858
Mol. Biol. Evol. 19(6):858?864. 2002
q 2002 by the Society for Molecular Biology and Evolution. ISSN: 0737-4038
Roles of Diversifying Selection and Coordinated Evolution in the Evolution
of Amphibian Antimicrobial Peptides
Thomas F. Duda, Jr.,* Damien Vanhoye,? and Pierre Nicolas?
*Naos Marine Lab, Smithsonian Tropical Research Institute, Balboa, Ancon, Republic of Panama; and ?Laboratoire de
Bioactivation des Peptides, Institut Jacques Monod, France
Antimicrobial peptides are expressed in the skin of amphibians and are used to prevent infection by microorganisms.
Frog species store distinct collections of antimicrobial peptides that show variation in size, charge, conformation,
and bactericidal activity, and so the evolution of antimicrobial peptide gene families may reflect the adaptive
diversification of these loci. We examined the molecular evolution of antimicrobial peptide transcripts from hylid
and ranid frog species. Our results show that after the gene family arose in the common ancestor of the Hylidae
and Ranidae, before the divergence of these families in the Mesozoic, it subsequently diversified within these groups
with numerous duplication events and divergence of loci. Moreover, we provide evidence that suggests that members
of the antimicrobial peptide gene family have been subject to diversifying selection within both propiece and mature
domains of hylids and solely within the mature domain of ranids. Finally, our results suggest that coordinated and
compensatory amino acid replacements have occurred within the acidic propiece and cationic mature domain of
hylid antimicrobial peptide precursors, as has been observed for mammalian defensin genes, but not among those
of ranid precursors.
Introduction
Vertebrate immune system genes exhibit exception-
ally high levels of polymorphism that is driven by se-
lective pressure to detect a diversity of quickly evolving
pathogens (Ota, Sitnikova, and Nei 2000). Excess of
nonsynonymous substitutions among loci of gene fam-
ilies encoding the major histocompatibility complex
(MHC) and immunoglobulin genes exemplify cases of
diversifying selection favoring elevated amino acid se-
quence diversity (Hughes 1997; Hughes and Yeager
1998).
Innate (nonadaptative) immunity uses gene-encod-
ed antimicrobial peptides to form a first line of host
defense against noxious microorganisms (Nicolas and
Mor 1995; Boman 1995, 1998; Andreu and Rivas 1998).
Most antimicrobial peptides, either inducible or consti-
tutive, are lethal against a broad array of microorgan-
isms, by permeating and disrupting the membrane of
target cells (Andreu and Rivas 1998; Shai 1999). Dif-
ferent mammalian species are equipped with sets of an-
timicrobial peptides (??defensins??) that appear to have
diversified in a species-specific manner as a result of
recent gene duplication followed by evolutionary diver-
gence (Hughes and Yeager 1997).
Dermatous glands of amphibians synthesize and
store wide-spectrum antimicrobial peptides, 10?50 res-
idues in length, that are released onto the outer layer of
the skin to provide an effective and fast-acting defense
Abbreviations: dN, proportion of nonsynonymous substitutions per
site; dS, proportion of synonymous substitutions per site; pNR, propor-
tion of radical nonsynonymous substitutions per site; pNC, proportion
of conservative nonsynonymous substitutions per site; HKY, Hasega-
wa, Kishino, and Yano model; v, ratio of dN to dS.
Key words: antimicrobial peptides, gene family, diversifying se-
lection, coordinated evolution.
Address for correspondence and reprints: Thomas F. Duda Jr.,
Naos Marine Lab, Smithsonian Tropical Research Institute, Apar-
tado 2072, Balboa, Ancon, Republic of Panama. E-mail:
dudat@naos.si.edu.
against harmful microorganisms (Nicolas and Mor 1995;
Simmaco, Mignogna, and Barra 1998; Amiche et al.
1999). A considerable degree of peptide polymorphism
is associated with antimicrobial activity in amphibian
hosts. Frogs belonging to different species or even sub-
species store distinct repertoires of between 6 and 20
antimicrobial peptides, with a differing size, charge, hy-
drophobicity, conformation, and spectrum of action, and
no two species have yet been found with the same pan-
oply of peptide antibiotics. These patterns suggest that
the amphibian antimicrobial peptide genes may be mem-
bers of gene families that have been subject to diversi-
fying selection similar to that of other immune-related
genes.
