Abstract

Motivation: Functional divergence among proteins is often assumed to be strongly influenced by natural selection, as inferred from the ratio of non-synonymous nucleotide divergence (dN) to synonymous nucleotide divergence (dS). That is, the more a mutation changes protein function, the more likely it is to be either selected against or selectively favored, and because the dN/dS ratio is a measure of natural selection, this ratio can be used to predict the degree of functional divergence (dF). However, these hypotheses have rarely been experimentally tested.

Results: I present a novel method to address this issue, and demonstrate that divergence in bacteria-killing activity among animal antimicrobial peptides is positively correlated with the log of the dN/dS ratio. The primary cause of this pattern appears to be that positively selected substitutions change protein function more than neutral substitutions do. Thus, the dN/dS ratio is an accurate estimator of adaptive functional divergence.

Contact:tennessj@science.oregonstate.edu

Supplementary information:Supplementary data, including GenBank Accession numbers, are available at Bioinformatics online.

1 INTRODUCTION

Although the neutral theory of molecular evolution (Kimura, 1983) has been the standard null hypothesis in evolutionary genetics, it is clear that natural selection does have an effect on the evolution of many genes (Hey, 1999). Much current research seeks to infer either positive selection or negative selection from sequence data (Nielsen, 2005). Often, such studies test whether rates of non-synonymous substitution per non-synonymous site (dN) differ from rates of synonymous substitution per synonymous site (dS) (Yang and Bielawski, 2000). Synonymous substitution is thought to be largely neutral, while non-synonymous substitution is influenced by selection. Thus, when the dN/dS ratio is high between genes, functional divergence is assumed to be high, due to positive selection and/or relaxed selective constraint (e.g. Hughes and Friedman, 2004; Wilkinson et al., 2005). When the dN/dS ratio is low, the functional properties of the gene products involved are thought to be conserved, because selective constraint is high and there is little or no positive selection (e.g. Sehgal and Lovette, 2003; Wilkinson et al., 2005).

However, the relationship between selection, the dN/dS ratio, and functional divergence (dF) has rarely been experimentally studied, and a tight correlation might not necessarily exist. First, selection does influence dS via effects on translational efficiency and mRNA folding, and this influence might not be truly negligible compared to the influence of selection on dN (Chamary and Hurst, 2005; Hey, 1999; Tennessen, 2005a). Second, not all functional changes affect fitness, so dF can theoretically be non-zero even under purifying selection and a low dN/dS ratio (Dykhuizen and Hartl, 1980; Kimura, 1983). Third, a non-synonymous change that alters function could be followed by a compensatory substitution that restores the original function, resulting in a low net dF despite a high dN/dS ratio (DePristo et al., 2005). Fourth, amino acid residues essential for intramolecular structure, such as cystines, might remain highly conserved and decrease the dN/dS ratio, even while positive selection acts on surface residues influencing charge, hydrophobicity, shape or other mediators of intermolecular interaction. Finally, non-synonymous substitutions might change function in very specific ways that cannot be detected with general in vitro assays, resulting in no apparent functional change. Although the functional effects of amino acid substitutions have been modeled (DePristo et al., 2005; Rastogi et al., 2006), and the functional effects of particular positively selected substitutions have been evaluated for individual proteins (Antcheva et al., 2004; Frentiu et al., 2007; Trabesigner-Ruef et al., 1996; Zhang et al., 2004), data on functional effects have never been integrated with statistical evidence for selection over a set of proteins to determine significant trends.

Genes encoding antimicrobial peptides (AMPs) are ideal for studying the relationship between nucleotide divergence and functional divergence (Tennessen, 2005b). These short, amphipathic, cationic molecules kill pathogens through a non-catalytic mechanism, either forming pores in the microbial membrane or carpeting the membrane until it is disrupted. Gene sequences of AMPs have been obtained from many taxa. Some AMPs clearly evolve adaptively, while in others evolution is selectively constrained (Tennessen, 2005b). Many studies have investigated the minimal concentrations necessary for different AMPs to kill various microbes, and results suggest remarkable functional diversity. Such extensive functional and genetic data, combined with substantial variation in selective pressures, provide an opportunity to examine the relationship between the dN/dS ratio and functional change.

