Dynamic evolution of venom proteins in squamate reptiles. Nicholas R. Casewell, Gavin A. Huttley and Wolfgang Wüster

Transcription

1 Dynamic evolution of venom proteins in squamate reptiles Nicholas R. Casewell, Gavin A. Huttley and Wolfgang Wüster Supplementary Information Supplementary Figure S1. Phylogeny of the Toxicofera and evolution of the venom system under the single early origin (SEO) hypothesis of Fry et al. [10]. Simplified from Fry et al. [10] and Vidal and Hedges [52]. Key: LJ = lower jaw, UJ = upper jaw, red = venom, blue = non-venom.

2 Supplementary Figure S2. Bayesian DNA gene tree of the crotamine toxin family. Multiple support values are given at key nodes in the following order: Bayesian DNA, maximum likelihood (ML) DNA, Bayesian amino acid (aa), ML aa. x indicates no support for the node in that analysis. Tips of the tree coloured in red indicate Toxicoferan sequences sourced from the venom gland and blue those sourced from non-venom gland tissues ( physiological non-toxins). Pie charts represent the bpp of ancestral state reconstructions at that node: red = venom, blue = non-venom. The numbered codes for each sequence presented in the genetree represent GenBank GI accession numbers.

3 Supplementary Figure S3. Bayesian DNA gene tree of the CVF toxin family. Multiple support values are given at key nodes in the following order: Bayesian DNA, ML DNA, Bayesian aa, ML aa. x indicates no support for the node in that analysis. Tips of the tree coloured in red indicate Toxicoferan sequences sourced from the venom gland and blue those sourced from non-venom gland tissues ( physiological non-toxins). Pie charts represent the bpp of ancestral state reconstructions at that node: red = venom, blue = non-venom. The numbered codes for each sequence presented in the genetree represent GenBank GI accession numbers.

4 Supplementary Figure S4. Bayesian DNA gene tree of the hyaluronidase toxin family. Multiple support values are given at key nodes in the following order: Bayesian DNA, ML DNA, Bayesian aa, ML aa. x indicates no support for the node in that analysis. Tips of the tree coloured in red indicate Toxicoferan sequences sourced from the venom gland and blue those sourced from nonvenom gland tissues ( physiological non-toxins). Pie charts represent the bpp of ancestral state reconstructions at that node: red = venom, blue = non-venom. The numbered codes for each sequence presented in the genetree represent GenBank GI accession numbers.

5 Supplementary Figure S5. Bayesian DNA gene tree of the NGF toxin family. Multiple support values are given at key nodes in the following order: Bayesian DNA, ML DNA, Bayesian aa, ML aa. x indicates no support for the node in that analysis. Tips of the tree coloured in red indicate Toxicoferan sequences sourced from the venom gland and blue those sourced from non-venom gland tissues ( physiological non-toxins). Pie charts represent the bpp of ancestral state reconstructions at that node: red = venom, blue = non-venom. The numbered codes for each sequence presented in the genetree represent GenBank GI accession numbers.

6 Supplementary Figure S6. Bayesian DNA gene tree of the veficolin toxin family. Multiple support values are given at key nodes in the following order: Bayesian DNA, ML DNA, Bayesian aa, ML aa. x indicates no support for the node in that analysis. Tips of the tree coloured in red indicate Toxicoferan sequences sourced from the venom gland and blue those sourced from non-venom gland tissues ( physiological non-toxins). Pie charts represent the bpp of ancestral state reconstructions at that node: red = venom, blue = non-venom. The numbered codes for each sequence presented in the genetree represent GenBank GI accession numbers.