On the basis of broad structural characteristics, am-
phibian antimicrobial peptides have been grouped into
superfamilies, each being differentiated into various
families. Most antimicrobial peptides from amphibians
of the genus Rana (family Ranidae; subfamily Raninae)
share a conserved disulfide bridged heptapeptide seg-
ment at the C-terminal end. Peptide families with this
motif include gaegurins (24?37 residues), brevinins-1
(17?24 residues) and -2 (30?34 residues), ranalexin (20
residues), ranatuerins-1 (25 residues) and -2 (33 resi-
dues), esculentins-1 (46 residues) and -2 (37 residues),
and rugosins (33?37 residues) isolated from Rana spe-
cies from Europe, Asia, and North America (Morikawa,
Hagiwara, and Nakajima 1992; Clark et al. 1994; Park,
Jung, and Lee 1994; Simmaco et al. 1994; Park et al.
1995; Suzuki et al. 1995; Goraya, Knoop, and Conlon
1998). Another peptide superfamily of very short pep-
tides composed of 10?13 residues called temporins,
which do not contain the C-terminal ring, have also been
characterized from European and North American Rana
(Simmaco et al. 1996).
South American and Neotropical frogs of the Phyl-
lomedusinae subfamily (family Hylidae) produce a rich
array of linear antimicrobial peptides that adopt an am-
phipathic a-helical structure. They include dermaseptins
B and S (24?34 residues), phylloxin (19 residues), and
Evolution of Amphibian Antimicrobial Peptides 859
dermatoxin (32 residues) from the genus Phyllomedusa
(Amiche et al. 1994; Mor and Nicolas 1994; Charpentier
et al. 1998; Amiche et al. 2000; Pierre et al. 2000) and
peptides of 24?33 residues called dermaseptin-related
peptides AA and PD from Agalychnis and Pachymedusa
(Wechselberger 1998).
For most of the peptides described above, the
cDNA-encoding precursors are known to code for a sin-
gle copy of the mature antimicrobial peptide at the C-
terminus of the precursor sequence. A comparison of
peptide precursor sequences reveals that they have a
common N-terminal preproregion, which is highly con-
served both intra- and interspecifically, followed by a
markedly different C-terminal domain that corresponds
to the mature antimicrobial peptides (Amiche et al.
1999; Nicolas and Amiche 1999). The conserved pre-
proregion comprises a hydrophobic signal peptide of 22
residues followed by a 16?25 residue acidic propiece
which terminates by a typical prohormone processing
signal Lys-Arg. The remarkable similarity of preprore-
gions of precursors that give rise to very different an-
timicrobial peptides in distantly related frog species sug-
gests that the corresponding genes form a multigene
family originating from a common ancestor. The diver-
sification of antimicrobial peptide loci could thus be part
of an optimum evolutionary strategy developed by these
frog species as a result of shifts to novel ecological nich-
es when microbial predators change very rapidly.
Mammalian defensins are similar to amphibian an-
timicrobial peptides in that they lyse bacterial cells and
are translated with signal-propiece-mature domains
(Hughes and Yeager 1997). It was postulated that the
cytotoxicity of mature defensins is caused by the posi-
tive net charges of these peptides (although other factors
also likely play a role) and that the anionic properties
of the propiece neutralize the cytotoxicity of the defen-
sin before use (Michaelson et al. 1992). Hughes and
Yeager (1997) showed that nonsynonymous substitu-
tions that affect the net charge of the mature defensin
are often associated with coordinated substitutions that
affect the net charge of the propiece; for example, sub-
stitutions in the mature domain that cause an increase
in the net charge of the defensin are compensated by
substitutions that cause a decrease in the net charge of
the propiece. Do antimicrobial peptides from amphibi-
ans show similar patterns?
In this paper we analyzed the molecular evolution
of antimicrobial peptide gene families of hylid and ranid
frogs. We specifically tested the hypothesis that the three
domains of antimicrobial peptide transcript sequences
evolve neutrally by comparing proportions of synony-
mous and nonsynonymous substitutions per respective
site among potential orthologous or recently duplicated
loci. We also assessed whether coordinated amino acid
changes characterize the evolution of amphibian anti-
microbial peptides by examining patterns of charge-al-
tering nonsynonymous substitutions among sequences
and ancestral sequence predictions.
Methods
Nucleotide sequences of 18 dermaseptins, derma-
septin-related peptides, dermatoxins, and phylloxins
from the hylids Agalychnis annae (GenBank accession
numbers AJ005183?AJ005188), Pachymedusa dacni-
color (AJ005189?AJ005193), and Phyllomedusa bicol-
or (AJ251875, AJ251876, X72387, X70278, and
Y16564?Y16566) were obtained from GenBank. Nucle-
otide sequences of 11 brevinins, esculentins, gaegurins,
ranalexins, and temporins from the ranids Rana cates-
beiana (GenBank accession number S69903) R. escu-
lenta (X77831?X77833), R. rugosa (U22392 and
U22393), and R. temporaria (AJ251566, AJ251567, and
Y09393?Y09395) were also obtained from GenBank.