Here, I present a method to test whether the dN/dS ratio is a predictor of functional divergence. If so, then phylogenetic branches showing higher dN/dS ratios have been subject to more positive selection or less negative selection, and thus should show more divergence in function. I compare functional divergence with nucleotide divergence in animal AMP genes. By statistically incorporating functional data with sequence data, novel questions in molecular evolution can be addressed.

2 METHODS

I chose 71 animal AMPs from 26 families (Table 1; Supplementary Table 1; Supplementary Fig. 1). All of the AMPs within each family were identical in length and could be aligned without gaps. There is no evidence that all 71 AMPs are homologous, but there is clear homology within all families of AMPs. All comparisons are within families, not between, so homology among families is not relevant to this study. Comparisons included both orthologs and paralogs, which were treated equivalently because this analysis only depends on total genetic divergence and AMP function in vitro, not mode of evolutionary divergence or gene role in vivo.

Table 1.

AMP gene families analyzed in this study

AMP family Na Nc Taxon Reference 
Cathelicidin Primate Zelezetsky et al. (2006
α defensin Primate Tanabe et al. (2004) 
β defensin 2 Primate Antcheva et al. (2004
β defensin 3 Primate Boniotto et al. (2003a
θ defensin Primate Tran et al. (2002
CRAMP Rodent Travis et al. (2000
Cryptdin Rodent Ouellette et al. (1994, 2000
Lactoferrin Bovid Vorland et al. (1998
β defensin Bovid Selsted et al. (1993
Esculentin-1 Anuran Chen et al. (2006
Esculentin-2 Anuran Chen et al. (2006
Magainin Anuran Zasloff et al. (1988
Nigrosin Anuran Li et al. (2007
Odorranain Anuran Li et al. (2007
Dermaseptin Anuran Charpentier et al. (1998
Brevinin-1 Anuran Basir et al. (2000
Phylloseptin Anuran Thompson et al. (2007
Maximin Anuran Lai et al. (2002
Maximin H Anuran Lai et al. (2002
Pleurocidin 1 Flatfish Patrzykat et al. (2003
Pleurocidin 2 Flatfish Patrzykat et al. (2003
Hepcidin Tilapia Huang et al. (2007) 
IsCT Scorpion Dai et al. (2002
Cecropin Moth Choi et al. (2000
Ceratotoxin Fly Bessin et al. (2004
Cecropin P Nematode Pillai et al. (2005
AMP family Na Nc Taxon Reference 
Cathelicidin Primate Zelezetsky et al. (2006
α defensin Primate Tanabe et al. (2004) 
β defensin 2 Primate Antcheva et al. (2004
β defensin 3 Primate Boniotto et al. (2003a
θ defensin Primate Tran et al. (2002
CRAMP Rodent Travis et al. (2000
Cryptdin Rodent Ouellette et al. (1994, 2000
Lactoferrin Bovid Vorland et al. (1998
β defensin Bovid Selsted et al. (1993
Esculentin-1 Anuran Chen et al. (2006
Esculentin-2 Anuran Chen et al. (2006
Magainin Anuran Zasloff et al. (1988
Nigrosin Anuran Li et al. (2007
Odorranain Anuran Li et al. (2007
Dermaseptin Anuran Charpentier et al. (1998
Brevinin-1 Anuran Basir et al. (2000
Phylloseptin Anuran Thompson et al. (2007
Maximin Anuran Lai et al. (2002
Maximin H Anuran Lai et al. (2002
Pleurocidin 1 Flatfish Patrzykat et al. (2003
Pleurocidin 2 Flatfish Patrzykat et al. (2003
Hepcidin Tilapia Huang et al. (2007) 
IsCT Scorpion Dai et al. (2002
Cecropin Moth Choi et al. (2000
Ceratotoxin Fly Bessin et al. (2004
Cecropin P Nematode Pillai et al. (2005

For each family, the number of AMPs (Na) and the number of independent comparisons (Nc) that were analyzed are shown.

