Evolution of An Arsenal

Research
Evolution of an Arsenal
STRUCTURAL AND FUNCTIONAL DIVERSIFICATION OF THE VENOM SYSTEM IN THE ADVANCED SNAKES
(CAENOPHIDIA)*
Bryan G. Fry‡§, Holger Scheib¶储, Louise van der Weerd**, Bruce Young‡‡,
Judith McNaughtan§§, S. F. Ryan Ramjan‡, Nicolas Vidal¶¶, Robert E. Poelmann**,
and Janette A. Norman‡储储
Venom is a key innovation underlying the evolution of systems in three front-fanged lineages is associated with
advanced snakes (Caenophidia). Despite this, very little is recruitment of new toxin types or explosive diversification
known about venom system structural diversification, of existing toxin types. These results support the role of
toxin recruitment event timings, or toxin molecular evolu- venom as a key evolutionary innovation in the diversifi-
tion. A multidisciplinary approach was used to examine cation of advanced snakes and identify a potential role
the diversification of the venom system and associated for non-front-fanged venom toxins as a rich source for
toxins across the full range of the ⬃100 million-year-old lead compounds for drug design and development.
Downloaded from www.mcponline.org at CNRS on February 7, 2008

advanced snake clade with a particular emphasis upon Molecular & Cellular Proteomics 7:215–246, 2008.
families that have not secondarily evolved a front-fanged
venom system (⬃80% of the 2500 species). Analysis of
cDNA libraries revealed complex venom transcriptomes It has only become evident recently that venom in snakes is
containing multiple toxin types including three finger tox- a basal characteristic and that the three front-fanged venom
ins, cobra venom factor, cysteine-rich secretory protein, delivery system architectures are each independent second-
hyaluronidase, kallikrein, kunitz, lectin, matrix metallopro- ary derivations (1– 4). In earlier schemes, the “colubrid,” or
tease, phospholipase A2, snake venom metalloprotease/a “non-front-fanged,” snakes were seen as a monophyletic
disintegrin and metalloprotease, and waprin. High levels
transitional group to the presumably advanced front-fanged
of sequence diversity were observed, including mutations
lineages Atractaspis, Elapidae, and Viperidae (e.g. Kardong
in structural and functional residues, changes in cysteine
spacing, and major deletions/truncations. Morphological (5)), and all front-fanged snakes were assumed to share a
analysis comprising gross dissection, histology, and mag- common ancestor. Resolution of higher order relationships
netic resonance imaging also demonstrated extensive revealed not only the colubrid snakes to be paraphyletic but
modification of the venom system architecture in non- the front-fanged snakes to be polyphyletic with viperids being
front-fanged snakes in contrast to the conserved struc- one of the earliest advanced snake radiations and elapids only
ture of the venom system within the independently recently derived (1, 6).
evolved front-fanged elapid or viperid snakes. Further, a Previous studies indicate significant morphological varia-
reduction in the size and complexity of the venom system tion in the venom gland (7–9) and dentition of snakes (10). The
was observed in species in which constriction has been fang may or may not be enlarged and can range from a solid
secondarily evolved as the preferred method of prey cap-
tooth with or without grooving to an enclosed canaliculate
ture or dietary preference has switched from live prey to
channel as present in elapids and viperids (10). Few com-
eggs or to slugs/snails. Investigation of the timing of toxin
recruitment events across the entire advanced snake ra- parative studies of the caenophidian venom system have
diation indicates that the evolution of advanced venom been performed, although West (11, 12) and Sarkar (13)
described several different structural and topological fea-
From the ‡Department of Biochemistry and Molecular Biology,
tures. A distinction was attempted between the venom
Bio21 Molecular Science and Biotechnology Institute and §§Depart- glands of the front-fanged and non-front-fanged snakes
ment of Oral Medicine and Surgery, School of Dental Science, Uni- with the glands in most non-front-fanged snakes termed
versity of Melbourne, Parkville, Victoria 3010, Australia, ¶SBC Lab “Duvernoy’s glands” (7).
AG, Seebüelstrasse 26, 8185 Winkel, Switzerland, **Molecular Imag- Similarly, very little has been revealed about the composi-
ing Laboratories Leiden, MRI Facility, Departments of Anatomy and
Radiology, 2300 RC, Leiden, The Netherlands, ‡‡Department of Bi-
tion of venoms from advanced snake lineages other than the
ology, Washburn University, Topeka, Kansas 66621, ¶¶UMR 7138, three clades with high pressure, hollow front-fang venom
Département Systématique et Evolution, Muséum National d’Histoire delivery systems (Atractaspis, elapids, and viperids). Despite
Naturelle, CP 26, 57 rue Cuvier, 75005 Paris, France, and 储储Molecular accounting for the majority of the families (Fig. 1 and Ref. 6)
Biology, Museum Victoria, G. P. O. Box 666, Melbourne, Victoria and ⬃1900 of the 2500 advanced snake species, the multiple
3001, Australia
Received, March 2, 2007, and in revised form, September 10, 2007
non-front-fanged families have received scant attention. Only
Published, MCP Papers in Press, September 17, 2007, DOI a few studies have been undertaken, and even fewer have
10.1074/mcp.M700094-MCP200 sequenced or bioactivity-tested individual toxins (2, 14 –18).
© 2008 by The American Society for Biochemistry and Molecular Biology, Inc. Molecular & Cellular Proteomics 7.2 215
This paper is available on line at http://www.mcponline.org

FIG. 1. Cladogram of evolutionary relationships of advanced snakes (1, 6, 58 – 61) showing relative timing of toxin recruitment events
and derivations of the venom system. MRI images are shown for representatives. Acn, Acetylcholine esterase; LAO, L-amino oxidase; C3B,
FAMC3B cytokine; CNP-BPP, c-type natriuretic peptide-bradykinin-potentiating peptide; GrTx, glycine-rich toxin; Hya, hyaluronidase; RAP,
renin-like aspartic protease; VEGF, vascular endothelial growth factor.
Studies of the first full-length toxins from non-front-fanged molecular phylogenetic studies demonstrated the shared or-
snakes were revealing, particularly the isolation and charac- igin of 3FTx and other toxin types across the entire advanced
terization of a potently neurotoxic three-finger toxin (3FTx)1 snake radiation (3).
from the colubrid snake Coelognathus radiatus (2). This toxin In view of the homology of the venom toxins and toxin-
type had long been considered the hallmark of elapid venoms secreting glands of all advanced snakes, the fact that Duver-
and had been the subject of intense study (19). Follow-up noy’s glands represented a primitive condition and that the
derived glands of the front-fanged snakes were independently
1
The abbreviations used are: 3FTx, three-finger toxin; CRISP, cys- evolved from these, the distinction between Duvernoy’s
teine-rich secretory protein; SVMP, snake venom metalloprotease; glands and venom glands was revealed to be an artificial one
ADAM, a disintegrin and metalloprotease; BLAST, Basic Local Align- that impeded the understanding of the evolution of the venom
ment Search Tool; 3D, three-dimensional; SPDBV, Swiss-PdbViewer;
MRI, magnetic resonance imaging; T, tesla; CVF, cobra venom factor;
apparatus of snakes. For this reason, the term Duvernoy’s
MMP, matrix metalloprotease; PLA2, phospholipase A2; CRD, cys- gland was abandoned, and the term “venom gland” was used
teine-rich domain. for the toxin-secreting oral glands of all snakes regardless of
216 Molecular & Cellular Proteomics 7.2

the degree of anatomical specialization in the venom delivery screened for vector sequences, and those parts were removed prior
apparatus (14). to analysis and identification. Toxin sequences were identified by
homology of the translated cDNA sequences with previously charac-
It has been shown previously that snake venoms evolve via
terized toxins using a BLAST search of the Swiss-Prot protein data-
a process by which a gene encoding for a normal body base (www.expasy.org/tools/blast/).
protein, typically one involved in key regulatory processes or Molecular Modeling of Toxins—3D models for caenophidian toxins
bioactivity, is duplicated, and the copy is selectively ex- were generated based on the assumption that homologous proteins
pressed in the venom gland (20). The newly created toxin type share similar 3D structures (25). In other words, the three-dimensional
structure of a target protein can be modeled if its sequence is ho-
evolves via the birth-and-death model of protein evolution in
mologous to at least one template protein whose 3D structure has
which a toxin multigene family is created by further gene been determined experimentally by applying either x-ray or NMR
duplication events followed by the deletion of some copies techniques (26). In this work aligning the protein sequences of target
and conversion of others to non-functional copies or pseudo- and template(s) was carried out in SPDBV (27), and the initial align-
genes (19). ments were refined manually. From these alignments 3D models were
built directly in SPDBV applying the “Build Preliminary Model” option,
In addition to gene duplication, mutation is an important
which is disabled in the currently distributed public version of the
process that generates a tremendous diversity of venom tox- software. Loops were built by scanning a database of known loop
ins within these multigene families. The newly created toxin structures using the same software, and suitable specimens were
multigene families preserve the molecular scaffold of the an- selected after visual inspection. The enthalpy of the resulting models
cestral protein but modify key functional residues at the tips of was minimized applying two times 200 steps of Steepest Descent

minimization. Finally the quality of each model structure was as-
loops to acquire a myriad of newly derived activities (19, 20).
sessed in iMolTalk (28), and a Van der Waals surface was calculated
Other mutations can include the selective expression of a in MolMol (29). In MolMol electrostatic potentials were calculated
particular domain, such as the expression of the disintegrin applying the “simplecharge” command and mapped on the model
domain from snake venom metalloprotease, ADAM-type structure surface. The families of venom proteins were analyzed by
(SVMP/ADAM) toxins in viperid venoms (21). These toxins superimposing the structures in SPDBV (27), and conserved and
variable structural regions were identified. In the molecular modeling
have an unusual combination of precise specificity and ex-
of representative proteins, blue surface areas indicate positive
treme potency, characteristics that make them particularly charges, red surface areas indicate negative charges, and model
amenable for use as investigational ligands or as leads for pairs show sides of the protein rotated by 180°.2
drug design and development (22–24). Molecular Phylogeny of Toxin Sequences—Molecular phylogenetic
In this study we used a multidisciplinary approach to (a) analyses of toxin transcripts were conducted using the translated
amino acid sequences. Comparative sequences from other venom-
characterize the venom transcriptomes of representative cae-
ous reptiles and outgroups were obtained through BLAST searching
nophidian snakes, (b) determine the timing of toxin recruit- (www.expasy.org/tools/blast/) using representative toxin sequences.
ment events and patterns of toxin diversification, and (c) To minimize confusion, all sequences obtained in this study are
characterize changes in the venom delivery architecture. referred to by their GenBankTM accession numbers (www.ncbi.nlm-
From comparison with the more widely studied elapids and .nih.gov/sites/entrez?db⫽Nucleotide), and sequences from previous
studies are referred to by their UniProt/Swiss-Prot accession num-
viperids, we determined whether there are significant struc-
bers (www.expasy.org/cgi-bin/sprot-search-ful). Resultant sequence
tural or functional differences in the evolution of the venom sets were aligned using the program ClustalX followed by visual
system and their associated toxins in the poorly characterized inspection for errors. When presented as sequence alignments, the
non-front-fanged caenophidian snakes. leader sequence is shown in lowercase, the prepro region is under-
lined, cysteines are highlighted in black, and functional residues are in
MATERIALS AND METHODS
bold. Datasets were analyzed using Bayesian inference implemented
on MrBayes, version 3.0b4. The analysis was performed by running a
Species Studied and Molecular Phylogeny of Advanced Snakes—A minimum of 1 ⫻ 106 generations in four chains and saving every 100th
total of 107 species representing each of the major lineages of ad- tree. The log likelihood score of each saved tree was plotted against
vanced snakes were included in the study and examined using one or the number of generations to establish the point at which the log
more of the following techniques: clone sequencing of cDNA libraries, likelihood scores of the analysis reached their asymptote, and the
gross dissection and examination of dentition, histology of venom posterior probabilities for clades were established by constructing a
glands and ducts, and magnetic resonance imaging (see “Appendix I” majority rule consensus tree for all trees generated after the comple-
for details). In most cases different individuals were used as a source tion of the burn-in phase.3
of material for each of these analyses. Gross Dissection and Analysis of Dentition—Gross dissection was
cDNA Library Construction and Analysis—RNA was isolated from performed on freshly euthanized and formalin-fixed specimens to
venom glands of 13 species spanning the full taxonomical diversity document the relative size and position of the venom gland and
(10 non-front-fanged snakes, one elapid, and two viperids; see “Ap- associated skeletal musculature. Features of the maxillary dentition
pendix I”) using the Qiagen RNeasy Midi kit with subsequent selection were scored using data from a previous study (10) in which five
of mRNAs using the Oligotex Midi kit. cDNA libraries were con- morphological states were defined: 1) smooth surface and no en-
structed using the Clontech Creator SMART cDNA Library Construc-
tion kit and transformed into One Shot Electrocompetent GeneHogs
2
(Invitrogen) as described previously (4). Isolation and sequencing of Homology model coordinates can be obtained from H. Scheib.
inserts was undertaken at the Australian Genome Research Facility E-mail: holger@moltalk.org.
3
using BDTv3.1 chemistry with electrophoretic separation on an Sequence alignments can be obtained from B. G. Fry. E-mail:
AB330xl. Up to 384 colonies were sequenced per library, inserts were bgf@unimelb.edu.au.
Molecular & Cellular Proteomics 7.2 217

