s E-test [42], which contrasts the very high- and very low-frequency variants, rejected the null hypothesis (Table 1), thus suggesting that the excess of highfrequency derived polymorphisms within myticin-C loci have evolved in a non-neutral fashion. This pattern has also been observed in several antibacterial peptide genes in Drosophila melanogaster [14]. By contrast, although AMPs in frogs have been shown to evolve under positive selection, polymorphisms within loci are reported to be neutral; therefore, frequent gene duplication is the possible alternative explanation for such diversity of these peptide variants [32]. Although the M0–M3 comparison suggests that M3 is a better fit to the data, indicating significant variation in the distribution of v, the M0–M3 comparison is not the ideal approach to precisely estimate the percentage of positively selected codons [39]. Most of the positively selected codons under the M2a and M8 models have v  5, with BEB posterior probability greater than 0.95 (Table 2). Consistent with earlier studies on molecular evolution of several AMPs (e.g., [3,6,23,24,28,33]), the results of the present study also provide convincing evidence that positive selection is the major driving force in generating high diversity in myticin-C. The direct involvement of these peptides with the altered/new pathogens in a changing environment is the possible cause of such adaptive molecular evolution of these AMPs [3,6,23,24,28,33]. Under the M2a and M8 models, while only one site (site-59) is positively selected in the mature peptide region, the majority of the positively selected sites are located in the C-terminal anionic rich residues region. The functional significance of the C-terminal region is unknown. However,

peptides 29 (2008) 1094–1101

1099

Fig. 5 – (a) Amino acid variable sites among 74 myticin-C variants. Amino acid residues from site 1–20 are signal peptides, from sites 21–60 are mature peptides, and from sites 61–100 are the C-terminal anionic rich peptides. Sites shaded in both light and dark grey are the positively selected sites selected under the M3 model with naı¨ve posterior probability I0.95; whereas, sites shaded with only dark grey are the positively selected sites selected by the M2a and M8 models. The predicted disulphide bridge among cysteine residues is also shown and (b) secondary structure prediction for myticin-A, -B, and -C mature peptide regions. Structure prediction and confidence values (0, low; 9, high) were made by using the PSIPRED server [18]. Boxes indicate the a-helix, arrows indicate b-strands, and lines indicate coils. this region might interact with the active peptide (i) to neutralize positive charges of the peptide, allowing suitable proteolytic processing, and/or addressing to a particular hemocyte compartment, or (ii) to protect the cells from

eventual cytotoxic effects [21]. Nevertheless, our results suggest a putative important role for these positively selected codon sites in the evolution of myticin-C, and this information might be useful when undertaking functional analyses to

1100

peptides 29 (2008) 1094–1101

Fig. 6 – (a) The distribution of hydrophobic and hydrophilic regions along the predicted amino acid sequences of myticin-A, B and -C as displayed above and below the central axes, respectively, was obtained using the Kyte and Doolittle [13] algorithm and (b) the distribution of flexibility index along the predicted amino acid sequences of myticin-A, -B and -C was obtained using the Bhaskaran and Ponnuswamy [2] algorithm. The hydropathy and flexibility indices of each amino acid were calculated over a window of 9 residues.

determine whether these positively selected amino acid sites play a crucial role in antimicrobial activity. Although under the M2a and M8 selection models only one site was positively selected within the mature peptide region, under the M3 model seven additional sites within the mature peptide region showed higher nonsynonymous substitutions, and some of these sites can be mapped into the predicted a-helix and bsheets (Fig. 5a). Thus, the findings of several hotspots for nonsynonymous substitutions within the domains of this functionally important region suggest that molecular interaction between hosts and the invading pathogens might be the possible explanation for such adaptive molecular evolution. Previous studies (reviewed in [43]) have also demonstrated that many members of the cysteine-stabilized a-helix and bsheet (CSab) superfamily showed adaptive molecular evolution driven by positive Darwinian selection. Despite limited amino acid identity [25], the predicted secondary structures of myticin-A, -B and -C proteins are remarkably similar (Fig. 5b), and thus indicate that these three proteins have been conserved during the course of evolution. Previously, Zhu et al. [43] reported that although many members of the CSab superfamily exhibit diverse biochemical and biological functions, they still retain the conserved structural motifs, and