I allowed AMPs to carry post-translational modifications, including amidation, cyclization and disulfide bonds, but only if all AMPs in a family carried the same modifications, or else if the modification were attached to an amino acid which differed within the family, such that it could be considered part of that amino acid difference. I did not include AMPs for which no nucleotide sequence was available. I only selected families for which antimicrobial activities of all AMPs against both the gram-negative bacterial species Escherichia coli and the gram-positive bacterial species Staphylococcus aureus had been reported in the same published study, apparently tested against the same strains under identical conditions. Strains and other test conditions were allowed to vary among the studies, though, including both radial diffusion assays and tests of bacterial growth in liquid media after incubation, as long as the study measured the lowest concentration necessary for antimicrobial activity, known as the minimal inhibitory concentration. Any family for which no activity could be demonstrated against either microbe by any AMP was excluded from analysis. I know of no other AMP families that fit my criteria.

I recorded the minimal inhibitory concentration of every AMP. When data were presented graphically, with concentration on the x-axis, I followed Lehrer et al. (1991) and estimated the minimal inhibitory concentration from the figure as the x-axis intercept. If an AMP had no demonstrated activity against a microbe at any concentration, I estimated minimal inhibitory concentration to be twice the highest concentration tested. I generated maximum likelihood phylogenies of AMP families using DNAML (Felsenstein, 1981). I used the Parsimony Ancestral States option in Mesquite (version 1.06; http://mesquiteproject.org) to estimate the ancestral minimal inhibitory concentrations at internal nodes in the phylogenies. In several cases, the predicted amino acid sequence at an internal node was identical to the sequence of an extant AMP (e.g. there were no non-synonymous substitutions along the branch leading to the AMP), and thus the minimal inhibitory concentration at the internal node could be accurately known. I selected 48 independent within-family comparisons between extant AMPs and/or internal nodes (Supplementary Fig. 1). No branch was included in more than one comparison. All comparisons included at least four nucleotide substitutions, because dN/dS cannot be estimated accurately if genetic divergence is small. Thus, comparisons were chosen to maximize the number of independent data points while combining adjacent branches into a single comparison when the number of substitutions was low.

For each comparison I measured the multiplicative difference in minimal inhibitory concentration, in micromolars, against both E.coli and S.aureus (Supplementary Table 1). I defined dF, or functional difference in antimicrobial activity, as the mean of the log multiplicative differences for these two bacterial species. For example, a 2-fold difference in minimal inhibitory concentration against both bacterial species would indicate dF = 0.69. If the same AMPs were tested against multiple strains or under various conditions, I calculated the mean difference in the log minimal inhibitory concentration of the AMPs under all tests. The multiplicative difference in minimal inhibitory concentration was the ratio of the less effective AMP to the more effective AMP for each bacterial species separately. Thus, dF is always positive and does not depend on which AMP is more effective or whether the same AMP is more effective against both microbes.

I used the codeml package in PAML (version 3.13a; http://abacus.gene.ucl.ac.uk/software/paml.html) to calculate dN and dS along the branch(es) of each AMP comparison. When calculating dN/dS, I added 0.01 to both the numerator and the denominator, in order to avoid a denominator of zero when there are no synonymous substitutions. I tested whether dN/dS values were significantly different than those expected under neutrality. For each AMP comparison, I ran both a model which estimates dN/dS, and a model which constrains dN/dS to be one along those branch(es), and I used the log likelihoods to obtain the P-value for the test that dN/dS equals one.

I predicted that dF is correlated with log(dN/dS), because when dN/dS is high there is less selective constraint and/or more positive selection. Using simple linear regression, I tested if log(dN/dS) along each phylogenetic branch predicts dF along that branch. To investigate whether positive or negative selection was causing this correlation, I also identified AMP comparisons which PAML indicated as having significantly high or low dN/dS values and used a t-test to evaluate if dF differed between these AMP comparisons and the more neutrally evolving AMP comparisons.

3 RESULTS AND DISCUSSION

Functional divergence (dF) is positively correlated with log(dN/dS) (R2 = 0.26, P < 0.001; Fig. 1). In contrast, dF is not correlated with non-synonymous divergence (dN) alone (P > 0.1). Only when both dN and dS, and therefore the evidence for selection, are included is there a relationship. In 24 of the 48 comparisons mean dN/dS > 1, and in the remaining 24 comparisons mean dN/dS < 1 (Supplementary Table 1). The dN/dS ratio ranged from 0.004 to 22.5 (logarithmic mean = 1.16). Non-synonymous divergence, dN, ranged from 0.03 to 0.64 (mean = 0.19). Functional divergence, dF, ranged from 0 to 2.92 (mean = 0.85). Neither dN/dS, dN, nor dF differed significantly between orthologs and paralogs (P > 0.1).