We aligned the nucleotide sequences of the anti-
microbial peptide transcripts from the two frog families
with ClustalX (Thompson et al. 1997) and by eye. First,
we aligned the predicted amino acid sequences of the
different domains of the peptides and the nucleotide se-
quences of the 59 and 39 untranslated regions (UTR)
separately with ClustalX. Then, the nucleotide sequenc-
es of the different regions were joined, and final adjust-
ments to the alignment were made manually. The align-
ments are available as Supplementary Material.
We used Modeltest 3.0 (Posada and Crandall 1998)
to determine the models of nucleotide substitutions that
best fit our data sets for phylogenetic reconstruction.
Molecular phylograms from the alignments were deter-
mined with Neighbor-Joining using PAUP* (Swofford
1999). Levels of support for branches were estimated
with bootstrapping methods (1,000 replicates) also with
PAUP* (Swofford 1999). To interpret the origins of
these gene families, we examined the topologies of the
phylogram; we assume that the sequences represent dis-
tinct loci in the species sampled.
To determine if diversifying selection operates
among members of antimicrobial gene families in frog
species, we estimated the proportions of nonsynony-
mous substitutions (dN) and synonymous substitutions
(dS) per respective site among sequences with maxi-
mum-likelihood (ML) methods (Yang 1998) among po-
tential recently duplicated or orthologous loci. These
values were calculated among terminal and predicted
ancestral node sequences for each peptide domain using
PAML (Yang 1997). Ancestral nodes were also predict-
ed using parsimony and ML methods with PAUP*; these
predictions were compared with those determined with
PAML. We performed likelihood ratio tests to determine
if ratios of dN to dS (v) exceed a value of 1 by com-
paring twice the difference between the log-likelihoods
of the null model (v is fixed at 1) and the alternative
model (v is a free parameter) to a x2 distribution with
one degree of freedom (Yang and Bielawski 2000).
Many of the mature antimicrobial peptides are known
to be or suspected of being carboxamidated at the C-
terminus during processing; this results in the exclusion
of two or three of the terminal amino acid residues of
ranid and hylid sequences, respectively. Therefore, the
final two or three codons of the mature domain of ranid
and hylid sequences, respectively, were excluded from
these analyses.
We estimated the proportions of radical (pNR) and
conservative (pNC) nonsynonymous substitutions with
respect to net charge among terminal and ancestral pre-
860 Duda et al.
FIG. 1.?Molecular phylograms of nucleotide sequences of hylid
and ranid antimicrobial peptide transcripts as reconstructed with neigh-
bor-joining methods of HKY distances. The HKY model with gamma
correction was used for the reconstruction of each data set. Parameters
used for the hylid phylogram reconstruction are transition to transver-
sion ratio (Ti/Tv) 5 1.0011, proportion of invariable sites (I) 5 0, and
the shape parameter of the gamma distribution (g) 5 0.8838. Param-
eters used for the hylid phylogram reconstruction are Ti/Tv 5 1. 1980,
I 5 0, and g 5 0.6694. Phylograms are midpoint rooted. Bootstrap
values from 1,000 replicates greater than 50% are indicated on branch-
es. Among the hylid sequences, D 5 dermaseptin; DRP 5 dermasep-
tin-related peptide; sequence names were appended with AA, PD, and
PB to show that the sequences were identified from A. annae, P. dac-
nicolor, and P. bicolor, respectively. Among the ranid sequences, the
brevinins are from R. temporaria (RT) (2Ta and 2Tb) and R. esculenta
(RE) (1Ef and 2Ef); esculentin 1B is from R. esculenta, gaegurins 4
and 5 are from R. rugosa (RR); ranalexin is from R. catesbeiana (RC);
and the temporins are from R. temporaria (RT); species from which
the sequences were identified are indicated with abbreviated sequence
names in parentheses.
dicted node sequences among the same sets of sequenc-
es as analyzed for dS and dN within propiece and mature
domains, according to the methods of Hughes, Ota, and
Nei (1990). We omitted the final two or three codons
from these analyses when they were suspected of being
excluded during processing. We tested whether pNR was
significantly greater than pNC by conducting a one-tailed
t-test with infinite degrees of freedom.
Net charges of propiece and mature peptides were
compared to determine if the charges of these domains
showed a negative relationship. Net charges were cal-
culated based on numbers of positively (arginine, histi-
dine, and lysine) and negatively charged residues (as-
partic and glutamic acid) in the peptides. As with esti-
mates of dN, dS, pNR, and pNC, we calculated the net
charges of mature domains while excluding codons cor-
responding to the two or three terminal residues of ranid
and hylid sequences, respectively.
