Article

Venom-Related Transcripts from Bothrops jararaca TissuesProvide Novel Molecular Insights into the Production andEvolution of Snake VenomInacio LM Junqueira-de-Azevedo12 Carolina Mancini Val Bastos12 Paulo Lee Ho3

Milene Schmidt Luna4 Norma Yamanouye4 and Nicholas R Casewell51Laboratorio Especial de Toxinologia Aplicada Center of Toxins Immune-Response and Cell Signaling (CeTICS) Instituto ButantanS~ao Paulo Brazil2Departamento de Genetica e Biologia Evolutiva Instituto de Biociencias Universidade de S~ao Paulo S~ao Paulo Brazil3Centro de Biotecnologia Instituto Butantan S~ao Paulo Brazil4Laboratorio de Farmacologia Instituto Butantan S~ao PaulomdashSP Brazil5Alistair Reid Venom Research Unit Liverpool School of Tropical Medicine Liverpool United Kingdom

Corresponding author E-mail inacioazevedobutantangovbr

Associate editor Katja Nowick

Abstract

Attempts to reconstruct the evolutionary history of snake toxins in the context of their co-option to the venom glandrarely account for nonvenom snake genes that are paralogous to toxins and which therefore represent importantconnectors to ancestral genes In order to reevaluate this process we conducted a comparative transcriptomic surveyon body tissues from a venomous snake A nonredundant set of 33000 unigenes (assembled transcripts of referencegenes) was independently assembled from six organs of the medically important viperid snake Bothrops jararaca pro-viding a reference list of 82 full-length toxins from the venom gland and specific products from other tissues such aspancreatic digestive enzymes Unigenes were then screened for nontoxin transcripts paralogous to toxins revealing 1) lowlevel coexpression of approximately 20 of toxin genes (eg bradykinin-potentiating peptide C-type lectin snake venommetalloproteinase snake venom nerve growth factor) in body tissues 2) the identity of the closest paralogs to toxin genesin eight classes of toxins 3) the location and level of paralog expression indicating that in general co-expression occursin a higher number of tissues and at lower levels than observed for toxin genes and 4) strong evidence of a toxin genereverting back to selective expression in a body tissue In addition our differential gene expression analyses identifyspecific cellular processes that make the venom gland a highly specialized secretory tissue Our results demonstrate thatthe evolution and production of venom in snakes is a complex process that can only be understood in the context ofcomparative data from other snake tissues including the identification of genes paralogous to venom toxins

Key words Bothrops jararaca transcriptome venom gland snake toxin differential expression Serpentes

IntroductionProtein neofunctionalization is one of the major processesdriving snake venom evolution Several snake venom toxinshave seemingly evolved from the recruitment of physiologicalmolecules that underwent convergent or divergent pathwaysto play a role in killing and paralyzing prey or defending theorganism (Ivanov CP and Ivanov OC 1979 Fry et al 2006Lynch 2007 Brust et al 2013)

These selected molecules are reasonably well knownas snake venoms have long been studied by biochemicalproteomic and transcriptomic approaches The latter haveprovided an important contribution to the field of toxinol-ogy by elucidating the diversity of molecules from differentprotein classes that are present in venom (Junqueira-de-Azevedo and Ho 2002 Ching et al 2006 Pahari et al 2007Casewell et al 2009 2014 Le~ao et al 2009 Durban et al 2011Rokyta et al 2012) Nevertheless with the exception of a fewnonvenomous species (Schwartz et al 2010 Castoe et al

2011 Reyes-Velasco et al 2014) no global view of geneexpression in the other body tissues of a venomous snakehas been described Genomic characterizations of snakes arealso neglected and are only recently becoming available(Castoe et al 2013 Vonk et al 2013) Thus the repertoireof body proteins and the tissue location of their expressionremain predominately unknown

A remarkable feature of many snakes is the ability of somespecies to produce in a controlled way high amounts ofcomplex protein solution used for an external purpose (ieinjection into prey)mdashvenom Snake venom glands (VGs) arevery specialized tissues that possess a high capacity of proteinsynthesis storage and secretion The proportion of secretorgranules in viperid VG cells has been shown to account foronly 4 of the cell volume and mRNA and protein synthesisare rapidly initiated suggesting that secretion constitutes afast process (Rotenberg et al 1971 De Lucca and Imaizumi1972 Oron and Bdolah 1973) However the cellular

The Author 2014 Published by Oxford University Press on behalf of the Society for Molecular Biology and EvolutionThis is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License(httpcreativecommonsorglicensesby-nc40) which permits non-commercial re-use distribution and reproduction in anymedium provided the original work is properly cited For commercial re-use please contact journalspermissionsoupcom Open Access754 Mol Biol Evol 32(3)754ndash766 doi101093molbevmsu337 Advance Access publication December 9 2014

mechanisms regulating protein productionsecretion in theVG are essentially unknown To assess whether these pro-cesses are specialized or shared with other secretory tissues itis necessary to first gain a greater understanding of the genesexpressed in different snake tissues

The lack of information about nonvenom genes alsoimpairs some important understandings relating to theprocess underpinning the evolution of venom This is be-cause without an adequate knowledge about the physio-logical counterparts that gave rise to toxins during theirevolutionary history phylogenetic analyses aimed at deter-mining the origin of toxins are deprived of a fundamentalcomponent and instead have to rely on genes from otherkinds of organisms This important aspect began to beaddressed recently (Casewell et al 2012) by analyzing theevolution of snake toxins in the context of snake bodyproteins related to those toxins The evolutionary pathwaysobserved in this study revealed an unexpected scenario ofplasticity represented by events of venom recruitment fol-lowed by reverse recruitment to nonvenom functions insome cases However this study used nonvenom genesfrom several nonvenomous snake species to comparewith venom genes from venomous species There are fewanimals for which toxin-related transcripts were investi-gated in other tissues to be compared with theVG-expressed toxins including the first know venomouscrustacean (von Reumont et al 2014) It would be ex-pected that comparing the venom and nonvenom genesfrom the same venomous species of snake would improvethe robustness of such an analysis

More recently a study analyzed the coexpression ofvenom toxins in a moderate number of snake tissues in-cluding the skin and scent gland (Hargreaves SwainHegarty et al 2014) The authors suggested that genesrelated to venom toxins are coexpressed in multiple tissuesand have subsequently been restricted to the VG(Hargreaves Swain Hegarty et al 2014) This is in contrastto a previous hypothesis whereby toxin antecedents werethought to be expressed in a specific tissue type before achange in expression location to the VG occurred (Fry2005) Unfortunately the expression level of toxin-relatedgenes in different snake tissues was not quantified byHargreaves Swain Hegarty et al (2014) thereby providinga major limitation to the interpretation of their results Bycontrast the analysis of Fry (2005) was limited by largelyutilizing tissue data from nonsnake taxa to predict theorigin of venom genesmdasha limitation inherent to the timeof study when few gene sequences from snake tissuesexisted