Fig. 1.

Positive correlation between functional divergence (dF) and the log of the non-synonymous/synonymous divergence ratio (dN/dS) among statistically independent AMP comparisons (R2 = 0.26, P < 0.001). The fitted slope is shown. One outlier has a very high dF value and another has a very low log(dN/dS) value; the correlation is still significant if these two points are removed (P < 0.001). Comparisons in which dN/dS is significantly greater than one are represented by white squares. Comparisons in which dN/dS is significantly less than one are represented by white circles. All other comparisons are represented by black circles.

Fig. 1.

Positive correlation between functional divergence (dF) and the log of the non-synonymous/synonymous divergence ratio (dN/dS) among statistically independent AMP comparisons (R2 = 0.26, P < 0.001). The fitted slope is shown. One outlier has a very high dF value and another has a very low log(dN/dS) value; the correlation is still significant if these two points are removed (P < 0.001). Comparisons in which dN/dS is significantly greater than one are represented by white squares. Comparisons in which dN/dS is significantly less than one are represented by white circles. All other comparisons are represented by black circles.

Among the 24 AMP comparisons in which dN/dS <1, in 5 the neutrality test P-value is <0.05, more than would be expected by chance alone (P < 0.001). Among the 24 AMP comparisons in which dN/dS > 1, in 5 the neutrality test P-value is <0.05, more than would be expected by chance alone (P < 0.001). Therefore, both positive and negative selection are acting among these AMP families. This conservative test likely underestimates the number of families under positive selection. More powerful tests, which allow dN/dS to vary among codons and/or examine more AMPs within each family, have revealed evidence of positive selection in several of the families I used that did not show dN/dS to be significantly greater than one in any of my comparisons (Boniotto et al., 2003b; Lee et al., 2005; Patil et al., 2004; Semple et al., 2006; Tennessen and Blouin, 2007).

Natural selection is the only explanation for a significant regression equation, but the selection could be positive, negative or both. That is, it could be that positively selected substitutions are associated with larger functional changes than neutral substitutions are, so branches where more positive selection has acted have both a higher dN/dS ratio and greater functional divergence. Alternatively, negative selection might purge mutations that affect function substantially, so high dN/dS and high dF only occur under relaxed selective constraint, for example, when a gene copy not needed for survival rapidly accumulates non-synonymous substitutions that degrade its function. In the five positively selected AMP comparisons shown by PAML to have dN/dS ratios significantly greater than one, mean dF (=1.69) is significantly higher than mean dF in the 38 more neutrally evolving AMP comparisons in which dN/dS did not significantly differ from one (mean dF = 0.79; P < 0.001). There is no difference in dN between these two groups (P > 0.1). In contrast, in the five negatively selected AMP comparisons shown by PAML to have dN/dS ratios significantly lower than one, mean dF (=0.48) is not significantly different than mean dF in the 38 AMP comparisons in which dN/dS did not significantly differ from one (P > 0.1). This result suggests that positive selection, not negative selection, is primarily responsible for the correlation between dF and dN/dS.

My results indicate that the dN/dS ratio can be used to help predict functional divergence. They support the hypothesis that the dN/dS ratio is an accurate predictor of natural selection, and the hypothesis that the probability of natural selection acting on a mutation is correlated with the degree to which the mutation affects protein function. These results might not be surprising, but they nevertheless represent an important step towards an experimentally validated comprehensive theory of molecular evolution. Even if natural selection acts on synonymous sites in antimicrobial peptide genes (Tennessen, 2005a), any effects of selection on dS do not completely disrupt the signature of selection in the dN/dS ratio. In addition, even if some function-altering substitutions are neutral, natural selection is more likely to act on mutations with greater functional impact. Furthermore, although compensatory substitutions might make net dF lower than the additive functional change conveyed by all non-synonymous substitutions (DePristo et al., 2005), net dF is still greater under positive selection than under neutrality. Finally, although most of these AMPs have probably evolved to target pathogens other than the human microbes E.coli and S.aureus, and thus many of the functional changes induced by adaptive non-synonymous substitutions would be more accurately measured with assays using different microbes, clearly E.coli and S.aureus are acceptable proxies for measuring dF. Indeed, these results indicate that AMPs which have adaptively diverged due to coevolution with one type of pathogen are more likely to differ in their ability to kill other, novel pathogens that an organism might encounter. Overall, the use of the dN/dS ratio is justified for inferring the action of natural selection on genes, and thus the degree of functional divergence.