positive Darwinian selection is the major driving force in generating such diverse biological functions. In agreement with previous studies on the molecular evolution of other members of the CSab superfamily (reviewed in [43]), the present study has also provided convincing evidence of positive natural selection on the myticin-C antimicrobial peptides of M. galloprovincialis. The direct interaction of this antimicrobial protein with the surrounding invading pathogens might be the possible explanation for such adaptive molecular evolution of myticin-C.

Acknowledgments We thank Dr. Peggy Hill and two anonymous reviewers for thoughtful comments towards improving the manuscript.

references

[1] Altschul SF, Madden TL, Scha¨ffer AA, Zhang J, Zhang Z, Miller W, et al. PSI-BLAST: a new generation of protein

peptides 29 (2008) 1094–1101

[2]

[3]

[4]

[5]

[6]

[7] [8] [9] [10]

[11]

[12] [13]

[14]

[15]

[16]

[17] [18] [19]

[20]

[21]

[22]

database search programs. Nucleic Acids Res 1997;25: 3389–402. Bhaskaran R, Ponnuswammy PK. Positional flexibilities of amino acid residues in globular proteins. Int J Pept Protein Res 1988;32:242–55. Bulmer M, Crozier R. Duplication and diversifying selection among termite antifungal peptides. Mol Biol Evol 2004;21:2256–64. Cellura C, Toubiana M, Parrinello N, Roch P. Specific expression of antimicrobial peptide and HSP70 genes in response to heat-shock and several bacterial challenges in mussels. Fish Shellfish Immunol 2007;22:340–50. Chisholm JRS, Smith VJ. Antibacterial activity in the hemocytes of the shore crab Carcinus maenas. J Mar Biol Assoc 1992;72:529–42. Duda TJ, Vanhoye D, Nicolas P. Roles of diversifying selection and coordinated evolution in the evolution of Amphibian antimicrobial peptides. Mol Biol Evol 2002;19:858–64. Fay JC, Wu C-I. Hitchhiking under positive Darwinian selection. Genetics 2000;155:1505–13. Fu YX, Li WH. Statistical tests of neutrality of mutations. Genetics 1993;133:693–709. Gosling E. The mussel Mytilus: ecology, physiology genetics and culture. Amsterdam: Elsevier; 1992. Hall T. BioEdit: a user friendly biological sequence alignment editor and analysis program for windows 95/98/ N.T. Nucleic Acids Symp Ser 1999;41:95–8. Hughes A, Nei M. Nucleotide substitution at major histocompatibility complex class II loci: evidence for overdominant selection. Proc Natl Acad Sci USA 1989;86:958–62. Kimura M. The neutral theory of molecular evolution. New York: Cambridge University Press; 1983. Kyte J, Doolittle RF. A simple method for displaying the hydropathic character of a protein. J Mol Biol 1982;157: 105–32. Lazzaro BP, Clark AG. Molecular population genetics of inducible antimicrobial peptide genes in Drosophila melanogaster. Mol Biol Evol 2003;20:914–23. Li H, Fu YX. NeutralityTest: novel software for performing tests of neutrality. Version 1.2, available at http:// hgc.sph.uth.tmc.edu/neutrality_test/; 2007. Maddison Y, Maddison D. Mesquite: a modular system for evolutionary analysis. Version 1.12. http:// mesquiteproject.org; 2006. Maynard Smith JM. Analyzing the mosaic structure of genes. J Mol Evol 1992;34:126–9. McGuffin LJ, Bryson K, Jones DT. The PSIPRED protein structure prediction server. Bioinformatics 2000;16:404–5. Mitta G, Hubert F, Noel T, Roch P. Myticin, a novel cysteinerich antimicrobial peptide isolated from hemocytes and plasms of the mussel Mytilus galloprovincialis. Eur J Biochem 1999;265:71–8. Mitta G, Vandenbulcke F, Roch P. Mussel defensins are synthesized and processed in granulocytes then released into the plasma after bacterial challenge. J Cell Sci 1999;112:4233–42. Mitta G, Vandenbulcke F, Roch P. Original involvement of antimicrobial peptides in mussel innate immunity. FEBS Lett 2000;486:185–90. Nei M. Molecular evolutionary genetics. New York: Columbia University Press; 1987.