Here we conducted a comparative transcriptomic surveyon the VG and other body tissues of the viperid snakeBothrops jararacamdashthe most medically important snake inBrazil and one of the most well studied in terms of venomWe profiled the diversity and expression of genes found in adiverse number of snake tissues and compared them with VGdata to provide a better understanding of snake physiologyvenom production and the evolutionary history of snakevenom toxins

Results and Discussion

Bothrops jararaca Reference Transcriptome

Transcriptomes from different tissues of B jararaca weregenerated by 454-pyrosequencing Sequencing of cDNAfragment libraries from the VG pancreas liver kidneybrain and heart generated a total of 884235 reads thatafter filtering against adaptors primers vertebrate ribosomalRNAs (rRNAs) short and low quality sequences resulted in560242 high-quality sequences These sequences were as-sembled into contigs in a hierarchical manner first fromeach tissue and then by combining the different tissuedata sets to generate a list of reference transcripts (unigenes)from B jararaca A total of 33000 unigenes were generatedusing this multitissue assembly approach

Considering that 1) the majority of snake toxin classes areenriched for paralogous gene forms sometimes with minorsequence differences and 2) the automatic assembly of nextgeneration sequencing data is subject to a certain level ofinaccuracy due to the nature of short reads derived fromfragment cDNA libraries we checked the contigs for theoccurrence of misassembled multigenes In some cases werebuilt the assembly of toxin gene types under rigorousmanual inspection in order to improve the distinction ofputative paralogous genes and to obtain full-length tran-scripts This approach permitted us to generate a refinedreference data set of toxin coding transcripts for subsequentdownstream analysis (see supplementary table S1Supplementary Material online)

VG Transcriptome

In the case of the B jararaca VG 116236 reads were mappedto 13642 unigenes (table 1) The annotation process providedputative identification of 8690 VG unigenes correspondingto 81 of reads that could be assigned to a functional cate-gory based on significant similarity with the GenBank nrdatabase InterProScan screenings and GO (gene ontology)mappingmdash4700 unigenes remained unidentified The identi-fied set was classified into the categories represented infigure 1 Transcripts coding for cellular proteins amountedto 7069 unigenes representing 25 of total gene expressionwhereas putative toxin coding transcripts were representedby 582 unigenes accounting for 53 of gene expression This isin accordance with the 50ndash58 toxin expression described inprevious expressed sequence tag studies on related BothropsVG transcriptomes (Valente et al 2009 Zelanis et al 2012)

The relative abundances of the major toxin classes identi-fied (fig 1) are also in agreement with these previous tran-scriptomic surveys In total 16 toxin types were identified inthe VG transcriptome (fig 1) represented by 92 referencetoxin transcripts of which 79 contained full-length openread frames (supplementary table S1 SupplementaryMaterial online) Fifteen corresponded to variant or identicalforms of previously identified mRNAs or proteins from Bjararaca and the remaining 77 corresponded to new putativevenom genes Interestingly some toxins (mRNA or proteins)known from this species were absent from our VG data set

755

Snake Venom-Related Transcripts doi101093molbevmsu337 MBE

including the metalloproteinase HF3 (GenBank AF149788)and the serine protease KN-BJ (GenBank AB004067)Presumably these genes are not present or are not tran-scribed in the single specimen sampled here and may repre-sent alleles unevenly distributed in the population

Body Tissue Transcriptomes

Five other body tissues (from liver pancreas kidney heartand brain) from B jararaca were sequenced and individuallyanalyzed in a similar fashion to the VGs described above Thenumbers of sequences obtained for each transcriptome are

described in table 1 In total 29888 unigenes were expressedin these five tissue types with 17978 successfully annotatedSupplementary table S2 Supplementary Material online de-tails the top ten most expressed unigenes in each of thesetissues

Pancreatic TranscriptomeThe pancreas is a secretory epithelium with a very specializedfluid secretion that includes several enzymes Thus it can beconsidered an appropriate tissue type for comparing geneexpression with the VG as both are secretory tissues albeit

FIG 1 Transcriptomic profile of the VGs of Bothrops jararaca (A) Relative expression indicated by the percentage of total expression values (RPKM) ofall transcripts (upper panel) and among venom-related transcripts (lower panel) (B) Gene diversity indicated by the percentage of the number of totaltranscripts (unigenes) (upper panel) and among venom-related transcripts (lower panel) The venom-related category corresponds to transcriptscoding for known or putative toxins classified based on their sequence similarities to known toxin superfamilies Numbers in parentheses indicate theexpression percentage and the number of reference transcripts in each category

Table 1 Bothrops jararaca Transcriptome Sequencing and Assemble

Raw Reads Filtered Reads Average Size (bp) Mapped Reads Expressed Contigs Annotated Contigs Unique Contigs

VG 219550 142726 3670 116236 13642 8690 3074

Liver 66870 41471 2007 31391 6030 4263 589

Pancreas 223070 26163 2504 19422 2728 1809 449

Kidney 120472 115958 3191 85090 14235 9786 1343

Brain 121245 117412 3742 75214 18354 11465 4646

Heart 133028 116512 3830 88784 15546 10092 2868

Total 884235 560242 3442 416137 33044 20486 ndash

756

Junqueira-de-Azevedo et al doi101093molbevmsu337 MBE

used for distinct purposes (internal and external functionsrespectively)

The pancreatic transcriptome revealed a core set of highlyexpressed proteinases related to digestion The mammalianpancreas typically secretes three forms of trypsin six of chy-motrypsin and three carboxypeptidases All these majorfamily members were identified in the B jararaca pancreas(fig 2) This is the first report of the whole set of pancreaticenzymes from a snake and most of those identified hererepresent full-length sequences A serine protease inhibitorof kazal-type containing a single kazal repeat was also foundamong the most expressed genes for pancreatic polypeptidesThis protein may function to prevent the action of the diges-tive enzymes described above during the long starvation pe-riods that snakes are often subjected to We also identifiedtwo lipases which are major components of pancreaticsecretions in our data set These genes correspond to a bilesalt-activated lipase and a triacylglycerol lipase (fig 2) Inorganisms that feed on relatively large prey such as manysnakes digestive enzymes are critical and therefore have anexpected high activity thus becoming an interesting modelfor enzymatic studies (Secor and Diamond 1998 Castoe et al2013) Moreover the putative production of high amounts ofa complete set of digestive enzymes in the pancreas of Bjararaca suggests that these secretions have the capacity topromote prey digestion These data therefore support thecurrent hypothesis that venom is used primarily for prey cap-ture and has no apparent effect on the ldquopredigestionrdquo of prey(McCue 2007) despite many venom enzymes being capableof promoting tissue digestion (Bottrall et al 2010)