Remarkably, the correlation is observed despite multiple sources of noise in the data. The peptides in this study were tested under different conditions, antimicrobial activity was measured in slightly different ways, antimicrobial activity was sometimes roughly estimated from graphical displays, minimal inhibitory concentration had to be estimated at internal nodes, and the formula for the calculation of dF was somewhat arbitrarily chosen. The relationship between dF and dN/dS is strong enough to be detectable in spite of all of this variation. As research in molecular biology proceeds, it is likely that cleaner, more extensive datasets will soon appear. Ideally, future applications of this method will provide identical conditions for all measurements of function, or its analogs such as in vivo activity or protein structure. This method should also be applied to more datasets in which dN/dS is consistently less than one, to test whether variation in negative selection alone can cause a correlation between dN/dS and functional divergence. This method might have practical uses as well, for proteins such as antimicrobial peptides where the identification of diverse functional variants could lead to therapeutic applications. For example, if multiple homologous gene sequences are obtained from one or more species, a frequent practice (e.g. Boniotto et al., 2003a,b; Lai et al., 2002; Patrzykat et al., 2003), the two sharing the highest pairwise dN/dS ratio are likely to code for the most functionally divergent peptides, and these could by synthesized and studied.

In conclusion, functional divergence among antimicrobial peptides is positively influenced by dN/dS and therefore by natural selection. Evidence suggests that this is at least partly due to increased positive selection with higher dN/dS, not merely relaxed functional constraint. The method presented here, or variations of it, can and should be applied to many different types of genes to thoroughly investigate the relationship between natural selection and the evolution of protein function.

ACKNOWLEDGEMENTS

The author would like to thank his advisor, M.S. Blouin; and C.D. Criscione, H. Araki, K. Georgitis and A. Hughes for conceptual and statistical help.

Funding: National Science Foundation Graduate Research Fellowship.

Conflict of Interest: none declared.

REFERENCES

Antcheva
N
, et al.  . 
Effects of positively selected sequence variations in humans and Macaca fascicularis β-defensins 2 on antimicrobial activity
Antimicrob. Agents Chemother
 , 
2004
, vol. 
48
 (pg. 
685
-
688
)
Basir
YJ
, et al.  . 
Multiple antimicrobial peptides and peptides related to bradykinin and neuromedin N isolated from skin secretions of the pickerel frog, Rana palustris
Biochim. Biophys. Acta
 , 
2000
, vol. 
1543
 (pg. 
95
-
105
)
Bessin
Y
, et al.  . 
Antibacterial activity and pore-forming properties of ceratotoxins: a mechanism of action based on the barrel stave model
Biochim. Biophys. Acta
 , 
2004
, vol. 
1667
 (pg. 
148
-
156
)
Boniotto
M
, et al.  . 
A study of host defense peptide β-defensin 3 in primates
Biochem. J
 , 
2003
, vol. 
374
 (pg. 
707
-
714
)
Boniotto
M
, et al.  . 
Evolution of the beta defensin 2 gene in primates
Genes Immun
 , 
2003
, vol. 
4
 (pg. 
251
-
257
)
Chamary
JV
Hurst
LD
Evidence for selection on synonymous mutations affecting stability of mRNA secondary structure in mammals
Genome Biol
 , 
2005
, vol. 
6
 pg. 
R75
 