TABLE I
Toxin types recovered from mRNA transcript sampling
CRI, CRISP; Fac X, Factor X; Ka, kallikrein; Ku, kunitz; Lec, lectin; NGF, nerve growth factor; Wap, waprin.
PLA2 PLA2
3FTx CRI CVF Fac X Ka Ku Lec NGF SVMP Wap
Type IA Type IB
Colubridae X
D. typus X X X
T. dhara X X X X
T. jacksonii X X X X
T. biscutatus X X X
Dipsadidae
L. poecilogyrus X X X X X
P. olfersii X X X X X X
Elapidae
O. microlepidotus X X X X X X
Homalopsidae
E. polylepis X X X X
Natricidae
R. tigrinus X X

Psammophiinae
P. mossambicus X X
Pseudoxyrhophiinae
L. madagascariensis X X X X X
Viperidae
Azemiops feae X X
C. rhombeatus X X X X X
closed venom canal, 2) no enclosed venom canal and surface with image acquisition. Anatomical images were acquired using a 3D
shallow furrow, 3) deep groove running the majority of the length of gradient echo sequence. The field of view and matrix were varied to
the tooth, 4) deep groove present but restricted to less than half the fit the individual samples, resulting in voxel sizes between (40)3 and
length of the tooth, and 5) enclosed venom canal. (70)3 mm3. Imaging parameters were: echo time ⫽ 8 ms; repetition
Histological Analysis—Histological sections were prepared from time ⫽ 40 ms; flip angle, 20°; four to eight averages; total scan time
the intact head and the excised venom delivery system. Whole heads between 3 and 9 h per sample, depending on size and resolution.
were removed, and a cut was made to the underside to allow fast Image segmentation of the glands was performed manually in Amira
penetration of the fixative (10% neutral buffered formalin). After a 4.1 (Mercury Computer Systems Inc.), and 3D surface renderings
minimum of 2 days excess tissue was removed, and specimens were were generated for all species.
immersed in Kristensen’s decalcification solution and placed on a
rotor for up to 3 weeks (depending on the size of the head). Before
RESULTS
processing the heads were bisected longitudinally for cutting trans-
versely, at 3 ␮m, in two separate blocks. The processing schedule Isolation and Characterization of Venom Toxin Transcripts—
was: 10% formalin, 2 h; absolute ethanol, 4 ⫻ 1 h; Histolene, 3 ⫻ 1 h; Analysis of venom gland cDNA from non-front-fanged snake
paraffin wax, 2 ⫻ 90 min. The sections were taken every 100 ␮m, and
libraries revealed the presence of transcripts with homology
matching sections were stained with periodic acid-Schiff stain and
Masson’s trichrome stain. In other specimens, the venom gland, to previously characterized venom toxins from front-fanged
venom duct, and the adjacent bony and muscular tissue were excised snakes and venomous lizards (helodermatids and varanids)
and placed in decalcifying solution (Cal-Ex, Fisher) for 72–168 h. Each (Tables I and II). Transcripts sequenced were 3FTx (Figs. 2
sample was dehydrated and cleared through a progressive ethanol and 3), C3/cobra venom factor (CVF) (Fig. 4), cysteine-rich se-
series and Cyto-Sol (Fisher) prior to embedding in Paraplast (Fisher).
cretory protein (CRISP) (Figs. 5 and 6), hyaluronidase (Fig. 7),
Serial sections were cut at 10 –12 ␮m. All species were sectioned in
the frontal plane; when available, the contralateral venom delivery kallikrein (Figs. 8 and 9), kunitz (Figs. 10 and 11), lectin (Figs. 12
system was sectioned either parasagittally or transversely. Sections and 13), matrix metalloprotease (MMP) (Fig. 14), phospholipase
were stained using a variant of Van Gieson’s stain, which provides A2 (PLA2) Type IB (Fig. 15), SVMP/ADAM (Figs. 16, 17, and 18),
clear distinction between connective tissue, muscle, and epithelium, and waprin (Fig. 19). Transcripts of five of these toxin types were
or with hematoxylin and eosin.
also recovered from the cDNA libraries of the representative
Magnetic Resonance Imaging—Magnetic resonance imaging (MRI)
was used to examine the three-dimensional shape and internal anat- elapid Oxyuranus microlepidotus (3FTx, CRISP, kunitz, and wa-
omy of the venom glands. Formalin-ethanol-fixed heads were first prin) and the viperid Causus rhombeatus (kallikrein).
submersed in Fomblin (Solvay Solexis) to prevent air artifacts. De- Alignment of the translated amino acid sequences revealed
pending on head size, imaging was performed on either 9.4-T (small/ extensive variation in the molecular structure of the transcripts
medium) or 17.6-T (large) vertical 89-mm-bore systems (Bruker Bio-
Spin, Rheinstetten, Germany) with a Bruker Micro2.5 gradient system
for most toxin types. There was less amino acid sequence
of 1 T/m and transmit/receive birdcage radiofrequency coil with di- divergence in C3/CVF, CRISP, hyaluronidase, kunitz, PLA2
ameter of 10 –30 mm. Bruker ParaVision 3.0 software was used for (Type IB), and MMP toxins than in the much more variable

TABLE II
Transcripts from non-front-fanged snakes and previously characterized basal and derived bioactivities from elapid or viperid venoms (19)
VWF, von Willebrand factor; GP, glycoprotein.
Toxin type Bioactivities
3FTx Ancestral toxic activity of ␣-neurotoxicity, antagonistically binding to the nicotinic acetylcholine receptor; ␣-
neurotoxicity greatly potentiated by the deletion of the second and third ancestral cysteines. Functional
derivations include binding to the postsynaptic muscarinic acetylcholine receptors, presynaptic neurotoxic
action upon the L-type calcium channels, cytotoxic interactions, acetylcholinesterase inhibition, and others.
C3/CVF Ancestral activity of unregulated activation of the complement cascade causing rapid and significant problems
such as anaphylactic-type problems and/or tissue damage via hemolysis/cytolysis. Derived activities not
currently documented.
CRISP Ancestral activity of paralysis of peripheral smooth muscle and induction of hypothermia due to the action
upon voltage-gated Ca2⫹ channels and resultant blockage of K⫹-induced contraction. Derived activities
include blockage of cyclic nucleotide-gated calcium channels.
Hyaluronidase Venom spreading factor.
Kallikrein Ancestral toxic activity of increase of vascular permeability and production of hypotension in addition to
stimulation of inflammation. Derived activities affect the blood, particularly targeting fibrinogen.
Kunitz Ancestral toxic activity of inhibition of circulating plasma serine proteases. Derivations include inhibition of
plasmin and thrombin and the blockage of L-type calcium channels. Structural derivatives form part of

neurotoxic complexes with PLA2 molecules.
Lectin Ancestral toxic activity of inhibition of platelet aggregation mediated by galactose binding. Derivations include
stimulation of platelet aggregation (binding GPVI, GPIb, GPIa/IIa, or VWF), platelet aggregation inhibition
(binding GPIb or GPIa/IIa), or anticoagulant actions by binding blood factors IX and X.
PLA2 (Type IB) Ancestral toxic activity of lipase activity resulting in inflammation and tissue destruction. Presynaptic
neurotoxicity is a basal derivation.
SVMP/ADAM Ancestral toxic activity of tissue edema and necrosis. Prothrombin activation is a basal derivation. In Viperidae
venoms, proteolytic cleavage of C-terminal domains results in a myriad of other activities including direct
acting fibrinolytic activity; liberated disintegrin domain inhibits platelets via GPIIb/IIIa integrin receptor.
Waprin Only antimicrobial activities currently described.
3FTx, kallikrein, lectin, SVMP/ADAM, and waprin toxins. For Even more so our in silico studies could not confirm sug-
most toxin types multiple transcripts containing significant gested functional motifs, namely EX2F and DVF, at least as a
molecular variations were also isolated from individual cDNA general principle of CRISP. Differences in CRISP function of
libraries; this is a characteristic previously attributed to acceler- toxic venom proteins and the non-toxic representative from
ated diversification in these toxin multigene families (19). cDNA mouse are subtle and likely to be found in CRD, i.e. Leu-220
sequencing revealed numerous transcripts in the venom glands and Leu-230 (numbering according to mouse; Fig. 5). The
of non-front-fanged snakes that preserve the key amino acids promiscuous susceptibility of CRISP toward different ion
essential for a particular bioactivity and potentially represent the channels is reflected in their high degree of sequence varia-
mRNA precursors of functional proteins. tion. In the absence of obvious sequence and structure dif-
Reflective of sequence variation of proteins in the CRISP ferences between toxic and non-toxic specimens (Figs. 5 and
family, its members have been found to interact with different 6) it will be necessary to acquire further data on the function
target proteins, i.e. cyclic nucleotide-gated ion channels as of the CRISP family members to identify the functionally dis-
well as L-type Ca2⫹ and BKCa K⫹ channels (30 –32). Binding criminating amino acids.
to the respective channels is speculated to be a general The hyaluronidase transcript sequenced from Liophis po-
property of CRISP and was attributed to a predominantly ecilogyrus showed significant sequence similarity to those
hydrophobic cavity involving residues from both domain PR-1 sequenced from viperid venoms (36), and the snake se-
and cysteine-rich domain (CRD), whereas specific channel quences all formed a clade (Fig. 7). Kallikrein transcripts iso-
block supposedly is taking place with residues of CRD (30 – lated from Philodryas olfersii retained residues essential for
35). Six amino acids were identified for the calcium channel the potent hypotensive action mediated by the liberation of
blocker Triflin (Protein Data Bank code 1WVR (30)) to bind to bradykinin from kininogen (Fig. 8).
ion channels (numbering according to Triflin): Glu-44, Tyr-54, Kunitz-type toxins belong to the superfamily of bovine pan-
Ser-65, Tyr-125 (all four in PR-1), Thr-184, and Arg-185 (both creatic trypsin-like inhibitors. Although they share the same
in CRD). An additional four amino acids were conserved and overall 3D fold, their antitrypsin activity may vary. It has been
postulated to affect ion channel function of this neurotoxin. long known that the residue in the so-called P1 site of the
They are all located in CRD: Phe-189, Leu-195, Tyr-205, and trypsin inhibitor is positively charged (arginine, lysine, or his-
Phe-215. However, this work reveals that none of these 10 tidine), whereas this residue in a chymotrypsin inhibitor usu-
positions is conserved throughout the CRISP toxin family. ally is large and hydrophobic (leucine, phenylalanine, tyrosine,