The most expressed unigenes in the pancreas also revealedsome unexpected pancreatic proteins (supplementary table

S2 Supplementary Material online) such as natterin Natterinis the major toxin of some venomous fishes and possesses aproteolytic kininogenase activity (Magalh~aes et al 2005)Interestingly natterin-related genes have an uneven distribu-tion across vertebrate taxa having been retrieved from thegenomes of reptiles birds and other fishes but seeminglyabsent from amphibians or mammalsmdashpresumably as theresult of independent gene loss during the evolution of tet-rapod lineages Nevertheless our identification of its selectiveexpression in the snake pancreas combined with its previ-ously determined proteolytic activity in fish venom suggests adigestive function of the ancestral natterin protein

Nontoxin Genes and Pathways Enhanced in the VGGenes with a potential role in the VG machinery were iden-tified by comparing unigene expression levels in the VG withother tissues The scatter plot displayed in figure 3 demon-strates that toxin genes (marked in red) have a strong prev-alence in the VG as expected However some nontoxinunigenes were found to have a considerably higher level ofexpression in the VG than on average in the other tissues(marked in green) with some exclusively expressed in the VG(marked in yellow and found on the x axis) Details of thesespecific unigenes are provided in supplementary tables S3 andS4 Supplementary Material online respectively In addition tocomparing gene expression we analyzed whether specific GOcategories are enriched in the subset of those nontoxingenes that we found unregulated in the VG when comparedwith the general occurrence of genes in all other tissues(fig 4)

The functions of the majority of the most differentiallyexpressed unigenes identified in the VG relate to processes

FIG 2 Transcriptomic profile of the pancreas of Bothrops jararaca (A) Relative expression indicated by the percentage of total expression values(RPKM) of all transcripts (upper panel) and those encoding digestive enzymes (lower panel) (B) Table describing the transcripts involved in digestionidentified in the transcriptome

757

Snake Venom-Related Transcripts doi101093molbevmsu337 MBE

involved in venom biosynthesis and pathways relevant toprotein secretion The identity of many of these enrichedgenes suggests roles relating to protein processing particularlypeptidases and modifying enzymes that are likely to be im-portant for the production of venom toxins Chaperones arealso highly represented in the VG as well as proteins involvedin disulfide bond formation Although few individual uni-genes were associated with glycosylation pathways thegene enrichment test highlighted the apparent importanceof this pathway for venom biosynthesis Vesicular traffic animportant process for a specialized secretory tissue was alsodetected as one of the most enriched processes for the VGdifferentially expressed unigenes

A more detailed description and discussion of the differ-entially expressed genes and their biological pathways in theVG are provided in the supplementary discussionSupplementary Material online In addition to providing ev-idences of the processing mechanisms that act to controlvenom production our profile of VG enhanced unigenes de-scribes from a molecular point of view the specialization ofthis tissue to produce a suite of stable proteins that are highlysecreted structurally diverse and hydrolytic in natureUnderstanding the molecular basis of this high-productionneuronally controlled secretory system that has no obviousparallel in other vertebrates therefore provides a novel insightinto the cellular regulation of glandular systems

Reserpine Treatment Does Not Impair Toxin ProductionThe release of venom from the snake VG during biting ormanual extraction leads to morphological changes in thevenom secretory cells and starts a new cycle of venom pro-duction (Kochva 1987) The sympathetic nervous system hasbeen shown to play an important role in the regulation of thisprocess in B jararaca Both alpha and beta adrenoceptorshave been implicated in venom production as the

stimulation of both receptors triggers a complex intracellularsignaling that culminates in the activation of NFkappaB andAP-1 transcription factors and regulation of the synthesis ofproteins in the VG (Yamanouye et al 1997 2000 Kerchoveet al 2004 2008 Luna et al 2009) The importance of thesympathetic system was demonstrated by treating snakeswith reserpine an alkaloid that depletes catecholaminefrom the peripheral sympathetic nerve After treatment itwas shown that 1) no venom could be collected by milkingand 2) the secretory cells of the VG remain in a quiescent-likestage with narrow rough endoplasmic reticulum cisternaeand no secretory vesicles indicative of little or no venomproduction (Yamanouye et al 1997) However reserpinetreatment does not seem to irreversibly damage the VGunlike for example rat salivary glands (Martinez et al 1975Johnson 1988) as adrenoceptor stimulation reverses its effect(Yamanouye et al 1997)

We therefore investigated the effect of reserpine treatmenton mRNA expression in the VG of B jararaca According tothe established model of the VG secretory process we wouldexpect this treatment to impair toxin production or alterna-tively block the secretory pathway When comparing the VGtranscriptomes of reserpine treated and untreated individualsour data showed a surprisingly high level of similarity In termsof the quantitative composition of the transcriptome toxin-related transcripts amounted to 58 of the total transcripts(fig 5A)mdashhighly comparable to the 53 obtained in theuntreated VG transcriptome and qualitatively all major clas-ses of toxins were detected in similar amounts in comparisonto the untreated specimen We confirmed these results byquantitative polymerase chain reaction (qPCR) analysis ofrepresentative toxin mRNAs which demonstrated thattoxins are not downregulated in reserpine-treated specimensIn fact in some cases the treated animals presented with ahigher level of toxin expression (fig 5B) These results suggestthat noradrenergic innervation does not influence thetranscription of toxin genes In agreement with this findingour recent study of the VG proteome showed that severaltoxin types are present in the secretory cells of the quiescentVG of B jararaca and some of them were present only in thisstage (Luna et al 2013) Taken together these data suggestthat venom production may be more dependent on a fineregulation of other cellular processes than solely on generalprotein synthesis Further investigation of specific genes re-lated to these pathways using more biological replicateswould aid the future identification of the mechanismsinvolved

The Evolution of Snake Venom Toxins and Their Nontoxin

ParalogsTo investigate the evolutionary history of snake venom toxinswe constructed phylogenetic trees using our toxin andnontoxin paralog data for the following gene families brady-kinin-potentiating peptide (BPP) cobra venom factor C-typelectin (CTL) cysteine-rich secretory protein (CRISP) cystatinhyaluronidase LAAO phospholipase A2 (PLA2) snake venomnerve growth factor (NGF) snake venom metalloproteinase(SVMP) snake venom serine proteinase (SVSP) snake venom