Results and Discussion
Our results show that the genes encoding antimi-
crobial peptides of hylid and ranid frogs are members
of a large gene family whose history has been charac-
terized by numerous duplication events and the subse-
quent evolutionary divergence of these loci. Results
from comparisons of the proportions of nonsynonymous
and synonymous substitutions among antimicrobial pep-
tide transcript sequences suggest that diversifying selec-
tion operates among these loci as it does among other
immunodefense-related genes (Hughes 1997; Hughes
and Yeager 1997, 1998; Ota, Sitnikova and Nei 2000),
particularly in the mature domain. Moreover, coordinat-
ed amino acid changes appear to have occurred within
propiece and mature domains of antimicrobial peptide
genes in hylids but not in ranids, suggesting different
roles of propiece peptides in these families.
Gene Family History
The Hasegawa, Kishino, and Yano (1985) (HKY)
model of nucleotide substitutions with gamma correc-
tion was the model of sequence evolution that best fit
both the hylid and ranid data sets (fig. 1). The phylo-
gram of hylid sequences is not completely resolved, but
several distinct clades are apparent. Sequences from A.
annae and P. dacnicolor cluster tightly, together with
strong bootstrap support in three cases, DRP AA-1-1?
DRP PD-1-5, DRP AA-2-5?DRP PD-3-6, and DRP AA-
3-6?DRP PD-3-3 (fig. 1), suggesting that these sequenc-
es may represent orthologous loci in these species.
Within the ranid phylogram, there are two distinct
clades of sequences: a clade consisting of a brevinin
from R. esculenta, a gaegurin from R. rugosa, ranalexin
from R. catesbeiana, and temporins from R. temporaria
supported by a bootstrap value of 98%; and a clade con-
sisting of brevinins from R. esculenta and R. temporaria
and a gaegurin from R. rugosa supported by a bootstrap
value of 96% (fig. 1). Relationships of the three brevi-
nins in the first clade and of the temporins in the second
suggest that brevinin 2Ta and 2Tb and temporin B and
H in R. temporaria are recently duplicated loci in this
species.
Hylid and ranid families diverged during the Me-
sozoic, though precise dating of this event is controver-
sial (see Duellman and Trueb 1994, pp. 472?495; Feller
and Hedges 1998), giving rise to very distinct evolu-
tionary histories and geographical distributions. The di-
versity of modern antimicrobial peptide loci present in
these families reflects the origination and divergence of
Evolution of Amphibian Antimicrobial Peptides 861
Table 1
Maximum-likelihood Estimates of Synonymous (dS) and Nonsynonymous (dN) Substitutions Per Respective Site and
Estimated Proportions of Conservative (pNC) and Radical (pNR) Nonsynonymous Substitutions (with respect to charge)
Among Terminal and Predicted Node Sequences (see fig. 1) Within Signal, Propiece, and Mature Domains.
COMPARISON TO
NODE SEQUENCE
SIGNAL
dS dN pNC pNR
PROPIECE
dS dN pNC pNR
MATURE
dS dN pNC pNR
Hylid sequences
DRP AA-1-1 . .
DRP PD-1-5 . .
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
10.2
0.0
0.0
0.0
0.0
0.0
0.0
DRP AA-2-5 . .
DRP PD-3-6 . .
0.2
4.3
2.3
0.0
2.6
0.0
0.0
0.0
0.0
0.0
0.0
3.9
0.0
0.0
0.0
4.5**
10.6
0.3
4.7
7.5*
3.0
6.0
0.0
14.5*
Ranid sequences
Brevinin 2Ta . .
Brevinin 2Tb . .
Brevinin 2Ef . .
0.0
0.0
0.0
0.0
0.0
2.2
0.0
0.0
2.4
0.0
0.0
0.0
0.0
0.0
10.0
0.0
0.0
2.2
0.0
0.0
6.3
0.0
0.0
0.0
11.8
0.3
0.1
5.5
6.6***
4.5***
5.3
4.3
4.5
3.4
8.8
3.4
Temporin B . . .
Temporin H . . .
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.1
0.0
5.6***
0.0
6.3
0.0
0.0
NOTE.?Values are presented as percentages. Cases where dN or pNR is greater than dS or pNC, respectively, are indicated with values in bold. Asterisks show
cases in which the dN to dS ratio is significantly greater than 1, or pNR is significantly greater than pNC (* 5 P , 0.05, ** 5 P , 0.01, *** 5 P , 0.005)
these genes since the Mesozoic. As shown in the phy-
lograms, sequences do not cluster according to species;
for example, sequences of A. annae occur throughout
the hylid phylogram (fig. 1). This pattern implies that
many antimicrobial peptide loci originated before the
divergence of the species sampled and that concerted
evolution has played little role in the evolution of this
gene family. Only in a few cases do loci appear to be
the result of recent duplications; that is, only in two
cases do sequences from the same species uniquely clus-
ter together, brevinin 2Ta and 2Tb and temporin B and
H from R. temporaria (fig. 1).