FIG. 2. Sequence alignment of rep-

resentative 3FTx. Species shown are E.
polylepis (1, EU029668), P. mossambi-
cus (2, EU029669), L. poecilogyrus (3,
EU029670; 6, EU029672; 7, EU029673),
T. dhara (4, EU029671), D. typus (5,
EU029674; 9, EU036636; 17, EU029681;
19, EU029683), T. jacksonii (8, EU036635;
18, EU029682; 20, EU029684; 21,

EU029685), T. biscutatus (10, EU029675;
12, EU029677; 13, EU029678), Leioheter-
odon madagascariensis (11, EU029676),
T. dhara (14, EU029686; 15, EU029679;
16, EU029680), C. radiatus (22, P83490),
Dendroaspis jamesonii (23, P25682),
from B. multicinctus (24, Q9PW19; 25,
Q9YGJ0), Bungarus candidus (26,
P81783), and Naja sputatrix (27,
Q9W7I3). Also included is the represent-
ative non-toxin peptide brain ␣-neu-
ropeptide (28, Q9WVC2) from Mus
musculus.
or asparagine) (37). From the multiple sequence alignment pre- Within the lectin toxins, variation of a key tripeptide motif
sented in Fig. 10 it can be concluded that kunitz-type toxins has been shown to have significant impact upon functionality
from P. olfersii, Ophiophagus hannah, Naja naja, and Bungarus (38) with EPN conferring mannose binding ability, whereas
multicinctus are likely to inhibit chymotrypsin, whereas the oth- QPD confers galactose binding. In this study, both were ob-
ers putatively are trypsin inhibitors. Our 3D modeling efforts tained as well as new variants of this motif containing EAP,
showed that the kunitz toxins vary, i.e. in the length of their C- QAP, and LTD (Figs. 12 and 13). The EPN motif appears to be
but also their N-terminal tails (Figs. 10 and 11). The N-terminal basal, whereas the QPD motif is an early emerging variant, and
tail is held in place by formation of two disulfide bonds involving the other variants evolved subsequently at different times during
the first and sixth as well as the second and fourth cysteines. the evolution of the animals themselves. The viper venom-
The inhibitory residues are solvent-exposed and reside in a long specific heterodimeric forms have lost this motif entirely.
surface loop immediately after the second cysteine, which indi- A novel MMP toxin had been reported previously from the
cates that the position of these residues is rather conserved venom of the lethal natricid snake Rhabdophis tigrinus (39).
among kunitz toxins (Figs. 10 and 11). Therefore, changes in However, only a very small fragment was obtained so the
polarity and charge are expected to affect the physicochemical MMP-subtype relationship remained unclear. In this study, a
properties of this region. full-length MMP was obtained from the L. poecilogyrus library

FIG. 3. Bayesian molecular phylog-

eny of 3FTx. The outgroup is the non-
toxic brain ␣-neuropeptide (Q9WVC2)

from M. musculus. The ribbon model of
EU029668 E. polylepis shows ␤-strands
in yellow.
FIG. 4. Molecular evolution of C3/

CVF toxins. A, partial sequence align-
ment of the representative toxin forms
shown from top to bottom: Q2XXR5 from
L. madagascariensis, Q49HM6 from Aus-
trelaps superbus, Q01833 from N. naja,
and Q91132 from Naja kaouthia as well
as the non-toxin forms P23667 from Xe-
nopus laevis and Q90633 from Gallus
gallus. B, Bayesian molecular phyloge-
netic analysis of representative toxin and
non-toxin body forms. Outgroups are
the non-toxin sequences P23667 X. lae-
vis and Q90633 G. gallus.
and was shown to be closest to MMP2 (Fig. 14) rather than well as two conserved tyrosine residues that together with
showing the MMP9 relationship hypothesized in the previous structural water form the catalytic center (Fig. 15). This toxin is
R. tigrinus study. most likely responsible for conferring the presynaptic neuro-
A Type IB PLA2 transcript was isolated from Trimorphodon toxicity observed for this venom (40). The electrostatic sur-
biscutatus that retained not only the ancestral pancreatic loop faces of PLA2 IB clearly indicate the overall similarity of the
but also the catalytic diad residues histidine and aspartate as five toxins studied both in terms of size and shape as well as

FIG. 5. Sequence alignment of rep-

resentative full-length (unless other-
wise indicated) CRISP toxins. 1,
Q2XXQ5 from D. typus; 2, partial se-
quence Q2XXP5 from T. dhara; 3, partial
sequence Q2XXP4 from T. biscutatus; 4,
partial sequence Q2XXP7 from P. olfer-
sii; 5, partial sequence Q2XXQ0 from L.
poecilogyrus; 6, Q2XXQ3 from E. polyl-
epis; 7, partial sequence Q2XXQ1 from

L. madagascariensis; 8, Q2XXP9 from O.
microlepidotus; 9, Q3SB03 from Hoplo-
cephalus stephensii; 10, Q3SB05 from
P. textilis; 11, Q8UW11 from L. curtus;
12, Q8AVA3 from Pseudechis porphyria-
cus; 13, Q8AVA4 from Pseudechis aus-
tralis; 14, Q8JI38 from Laticauda semi-
fasciata; 15, Q8JI39 from Trimeresurus
flavoviridis; 16, P79845 from P. mucros-
quamatus; 17, Q8JGT9 from R. tigrinus;
18, Q2XXP1 from Varanus varius; 19,
Q91055 from H. horridum; 20, the non-
toxin representative Q91XA3 from M.
musculus.
with regard to the physicochemical properties. It can be con- (Figs. 4, 5, 8, 10, and 15). Large deletions were detected in
cluded that all PLA2 IB toxin specimens exhibit very similar transcripts of the lectin and SVMP/ADAM toxins. Multiple tran-
neurotoxic function. scripts of a deleted form of the lectin toxin were isolated from
The SVMP/ADAM sequences were shown to be of the PIII Enhydris polylepis (Fig. 12) in which a large stretch of residues
type, consistent with previously reported fragments from including an ancestral cysteine is deleted, leaving a free cys-
Dispholidus typus (41). Numerous transcripts were also recov- teine, potentially facilitating dimerization. Another lectin toxin
ered with significant variations including changes in the number version from the E. polylepis cDNA (again for which multiple
and spacing of cysteine residues and large scale deletions. transcripts were obtained) had a significant change in sequence
Caenophidian 3FTx, lectin, MMP, SVMP, and waprin transcripts to the second half of the protein as a consequence of a frame-
were all characterized by changes in ancestral cysteines in shift mutation. The resulting new transcript contains overall four
addition to the evolution of new cysteines (Figs. 2, 12, 16, 17, cysteine bonds and therefore may fold into a stable, bioactive
and 19). In contrast, transcripts of the caenophidian C3/CVF, molecule. A major truncation was detected in transcripts of the
CRISP, hyaluronidase, kallikrein, kunitz, and PLA2 Type IB tox- SVMP/ADAM toxins isolated from Psammophis mossambicus
ins preserved the ancestral cysteine numbers and spacing (Fig. 16). These transcripts consisted solely of the propeptide

FIG. 6. A, Bayesian molecular phylog-

eny of CRISP toxins. Outgroups are non-
toxin forms Q034O1 and Q91XA3 from
M. musculus. B, ribbon view of Q2XXQ5
from D. typus. ␣-Helices are in red, and
␤-strands are in yellow. Residues postu-
lated to bind to ion channels (without
blocking them) are in white (Glu-55, Tyr-
65, Ser-76, and Tyr-136 in PR-1; Ser-
195 and Arg-196 in CRD); Ser-195 and
Arg-196 are colored orange because

they are part of a putative ion channel
blocking site. The EX2F motif is in ma-
genta, i.e. Glu-197 and Phe-200. Chan-
nel blocking motif hypothesized in this
study is shown: part I in pink containing
Val-199 and Phe-200 (of the DVF, EX2F
motifs) but also Asn-202 and Asn-220;
part II in blue containing Gln-209, Ser-
210, and Asn-211; part III in orange con-
sisting of Ser-195, Arg-196, and Phe-
226; and part IV in green consisting of
the C-terminal tail Pro-228 and Asn-229.
region normally post-translationally cleaved from the functional Pareas carinatus. The course of the venom ducts ranged from
enzymatic region, and some isoforms had evolved new cys- nearly craniad in the elapid and viperid to varying degrees of
teines within this domain. medial or craniomedial in the non-front-fanged species.
Structural Variation of the Venom System—The combina- In histological analyses the serous (protein-secreting) sec-
tion of MRI, histological analysis, gross dissection, and ex- tion of the venom gland was isolated and easily distinguished
amination of dentition revealed extensive variations in the from the mucus-secreting supralabial glands of all species. In
relative size of the venom gland and lumen and variation in the the majority of the taxa examined the venom duct was lined
course of the venom duct (Figs. 1, 20, and 21 and see “Ap- with a combination of stratified squamous and mucoid cells.
pendix II”). Large venom glands, similar in size to those of The relative amount of mucoid cells was far more variable
some elapids, were found in representatives from each of the within the non-front-fanged snakes; mucoid cells were absent
different non-front-fanged families, such as the colubrid in some taxa (e.g. Diadophis punctatus). In Atractaspis bibro-
snake Telescopus dhara, the homalopsid snake Cerberus ryn- nii the radially arranged secretory tubules have a mucoid
chops, the psammophiine snake P. mossambicus, and the section at their opening into the central lumen. Irrespective of
dipsadid snake Helicops leopardinus. The lethal colubrid the epithelial lining, the majority of the taxa examined had a
snake D. typus had the most robust glands of all non-front- localized expansion of the venom duct, termed a venom
fanged snakes studied. In contrast, the venom glands of two vestibule. A venom vestibule was located adjacent to the fang
colubrid snakes (Pituophis guttatus and Dasypeltis scabra) sheath of all front-fanged species as well as many non-front-
were found to be greatly atrophied as was that of the pareatid fanged snakes. However, unique among the non-front-fanged

FIG. 7. Hyaluronidase. A, partial se-

quence alignment of toxin sequences. 1,
EU091710 from L. poecilogyrus; 2,
A3QVP2; 5, A3QVP3; 13, A3QVP1; 14,
A3QVP0; 15, A3QVN9, all from Bitis ari-
etans; 3, A3QVN8; 4, A3QVN7; 6,

A3QVP5; 9, A3QVN6, all from Echis
pyramidum leakeyi; 7, A3QVN5; 8,
A3QVN3; 10, A3QVN4, all from Cerastes
cerastes; 11, A3QVN2 Echis ocellatus;
12, A3QVP4 from Echis carinatus soch-
ureki. Also shown are the representative
non-toxins: 16, Q12794 from Homo sa-
piens; 17, Q05A56 from M. musculus. B,
Bayesian molecular phylogeny; the out-
group is Q9W6K5 from Xenopus laevis.
snakes was the presence of a venom vestibule located along tween these two structures as it is in all front-fanged taxa.
the medial surface of the venom gland. The proximal portion Most commonly in non-front-fanged snakes the venom duct
of the venom duct was septate in the front-fanged taxa and opened both into the fang sheath and directly to the roof of
generally ovoid and largely free of surrounding connective the mouth immediately adjacent to the fang sheath; this dual
tissue in the non-front-fanged taxa. In a few of them (most opening was achieved by the venom duct, or in some cases
notably in D. typus, Stenorrhina freminvillei, and Thelotornis the associated venom vestibule, paralleling the long axis of
capensis) the venom duct was more crenate in cross-section the fang sheath forming a prominent diverticulum. In a few
and was surrounded by concentric layers of connective tissue non-front-fanged taxa there was a secondary venom duct that
suggestive of pressure regulation. arose from the distal portion of the fang sheath and extended
The greatest level of morphological variation was found in peripherally to contact the roof of the mouth. Combining
the topographical relationship between the venom duct and information for the different structural features revealed the
the fang. In some non-front-fanged taxa (e.g. Heterodon) the non-front-fanged snakes to possess a highly variable venom
venom duct opens directly into the oral cavity rather than to system architecture with all combinations of epithelial lining,
the lumen of the fang sheath and the surface of the fang. In venom vestibule, and venom duct topography identified (see
other cases the venom duct courses directly to the fang “Appendix II”).
sheath with no venom vestibule located at the juncture be- The maxillary dentition was also revealed to be extremely