FIG 3 Scatter plot showing the relationship of gene expression values(normalized RPKM Log10) in the VG (VGLA) and the average of allother tissues (TISSUES) Red spots represent toxin annotated genesGreen spots represent nontoxin genes that have a 20 fold changein expression level andor are highly expressed in the VG Yellow spotsrepresent the top ten most expressed nontoxin genes with selectiveexpression in the VG Blue spots represent the remaining annotatedgenes

758

Junqueira-de-Azevedo et al doi101093molbevmsu337 MBE

vascular endothelial growth factor (VEGF) and three-fingertoxins Venom toxins were defined as those found placedwithin clades containing previously described snake venomtoxins (highlighted in fig 6 and supplementary figs S1ndashS12Supplementary Material online) Only one B jararaca genethat exhibited its greatest abundance of expression in the VGwas found placed outside these previously defined snakevenom toxin clades (cystatin gene BJARALL22588mdashsupple-mentary fig S5 Supplementary Material online) For the vastmajority of other toxin types we found VG expressed genesgrouping phylogenetically with known snake venom toxinsNotably we recovered multiple toxin isoforms for many ofthese gene families (eg BPP CTL L-amino acid oxidase[LAO] PLA2 SVSP and SVMP) indicative of duplicationevents giving rise to new gene copies following their recruit-ment for expression in the VG of snakes as previously de-scribed (eg Fry et al 2003 Casewell et al 2011 Vonk et al2013) For other toxin families such as CRISP hyaluronidaseand NGF we only identified a single B jararaca gene nestedwithin toxin clades

Recently Hargreaves Swain Hegarty et al (2014) suggestedthat the location of expression of snake toxins has been re-stricted to the VG following their recruitment to this tissuetype However an absence of data quantifying gene expres-sion in different tissues limits our interpretation of this

seemingly rational hypothesis particularly as the pathogenicviper toxin families analyzed by the authors do not appear tobe selectively expressed in the VG of the venomous snakesampled (Hargreaves Swain Hegarty et al 2014) We there-fore investigated the location of expression of B jararacatoxin genes using our transcriptomic data sets Our resultsdemonstrate that the majority of toxins are indeed selectivelyexpressed in the VG and typically at high levels (fig 7Amdashnotethe logarithmic scale) Although these results are novel thisobservation was expected and corresponds with previoustheories relating to venom evolution (eg Casewell et al2012 Fry et al 2012 Hargreaves Swain Hegarty et al 2014Reyes-Velasco et al 2014) as the expression of toxic moleculesin internal tissues would be expected to be deleterious to theproducing animal

However we were surprised to find that approximately20 of B jararaca toxin unigenes are actually coexpressedin at least one other tissue type (fig 7A and B) For examplemultiple BPP CTL and SVMP toxins were coexpressed in thepancreas (supplementary figs S1 and S3 SupplementaryMaterial online and fig 6) some SVMP genes were alsofound coexpressed in the kidney (fig 6) whereas NGF ex-hibited expression in the VG liver and heart (supplementaryfig S8 Supplementary Material online) These results suggestthat the premise that snake venom toxins are only expressed

FIG 4 Histogram showing the GO biological process categories enriched in the VG Gene Enrichment analysis was performed by BLAST2Go software bycomparing the number of unigenes annotated with a given GO category in a test set (VG differentially expressed unigenes) against the number of thosegenes in a reference set (all annotated unigenes) and selecting enriched categories by statistical significance (P 005) using Fisherrsquos exact test

759

Snake Venom-Related Transcripts doi101093molbevmsu337 MBE

in the VG is not unequivocal and correlates with findingsfrom another venomous vertebrate the platypus(Whittington and Belov 2009) However in all cases observedhere the level of expression detected in non-VG tissues wassignificantly lower (typically 4 50 times) than expression inthe VG (fig 7D) It therefore seems likely that on the occasionswhere toxins are coexpressed perhaps through leaky expres-sion (Ramskeuroold et al 2009) their expression level is sufficientlylimited that any resulting proteins are tolerated We shouldnote however that we used 454 data in this study and itproduces limited amounts of reads compared with othertechnologies such as Illumina It means that the level of ex-pression of low expressed genes is not precise and could beunderestimated

Our observations of toxin coexpression in non-VG tissuesmay partially explain the unusual phylogenetic placement ofsome toxin orthologs previously detected in non-VG tissuesof Thamnophis elegans and Python bivittatus (Casewell et al2012) This process termed ldquoreverse recruitmentrdquo suggestedthat some genes recruited early in venomous snake ancestorsfor expression in the VG may subsequently have evolvedprofiles for expression in nonvenom tissues whereas the ex-pression location of the other descendants remained re-stricted to the VG A recent study dismissed thismechanism of gene evolution (Hargreaves Swain Loganet al 2014) as ldquoless parsimoniousrdquo than a situation where

the gene of interest has always been expressed in non-VGtissues However this interpretation ignores the shared ances-try of the reverse recruited genes with toxin orthologs foundexpressed in the VGs of other species Utilizing the B jararacatissue transcriptomes we identified the strongest case for re-verse recruitment identified to date where an SVMP genenested within the snake toxin clade was found to be exclu-sively expressed in the pancreas (fig 6) The placement of thisgene within the venom toxin clade where it groups withelapid SVMPs (sharing 80 similarity) at the base of thevenom toxin radiation was validated by Bayes factor analysis(supplementary table S5 Supplementary Material online)Furthermore stochastic mutational mapping character anal-ysis robustly supported (Bayesian Posterior Probability = 099)the premise that this gene is derived from venom toxins(fig 6)mdashtherefore suggesting that the location of gene ex-pression has changed from the VG to the pancreas Furtherdiscussion of other non-VG expressed B jararaca genes form-ing monophyletic groups with venom toxins can be found inthe supplementary discussion Supplementary Materialonline

Our phylogenetic analyses also provided information re-garding the location of expression of nontoxin paralogs re-lated to toxin genes We identified these genes in 8 of the 13toxin families analyzed (fig 7C) No discernible pattern oftissue expression was observed for these paralogous genes

FIG 5 Transcriptomic profile of the VGs of Bothrops jararaca specimens treated with reserpine demonstrating the high expression of toxin encodinggenes (A) Relative expression indicated by the percentage of total expression values (RPKM) of all transcripts (B) qPCR expression levels of selectedtoxin genes in treated and untreated individuals (N = 3 Plt 005 Plt 001)