Adaptive Evolution
We examined patterns of nucleotide substitutions
within each of the three peptide domains among four
sets of sequences that may represent orthologous or re-
cently duplicated loci: DRP AA-1-1 and PD-1-5 (pair-
wise HKY distance 5 0.031) and DRP AA-2-5 and PD-
3-6 (HKY distance 5 0.056) from the hylids A. annae
and P. dacnicolor; brevinins 2Ta, 2Tb, and 2Ef from the
ranids R. esculenta and R. temporaria (HKY distances
range between 0.041 and 0.052); and temporins B and
H from R. temporaria (HKY distance 5 0.013). We did
not include DRP AA-3-6 and PD-3-3 in these analyses
because the level of divergence among these sequences
was more than two times greater (HKY distance 5
0.116) than those observed among the other pairs from
these species, suggesting that DRP AA-3-6 and PD-3-3
are not orthologous.
We estimated dS and dN along branches among ter-
minal and predicted ancestral node sequences for each
domain with PAML (Yang 1997). Because our phylo-
grams were not completely resolved, we did not estimate
parameters with PAML for complete data sets; dS and
dN were calculated using three subsets of sequences as
described subsequently. The first subset of sequences in-
cluded sequence pairs DRP AA-1-1?PD-1-5 and DRP
AA-2-5?PD-3-6 and dermatoxin PB (see fig. 1). The
second subset included brevinins 2Ta, 2Tb, 2Ef, and
gaegurin 4. The final subset included the sequence pair
temporin B and H plus brevinin 1E, gaegurin 5, ranal-
exin, and temporin G; because the topology of the last
four sequences is not well supported (see fig. 1), the
user-supplied tree was constructed as a polytomy with
regard to these sequences. Where differences occurred
between terminal and predicted ancestral node sequenc-
es with PAML, we compared the ancestral predictions
with those generated by parsimony and ML methods
with PAUP* (Swofford 1999), and in all cases the meth-
ods gave similar if not identical predictions of ancestral
node sequences.
Results from analyses of ML estimates of dS and
dN show that patterns of substitutions are not equivalent
among the three peptide domains (table 1). Among hylid
sequences, dS and especially dN increase from the signal
to propiece to mature domain. Among the first set of
sequences from hylids, DRP AA-1-1 and PD-1-5, there
are no nonsynonymous substitutions within any domain
among these and the predicted ancestral node sequence;
the only differences are synonymous substitutions with-
in the mature domain among DRP PD-1-5 and the pre-
dicted ancestral node sequence. The lack of nonsynon-
ymous divergence among these sequences shows that
some antimicrobial loci are under strong purifying
selection.
On the contrary, among the second set of hylid se-
quences, DRP AA-2-5 and PD-3-6, dN exceeds dS and
dN to dS ratios (v) are greater than 1 within the propiece
and mature domains among DRP PD-3-6 and the an-
cestral node sequence (table 1); v is significantly greater
than 1 within the mature domain. These results suggest
that the DRP PD-3-6 locus has been subject to diversi-
fying selection potentially within the propiece domain
and significantly within the mature peptide domain.
All nonsynonymous substitutions within the pro-
piece and most within the mature domain among DRP
PD-3-6 and the ancestral node sequence are radical nu-
862 Duda et al.
FIG. 2.?Nucleotide and amino acid differences among DRP PD-3-6 and the predicted ancestral node sequence. Single-letter amino abbre-
viations are used. All amino acid differences are underlined; amino acid differences that result in a change in net charge of the peptide are also
in bold.
cleotide substitutions that affect the net charges of pro-
piece and mature peptides (table 1). In both these cases,
the proportion of radical nonsynonymous substitutions
(pNR) is significantly greater than that of conservative
nonsynonymous substitutions (pNC) (table 1). These sub-
stitutions account for two charge-altering amino acid
differences within both the propiece and the mature pep-
tides that essentially result in no change of the net charg-
es of these peptides (fig. 2). The pattern and mode of
nucleotide substitutions in the propiece and mature do-
mains suggest that substitutions in these domains may
be coordinated and compensatory among hylid antimi-
crobial peptides (see subsequently).
Among ranid sequences, both dS and dN also gen-
erally increase from the signal to propiece to mature
domains; the largest values of dN were measured exclu-
sively within the mature domain (table 1). In three cases,
dN is greater than dS within the mature domain, and v
is significantly greater than 1. In only one case, dN ex-
ceeds dS within the signal domain, but v is not signifi-
cantly greater than 1. Values of dN do not exceed dS in
any case, and in most cases dN equals 0 within the pro-
piece domain for ranid sequences. These results suggest
that diversifying selection has operated within the ma-
ture domain of some antimicrobial peptide loci in ranids,
but propiece domains appear to be strictly under puri-
fying selection.