FIG. 8. Sequence alignment of the

representative full-length (unless oth-
erwise indicated) kallikrein toxins. 1,
partial sequence Q2XXM2 from P. olfer-
sii; 2, Q5MCS0 from L. curtus; 3, Q6T6S7
from Bitis gabonica; 4, P81824; 5,
Q9PTU8, both from Bothrops jararaca;
6, P33589 from Lachesis muta; 7,
Q2XXN0 from V. varius; 8, P43685 from
H. horridum; 9, the non-toxin represent-

ative Q9POG3 from H. sapiens.
variable among the non-front-fanged snakes, ranging from quenced from non-front-fanged species. For 3FTx, CRISP,
solid smooth fangs (type 1) to grooved fangs of varying depth lectin, and kallikrein, taxonomically limited cDNA sequences
and length (types 2– 4) (Table III). Tooth size was also variable were previously reported, but comprehensive comparisons
with significant enlargement occurring on multiple independ- had not been undertaken.
ent occasions. Particularly enlarged teeth were observed in In the sequencing surveys undertaken in this study, tran-
representatives from all families, such as Dispholidus and scripts of 3FTx, CRISP, and SVMP/ADAM were the most
Oligodon in Colubridae, Tomodon and Waglerophis in Dipsa- phylogenetically widespread dominantly secreted toxin types,
didae, Macropisthodon in Natricidae, Malpolon and Rham- occurring in at least seven of the 10 non-front-fanged snakes
phiophis in Psammophiinae, and Homalopsis in Homalopsi- examined. Other phylogenetically widespread dominant toxin
dae. In contrast, Atractaspis plus all elapids and viperids transcripts were lectin and waprin. Identification of 3FTx tran-
possessed type 5 dentition with an enclosed venom canal, scripts as a common component of caenophidian venom is
and relative tooth size was also much less variable. consistent with previous studies showing active expression of
potent 3FTx in a range of non-front-fanged snakes (2, 14, 40).
DISCUSSION Similarly for the other toxin types, the congruence with pre-
Molecular Characterization of Caenophidia Venoms—Com- vious LC/MS data on the same species provides additional
plex venoms are a feature of the well studied elapids, viperids, evidence of active expression of bioactive proteins in the
and Atractaspis and have also recently been shown to be a venom glands of non-front-fanged snakes (14). Multiple tran-
feature of the helodermatid and varanid venomous lizards (4). scripts of the majority of toxins were recovered from individual
In this study, multiple toxin types were sequenced from all cDNA libraries; this pattern is consistent with accelerated
non-front-fanged snake cDNA libraries, indicating the com- diversification in toxin multigene families as observed in elap-
mon presence of complex venom transcriptomes containing ids and viperids (19). Numerous transcripts were recovered
multiple bioactive components. For hyaluronidase, kunitz, and with significant variations including changes in key functional
waprin these were the first sequences, whether protein or residues, changes in the number and spacing of cysteine
nucleotide, obtained for non-front-fanged species. For 3FTx, residues, and large scale deletions. These modifications rep-
CRISP, kallikrein, lectin, MMP, PLA2, and SVMP/ADAM, only resent potential neofunctionalization (evolution of novel bio-
a very few proteins had been previously even partially se- activities). These results support previous suggestions that

the venom system is a basal characteristic of the advanced Despite their relative presence, these toxin types in different
snakes (3, 4) and also previous crude venom studies (14, 16, lineages should not be considered the only classes present in
17). This has important implications for understanding the a particular venom. This caveat is due to the limited sequenc-
evolution and ecology of the advanced snake radiation and ing that detected only the major toxin transcripts for each
identifies venoms from non-front-fanged snakes as an impor- species. Other types may be present at lower expression
tant bioresource. levels. It is likely that more detailed exploration of their ven-
oms will reveal the presence of additional toxin types. This is
demonstrated by this study recovering only 3FTx, CRISP,
factor X, kunitz, and Type IB PLA2 from O. microlepidotus but
not other toxin types previously sequenced from this species
such as factor V, natriuretic peptides, or nerve growth factor.
Toxin Structure-Function Relationships—Structure-function
relationships of a number of the toxin types sequenced in this
study have been well characterized in venoms from snakes
(Atractaspis, elapids, or viperids) or helodermatid lizards (Table
II) (20). This includes information on the key amino acid residues
necessary for conferring toxicity (e.g. CRISP proteins (30 –35)),

the role of cysteine spacing in determining folding structure and
toxicity, and a demonstrated role for novel mutations such as
large scale deletions in conferring new bioactivities.
CRISP toxin transcripts were present in the majority of the
venom glands examined. Previous analysis of CRISP toxin
sequences from the venom of the natricid snake R. tigrinus
showed that it lacked the EX2F motif thought to be responsi-
ble for the smooth muscle paralytic effect as well as the
KX6KR motif hypothesized to be essential for the inhibition of
cyclic nucleotide-gated calcium channels. Subsequent bioac-
tivity testing confirmed that R. tigrinus CRISP toxin indeed
lacked these activities (31, 32, 42, 43), but it was unclear
whether this was representative of CRISP toxins from the
venoms of other non-front-fanged snakes. In this study, forms
containing the EX2F motif were present in some venoms (Figs.
5 and 6). Translation and expression of these encoded proteins
would thus likely induce smooth muscle paralysis. It is notable
that CRISP toxins are particularly rich in the venoms of species
that are reptilian feeders (14) such as T. dhara and T. biscutatus.
FIG. 9. Bayesian molecular phylogeny of kallikrein toxins. Out- It may be that the toxic hypothermic effect is useful in slowing
groups are non-toxin forms Q9GOG3 and Q9UBX7 from H. sapiens. down the movement of exothermic prey. Studies on this aspect
FIG. 10. Sequence alignment of

kunitz toxins with functional residues
(protease-inhibiting reactive bond)
shown in bold and the leader se-
quence shown in lowercase. 1,
EU029687 from T. dhara; 2, EU029688
from P. olfersii; 3, P24541 from Eristoco-
phis macmahonii; 4, P82966 from O.
hannah; 5, P19859 from N. naja; 6,
P00981 from Dendroaspis polylepis; 7,
Q7LZE4 from O. scutellatus; 8, Q90WA0
from P. textilis; 9, P00989 from B. mul-
ticinctus; 10, the representative non-
toxin P04815 from Bos taurus.


FIG. 11. A, Bayesian molecular phy-
logeny of kunitz toxins. The outgroup is
the non-toxin P04815 from B. taurus. B,
ribbon view of EU029687 from T. dhara.
␣-Helices are in red, ␤-strands are in
yellow, and the functional residue Ala-18
is shown in green.
may shed significant light into the evolutionary pressures driving nels). To attract positively charged ions this pore contains
the molecular diversification of this toxin type. both acidic and aromatic residues. Therefore, extracellular
Cation-permeable channels (sodium-, potassium-, and cal- channel blockers often contain positive charges, i.e. guani-
cium-selective channels as well as cyclic nucleotide-gated dinium groups as in arginine, which are flanked by H-bond
channels, which are unselective for monovalent Na⫹ and K⫹ donors, like in conotoxins and dendrotoxins. Due to their
but also allow transfer of Ca2⫹ cations) often contain more size, venom proteins can affect channel gating only from the
than one binding site for channel blockers, one generally extracellular side. We postulate that to occlude the channel
within the transmembrane bundle to bind e.g. anesthetics and pore these proteins require a 3D structure motif that (i) is
one or more at their extracellular side that interact with toxins. accessible to interact with an ion channel, (ii) consists of one
A most effective channel block occurs when the ion perme- (ideally positively) charged residue to interact with the
ation pore is directly occluded by a toxin from the extracellular charged amino acids inside the ion permeation pore, (iii)
side. This pore in many cases is bevelled, narrowing toward a exhibits one or more H-bond donor(s) in vicinity to the
selectivity filter (i.e. in potassium, calcium, and sodium chan- charged residue, and (iv) possibly contains an aromatic

FIG. 12. Sequence comparison of

representative lectin toxins. The homo-
meric forms shown are: 1, EU029691; 2,
EU029689; 3, EU091713, all from E. polyl-
epis; 4, EU029699 from L. madagascar-
iensis; 5, EU029697; 6, EU029702, both
from L. poecilogyrus; 7, EU029700 from
P. olfersii; 8, EU029696 from T. jacksonii;
9, Q90WI6 from B. multicinctus; 10, Mi-
crurus corallinus sequence published
but not database-curated (62); 11,
Q6TRS6 from Bothrops jararacussu; and

12, Q6T7B7 from B. gabonica. Repre-
sentative heterodimeric forms (␣- and
␤-chains, respectively) shown are: 13,
Q8JIV6; and 14, Q8JIV7A, both from D.
acutus. Also included is the representa-
tive non-toxin form: 15, P83300 from
Anser anser.
amino acid to form either ␲-␲ or cation-␲ interactions with date and were mainly postulated from multiple sequence align-
the channel pore. ments. To our understanding, the cysteine-rich domain contains
Structural studies revealed that CRISP is a two-domain three sites, which may putatively block ion channels: (i) the loop
protein of conserved fold consisting of a pathogenesis-related between ␤11 and ␣7 spanning eight residues between the first
PR-1 and a CRD, which are linked together via a flexible hinge and second cysteine in CRD, (ii) the loop connecting ␣7 and ␣8
(30, 44). Although significantly smaller than PR-1, the CRD containing up to six amino acids, and (iii) the C-terminal tail
spans the C-terminal six of a total of 16 conserved cysteines. following the sixth conserved CRD cysteine.
The resulting three disulfide bonds stabilize the CRD fold and A hypothesized functional EX2F motif (34, 35) was found
leave only marginal room for movement in this domain. It is three to six residues C-terminal to the first CRD cysteine. In
widely assumed that hydrophobic residues of a concave cleft this study we could not confirm this motif to be conserved and
including both the PR-1 and the cysteine-rich domain would thus functional throughout the CRISP family. However, the
bind the CRISP to the ion channel (30), whereas actual chan- widespread taxonomical presence is suggestive that if this
nel block is hypothesized to involve amino acids of CRD only. motif is not basal it is at least early emerging. Whereas the
The fact that CRISP toxins block a variety of ion channels may aromatic residue was conserved (either phenylalanine or ty-
indicate that different residues located at different positions of rosine), the glutamate occurred in only five of 16 sequences.
CRD are responsible for occluding the extracellular entry to At the same position five asparagines, one glutamine, and two
the channel. This hypothesis is supported by the observation lysines were found, but interestingly no aspartate was found.
that CRD can move relative to the PR-1 domain, which may In contrast, the C-terminal neighbor to glutamate in position
even lead to exposure of residues that in the experimental ⫹3 of the first CRD cysteine is highly conserved exhibiting
x-ray structures (Protein Data Bank codes 1RC9 (44) and only aspartate (nine occurrences) and asparagine (11 occur-
1XX5 (31)) are found at the interface to PR-1. Yet the amino rences). This position corresponds to the first residue in the
acids actually blocking ion channels remained unidentified to DVF motif that has been identified to selectively interact with