760

Junqueira-de-Azevedo et al doi101093molbevmsu337 MBE

with some selectively expressed in individual tissues (SVSPhyaluronidase) whereas others were found to be coexpressedin two (SVMP VEGF) three (cobra venom factor) or five(3FTX cystatin) different tissue types (fig 7B) Notably theexpression levels of all of these paralogs were low irrespective

of their tissue location and significantly lower than the ex-pression level of venom toxins in the VG (Plt 0001 two-tailed t-test) The sole exception to this were the nearestrecovered paralogs to the SVSPs which were representedby four trypsin genes found abundantly expressed in the

FIG 6 Bayesian DNA gene tree of the SVMP toxin family Bothrops jararaca sequences are labeled in bold with the tissue location and expression levelannotated on the tip label The tips of the tree are colored to indicate sequences sourced from the VG (red) and other snake tissues (blue) Pie chartsrepresent the Bayesian posterior probabilities (Bayesian pp) of ancestral state reconstructions at that node Red venom blue nonvenom Asteriskshighlight significant support (Bayesian pp 095) for a character state at ancestral nodes Circles placed at internal tree nodes indicate the Bayesian ppfor that node Black = 100 gray 095

761

Snake Venom-Related Transcripts doi101093molbevmsu337 MBE

pancreas of B jararaca at comparative expression levels totheir VG toxin counterparts (fig 7E) In general howevernontoxin paralogs were found coexpressed in a highernumber of tissues than venom toxins (fig 7B) correlatingwith recent reports that python nontoxin gene orthologsare often coexpressed in multiple tissues at low levels(Reyes-Velasco et al 2014)

We harnessed this recently published information onvenom toxin orthologs recovered from the Burmese python(Reyes-Velasco et al 2014) to compare changes in the loca-tion of gene expression in venom toxin gene families We usedB jararaca data to describe 1) the location of expression oftoxin genes and nontoxin paralogs in the liver kidney pan-creas brain heart and VG and 2) the location of expression ofthe nontoxin orthologs identified from the python in thesesame tissue types Comparisons of these data provide a com-plicated picture of change in the location of expression ofgenes associated with toxin families (fig 8) For example someappear to show a transition from wide levels of coexpressionin paralogous genes to lesser coexpression in python ortho-logs to abundant expression in the VG (eg CTL) whereasothers show more selective expression of nontoxin genes infewer tissues types and an apparent switch to selective

expression in the VG (eg hyaluronidase PLA2) It is clearthat changes in the location of expression of these gene fam-ilies have occurred over evolutionary time and as suggestedby the recently proposed SINNER model of venom evolution(Reyes-Velasco et al 2014) the evolution of venom toxinsappears to have involved the switching of low level physio-logical expression to high levels of tissue-specific expression inthe VG followed by a reduction or perhaps little change inexpression levels in nonvenom-related tissues

ConclusionsUtilizing comparative data sourced from non-VG tissuesgreatly improves our understanding of the evolution andproduction of snake venom For example our analyses ofdifferentially expressed nontoxin genes detected in the VGof B jararaca provide us novel information relating to thebiological processes that are of critical importance for theproduction of venom Notably we find clear distinctionsbetween the genes and pathways enriched in the VG incomparison with other snake secretory tissues such as thepancreas This is perhaps unsurprising as the evolution of adistinct secretory gland that functions to produce highlybioactive and toxic proteins for injection into other animals

FIG 7 Expression analysis of snake venom toxins and their related nontoxin paralogs (A) The location and abundance of expression (normalized RPKMLog10) of venom toxins in each of the tissue transcriptomes Note the logarithmic scale (B) A comparison of the percentage of venom toxin genes andtheir nontoxin paralogs found coexpressed in the tissue transcriptomes The percentage of genes is defined as the percentage of all toxins and allnontoxins respectively (C) The location and abundance of expression of nontoxin genes that are paralogous to venom toxins as determined byphylogenetic analysis (D) Comparisons of the mean level of venom toxin gene expression in the VG and when coexpressed in one or two other tissuetypes (non-VG) (E) Comparisons of the mean level of nontoxin paralog gene expression when found selectively expressed in one tissue type orcoexpression in multiple tissues The selective expression of SVSP nontoxin paralogs in the pancreas has been separated from the ldquoonerdquo tissue bar suchas not to skew the data

762

Junqueira-de-Azevedo et al doi101093molbevmsu337 MBE

represents an atypical anatomical occurrence These datareinforce our belief that venom systems have a number ofunique characteristics (eg high production of toxic pro-teins restriction of abundant levels of toxins to the VGavoidance of self-toxicity and intense use of posttransla-tional modifications) Our results also provide a rationalbasis to holistically investigate how venom systems (ienot only the snake venom toxins) have evolved As compar-ative gene data relating to other animal venom systems (egspiders scorpions etc) become available it will be particu-larly interesting to identify whether the specific peptidasesmodifying enzymes and chaperones involved in snake toxinproduction are shared with other animal lineages therebymirroring previous reports of toxin convergence across di-verse venomous taxa (Fry et al 2009 Casewell et al 2013)

Our analyses of gene family evolution revealed that manytoxin families have evolved by a process of gene duplication

with nontoxin paralogs typically being coexpressed in non-VG tissues at low levels By contrast the majority of venomtoxins are found selectively expressed in the VG with someexhibiting leaky low-level expression in other body tissuesdemonstrating that changes in the location and abundanceof gene expression have occurred during their co-option tovenom This process may be controlled by the up or downregulation of gene transcription in different tissues andorthrough the loss or gain of core promoters influencing genetranscription in different tissues types Critically identifyingthe specific mechanisms that underpin this process remainsone of the major barriers limiting our understanding of howanimal venom systems have evolved It is apparent from ouranalyses here that additional non-VG data from otherToxicoferan reptile taxa are required to robustly reconstructthe evolutionary history of these gene families in the contextof 1) changes in their location of expression and 2) the gene

FIG 8 Reconstructing changes in the location of expression of venom toxin genes and their orthologs and paralogs For each viper toxin family thelocations of expression of nontoxin paralogs (identified here from Bothrops jararaca) nontoxin orthologs (identified from the nonvenomous snakePython bivittatus [Reyes-Velasco et al 2014]) and the toxins themselves (identified here from B jararaca) are displayed Oral gland refers to the VG forB jararaca and the rictal gland for P bivittatus Colors indicate presenceabsence and level of expression Blue not expressed yellow low level expressionred high level expression NS indicates that pancreatic data were not sampled for P bivittatus Not identified indicates toxin families where we failed todetect nontoxin paralogs in our B jararaca transcriptomes