1101

[23] Nicolas P, Vanhoye D, Amiche M. Molecular strategies in biological evolution of antibacterial peptides. Peptides 2003;24:1669–80. [24] Padhi A, Verghese B, Otta SK, Varghese B, Ramu K. Adaptive evolution after duplication of penaeidin antimicrobial peptides. Fish Shellfish Immunol 2007;23:553–66. [25] Pallavicini A, Costa MM, Gestal C, Dreos R, Figueras A, Venier P, et al. High sequence variability of myticin transcripts in hemocytes of immune-stimulated mussels suggests ancient host-pathogen interactions. Dev Comp Immunol 2008;32:213–26. [26] Posada D, Crandall K. Modeltest: testing the model of DNA substitution. Bioinformatics 1998;14:817–8. [27] Rozas J, Sa´nchez-DelBarrio JC, Messeguer X, Rozas R. DnaSP, DNA polymorphism analyses by the coalescent and other methods. Bioinformatics 2003;19:2496–7. [28] Semple C, Rolfe M, Dorin J. Duplication and selection in the evolution of primate b-defensin genes. Genome Biol 2003;4:R31. [29] Swanson W, Nielsen R, Yang Q. Pervasive adaptive evolution in mammalian fertilization proteins. Mol Biol Evol 2003;20:18–20. [30] Swofford DL. PAUP*: phylogenetic analyses using Parsimony (*and other methods) 4.0 b. Sunderland (MA), USA: Sinauer Associates; 2002. [31] Tajima F. Statistical method for testing the neutral mutation hypothesis by DNA polymorphism. Genetics 1989;123:585–95. [32] Tennessen JA, Blouin MS. Selection for antimicrobial peptide diversity in frogs leads to gene duplication and low allelic variation. J Mol Evol 2007;65:605–15. [33] Tennessen JA. Molecular evolution of animal antimicrobial peptides: widespread moderate positive selection. J Evol Biol 2005;18:1387–94. [34] Tinku JA, Taylor SW. Antimicrobial peptides from marine invertebrates. Antimicrob Agents Chemother 2004;48: 3645–54. [35] Vanhoye D, Bruston F, Nicolas P, Amiche M. Antimicrobial peptides from hylid and ranid frogs originated from a 150million-year-old ancestral precursor with a conserved signal peptide but a hypermutable antimicrobial domain. Eur J Biochem 2003;270:2068–81. [36] Watterson GA. On the number of segregating sites in genetical models without recombination. Theor Popul Biol 1975;7:256–76. [37] Xia X, Xie Z. DAMBE: data analysis in molecular biology and evolution. J Heredity 2001;92:371–3. [38] Xia X. Data analysis in molecular biology and evolution. Boston: Kluwer Academic Publishers; 2000. [39] Yang Z, Nielsen R, Goldman N, Pederson A. Codonsubstitution models for heterogeneous selection pressures at amino acid sites. Genetics 2000;155:431–49. [40] Yang Z, Wong WSW, Nielsen R. Bayes empirical bayes inference of amino acid sites under positive selection. Mol Biol Evol 2005;22:1107–18. [41] Yang Z. PAML: a program package for phylogenetic analysis by maximum likelihood. Comput Appl Biosci 1997;13:555–6. [42] Zeng K, Fu YX, Shi S, Wu CI. Statistical tests for detecting positive selection by utilizing high-frequency variants. Genetics 2006;174:1431–9. [43] Zhu S, Gao B, Tytgat J. Phylogenetic distribution, functional epitopes and evolution of the CSab superfamily. Cell Mol Life Sci 2005;62:2257–69.