FIG. 13. Bayesian molecular phylogeny of lectin toxins. The outgroup is the non-toxin P83300 from A. anser. *, M. corallinus sequence
published but not database-curated (62).
L-type calcium channels (35). It ranges from position 4 to 6 Alternatively, interactions with L-type calcium channels may
after the first CRD cysteine. This motif was first identified in involve one of the following motifs. (i) The loop tip between
␻-conotoxin TxVII (45) and was later found in the venom helices ␣7 and ␣8 may be involved. Here the first residue of
protein ablomin (35). Our structural studies showed that the the loop is either polar (Gln or Asn) or charged (Lys or Glu)
side chain of the aspartate/asparagine in DVF was oriented followed by a mostly polar (Asn, Ser, or Thr) and an either
toward the core of CRD forming several H-bonds with the polar (Ser, Asn, Thr, or His) or charged amino acid (Lys or Asp)
third and most C-terminal a-helix ␣9. The second and third just before the third conserved cysteine. This region contains
residues of this motif are located at the surface of CRD with charged residues in many toxins and is structurally exposed.
their side chains pointing toward the solvent. Yet valine was However, it lacks an aromatic amino acid in its vicinity. (ii) The
found only three times in this study, whereas in 14 instances residues in position ⫹1 or ⫹2 of the first and in ⫹1 of the fifth
this residue was charged (four times lysine and 10 times conserved CRD cysteine may be involved. The former was
glutamate and aspartate). The third residue of this motif was postulated to be involved in cyclic nucleotide-gated channel
large and hydrophobic and, with the exception of the se- block and often contains a basic amino acid (see below and
quence Q3SB05 from Pseudonaja textilis, aromatic. We hy- Fig. 5); the latter is predominantly phenylalanine. This sub-
pothesize that these two residues may dock to ion channels structure is also accessible and is held in place by a nearby
because they satisfy, in most cases, the four criteria postu- disulfide bond. (iii) Positions ⫹1, ⫹2, and ⫹3 after the sixth
lated earlier: (i) spatial accessibility, (ii) a charged amino acid, CRD cysteine may be involved. This part of the C terminus is
(iii) nearby H-bond donors, and (iv) an aromatic residue in mostly positively charged and polar. It is relatively flexible and
spatial proximity to the charged amino acid. Exceptions to accessible and may therefore interact with L-type calcium
criterion ii are several hydrophobic or polar residues found in channels.
D. typus, L. poecilogyrus, E. polylepis, Lapemis curtus, Pro- Morita and co-workers (32, 33, 35) proposed one or two
tobothrops mucrosquamatus, and Heloderma horridum. Inter- residues immediately C-terminal of the first conserved CRD
estingly potential H-bond donors are available in positions ⫺1 cysteine to be critical for interaction with cyclic nucleotide-
and ⫺2 of the fourth CRD cysteine (Fig. 5). gated ion channels. Their work revealed that the C-terminal


FIG. 15. PLA2 (Type IB) analyses. A, sequence alignment of rep-
FIG. 14. Matrix metalloprotease. A, partial sequence alignment of resentative toxins. 1, EU029703 from T. biscutatus; 2, P80966 from O.
EU091709 from L. poecilogyrus (1), Q9DE15 from G. gallus (2), hannah; 3, Q9PSN5 from Notechis scutatus; 4, Q9W7J3 from P.
A2VCV4 from X. laevis (3), Q98TC6 from Cyprinus carpio (4), P50757 textilis; 5, the non-toxin representative Q8JFB2 from L. semifasciata
from Oryctolagus cuniculus (5), and Q9TUL8 from Equus caballus (6). pancreas. B, Bayesian molecular phylogeny; the outgroup is Q8JFB2
B, Bayesian molecular phylogeny of lectin toxins. The outgroup is the from L. semifasciata liver. The ribbon model of TRI002F09 from T.
non-toxin P52176 from B. taurus. biscutatus shows the conserved catalytic residues (63). ␣-Helices are
in red, and ␤-strands are in yellow.
domains of pseudechetoxin and pseudecin, which are very
similar to CRD, were identical but for two residues. The KR that have specificities or potencies differing radically from
motif in pseudechetoxin made this protein more susceptible previously characterized forms, including the potential emer-
to cyclic nucleotide-gated block than the NY of pseudecin. In gence of neofunctionalizations.
this work, seven of 19 specimens had a positively charged Most of the 3FTxs sequenced in this study retained the
residue in either position ⫹1 or ⫹2 of the first cysteine (Fig. 5). ancestral 10-cysteine arrangement (19) (Fig. 2) as had been
For these specimens, we therefore postulate some activity to shown previously for the ␣-neurotoxic 3FTx from the venom
block cyclic nucleotide-gated channels. of the colubrid C. radiatus (2). However, the conservation of
Modification of structural residues was also evident in nu- the ancestral cysteines does not preclude bioactivities other
merous caenophidian venom toxin transcripts. It has been than ␣-neurotoxicity as typified by a form from the venom of
shown that changes in the spacing of ancestral cysteines, the the colubrid snake Boiga dendrophila that retained the ances-
occurrence of newly evolved cysteines, and modifications in tral cysteines yet was only weakly ␣-neurotoxic but had a
structural residues flanking the cysteines can alter the three- newly derived presynaptic mode of neurotoxicity (16). As the
dimensional structure of the molecule and thus the residues sequences obtained in this study were even more divergent,
contributing to the surface chemistry with consequences for including the evolution of alternate cysteines, it is quite likely
bioactivity (20). Such modifications can produce toxin variants that a multiplicity of novel activities has been derived.

FIG. 16. N-terminal sequence align-

ment of SVMP/ADAM toxins. The partial
sequences from non-front-fanged ad-
vanced snakes are: 1, EU029705 from P.
olfersii; and 2, EU029707 from L. poecilo-
gyrus. The full-length unique truncated
forms are: 3, EU029708; 4, EU029713; 5,
EU029717; and 6, EU029727, all from P.
mossambicus. The representative viperid
venom sequences are: 7, O42138 from

Agkistrodon contortrix laticinctus; and 8,
Q4VM08 from Macrovipera lebetina. 9,
the representative elapid sequence
Q8JGN1 from Naja mossambica. 10, the
representative atractaspidid venom form
Q9PT48 from A. engaddensis. The rep-
resentative non-toxin forms are: 11,
Q9UKQ2 (ADAM28 type); and 12,
Q9H2U9 (ADAM7 type) from H. sapiens.
In addition to variations in functional residues and cys- Venom toxins from non-front-fanged snakes may also
teines, mutants were also sequenced with frameshifts, trun- prove useful for investigations of the structure-function rela-
cations, and putative exon deletions. Novel transcripts of the tionships of the normal body proteins from which they were
SVMP/ADAM toxin type were isolated from P. mossambicus derived (20). In some cases, much more is known about the
that were comprised solely of the propeptide domain (Fig. 16). toxic forms than the ancestral body homologues. For exam-
Although forms of SVMP/ADAM have been identified previ- ple, the normal body forms of the CRISP proteins are poorly
ously that selectively express a particular domain (e.g. com- characterized with virtually nothing known of their bioactivi-
prised solely of the disintegrin domain (19)) this is the first time ties. Even the activities of forms that may play important roles,
that the prepro domain has been discovered as being selec- such as those with high expression levels in specific tissue or
tively expressed as the sole domain. The putative exon dele- organs, remain unknown. The basal activity of the CRISP
tion in the lectin toxin variant from E. polylepis not only re- venom protein is likely to be relaxation of peripheral smooth
moved a large stretch of residues but also one of the ancestral muscle, such as helothermine from H. horridum (46), and may
cysteines, leaving a free cysteine, potentially facilitating indicate a tissue-specific role for the ancestral normal body
dimerization (Fig. 12). Similarly the putative frameshift in the protein. The kallikrein toxins are an example of where an
lectin toxin variant from E. polylepis produced transcripts with ancestral body action (liberation of bradykinin from kininogen)
even numbers of cysteines, and therefore these sequences is preserved, but some toxin isoforms have derived additional
may be able to fold to produce stable frameworks (Fig. 12). activities (e.g. cleavage of fibrinogen). Similarly the two inde-
Thus, genomic mutations in both the caenophidian SVMP/ pendently recruited types of PLA2 proteins in snake venoms
ADAM and lectin toxins have the potential to produce func- have derived activities that the normal body form lacks (e.g.
tional proteins with significantly altered bioactivities. antiplatelet toxicity, myotoxicity, and neurotoxicity). Consist-

FIG. 17. C-terminal sequence align-

ment of SVMP/ADAM toxins. The partial
sequences from non-front-fanged snakes
are: 1, EU029734; 2, EU029736; 3,
EU029737, all from D. typus; 4, EU029735
from T. jacksonii; 5, EU029738; 6,
EU029739, both from L. madagascarien-
sis; and 7, EU036637 from T. dhara. The

typical viperid venom forms are: 8,
O42138 from A. contortrix laticinctus; and
9, Q4VM08 from M. lebetina. 10, the rep-
resentative truncated disintegrin form
Q6T6T2 from B. gabonica. 11, the repre-
sentative elapid sequence Q8JGN1 from
N. mossambica. 12, the representative
atractaspidid venom form Q9PT48 from
A. engaddensis. The representative non-
toxin forms are: 13, Q9UKQ2 (ADAM28
representative); and 14, Q9H2U9 (ADAM7
representative) from H. sapiens.
ent with this, the toxic forms contain a positively charged are either shared with the two other Toxicofera lineages (Igua-
hotspot on the surface that is lacking in the ancestral form from nia and Anguimorpha) (7 of 27) or occur near the base of the
the pancreas (Fig. 15). The venom proteins therefore represent Caenophidia snake clade (7 of 27). The independent evolu-
exquisite natural “knock-out” studies of tremendous usefulness tions of advanced front-fang architectures in Atractaspis and
in elucidating the structure-function relationships of the non- Viperidae are linked with recruitments of new toxin types (Fig.
toxin, body homologues of physiological importance. The rela- 1). In the viperids, two new toxin types, c-type natriuretic
tive assignment of conserved/variable functional residues peptide-bradykinin-potentiating peptide and PLA2 (Type IIA),
should not be considered definitive as not all sites/proteins are are closely linked to the evolution of advanced venom delivery
well characterized, and other transcripts/toxin types may con- systems and ambush feeding, whereas four other toxin types
tain functional residues that confer ancestral or novel bioactiv- were evolved later on (renin-like aspartic protease, cytokine
ities, especially for the toxin types not listed above. FAM3B, glycine-rich toxin, and waglerin). The evolution of
Timing of Toxin Recruitment Events—The phylogenetic dis- advanced venom architecture in viperids also coincides with
tribution of venom toxins in the Toxicofera (venomous squa- significant molecular diversification of two existing toxin types
mates) provides insights into the timing of toxin recruitment (SVMP/ADAM and kallikrein). The evolution of sarafotoxins in
events (Fig. 1). The majority of snake venom toxins (14 of 27) Atractaspis is also closely timed with the evolution of ad-

FIG. 18. Bayesian molecular phylog-

eny of SVMP/ADAM toxins. The out-
group is the non-toxin ADAM28 se-
quence Q9UKQ2 from H. sapiens. *, a
published but uncurated sequence from
Micropechis ikaheka (64).