763

Snake Venom-Related Transcripts doi101093molbevmsu337 MBE

duplication events that are responsible for generating thediversity of toxin-related genes observed in venomous snakes

Materials and Methods

Specimens

One adult specimen of B jararaca from S~ao Paulo State wasobtained by the Herpetology Laboratory Instituto ButantanS~ao Paulo Brazil and used for body organ dissection (exceptfor brain which was removed from a second animal) Firstvenom was extracted and 4 days later the animal wasanesthetized with sodium pentobarbital (30 mgkg of bodymass sc) The organs used for tissues sampling in this study(VG liver pancreas brain heart and kidney) were carefullydissected frozen in liquid nitrogen and stored at80C untiluse

For the reserpine treatment experiment the same proce-dure was adopted except that three specimens were injectedwith reserpine (20 mgkg of body mass sc) to deplete en-dogenous catecholamine stores and 24 h later venom wasextracted The animals received additional doses of reserpine(5 mgkg) on each of the following 4 days before they wereanesthetized for VG dissection

mRNA Extraction and 454-Pyrosequencing

For total RNA isolation the tissues were ground into apowder in liquid nitrogen and homogenized using aPolytron Tissue Homogenizer Total RNA was extractedwith TRIZOL Reagent (Invitrogen Life Technologies Corp)and mRNA was prepared using the Dynabeads mRNADIRECT kit (Invitrogen Life Technologies Corp) mRNAwas quantified by the Quant-iT RiboGreen RNA reagentand kit (Invitrogen Life Technologies Corp) A cDNA libraryfrom each tissue was constructed using 500 ng of mRNA andthe cDNA Synthesis System (Roche Diagnostics) EmulsionPCR amplification and library sequencing were performedindividually using a GS Junior 454 Sequencing System(Roche Diagnostics) for liver brain kidney and heart VGand pancreas libraries were run on a GS FLX instrumentusing MID indexing on different sections of a PicoTiterPlate according to the manufacturerrsquos protocols (RocheDiagnostics)

Bioinformatic AnalysesRaw sequences from each tissue were first assembled withNewbler 27 (Roche Diagnostics) The Newbler assembler firstremoved adaptors and contaminating rRNA sequences Theassembly parameters were set to 1) a minimum overlaplength of 50 of the read and 2) a minimum overlap identityof 98 with all other parameters set as default Newbler isotigconsensus sequences from each tissue were combined usingCap3 software (Huang and Madan 1999) and the De Novoassembly tool of CLC Genomics Workbench software(Qiagen) to join contigs expressed in more than one tissueThe final unigene data set included consensus sequences ofcombined contigs from two or more tissues (namedBJARALL) and isotigs exclusively occurring in single tissues(BJAR[tissue])

Unigenes were automatically annotated using BLAST2Go(Conesa et al 2005) by performing a Basic Local AlignmentSearch Tool (BLAST) search against the UniProt database withthe algorithm BLASTX to identify similar sequences Thosewithout annotations were subsequently searched against theGenBank nr database using BLASTN GO categories were at-tributed by BLAST2Go and used for downstream analysissuch as gene enrichment Toxin categories were manuallyassigned by comparing the unigenes to a compiled list ofknown snake toxins Final manual curation of relevant isotigsequences was undertaken to improve the quality and exten-sion of the informatically assembled cDNAs

Expression values of the unigenes identified in the differenttissues were calculated using the RNA-Seq function of CLCGenomics Workbench software by mapping cleaned reads(without known contaminants and rRNAs) back to the uni-genes and normalizing by unigene length using RPKM (readsper kilobase per million of mapped reads)

Quantitative PCRRNA was isolated from the VGs of reserpine-treated anduntreated animals and 1mg of total RNA was reverse tran-scribed using SuperScriptTM III First Strand Synthesis Systemfor RT-PCR (Invitrogen) according to the manufacturerrsquos in-structions Quantitative PCR experiments were performed inStepOnePlusTM (Applied Biosystems) using PlatinumSYBRGreen qPCR SuperMix-UDG (Invitrogen) and SYBR GreenPCR Master Mix (Applied Biosystems) For normalizationCt values were calculated using the formulaCt = (Cttarget Ctcontrol) where control correspondsto the level of GAPDH (glyceraldehyde 3-phosphate dehydro-genase) transcript Fold differences in normalized gene expres-sion were calculated by dividing the level of expression of thereserpine-treated with the untreated sample The resultsgenerated from three replicates were corrected toreduce the weight of individual variations in the statisticalanalysis (Willems et al 2008) Significance was assessedusing a two-tailed Studentrsquos t-test and a P-value cutoffof 005

Evolutionary AnalysesBothrops jararaca toxin sequences were incorporated intogene family sequence alignments containing toxin andnontoxin gene homologs and paralogs sourced from othersnakes and representative outgroup taxa Sequences wereobtained from our previous work (Casewell et al 2012Vonk et al 2013) and supplemented by BLAST searchingthe GenBank nucleotide database DNA data sets weretrimmed to the open reading frame in MEGA6 (Tamuraet al 2013) with identical sequences and those containingtruncations or frameshifts (as the result of indels) excludedbefore they were aligned using MUSCLE (Edgar 2004) andchecked manually

For phylogenetic analysis we used MrBayes v32 (Ronquistet al 2012) to generate DNA gene trees for each toxin familyWe first selected models of sequence evolution for eachcodon using MrModelTest v23 (httpswwwabcse~nylandermrmodeltest2mrmodeltest2html last accessedDecember 15 2014) to determine the model favored by the

764

Junqueira-de-Azevedo et al doi101093molbevmsu337 MBE

Akaike Information Criterion (supplementary table S6Supplementary Material online) We implemented thesemodels in Bayesian Inference analyses using MrBayes onthe CIPRES Science Gateway (wwwphyloorg last accessedDecember 15 2014) Each data set was run in duplicateusing four chains simultaneously (three heated and onecold) for 1 107 generations sampling every 500th cyclefrom the chain and using default settings in regards topriors Tracer v15 (httpbeastbioedacukTracer lastaccessed December 15 2014) was used to estimate effectivesample sizes for all parameters (with a minimum of 200deemed acceptable) and to construct plots of ln(L) againstgeneration to verify the point of convergence (burnin) withall trees generated prior to this point (the first 25) discardedWe then annotated the resulting consensus gene trees withinformation relating to the location and level of expression ofthe B jararaca sequences identified in our different tissuetranscriptomes