Charpentier
S
, et al.  . 
Structure, synthesis, and molecular cloning of dermaseptins B, a family of skin peptide antibiotics
J. Biol. Chem
 , 
1998
, vol. 
273
 (pg. 
14690
-
14697
)
Chen
T
, et al.  . 
Cloning from tissue surrogates: antimicrobial peptide (esculentin) cDNAs from the defensive skin secretions of Chinese ranid frogs
Genomics
 , 
2006
, vol. 
87
 (pg. 
638
-
644
)
Choi
CS
, et al.  . 
Antibacterial properties and partial cDNA sequences of cecropin-like antibacterial peptides from the common cutworm. Spodoptera litura
Comp. Biochem. Physiol. C Toxicol. Pharmacol
 , 
2000
, vol. 
125
 (pg. 
287
-
297
)
Dai
L
, et al.  . 
Purification, structure-function analysis, and molecular characterization of novel linear peptides from scorpion Opsthacanthus madagascariensis
Biochem. Biophys. Res. Commun
 , 
2002
, vol. 
293
 (pg. 
1514
-
1522
)
DePristo
MA
, et al.  . 
Missense meanderings in sequences space: a biophysical view of protein evolution
Nat. Rev. Genet
 , 
2005
, vol. 
6
 (pg. 
678
-
687
)
Dykhuizen
D
Hartl
DL
Selective neutrality of 6PGD allozymes in E. coli and the effects of genetic background
Genetics
 , 
1980
, vol. 
96
 (pg. 
801
-
817
)
Felsenstein
J
Evolutionary trees from DNA sequences: a maximum likelihood approach
J. Mol. Evol
 , 
1981
, vol. 
17
 (pg. 
368
-
376
)
Frentiu
FD
, et al.  . 
Adaptive evolution of color vision as seen through the eyes of butterflies
Proc. Natl Acad. Sci. USA
 , 
2007
, vol. 
104
 (pg. 
8634
-
8640
)
Hey
J
The neutralist, the fly, and the selectionist
Trends Ecol. Evol
 , 
1999
, vol. 
14
 (pg. 
35
-
38
)
Huang
P-H
, et al.  . 
Three different hepcidins from tilapia, Oreochromis mossambicus: analysis of their expressions and biological functions
Mol. Immunol
 , 
2007
, vol. 
44
 (pg. 
1922
-
1934
)
Hughes
AL
Friedman
R
Recent mammalian gene duplications: robust search for functionally divergent gene pairs
J. Mol. Evol
 , 
2004
, vol. 
59
 (pg. 
114
-
120
)
Kimura
M
The Neutral Theory of Molecular Evolution.
 , 
1983
Cambridge, UK
Cambridge University Press
Lai
R
, et al.  . 
Antimicrobial peptides from skin secretions of Chinese red-belly toad Bombina maxima
Peptides
 , 
2002
, vol. 
23
 (pg. 
427
-
435
)
Lee
W-H
, et al.  . 
Variety of antimicrobial peptides in the Bombina maxima toad and evidence of their rapid diversification
Eur. J. Immunol
 , 
2005
, vol. 
35
 (pg. 
1220
-
1229
)
Lehrer
RI
, et al.  . 
Ultrasensitive assays for endogenous antimicrobial polypeptides
J. Immunol. Methods
 , 
1991
, vol. 
137
 (pg. 
167
-
173
)
Li
J
, et al.  . 
Anti-infection peptidomics of amphibian skin
Mol. Cell Proteomics
 , 
2007
, vol. 
6
 (pg. 
882
-
894
)
Nielsen
R
Molecular signatures of natural selection
Annu. Rev. Genet
 , 
2005
, vol. 
39
 (pg. 
197
-
218
)
Ouellette
AJ
, et al.  . 
Mouse paneth cell defensins: primary structures and antibacterial activities of numerous cryptdin isoforms
Infect. Immun
 , 
1994
, vol. 
62
 (pg. 
5040
-
5047
)
Ouellette
AJ
, et al.  . 
Characterization of luminal paneth cell alpha-defensins in mouse small intestine. Attenuated antimicrobial activities of peptides with truncated amino termini
J. Biol. Chem
 , 
2000
, vol. 
275
 (pg. 
33969
-
33973
)
Patil
A
, et al.  . 
Rapid evolution and diversification of mammalian.α-defensins as revealed by comparative analysis of rodent and primate genes
Physiol. Genomics
 , 
2004
, vol. 
20
 (pg. 
1
-
11
)
Patrzykat
A
, et al.  . 
Novel antimicrobial peptides derived from flatfish genes