Very few of the nonsynonymous substitutions are
radical among the ranid and ancestral node sequences
(table 1). Values of pNR exceed those of pNC in only one
case, within the mature domain among brevinin 2Tb and
the ancestral node sequence, but this difference is not
significant (table 1). This evidence implies that although
mature peptide domains are subject to diversifying se-
lection within both hylids and ranids, selection plays a
much different role among the antimicrobial peptide loci
of these two groups, and it does not appear that propiece
and mature peptides evolve in a coordinated manner
within Ranidae (see subsequently).
We clearly performed multiple tests in comparing
v and values of pNR and pNC (table 1), and so in some
cases the significant outcomes may simply be the result
of chance. However, the patterns we observed, particu-
larly the higher values of dN in the mature domain and
the lower values in the other domains among hylid and
ranid ancestral sequence comparisons (table 1), suggest
that amphibian antimicrobial peptides evolve adaptively.
Coordinated Evolution
The propiece and mature domains of mammalian
defensins have been shown to evolve in a coordinated
manner (Hughes and Yeager 1997). This phenomenon
is revealed by a negative relationship among the net
charges of propiece and mature peptides. If the evolution
of propiece and mature domains of amphibian antimi-
crobial peptides is also coordinated, we expect to find a
negative relationship among net charges of these pep-
tides. Such a relationship was observed among antimi-
crobial peptides from hylids but not from ranids (fig. 3).
Net charges of the amino acid sequences of propiece
and mature antimicrobial peptides from hylids range
from 211 to 24 and from 22 to 16, respectively.
Among ranids, they range from 29 to 24 and from 11
to 15 (fig. 3). The net charges of propiece and mature
peptides show a negative relationship among hylid se-
quences (slope 5 20.58 and is significantly different
from 0, P 5 0.014), but they show no relationship
among ranid sequences (slope 5 0.09 and is not signif-
icantly different from 0, P 5 0.802) (fig. 3). These re-
sults suggest that opposing charges of propiece and ma-
ture peptides are associated among hylid but not among
ranid antimicrobial peptide loci, as might be expected
from the patterns of substitutions of these loci men-
tioned previously.
If substitutions are coordinated and compensatory,
charge-altering nucleotide substitutions will have oc-
curred concurrently within both propiece and mature do-
mains. On the basis of the prediction of the ancestral
node sequence of hylid sequence DRP PD-3-6 from P.
dacnicolor (as determined with PAML), we show that
two charge-altering nucleotide substitutions occurred in
both propiece and mature domains in the gene lineage
that gave rise to this locus (fig. 2). Ancestral node pre-
dictions using parsimony and ML methods with PAUP*
show identical results. These substitutions are compen-
satory in the sense that in both domains they account
for no change in the net charges of propiece and mature
peptides. Moreover, v is greater than 1, and pNR is sig-
nificantly greater than pNC within both these domains
(table 1). It should be noted that the chronological order
Evolution of Amphibian Antimicrobial Peptides 863
FIG. 3.?A, Net charge of the propiece is negatively related to the
net charge of mature peptide for sequences from hylids (Y 5 22.64
2 0.58X; the regression line is plotted as a solid line, the dashed line
is plotted at a slope corresponding to 458). B, Net charge of the pro-
piece does not show any relationship to that of the mature peptide for
sequences from ranids (Y 5 23.89 1 0.09X; the regression line is
plotted as a solid line, the dashed line is plotted at a slope correspond-
ing to 458).
of the charge-altering substitutions within propiece and
mature domains is unknown, and so although the sub-
stitutions appear to be coordinated and compensatory,
they may not be so. Future work should be directed at
analyses of orthologous loci of DRP PD-3-6 in close
relatives of P. dacnicolor to verify the occurrence of
coordinated evolution at this locus.
Summary
Our results suggest that diversifying selection has
operated within the mature domain of some antimicro-
bial peptide loci of hylid and ranid amphibians. Because
the peptides that members of this gene family encode
are used to protect against noxious microbes, their adap-
tive evolution may be caused by several factors. Indeed,
results from functional assays previously conducted
show that amphibian antimicrobial peptides have differ-
ent bactericidal activities and are specific for different
types of microorganisms (Simmaco, Mignogna, and
Barra 1998). As with MHC receptors, immunoglobulins,
and defensins (Hughes 1997; Hughes and Yeager 1997,
1998; Ota, Sitnikova, and Nei 2000), antimicrobial pep-
tides may be under selection directed by the evolution
of pathogens. For antimicrobial peptides this selection
may be in response to the evolution of the cellular mem-
branes of microbes to prevent disruption by antimicro-
bial peptides. Alternatively, frog species may be ex-
posed to different microorganisms in different habitats
or environments such that the plethora of expressed an-
timicrobial peptides have evolved for the particular mi-
crobial biota that these species encounter. Such hypoth-
eses cannot be tested without further analyses of the
evolution and expression of antimicrobial peptide gene
families among closely related hylid and ranid species,
surveys of the communities of microbes with which
these frogs are associated, and further functional assays
and investigations of the activities of these peptides.