To test alternative hypotheses relating to the phylogeneticplacement of specific B jararaca sequences of interest withinthe gene trees we used Bayes factors (Kass and Raftery 1995)and stochastic mutational mapping character analyses(Huelsenbeck et al 2003 Bollback 2006) These analyseswere performed as previously described (Casewell et al 2012)

Supplementary MaterialSupplementary discussion figures S1ndashS13 and tables S1ndashS6are available at Molecular Biology and Evolution online (httpwwwmbeoxfordjournalsorg)

Acknowledgments

This work was supported by grants from Fundac~ao deAmparo a Pesquisa do Estado de S~ao Paulo (FAPESP grantnumbers 1200177-5 201307467-1) and Conselho Nacionalde Desenvolvimento Cientıfico e Tecnologico (CNPq) toILMJ NRC acknowledges support from the NaturalEnvironment Research Council United Kingdom(Fellowship NEJ0186781) The raw data generated in thisproject were deposited in GenBank BioProject section underaccession code PRJNA262899

ReferencesBollback JP 2006 SIMMAP stochastic character mapping of discrete

traits on phylogenies BMC Bioinformatics 788Bottrall JL Madaras F Biven CD Venning MG Mirtschin PJ 2010

Proteolytic activity of Elapid and Viperid Snake venoms and itsimplication to digestion J Venom Res 118ndash28

Brust A Sunagar K Undheim EAB Vetter I Yang DC Casewell NRJackson TNW Koludarov I Alewood PF Hodgson WC et al 2013Differential evolution and neofunctionalization of snake venommetalloprotease domains Mol Cell Proteomics 12651ndash663

Casewell NR Harrison RA Weurouster W Wagstaff SC 2009 Comparativevenom gland transcriptome surveys of the saw-scaled vipers(Viperidae Echis) reveal substantial intra-family gene diversity andnovel venom transcripts BMC Genomics 10564

Casewell NR Huttley GA Weurouster W 2012 Dynamic evolution of venomproteins in squamate reptiles Nat Commun 31066

Casewell NR Wagstaff SC Harrison RA Renjifo C Weurouster W 2011Domain loss facilitates accelerated evolution and neofunctionaliza-tion of duplicate snake venom metalloproteinase toxin genesMol Biol Evol 282637ndash2649

Casewell NR Wagstaff SC Weurouster W Cook DAN Bolton FMS King SIPla D Sanz L Calvete JJ Harrison RA 2014 Medically importantdifferences in snake venom composition are dictated by distinctpostgenomic mechanisms Proc Natl Acad Sci U S A 1119205ndash9210

Casewell NR Weurouster W Vonk FJ Harrison RA Fry BG 2013 Complexcocktails the evolutionary novelty of venoms Trends Ecol Evol 28219ndash229

Castoe T de Koning APJ Hall K Yokoyama K Gu W Smith E FeschotteC Uetz P Ray D Dobry J et al 2011 Sequencing the genome of theBurmese python (Python molurus bivittatus) as a model for studyingextreme adaptations in snakes Genome Biol 12406

Castoe TA de Koning APJ Hall KT Card DC Schield DR Fujita MKRuggiero RP Degner JF Daza JM Gu W et al 2013 The Burmesepython genome reveals the molecular basis for extreme adaptationin snakes Proc Natl Acad Sci U S A 11020645ndash20650

Ching ATC Rocha MMT Paes Leme AF Pimenta DC de Fatima DFurtado M Serrano SMT Ho PL Junqueira-de-Azevedo ILM 2006Some aspects of the venom proteome of the Colubridae snakePhilodryas olfersii revealed from a Duvernoyrsquos (venom) gland tran-scriptome FEBS Lett 5804417ndash4422

Conesa A Geurootz S Garcıa-Gomez JM Terol J Talon M Robles M 2005Blast2GO a universal tool for annotation visualization and analysisin functional genomics research Bioinformatics 213674ndash3676

De Lucca FL Imaizumi MT 1972 Synthesis of ribonucleic acid in thevenom gland of Crotalus durissus terrificus (Ophidia Reptilia) aftermanual extraction of tne venom Biochem J 130335ndash342

Durban J Juarez P Angulo Y Lomonte B Flores-Diaz M Alape-Giron ASasa M Sanz L Gutierrez JM Dopazo J et al 2011 Profiling thevenom gland transcriptomes of Costa Rican snakes by 454 pyrose-quencing BMC Genomics 12259

Edgar RC 2004 MUSCLE multiple sequence alignment with high accu-racy and high throughput Nucleic Acids Res 321792ndash1797

Fry BG 2005 From genome to ldquovenomerdquo molecular origin and evolu-tion of the snake venom proteome inferred from phylogenetic anal-ysis of toxin sequences and related body proteins Genome Res 15403ndash420

Fry BG Casewell NR Weurouster W Vidal N Young B Jackson TNW 2012The structural and functional diversification of the Toxicofera reptilevenom system Toxicon 60434ndash448

Fry BG Roelants K Champagne DE Scheib H Tyndall JDA King GFNevalainen TJ Norman JA Lewis RJ Norton RS et al 2009 Thetoxicogenomic multiverse convergent recruitment of proteins intoanimal venoms Annu Rev Genomics Hum Genet 10483ndash511

Fry BG Vidal N Norman JA Vonk FJ Scheib H Ramjan SFR Kuruppu SFung K Hedges SB Richardson MK et al 2006 Early evolution of thevenom system in lizards and snakes Nature 439584ndash588

Fry BG Weurouster W Kini RM Brusic V Khan A Venkataraman D RooneyAP 2003 Molecular evolution and phylogeny of elapid snake venomthree-finger toxins J Mol Evol 57110ndash129

Hargreaves AD Swain MT Hegarty MJ Logan DW Mulley JF 2014Restriction and recruitment-gene duplication and the origin andevolution of snake venom toxins Genome Biol Evol 62088ndash2095

Hargreaves AD Swain MT Logan DW Mulley JF 2014 Testing theToxicofera comparative transcriptomics casts doubt on the singleearly evolution of the reptile venom system Toxicon 92140ndash156

Huang X Madan A 1999 CAP3 a DNA sequence assembly programGenome Res 9868ndash877

Huelsenbeck JP Nielsen R Bollback JP 2003 Stochastic mapping ofmorphological characters Syst Biol 52131ndash158

Ivanov CP Ivanov OC 1979 The evolution and ancestors of toxic pro-teins Toxicon 17205ndash220

Johnson DA 1988 Changes in rat parotid saliva protein compositionfollowing chronic reserpine treatment and their relation to inani-tion Arch Oral Biol 33463ndash466