vanced front fangs, but whether as a cause or an effect non-front-fanged species have been shown to be just as toxic
(occurring before or after) the evolution of hollow fangs re- as well characterized extremely potent elapid venoms (40). A
mains to be elucidated. In the elapids, the only two additional unique nuance of this potent and fast acting toxin type is that,
toxin types currently known do not coincide with development unlike most other toxins, the lethality is almost the same
of a high pressure front-fanged venom system with the hole- whether injected subcutaneously, intramuscularly, or even
punching, short stubby fangs useful for penetrating tough intravenously (19), and therefore delivery just under the skin is
reptile scales. Instead the evolution of new venom systems as efficient as straight into the bloodstream. The highest
appears to also be linked to the explosive diversification of expression levels of this toxin type in the non-front-fanged
existing toxin types: 3FTx and PLA2 (Type IB) toxins in the snake families are in the gracile, fast moving forms that have
elapids with numerous new bioactivities developed for each only slightly or moderately enlarged rear maxillary teeth and
toxin type. The only documented newly evolved toxin types are predate upon fast moving, soft, thin skinned, non-dangerous
found within the rapidly radiated Australian elapid snakes. One prey items such as geckos and frogs (14). In contrast, the long
evolved at the base of this clade (factor X), and one evolved in fanged D. typus is an arboreal snake that includes birds as a
the common ancestor of Pseudonaja and Oxyuranus (factor V). major prey item, and the venom is rich in forms of the SVMP/
The very long, derived glands of C. rhombeatus do not ADAM toxins that are potently prothrombin-activating. The
appear to be linked to the evolution of new toxin types either fish-eating aquatic homalopsid snakes, which also have large
with the dominant toxin types being CRISP, kallikrein, Type IIA maxillary dentition, favor hemotoxic lectin toxins as well as
PLA2, and SVMP/ADAM (including RGD disintegrins). Within ADAM toxins. The elapid snakes, which puncture tough reptile
Atractaspis the venom differences between long and short skin with their short, strong hollow front fangs, also typically
glanded species is unclear. It is also unknown whether the express 3FTx transcripts as the dominant mRNA species.
long glands of some Calliophis species are linked to signifi- However, they favor the derived forms that lack the second
cant changes in venom composition. and third ancestral cysteines.
The early recruitment of a LYNX/SLUR-like gene to form the Within the elapids additional toxin types were evolved later
3FTx toxin multigene family was one of the most significant on but also with tremendous impact, such as factor X at the
developments in snake venom evolution, priming the Cae- base of the Australian elapid clade and factor V in the com-
nophidia for extensive diversification. Indeed, venoms from mon ancestor of Pseudonaja/Oxyuranus. The recruitment of

FIG. 19. A, sequence comparison of

representative waprin toxins from non-
front-fanged advanced snakes (1,
EU029745 from E. polylepis; 2, EU029744
from L. poecilogyrus; 3, EU029743; 4,
EU029746; 5, EU029742, all from P. olf-

ersii; 6, EU029741 from R. tigrinus; 7,
EU029740 from T. jacksonii) and from
front-fanged advanced snakes (8, P60589
from Naja nigricollis; 9, P83952 from O.
microlepidotus) and of the representative
non-toxin form (10, P19957 from H. sapi-
ens). B, Bayesian molecular phylogeny of
waprin toxins. The outgroup is the non-
toxin P83300 from A. anser. The ribbon
model of EU029740 from T. jacksonii
shows ␤-strands in yellow.
factor X and factor V in the Australian clade of elapids would tremely long fangs and high venom yields. Unlike most Aus-
have also tremendously aided prey capture with only a small tralian snakes, which often feed on reptiles, these snakes feed
amount of these very active enzymes needed to cause severe exclusively on mammalian prey including large rats and
disruption of blood chemistry. However, the activity of factor bandicoots, dangerous prey items capable of an adept de-
X toxins is rate-limited by the requirement for cofactors, par- fense potentially causing serious injury to the snake preda-
ticularly the essential need to form a 1:1 complex with factor tors. The snakes successfully feed on such dangerous prey by
V (47). Thus, the recruitment of factor V into the venom of the minimizing prey contact time through the use of a snap-
Oxyuranus/Pseudonaja clade resulted in a virtually complete, release form of striking along with a tendency for multiple
perfect toxin complex, greatly increasing the relative toxicity strikes, thus overwhelming the prey items with copious
of the venoms and thus significantly aiding in the prey cap- amounts of the extraordinarily potent venom delivered deep
turing ability. Subsequent to the recruitment of factor V, the into the tissues several times. A limitation of this form of prey
Oxyuranus/Pseudonaja common ancestor split into two very capture is that the snakes need to be very warm to success-
different forms despite the two genera remaining very closely fully use this energetically costly form of prey capture. Con-
genetically related. sequently their habitat is limited only to the tropical north of
The Oxyuranus species are very large snakes with ex- Australia and in New Guinea (Oxyuranus scutellatus) and the

baking heat of the channel country in the Australian outback

(O. microlepidotus).
In contrast, the Pseudonaja species have taken a very dif-
ferent evolutionary path, taking advantage of possessing such
toxic venom to greatly minimize the amount of biological
energy expended in venom production. Pseudonaja species
are smaller snakes with very short fangs and much smaller
venom yields, utilizing constriction to hold non-dangerous
prey items (such as reptiles) in place while envenomating. As
these snakes do not expend as much energy in prey capture,
they do not have to be as warmed up as the Oxyuranus
species. These snakes consequently have taken advantage of
their extremely toxic venom to become one of the most
adaptable and successful snake types of the entire Australian
continent.
Another conspicuous finding of this study was the resolu-

tion of the lectin toxins into a single clade with homomeric
forms not reciprocally monophyletic in relation to the het-
erodimeric form found exclusively in viperid venoms (Fig. 13).
The homomeric and the heterodimeric lectin forms had been
FIG. 20. Transverse histology of Masson’s trichrome-stained considered previously to be the result of two separate toxin
sections showing the relative size of venom glands (proportional
recruitment events (3) or the result of a single recruitment
to head size) for H. leopardinus (A), D. typus (B), Homalopsis
buccata (C), P. guttatus (D), and D. scabra (E). VG, venom gland; event (18). However, reconstructing the evolutionary history
SG, supralabial gland. had been hampered by the scarcity of homomeric sequences.
FIG. 21. Histology using modified

Van Gieson’s stain. A, frontal section
through Ahaetulla nasuta showing the
structural differences between the su-
pralabial gland and venom gland. B,
transverse section showing a proximal to
distal transition in the epithelial lining of
the venom duct of Psammophis
schokari. C, frontal section showing the
relatively small venom vestibule of D.
punctatus located adjacent to the fang.
D, frontal section through R. tigrinus
showing the venom vestibule located
immediately adjacent to the venom
gland. E, frontal section through the
venom duct of T. capensis showing the
concentric rings of surrounding connec-
tive tissue. F, transverse section through
the venom duct of Coluber constrictor
showing the venom duct opening to the
oral cavity. G, frontal section through
Psammodynastes pulverulentus show-
ing the direct course of the venom duct
from the venom gland to the fang
sheath. H, frontal section through the
fang sheath and diverticulum of Malpo-
lon monspessulanus. I, frontal section
through Oligodon smithii. showing the
accessory venom duct extending away
from the fang sheath. D, venom duct; F,
fang; O, oral cavity; S, fang sheath; SG,
supralabial gland; VG, venom gland; V,
venom vestibule.

TABLE III
Summary of venom dentition variation in genera studied (for full species list see ⬙Appendix III⬙)
Smooth surface and no enclosed venom canal Colubridae: Coluber, Dasypeltis, Lycodon, Masticophis, Oligodon, Pituophis,
Philothamnus, Spalerosophis. Dipsadidae: Diadophis, Heterodon,
Hydrodynastes, Hypsiglena. Lamprophiinae: Lamprophis, Lycophidion.
Natricidae: Macropisthodon, Nerodia, Rhabdophis.
No enclosed venom canal and surface with Dipsadidae: Erythrolamprus. Natricidae: Amphiesma.
shallow furrow
Deep groove present but restricted to less than Colubridae: Ahaetulla, Chrysopelea, Dispholidus, Trimorphodon. Dipsadidae:
half the length of the tooth Coniophanes, Leptodeira, Oxyrhopus, Pseudoboa, Tachymenis, Tomodon.
Homalopsidae: Cerberus, Enhydris, Erpeton. Lamprophiinae:
Psammodynastes. Psammophiinae: Psammophis. Pseudoxyrhophiinae:
Madagascarophis.
Deep groove running the majority of the length Atractaspidinae: Amblyodipsas, Aparallactus. Colubridae: Boiga, Chionactis,
of the tooth Crotaphopeltis, Oxybelis, Stenorrhina, Thelotornis. Dipsadidae: Philodryas.
Psammophiinae: Malpolon, Psammophylax.
Enclosed venom canal Atractaspidinae: Atractaspis only. Elapidae: ubiquitous. Viperidae: ubiquitous.
In this study, we sequenced numerous lectin toxin transcripts gland in non-front-fanged snakes (54), the relatively weak wall
from four additional non-front-fanged snake families. This of the venom duct, and the presence of a venom vestibule

allowed for a more robust reconstruction of the molecular would all oppose the pressurization of the venom delivery
evolutionary history and thus the confirmation that all forms of system (but see Jansen and Foehring (55)). The multiple to-
the snake venom lectin toxins are the result of a single toxin pographic relationships between the venom duct and the
recruitment event (Fig. 1) and that the viperid venom het- fang, including taxa in which the venom duct opens to the roof
erodimeric forms arose from within the homomeric basal form of the mouth rather than the fang sheath, clearly indicate that
(Fig. 13). This study also allowed for the inference that the deformation of the fang sheath need not be an important part
EPN mannose-binding motif is basal whereas the QPD galac- of venom delivery in non-front-fanged snakes.
tose-binding variant was an early emerging derivation that Interestingly the colubrid family also contained a clade in
predated the viperid radiation. which the venom gland was greatly atrophied, the American
Derivations of the Venom System—Taub (7) described ex- “rat snake” clade of colubrid snakes, as typified by P. guttatus
tensive morphological variation within the venom gland (his (Figs. 1 and 20). The American rat snakes have secondarily
“Duvernoy’s” gland) of non-front-fanged snakes. The present evolved a new form of prey capture (powerful constriction)
study demonstrates that similar morphological variation exists and prey preference (rodents). Subsequently the gland has
in the venom duct as well as in the topographic relationship become greatly atrophied. Similarly the African D. scabra,
between the venom duct and the fang sheath (see “Appendix which feeds exclusively on bird eggs, also has greatly atro-
I”). These morphological variations do not assort along estab- phied venom glands (Fig. 20). This is paralleled in the sea
lished evolutionary lines, nor do they correlate well with the snake Aipysurus eydouxii that, subsequent to switching from
attributes of the venoms detailed in this study. This is in feeding on fish to feeding exclusively on fish eggs, has greatly
contrast to the front-fanged Elapidae and Viperidae families, atrophied venom glands (Fig. 1) and almost entirely lost fangs
which each have a well defined, characteristic venom appa- accompanied by significant accumulation of deleterious mu-
ratus at the familial level with some minor variations (Fig. 1), tations to the toxins still transcribed (56, 57). Similarly the
such as the size of venom glands and connections of the lizard egg-eating Brachyurophis species of Australian terres-
compressor and other head muscles. The same is true for the trial elapids also have significant reduction of the venom
front-fanged genus Atractaspis but not for the non-front- system. Finally the other few lineages displaying an absence
fanged snakes now included in the Atractaspidinae subfamily, or reduction of the glands are malacophagous (Asian
which were shown in this study to be quite variable. pareatids and some American dipsadids) (1).
The mechanics of venom injection in front-fanged snakes Changes in gland size were not restricted to atrophying but
relies on pressurization of the venom delivery system (48 –51). also included spectacular lengthening. In each of the front-
The colubrid snake D. typus is the only species with well fanged clades, tremendous elongation has occurred inde-
characterized rudimentary compressor musculature (52), al- pendently: Atractaspis engaddensis, Atractaspis microlepi-
though there is evidence that this has been convergently dota, Atractaspis micropholis, and Atractaspis scortecci in the
evolved in the lamprophiine snake Mehelya capensis (52) and Atractaspinae, Calliophis intestinalis and Calliophis bivirgata in
in the non-front-fanged atractaspidine Brachyophis revoili the Elapidae, and Causus resimus and C. rhombeatus in the
(53). The general absence of these features from the venom Viperidae. The elongations extend to a quarter of the body
delivery system of non-front-fanged snakes suggests that a length. The biological advantage gained by these elongations
substantially different mechanism(s) is present within this remains to be elucidated. In at least the case of Causus the
group. The limited skeletal muscle contact with the venom change in gland length was not accompanied by a significant

shift in venom profile or recruitment of new toxin types. In- treat this species with the utmost caution until studies of the
stead, having typical viperid venom composition with the action of T. jacksonii venom upon the hemostatic system are
major toxin types sequenced (SVMP, kallikrein, and PLA2 undertaken, including the relative neutralization by D. typus
(Type IIA)) in this study being consistent with molecular antivenom. In at least this case, absence of evidence should
weights obtained in a previous LC/MS study by us (14). It not be taken as evidence of absence.
remains to be investigated whether there are significant Conclusion—Collectively these findings strongly support
venom compositional differences between long and short that there has been extensive evolutionary modification of the
glanded forms within the Atractaspis and Calliophis genera. venom system in the advanced snake radiation (Caenophidia).
Snakes of Potential Medical Concern—This study has im- All variables appear to change independently ranging from the
portant implications in regard to human medicine by providing biochemical variation and specialization of the venoms to the
new information regarding snake venom composition and dentition and glandular morphology. The long evolutionary
thus potential clinical effects of envenomation. It also high- history of the advanced snake radiation (⬃100 million years)
lights the potential danger of snakes regarded as of trivial has resulted in extensive diversification of the venom system
concern based solely upon the lack of severe bites, but in and shows how little we really know about venom macroevo-
some cases the rarity of reports is due to the rarity of the lution and the potential role in the ecology/evolution of the
snake in captivity or encounters in the wild. animals themselves. Furthermore the toxin types identified
This study has shown that the venom transcriptome of here should not be regarded as the only toxins present in