Our analyses also show that antimicrobial peptide
loci have evolved differently among hylids and ranids.
Within both groups, diversifying selection has operated
within the mature domain; within hylids, the propiece
domain potentially appears to have been subject to di-
versifying selection. The results also suggest that coor-
dinated and compensatory amino acid replacements in
the propiece and mature domains may have occurred
among antimicrobial peptide loci from hylids but not in
ranids.
Acknowledgments
We wish to thank L. Trueb, B. Hedges, H. Lessios,
B. Kessing, the reviewing editor E. Holmes and two
anonymous reviewers for helpful comments and criti-
cisms. We also wish to thank A. L. Hughes for a copy
of his program that estimates proportions of radical and
conservative nonsynonymous substitutions. T.F.D. is
supported by a Tupper Fellowship from the Smithsonian
Tropical Research Institute.
LITERATURE CITED
AMICHE, M., F. DUCANCEL, A. MOR, J. C. BOULAIN, A. MENEZ,
and P. NICOLAS. 1994. Precursors of vertebrate peptide an-
tibiotics dermaseptin b and adenoregulin have extensive se-
quence identities with precursors of opioid peptides der-
morphin, dermenkephalin and deltorphins. J. Biol. Chem.
269:17847?17853.
AMICHE, M., A. SE? ON, T. PIERRE, and P. NICOLAS. 1999. The
dermaseptin precursors: a protein family with a common
preproregion and a variable C-terminal antimicrobial do-
main. FEBS Lett. 456:352?356.
AMICHE, M., A. SE? ON, H. WROBLEWSKI, and P. NICOLAS. 2000.
Isolation of dermatoxin from frog skin, an antibacterial pep-
tide encoded by a novel member of the dermaseptin gene
family. Eur. J. Biochem. 267:4583?4592.
ANDREU, D., and L. RIVAS. 1998. Animal antimicrobial pep-
tides: an overview. Biopolymers 47:415?433.
BOMAN, H. G. 1995. Peptide antibiotics and their role in innate
immunity. Annu. Rev. Immunol. 13:61?76.
???. 1998. Gene-encoded peptide antibiotics and the con-
cept of innate immunity. Scand. J. Immunol. 48:15?25.
CHARPENTIER, S., M. AMICHE, Y. MESTER, V. VOUILLE, J. P.
LE CAER, P. NICOLAS, and A. DELFOUR. 1998. Structure,
synthesis and molecular cloning of dermaseptins B, a family
of skin peptide antibiotics. J. Biol. Chem. 273:14690?
14696.
CLARK, D. P., S. DURELL, W. L. MALOY, and M. ZASLOFF.
1994. Ranalexin. A novel antimicrobial peptide from bull-
864 Duda et al.
frog (Rana catesbeiana) skin, structurally related to the bac-
terial antibiotic, polymyxin. J. Biol. Chem. 269:10849?
10854.
DUELLMAN, W. E., and L. TRUEB. 1994. Biology of amphibi-
ans. Johns Hopkins University Press, London.
FELLER, A. E., and B. HEDGES. 1998. Molecular evidence for
the early history of living amphibians. Mol. Phylogenet.
Evol. 9:509?516.
GORAYA, J., F. C. KNOOP, and J. M. CONLON. 1998. Ranatuer-
ins: antimicrobial peptides isolated from the skin of the
American bullfrog, Rana castesbetana. Biochem. Biophys.
Res. Commun. 250:589?592.
HASEGAWA, M., H. KISHINO, and T. YANO. 1985. Dating the
human-ape split by a molecular clock of mitochondrial
DNA. J. Mol. Evol. 22:160?174.
HUGHES, A. L. 1997. Rapid evolution of immunoglobulin su-
perfamily C2 domain expressed in immune system cells.
Mol. Biol. Evol. 14:1?5.
HUGHES, A. L., T. OTA, and M. NEI. 1990. Positive Darwinian
selection promotes charge profile diversity in the antigen-
binding cleft of class I major histocompatibility complex
genes in mammals. Mol. Biol. Evol. 7:515?524.
HUGHES, A. L., and M. YEAGER. 1997. Coordinated amino acid
changes in the evolution of mammalian defensins. J. Mol.
Evol. 44:675?682.