Junqueira-de-Azevedo IDLM Ho PL 2002 A survey of gene expressionand diversity in the venom glands of the pitviper snake Bothropsinsularis through the generation of expressed sequence tags (ESTs)Gene 299279ndash291

Kass RE Raftery A 1995 Bayes factors J Am Stat Assoc 90773ndash795

765

Snake Venom-Related Transcripts doi101093molbevmsu337 MBE

Kerchove CM Carneiro SM Markus RP Yamanouye N 2004Stimulation of the alpha-adrenoceptor triggers the venom produc-tion cycle in the venom gland of Bothrops jararaca J Exp Biol 207411ndash416

Kerchove CM Luna MS Zablith MB Lazari MF Smaili SS Yamanouye N2008 1-adrenoceptors trigger the snake venom production cyclein secretory cells by activating phosphatidylinositol 45-bisphosphatehydrolysis and ERK signaling pathway Comp Biochem Physiol Part A150431ndash437

Kochva E 1987 The origin of snakes and evolution of the venom ap-paratus Toxicon 2565ndash106

Le~ao LI Ho PL Junqueira-de-Azevedo IDLM 2009 Transcriptomic basisfor an antiserum against Micrurus corallinus (coral snake) venomBMC Genomics 10112

Luna MS Valente RH Perales J Vieira ML Yamanouye N 2013Activation of Bothrops jararaca snake venom gland and venomproduction a proteomic approach J Proteomics 94460ndash472

Luna MSA Hortencio TMA Ferreira ZS Yamanouye N 2009Sympathetic outflow activates the venom gland of the snakeBothrops jararaca by regulating the activation of transcription fac-tors and the synthesis of venom gland proteins J Exp Biol 2121535ndash1543

Lynch VJ 2007 Inventing an arsenal adaptive evolution and neofunc-tionalization of snake venom phospholipase A2 genes BMC EvolBiol 72

Magalh~aes GS Lopes-Ferreira M Junqueira-de-Azevedo ILM Spencer PJAraujo MS Portaro FCV Ma L Valente RH Juliano L Fox JW et al2005 Natterins a new class of proteins with kininogenase activitycharacterized from Thalassophryne nattereri fish venom Biochimie87687ndash699

Martinez JR Adelstein E Quissel D Barbero GJ 1975 The chronicallyreserpinized rat as a possible model for cystic fibrosis I Submaxillarygland morphology and ultrastructure Pediatr Res 9463ndash469

McCue MD 2007 Prey envenomation does not improve digestive per-formance in western diamondback rattlesnakes (Crotalus atrox) JExp Zool A Ecol Genet Physiol 307568ndash577

Oron U Bdolah A 1973 Regulation of protein synthesis in the venomgland of viperid snakes J Cell Biol 56177ndash190

Pahari S Mackessy SP Kini RM 2007 The venom gland transcriptomeof the Desert Massasauga rattlesnake (Sistrurus catenatus edwardsii)towards an understanding of venom composition among advancedsnakes (Superfamily Colubroidea) BMC Mol Biol 8115

Ramskeuroold D Wang ET Burge CB Sandberg R 2009 An abundance ofubiquitously expressed genes revealed by tissue transcriptome se-quence data PLoS Comput Biol 5e1000598

Reyes-Velasco J Shaney KJ Card DC Andrew A Adams RA Schield DModahl C Casewell NR Mackessy SP Castoe TA 2014 Expression ofvenom homologs in the python suggests a model for venom generecruitment Mol Biol Evol 2173ndash183

Rokyta DR Lemmon AR Margres MJ Aronow K 2012 The venom-gland transcriptome of the eastern diamondback rattlesnake(Crotalus adamanteus) BMC Genomics 13312

Ronquist F Teslenko M van der Mark P Ayres DL Darling A Heuroohna SLarget B Liu L Suchard MA Huelsenbeck JP 2012 MrBayes 32efficient Bayesian phylogenetic inference and model choice acrossa large model space Syst Biol 61539ndash542

Rotenberg D Bamberger ES Kochva E 1971 Studies on ribonucleic acidsynthesis in the venom glands of Vipera palaestinae (OphidiaReptilia) Biochem J 121609ndash612

Schwartz TS Tae H Yang Y Mockaitis K Van Hemert JL Proulx SR ChoiJ-H Bronikowski AM 2010 A garter snake transcriptome pyrose-quencing de novo assembly and sex-specific differences BMCGenomics 11694

Secor SM Diamond J 1998 A vertebrate model of extreme physiologicalregulation Nature 395659ndash662

Tamura K Stecher G Peterson D Filipski A Kumar S 2013 MEGA6molecular evolutionary genetics analysis Version 60 Mol Biol Evol302725ndash2729

Valente RH Guimar~aes PR Junqueira M Neves-Ferreira AGC SoaresMR Chapeaurouge A Trugilho MRO Leon IR Rocha SLG Oliveira-Carvalho AL et al 2009 Bothrops insularis venomics a proteomicanalysis supported by transcriptomic-generated sequence dataJ Proteomics 72241ndash255

von Reumont BM Blanke A Richter S Alvarez F Bleidorn C Jenner RA2014 The first venomous crustacean revealed by transcriptomicsand functional morphology remipede venom glands express aunique toxin cocktail dominated by enzymes and a neurotoxinMol Biol Evol 3148ndash58

Vonk FJ Casewell NR Henkel CV Heimberg AM Jansen HJ McClearyRJR Kerkkamp HME Vos RA Guerreiro I Calvete JJ et al 2013 Theking cobra genome reveals dynamic gene evolution and adaptationin the snake venom system Proc Natl Acad Sci U S A 11020651ndash20656

Whittington CM Belov K 2009 Platypus venom genes expressed innon-venom tissues Aust J Zool 57199

Willems E Leyns L Vandesompele J 2008 Standardization of real-timePCR gene expression data from independent biological replicatesAnal Biochem 379127ndash129

Yamanouye N Britto LR Carneiro SM Markus RP 1997 Control ofvenom production and secretion by sympathetic outflow in thesnake Bothrops jararaca J Exp Biol 2002547ndash2556

Yamanouye N Carneiro SM Scrivano CN Markus RP 2000Characterization of beta-adrenoceptors responsible for venom pro-duction in the venom gland of the snake Bothrops jararaca Life Sci67217ndash226

Zelanis A Andrade-Silva D Rocha MM Furtado MF Serrano SMTJunqueira-de-Azevedo ILM Ho PL 2012 A transcriptomic view ofthe proteome variability of newborn and adult Bothrops jararacasnake venoms PLoS Negl Trop Dis 6e1554

766

Junqueira-de-Azevedo et al doi101093molbevmsu337 MBE