Thrasops jacksonii is very similar to the extremely potent those venoms. Additional experimental efforts will certainly
venom of the very closely related (and proven lethal) D. typus, reveal new isoforms of these toxins, and some classes will
particularly the abundance of PIII-type SVMP/ADAM toxins.
almost certainly be confirmed as evolving earlier than cur-
The lethality of D. typus venom is the result of high expression
rently appreciated. Further entirely new toxin types may yet be
levels of a potent prothrombin-activating form of the SVMP/
discovered in any one of the poorly studied snake lineages.
ADAM toxin type. In the clinical pathology, the prothrombin
Our major findings are of a multidisciplinary nature with uses
activation causes consumptive coagulopathy, resulting in the
ranging from evolutionary biology to protein evolution and
disappearance of fibrinogen from the blood and the elevated
highlight the underutilization of snake venoms as biore-
presence of d-dimer products with the ultimate net result of
sources in drug design and development.
incoagulable blood. Predation by T. jacksonii on live mamma-
lian prey results in rapid death in a manner identical to that Acknowledgments—The Naturalis Museum, Leiden, The Nether-
produced by D. typus: profound disruption of hemostasis lands generously loaned us preserved specimens for some of mag-
manifested by rapid and copious bleeding from the mouth netic resonance imaging studies. H. S. is grateful to GlaxoSmithKline
for an exclusive version of SPDBV. We are deeply grateful to Profes-
and nostrils.4 Clearly potent venom toxins are actively ex-
sor Elazar Kochva for patience in answering our many questions and
pressed in this species. We advise exotic animal keepers to for very helpful comments on the manuscript. We also thank the
various wildlife departments for the issuing of the relevant scientific
4
B. G. Fry, personal observations. collecting permits.

APPENDIX I
Species and mode examined
C, cDNA library; G, gross dissection; D, dentition; H, histology; M, magnetic resonance imaging.

APPENDIX I—continued

APPENDIX I—continued

APPENDIX II
Structural variation
Epithelium: 1, no mucus; 2, isolated mucoid cells or patches; 3, prominent mucoid region or gland. Contour: 1, relatively large ovate duct;
2, duct diameter reduced, surrounded by extensive circular connective tissue; 3, internal partitions to venom duct. Vestibule: 1, no vestibule;
2, vestibule present adjacent to venom gland; 3, vestibule present adjacent to fang sheath; 4, vestibule in contact with oral cavity. Topography
of venom duct opening: 1, venom duct opens only to oral cavity; 2, venom duct opens only to fang sheath; 3, venom duct opens to both oral
cavity and fang sheath.

APPENDIX II—continued

APPENDIX II—continued

APPENDIX III
Dentitional morphology

* This work was supported by grants (to B. G. F.) from the Austra- Snake Trimorphodon biscutatus lambda (family Colubridae). Toxicon 44,
lian Academy of Science, Australian French Association for Science 27–36
and Technology, Australia and Pacific Science Foundation, CASS 16. Lumsden, N. G., Fry, B. G., Ventura, S., Kini, R. M., and Hodgson, W. C.
(2005) Pharmacological characterisation of a neurotoxin from the venom
Foundation, Ian Potter Foundation, International Human Frontiers
of Boiga dendrophila (Mangrove snake). Toxicon 45, 329 –334
Science Program Organisation, Netherlands Organisation for Scien-
17. Lumsden, N. G., Banerjee, Y., Kini, R. M., Kuruppu, S., and Hodgson, W. C.
tific Research, and University of Melbourne (Faculty of Medicine and (2007) Isolation and characterization of rufoxin, a novel protein exhibiting
Department of Biochemistry) and a Department of Innovation, Indus- neurotoxicity from venom of the psammophiine Rhamphiophis oxyrhyn-
try and Regional Development Victoria Fellowship; by Australian Re- chus (Rufous beaded snake). Neuropharmacology 52, 1065–1070
search Council Grant DP0451663 (to B. G. F. and J. A. N.); and by an 18. Ching, A. T., Rocha, M. M., Paes Leme, A. F., Pimenta, D. C., de Fatima, M.,
Australian Government Department of Education, Science and Train- Furtado, D., Serrano, S. M. T., Ho, P. L., and Junqueira de Azevedo,
ing/EGIDE International Science Linkages grant (to B. G. F., J. A. N., I. L. M. (2006) Some aspects of the venom proteome of the Colubridae
and N. V.). The costs of publication of this article were defrayed in part snake Philodryas olfersii revealed from a Duvernoy’s (venom) gland tran-
by the payment of page charges. This article must therefore be hereby scriptome. FEBS Lett. 580, 4417– 4422; Correction (2006) FEBS Lett.
marked “advertisement” in accordance with 18 U.S.C. Section 1734 580, 5122–5123
19. Fry, B. G., Wüster, W., Kini, R. M., Brusic, V., Khan, A., Venkataraman, D.,
solely to indicate this fact.
and Rooney, A. P. (2003) Molecular evolution of elapid snake venom
The nucleotide sequence(s) reported in this paper has been sub- three finger toxins. J. Mol. Evol. 57, 110 –129
mitted to the GenBankTMEBI Data Bank with accession number(s) 20. Fry, B. G. (2005) From genome to ‘venome’: molecular origin and evolution
EU029668 –EU029765, EU036635–EU036637, and EU091709 – of the snake venom proteome inferred from phylogenetic analysis of
EU091713. toxin sequences and related body proteins. Genome Res. 15, 403– 420
§ To whom correspondence should be addressed. E-mail: 21. Nei, M., Gu, X., and Sitnikova, T. (1997) Evolution by the birth-and-death

bgf@unimelb.edu.au. process in multigene families of the vertebrate immune system. Proc.
储 Present address: Nestlé Research Center, Vers-chez-les-Blanc, Natl. Acad. Sci. U. S. A. 94, 7799 –7806
P. O. Box 44, 1000 Lausanne, Switzerland. 22. Harvey, A. L., Bradley, K. N., Cochran, S. A., Rowan, E. G., Pratt, J. A.,
Quillfeldt, J. A., and Jerusalinsky, D. A. (1998) What can toxins tell us for
REFERENCES drug discovery? Toxicon 36, 1635–1640
23. Menez, A. (1998) Functional architectures of animal toxins: a clue to drug
1. Vidal, N. (2002) Colubroid systematics: evidence for an early appearance of
design? Toxicon 36, 1557–1572
the venom apparatus followed by extensive evolutionary tinkering. J.
24. Cushman, D. W., Ondetti, M. A., Gordon, E. M., Natarajan, S., Karanewsky,
Toxicol. Toxin Res. 21, 21– 41
D. S., Krapcho, J., and Petrillo, E. W., Jr. (1987) Rational design and
2. Fry, B. G., Lumsden, N., Wüster, W., Wickramaratna, J., Hodgson, W. C.,
biochemical utility of specific inhibitors of angiotensin-converting en-
and Kini, R. M. (2003) Isolation of a neurotoxin (␣-colubritoxin) from a
zyme. J. Cardiovasc. Pharmacol. 10, S17–30
‘non-venomous’ colubrid: evidence for early origin of venom in snakes. J.
25. Chothia, C., and Lesk, A. M. (1986) The relation between the divergence of
Mol. Evol. 57, 446 – 452
sequence and structure in proteins. EMBO J. 5, 823– 826
3. Fry, B. G., and Wüster, W. (2004) Assembling an arsenal: origin and evo-
26. Greer, J. (1981) Comparative model-building of the mammalian serine
lution of the snake venom proteome inferred from phylogenetic analysis
proteases. J. Mol. Biol. 153, 1027–1042
of toxin sequences. Mol. Biol. Evol. 21, 870 – 883
27. Guex, N., and Peitsch, M. C. (1997) SWISS-MODEL and the Swiss-Pdb-
4. Fry, B. G., Vidal, N., Norman, J. A., Vonk, F. J., Scheib, H., Ramjan, R.,
Viewer: an environment for comparative protein modeling. Electrophore-
Kuruppu, S., Fung, K., Hedges, S. B., Richardson, M. K., Hodgson,
sis 18, 2714 –2723
W. C., Ignjatovic, V., Summerhayes, R., and Kochva, E. (2006) Early
evolution of the venom system in lizards and snakes. Nature 439, 28. Diemand, A. V., and Scheib, H. (2004) iMolTalk: an interactive, internet-
509 – 632 based protein structure analysis server. Nucleic Acids Res. 32,
5. Kardong, K. (1980) Evolutionary patterns in advanced snakes. Am. Zool. 20, W512–W516
269 –282 29. Koradi, R., Billeter, M., and Wuthrich, K. (1996) MOLMOL: a program for
6. Vidal, N., Delmas, A. S., David, P., Cruaud, C., Couloux, A., and Hedges, display and analysis of macromolecular structures. J. Mol. Graph. 14,
S. B. (2007) The phylogeny and classification of caenophidian snakes 51–55
inferred from seven nuclear protein-coding genes. C. R. Biol. 330, 30. Shikamoto, Y., Suto, K., Yamazaki, Y., Morita, T., and Mizuno, H. (2005)
182–187 Crystal structure of a CRISP family Ca2⫹-channel blocker derived from
7. Taub, A. M. (1967) Comparative histological studies on Duvernoy’s gland of snake venom. J. Mol. Biol. 350, 735–743
colubrid snakes. Bull. Am. Mus. Nat. Hist. 138, 1–50 31. Wang, J., Shen, B., Guo, M., Lou, X., Duan, Y., Cheng, X. P., Teng, M., Niu,
8. Gabe, M., and Saint Girons, H. (1969) Histological data on the salivary L., Liu, Q., Huang, Q., and Hao, Q. (2005) Blocking effect and crystal
glands of the Lepidosauria. Mém. Mus. Natn. Hist. Nat. Paris 58, 1–112 structure of natrin toxin, a cysteine-rich secretory protein from Naja atra
9. Kochva, E. (1978) Oral glands of the Reptilia, in Biology of the Reptilia venom that targets BKCa channel. Biochemistry 44, 10145–10152
(Gans, C., and Gans, K. A., eds) Vol. 8, pp. 43–161, Academic Press, 32. Yamazaki, Y., Brown, R. L., and Morita, T. (2002) Purification and cloning of
New York toxins from elapid venoms that target cyclic nucleotide-gated ion chan-
10. Young, B., and Kardong, K. (1996) Dentitional surface features in snakes nels. Biochemistry 41, 11331–11337
(Reptilia: Serpentes). Amphib.-Reptilia 17, 261–276 33. Yamazaki, Y., and Morita, T. (2004) Structure and function of snake venom
11. West, G. S. (1895) On the buccal glands and teeth of certain poisonous cysteine-rich secretory proteins. Toxicon 44, 227–231
snakes. Proc. Zool. Soc. Lond. 1895, 195–206 34. Yamazaki, Y., Koike, H., Sugiyama, Y., Motoyoshi, K., Wada, T., Hishinuma
12. West, G. (1898) On the history of the salivary buccal and harderian glands S., Mita, M., and Morita, T. (2002) Cloning and characterization of novel
of the colubroidae with notes on their tooth succession and the relation- snake venom proteins that block smooth muscle contraction. Eur. J. Bio-
ships of the poison ducts. J. Linn. Soc. Lond. Zool. 26, 517–526 chem. 269, 2708 –2715
13. Sarkar, S. C. (1923) A comparative study of the buccal glands and teeth of 35. Yamazaki, Y., Hyodo, F., and Morita, T. (2003) Wide distribution of cysteine-
the opisthoglypha, and a discussion on the evolution of the order from rich secretory proteins in snake venoms: isolation and cloning of novel
aglypha. Proc. Zool. Soc. Lond. 1923, 295–322 snake venom cysteine-rich secretory proteins. Arch. Biochem. Biophys.
14. Fry, B. G., Wüster, W., Ramjan, S. F. R., Jackson, T., Martelli, P., and Kini, 412, 133–141
R. M. (2003) LC/MS (liquid chromatography, mass spectrometry) analy- 36. Harrison, R. A., Ibison, F., Wilbraham, D., and Wagstaff, S. C. (2007)
sis of Colubroidea snake venoms: evolutionary and toxinological impli- Identification of cDNAs encoding viper venom hyaluronidases: cross-
cations. Rapid Commun. Mass Spectrom. 17, 2047–2062 generic sequence conservation of full-length and unusually short variant
15. Huang, P., and Mackessy, S. P. (2004) Biochemical characterization of transcripts. Gene (Amst.) 392, 22–33
phospholipase A2 (trimorphin) from the venom of the Sonoran Lyre 37. Chen, C., Hsu, C. H., Su, N. Y., Lin, Y. C., Chiou, S. H., and Wu, S. H. (2001)