7 Supplementary Table S1. Comparison of Bayes factors generated by Bayesian analysis of codon partitioned and unpartitioned DNA datasets. Codon Unpartitioned: Unpartitioned: Gene family partitioned (H 0 ) mixed model (H A ) model selected (H B ) Bayes factors 2(H 0 H A ) 2(H 0 H B ) Crotamine ** 9.40 * CVF *** 3.22 * Cystatin *** * Hyaluronidase N/A *** - Lectin *** 4.90 * Kallikrein *** * Natriuretic *** 3.84 * NGF N/A *** - Veficolin *** 3.24 * Codon partitions utilised are displayed in Supplementary Table S2. Unpartitioned datasets were analysed with: (i) mixed models of sequence evolution (H A ) and (ii) models selected by MrModelTest v2.3 (H B ) [15]. The GTR + I + Γ models was selected by MrModelTest for each gene family except crotamine where GTR + Γ was selected. The marginal log-likelihoods produced by the codon partitioned datasets (H 0 ) were compared with unpartitioned datasets to generate Bayes factors (2logB10) = 2(H 0 H A or H B ). Interpretation of the differences between Bayes factors is taken from Kass and Raftery [40] - *** very strong, ** strong, * positive, NE little to no evidence. In all gene families Bayes factors advocate the use of codon partitioned datasets implementing multiple models of sequence evolution. N/A indicates where codon partitioned models are identical at each position and also the same as the unpartitioned model.

8 Supplementary Table S2. Estimated models of sequence evolution for DNA and amino acid datasets determined by MrModelTest and ModelGenerator. Data Type Dataset Codon Position Model DNA Crotamine CVF Cystatin Hyaluronidase Kallikrein Lectin Natriuretic NGF Veficolin 1 SYM + Γ 3 HKY + Γ 1 GTR + I + Γ 3 GTR + Γ 1 HKY + Γ 3 HKY + Γ 1 GTR + I + Γ 3 GTR + I + Γ 1 GTR + I + Γ 3 HKY + I + Γ 1 GTR + I + Γ 3 GTR + Γ 1 GTR + Γ 2 GTR + Γ 3 GTR + Γ 1 GTR + I + Γ 3 GTR + I + Γ 1 GTR + Γ 3 GTR + Γ Amino acid Crotamine - WAG + Γ CVF - WAG + Ι + Γ Cystatin - WAG + Γ Hyaluronidase - WAG + Ι + Γ Kallikrein - WAG + Γ Lectin - WAG + Ι + Γ Natriuretic - WAG + Γ NGF - WAG + Ι + Γ Veficolin - WAG + Γ

9 Supplementary Table S3. Test statistics for positive selection analyses undertaken on non-toxin branches observed in the gene trees Gene family Non-toxin branch Test statistic (lnl) Free parameters Null Alt Null Alt LR Significance cut-off P Crotamine CVF FS3E17002I7SX contig79280 N. naja P. bivittatus contig Cystatin contig75753 FT7MHCY04JXDK contig09116 contig contig Hyaluronidase Kallikrein Lectin Natriuretic NGF contig26789 contig77142 P. bivittatus contig02483 contig03054 contig10774 B. jararaca P. bivittatus contig15144 FT7MHCY03HENJ Veficolin contig79448 contig Positive selection tests were calculated for non-toxin branches observed nested within toxin clades of the gene trees. The probability (P) of the test is calculated using likelihood ratio (LR) tests of the 0.207

10 test statistic generated from null and alternative (alt) models and three degrees of freedom. Significance of the test is indicated by bold and underline font where the P-value of the test falls below the sequential Bonferroni correction significance cut-off.

11 Supplementary Table S4. Additional non-venom gland sequences previously identified from snake physiological tissues that were incorporated into the toxin family datasets. Dataset Species Accession number Source of sequence Crotamine CVF Natriuretic Crotalus durissus Naja naja Bothrops jararaca Pseudonaja textilis Liver Liver Brain Heart

12 Supplementary Table S5. DNA sequences excluded from phylogenetic analyses due to significant (P<0.05) evidence of recombination by the RDP, GENECONV or Bootscan methods. Dataset Species Accession number Source of sequence Crotamine Lectin Kallikrein Crotalus durissus Enhydris polylepis Enhydris polylepis Crotalus durissus Liver Venom gland Venom gland Venom gland