Antimicrob. Agents Chemother
 , 
2003
, vol. 
47
 (pg. 
2464
-
2470
)
Pillai
A
, et al.  . 
Cecropin P1, and novel nematode cecropins: bacteria-inducible antimicrobial peptide family in the nematode Ascaris suum
Biochem. J
 , 
2005
, vol. 
390
 (pg. 
207
-
214
)
Rastogi
S
, et al.  . 
Evaluation of models for the evolution of protein sequences and functions under structural constraint
Biophys. Chem
 , 
2006
, vol. 
124
 (pg. 
134
-
144
)
Sehgal
RN
Lovette
IJ
Molecular evolution of three avian neurotrophin genes: implications for proregion functional constraints
J. Mol. Evol
 , 
2003
, vol. 
57
 (pg. 
335
-
342
)
Selsted
ME
, et al.  . 
Purification, primary structures, and antibacterial activities of β-defensins, a new family of antimicrobial peptides from bovine neutrophils
J. Biol. Chem
 , 
1993
, vol. 
268
 (pg. 
6641
-
6648
)
Semple
CA
, et al.  . 
The changing of the guard: molecular diversity and rapid evolution of β-defensins
Mol. Divers
 , 
2006
, vol. 
10
 (pg. 
575
-
584
)
Tanabe
H
, et al.  . 
Paneth cell alpha-defensins from rhesus macaque small intestine
Infect. Immun
 , 
1999
, vol. 
72
 (pg. 
1470
-
1478
)
Tennessen
JA
Enhanced synonymous site divergence in positively selected vertebrate antimicrobial peptide genes
J. Mol. Evol
 , 
2005
, vol. 
61
 (pg. 
445
-
455
)
Tennessen
JA
Molecular evolution of animal antimicrobial peptides: widespread moderate positive selection
J. Evol. Biol
 , 
2005
, vol. 
18
 (pg. 
1387
-
1394
)
Tennessen
JA
Blouin
MS
Selection for antimicrobial peptide diversity in frogs leads to gene duplication and low allelic variation
J. Mol. Evol
 , 
2007
, vol. 
65
 (pg. 
605
-
615
)
Thompson
AH
, et al.  . 
A combined mass spectrometric and cDNA sequencing approach to the isolation and characterization of novel antimicrobial peptides from the skin secretions of Phyllomedusa hypochondrialis azurea
Peptides
 , 
2007
, vol. 
28
 (pg. 
1331
-
1343
)
Trabesinger-Ruef
N
, et al.  . 
Pseudogenes in ribonuclease evolution: a source of new biomacromolecular function?
FEBS Lett
 , 
1996
, vol. 
382
 (pg. 
319
-
322
)
Tran
D
, et al.  . 
Homodimeric θ-defensins from Rhesus macaque leukocytes: isolation, synthesis, antimicrobial activities, and bacterial binding properties of the cyclic peptides
J. Biol. Chem
 , 
2002
, vol. 
277
 (pg. 
3079
-
3084
)
Travis
SM
, et al.  . 
Bactericidal activity of mammalian cathelicidin-derived peptides
Infect. Immun
 , 
2000
, vol. 
68
 (pg. 
2748
-
2755
)
Vorland
LH
, et al.  . 
Lactoferricin of bovine origin is more active than lactoferricins of human, murine, and caprine origin
Scand. J. Infect. Dis
 , 
1998
, vol. 
30
 (pg. 
513
-
517
)
Wilkinson
TN
, et al.  . 
Evolution of the relaxin-like peptide family
BMC Evol. Biol
 , 
2005
, vol. 
5
 pg. 
14
 
Yang
Z
Bielawski
JP
Statistical methods for detecting molecular adaptation
Trends Ecol. Evol
 , 
2000
, vol. 
15
 (pg. 
496
-
503
)
Zasloff
M
, et al.  . 
Antimicrobial activity of synthetic magainin peptides and several analogues
Proc. Natl Acad. Sci. USA
 , 
1988
, vol. 
85
 (pg. 
910
-
913
)
Zelezetsky
I
, et al.  . 
Evolution of the primate cathelicidin: correlation between structural variations and antimicrobial activity
J. Biol. Chem
 , 
2006
, vol. 
281
 (pg. 
19861
-
19871
)
Zhang
J
, et al.  . 
Evolving protein functional diversity in new genes of Drosophila
Proc. Natl Acad. Sci. USA
 , 
2004
, vol. 
101
 (pg. 
16246
-
16250
)

Author notes

Associate Editor: Dmitrij Frishman

Comments

0 Comments