???. 1998. Natural selection at major histocompability
complex loci of vertebrates. Annu. Rev. Genet. 32:415?435.
MICHAELSON, D., J. RAYNER, M. CONTO, and T. GANZ. 1992.
Cationic defensins arise from charge-neutralized propep-
tides: a mechanism for avoiding leukocyte autocytotoxicity?
J. Leukoc. Biol. 51:632?639.
MOR, A., and P. NICOLAS. 1994. Isolation and structure of nov-
el defensive peptides from frog skin. Eur. J. Biochem. 219:
145?154.
MORIKAWA, N., K. HAGIWARA, and T. NAKAJIMA. 1992. Brev-
inin-1 and -2, unique antimicrobial peptides from the skin
of the frog, Rana brevipoda porsa. Biochem. Biophys. Res.
Commun. 189:184?189.
NICOLAS, P., and M. AMICHE. 1999. The dermaseptin family
of antimicrobial peptide precursors. Curr. Top. Biochem.
Res. 1:243?248.
NICOLAS, P., and A. MOR. 1995. Peptides as weapons against
microorganisms in the chemical defense system of verte-
brates. Annu. Rev. Microbiol. 49:277?304.
OTA, T., T. SITNIKOVA, and M. NEI. 2000. Evolution of im-
munoglobulin variable gene segments. Curr. Top. Micro-
biol. Immunol. 248:221?245.
PARK, J. M., J. E. JUNG, and B. J. LEE. 1994. Antimicrobial
peptides from the skin of a Korean frog, Rana rugosa.
Biochem. Biophys. Res. Commun. 205:948?952.
PARK, J. M., J. Y. LEE, H. M. MOON, and B. J. LEE. 1995.
Molecular cloning of cDNAs encoding precursors of frog
skin antimicrobial peptides from Rana rugosa. Biochim.
Biophys. Acta 1264:23?28.
PIERRE, T. N., A. SE? ON, M. AMICHE, and P. NICOLAS. 2000.
Phylloxin, a novel peptide antibiotic of the dermaseptin
family of antimicrobial/opioid peptide precursors. Eur. J.
Biochem. 267:370?378.
POSADA, D., and K. A. CRANDALL. 1998. Modeltest: testing
the model of DNA substitution. Bioinformatics 14:817?818.
SHAI, Y. 1999. Mechanism of the binding, insertion and desta-
bilization of phospholipid bilayer membranes by helical an-
timicrobial and cell?non selective membrane lytic peptides.
Biochim. Biophys. Acta 1462:55?70.
SIMMACO, M., G. MIGNOGNA, and D. BARRA. 1998. Antimi-
crobial peptides from amphibian skin: what do they tell us?
Biopolymers 47:435?450.
SIMMACO, M., G. MIGNOGNA, D. BARRA, and F. BOSSA. 1994.
Antimicrobial peptides from skin secretions of Rana escu-
lenta. Molecular cloning of cDNAs encoding esculentin and
brevinins and isolation of new active peptides. J. Biol.
Chem. 269:11956?11961.
SIMMACO, M., G. MIGNOGNA, S. CANNOFENI, R. MIELE, M. L.
MANGONI, and D. BARRA. 1996. Temporins, antimicrobial
peptides from the European red frog Rana temporaria. Eur.
J. Biochem. 242:788?794.
SUZUKI, S., Y. OLIC, T. OKUBO, T. KAKEGAWA, and K. TATE-
MOTO. 1995. Isolation and characterization of novel anti-
microbial peptides, rugosins A, B and C, from the skin of
the frog, Rana rugosa. Biochem. Biophys. Res. Commun.
212:249?254.
SWOFFORD, D. L. 1999. PAUP*: phylogenetic analysis using
parsimony (*and other methods). Version 4. Sinauer Asso-
ciates, Sunderland, Mass.
THOMPSON, J. D., T. J. GIBSON, F. PLEWNIAK, F. JEANMOUGIN,
and D. G. HIGGINS. 1997. The ClustalX windows interface:
flexible strategies for multiple sequence alignment aided by
quality analysis tools. Nucleic Acids Res. 24:4876?4882.
WECHSELBERGER, C. 1998. Cloning of cDNAs encoding new
peptides of the dermaseptin family. Biochim. Biophys. Acta
1388:279?283.
YANG, Z. 1997. PAML: a program package for phylogenetic
analysis by maximum likelihood. CABIOS 13:555?556.
???. 1998. Likelihood ratio tests for detecting positive se-
lection and application to primate lysozyme evolution. Mol.
Biol. Evol. 15:568?573.
YANG, Z., and J. P. BIELAWSKI. 2000. Statistical methods for
detecting molecular adaptation. Trends Ecol. Evol. 15:496?
503.
EDWARD HOLMES, reviewing editor
Accepted January 24, 2002