Solution structure of a kunitz-type chymotrypsin inhibitor isolated from (Colubridae) and the venom glands of Elaps (Elapidae) in relation to the
the elapid snake Bungarus fasciatus. J. Biol. Chem. 276, 45079 – 45087 taxonomic status of this genus. Zool. J. Linn. Soc. 49, 217–224
38. Drickamer, K. (1992) Engineering galactose-binding activity into a C-type 53. Underwood, G., and Kochva, E. (1993) On the affinities of the burrowing
mannose-binding protein. Nature 360, 183–186 asps Atractaspis (Serpentes: Atractaspididae). Zool. J. Linn. Soc. 107,
39. Komori, K., Konishi, M., Maruta, Y., Toriba, M., Sakai, A., Matsuda, A., Hori, 3– 64
T., Nakatani, M., Minamino, N., and Akizawa, T. (2006 May) Character- 54. Haas, G. (1973) Muscles of the jaw and associated structures in the
ization of a novel metalloproteinase in Duvernoy’s gland of Rhabdophis Rhyncocephalia and Squamata, in Biology of the Reptilia (Gans, C., and
tigrinus tigrinus. J. Toxicol. Sci. 31, 157–168 Parsons, T., eds) Vol. 4, pp. 285– 490, Academic Press, New York
40. Lumsden, N. G., Fry, B. G., Kini, R. M., and Hodgson, W. C. (2004) In vitro 55. Jansen, W., and Foehring, R. (1983) The mechanism of venom secretion
neuromuscular activity of ‘colubrid’ snake venoms: clinical and evolu- from Duvernoy’s gland of the snake Thamnophis sirtalis (Serpentes:
tionary implications. Toxicon 43, 819 – 827 Colubridae). J. Morphol. 175, 271–277
41. Kamiguti, A. S., Theakston, R. D., Sherman, N., and Fox, J. W. (2000) Mass 56. Li, M., Fry, B. G., and Kini, R. M. (2005) Eggs only diet: the shift in preferred
spectrophotometric evidence for P-III/P-IV metalloproteinases in the prey by the Marbled sea snake (Aipysurus eydouxii) resulting in a loss of
venom of the Boomslang (Dispholidus typus). Toxicon 38, 1613–1620. postsynaptic neurotoxicity. J. Mol. Evol. 60, 81– 89
42. Brown, R. L., Lynch, L. L., Haley, T. L., and Arsanjani, R. (2003) Pseudeche- 57. Li, M, Fry, B. G., and Kini, R. M. (2005) Putting the brakes on snake venom
toxin binds to the pore turret of cyclic nucleotide-gated ion channels. evolution: the unique molecular evolutionary patterns of Aipysurus ey-
J. Gen. Physiol. 122, 749 –760 douxii (Marbled sea snake) phospholipase A2 toxins. Mol. Biol. Evol. 22,
43. Brown, R. L., Haley, T. L., West, K. A., and Crabb, J. W. (1999) Pseudeche- 934 –941
toxin: a peptide blocker of cyclic nucleotide-gated ion channels. Proc. 58. Keogh, J. S., Shine, R., and Donnellan, S. (1998) Phylogenetic relationships
Natl. Acad. Sci. U. S. A. 96, 754 –759 of terrestrial Australo-Papuan elapid snakes (subfamily Hydrophiinae)
44. Guo, M., Teng, M., Niu, L., Liu, Q., Huang, Q., and Hao, Q. (2005) Crystal based on cytochrome b and 16S rRNA sequences. Mol. Phylogenet.
structure of the cysteine-rich secretory protein stecrisp reveals that the Evol. 10, 67– 81
cysteine-rich domain has a K⫹ channel inhibitor-like fold. J. Biol. Chem. 59. Vidal, N., Kindl, S. G., Wong, A., and Hedges, S. B. (2000) Phylogenetic

280, 12405–12412 relationships of xenodontine snakes inferred from 12S and 16S riboso-
45. Kobayashi, K., Sasaki, T., Sato, K., and Kohno, T. (2000) Three-dimensional mal RNA sequences. Mol. Phylogenet. Evol. 14, 389 – 402
solution structure of ␻-conotoxin TxVII, an L-type calcium channel 60. Vidal, N., and Hedges, S. B. (2002) Higher-level relationships of caenophid-
blocker. Biochemistry 39, 14761–14767 ian snakes inferred from four nuclear and mitochondrial genes. C. R. Biol.
46. Mochca-Morales, J., Martin, B. M., and Possani, L. S. (1990) Isolation and 325, 987–995
characterization of helothermine, a novel toxin from Heloderma horridum 61. Lawson, R., Slowinski, J. B., Crother, B. I., and Burbrink, F. T. (2005)
horridum (Mexican beaded lizard) venom. Toxicon 28, 299 –309 Phylogeny of the Colubroidea (Serpentes): new evidence from mitochon-
47. Kini, R. M., Rao, V. S., and Joseph, J. S. (2001) Procoagulant proteins from drial and nuclear genes. Mol. Phylogenet. Evol. 37, 581– 601
snake venoms. Haemostasis 31, 218 –224 62. Ho, P. L., Soares, M. B., Yamane, T., and Raw, I. (1995) Reverse biology
48. Rosenberg, H. (1967) Histology, histochemistry and emptying mechanism applied to Micrurus corallinus, a South American coral snake. J. Toxicol.
of the venom gland of some elapid snakes. J. Morphol. 122, 133–156 Toxin Rev. 14, 327–337
49. Young, B., Blair, M., Zahn, K., and Marvin, J. (2001) Mechanics of venom 63. Rouault, M., Rash, L. S., Escoubas, P., Boilard, E., Bollinger, J., Lomonte,
expulsion in Crotalus, with special reference to the role of the fang B., Maurin, T., Guillaume, C., Canaan, S., Deregnaucourt, C., Schrevel,
sheath. Anat. Rec. 264, 415– 426 J., Doglio, A., Gutierrez, J. M., Lazdunski, M., Gelb, M. H., and Lambeau,
50. Young, B. A., Lee, C. E., and Daley, K. M. (2002) Do snakes meter venom? G. (2006) Neurotoxicity and other pharmacological activities of the snake
Biosciences 52, 1121–1126 venom phospholipase A2 OS2: the N-terminal region is more important
51. Young, B. A., Dunlap, K., Koenig, K., and Singer, M. (2004) The buccal than enzymatic activity. Biochemistry 45, 5800 –5816
buckle: the functional morphology of venom spitting in cobras. J. Exp. 64. Gao, R., Kini, R. M., and Gopalakrishnakone, P. (2002) A novel prothrombin
Biol. 207, 3483–3494 activator from the venom of Micropechis ikaheka: isolation and charac-
52. Kochva, E., and Wollberg, M. (1970) The salivary glands of Aparallactinae terisation. Arch. Biochem. Biophys. 408, 87–92

Evolution of An Arsenal

Uploaded by

Document Informationclick to expand document information

Document Informationclick to expand document information

Copyright:

Available Formats

Evolution of An Arsenal

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Evolution of An Arsenal

Uploaded by

Copyright:

Available Formats

Research

Downloaded from www.mcponline.org at CNRS on February 7, 2008

Downloaded from www.mcponline.org at CNRS on February 7, 2008

216 Molecular & Cellular Proteomics 7.2

Downloaded from www.mcponline.org at CNRS on February 7, 2008

Molecular & Cellular Proteomics 7.2 217

Downloaded from www.mcponline.org at CNRS on February 7, 2008

218 Molecular & Cellular Proteomics 7.2

Downloaded from www.mcponline.org at CNRS on February 7, 2008

Molecular & Cellular Proteomics 7.2 219

FIG. 2. Sequence alignment of rep-

Downloaded from www.mcponline.org at CNRS on February 7, 2008

220 Molecular & Cellular Proteomics 7.2

FIG. 3. Bayesian molecular phylog-

Downloaded from www.mcponline.org at CNRS on February 7, 2008

FIG. 4. Molecular evolution of C3/

Molecular & Cellular Proteomics 7.2 221

FIG. 5. Sequence alignment of rep-

Downloaded from www.mcponline.org at CNRS on February 7, 2008

222 Molecular & Cellular Proteomics 7.2

FIG. 6. A, Bayesian molecular phylog-

Downloaded from www.mcponline.org at CNRS on February 7, 2008

Molecular & Cellular Proteomics 7.2 223

FIG. 7. Hyaluronidase. A, partial se-

Downloaded from www.mcponline.org at CNRS on February 7, 2008

224 Molecular & Cellular Proteomics 7.2

FIG. 8. Sequence alignment of the

Downloaded from www.mcponline.org at CNRS on February 7, 2008

Molecular & Cellular Proteomics 7.2 225

Downloaded from www.mcponline.org at CNRS on February 7, 2008

FIG. 10. Sequence alignment of

226 Molecular & Cellular Proteomics 7.2

Downloaded from www.mcponline.org at CNRS on February 7, 2008

Molecular & Cellular Proteomics 7.2 227

FIG. 12. Sequence comparison of

Downloaded from www.mcponline.org at CNRS on February 7, 2008

228 Molecular & Cellular Proteomics 7.2

Downloaded from www.mcponline.org at CNRS on February 7, 2008

Molecular & Cellular Proteomics 7.2 229

Downloaded from www.mcponline.org at CNRS on February 7, 2008

230 Molecular & Cellular Proteomics 7.2

FIG. 16. N-terminal sequence align-

Downloaded from www.mcponline.org at CNRS on February 7, 2008

Molecular & Cellular Proteomics 7.2 231

FIG. 17. C-terminal sequence align-

Downloaded from www.mcponline.org at CNRS on February 7, 2008

232 Molecular & Cellular Proteomics 7.2

FIG. 18. Bayesian molecular phylog-

Downloaded from www.mcponline.org at CNRS on February 7, 2008

Molecular & Cellular Proteomics 7.2 233

FIG. 19. A, sequence comparison of

Downloaded from www.mcponline.org at CNRS on February 7, 2008

234 Molecular & Cellular Proteomics 7.2

baking heat of the channel country in the Australian outback

Downloaded from www.mcponline.org at CNRS on February 7, 2008

FIG. 21. Histology using modified

Molecular & Cellular Proteomics 7.2 235