Untangling spider silk evolution with spidroin terminal domains

Abstract

Background: Spidroins are a unique family of large, structural proteins that make up the bulk of spider silk fibers. Due to the highly variable nature of their repetitive sequences, spidroin evolutionary relationships have principally been determined from their non-repetitive carboxy (C)-terminal domains, though they offer limited character data. The few known spidroin amino (N)-terminal domains have been difficult to obtain, but potentially contain critical phylogenetic information for reconstructing the diversification of spider silks. Here we used silk gland expression data (ESTs) from highly divergent species to evaluate the functional significance and phylogenetic utility of spidroin N-terminal domains.

Results: We report 11 additional spidroin N-termini found by sequencing approximately 1,900 silk gland cDNAs from nine spider species that shared a common ancestor > 240 million years ago. In contrast to their hyper-variable repetitive regions, spidroin N-terminal domains have retained striking similarities in sequence identity, predicted secondary structure, and hydrophobicity. Through separate and combined phylogenetic analyses of N-terminal domains and their corresponding C-termini, we find that combined analysis produces the most resolved trees and that N-termini contribute more support and less conflict than the C-termini. These analyses show that paralogs largely group by silk gland type, except for the major ampullate spidroins. Moreover, spidroin structural motifs associated with superior tensile strength arose early in the history of this gene family, whereas a motif conferring greater extensibility convergently evolved in two distantly related paralogs.

Conclusions: A non-repetitive N-terminal domain appears to be a universal attribute of spidroin proteins, likely retained from the origin of spider silk production. Since this time, spidroin N-termini have maintained several features, consistent with this domain playing a key role in silk assembly. Phylogenetic analyses of the conserved N- and C-terminal domains illustrate dramatic radiation of the spidroin gene family, involving extensive duplications, shifts in expression patterns and extreme diversification of repetitive structural sequences that endow spider silks with an unparalleled range of mechanical properties.

Figure 1

Spidroin molecular organization and comparison of domain sequences from two paralogs. A. Schematic of spidroin primary structure showing short, non-repetitive N- and C-terminal domains flanking a long region of tandem sequence repeat modules. B. A comparison of full-length, divergent spidroin paralogs encoding the dragline silk protein MaSp1 from Latrodectus hesperus (L.h.) [Genbank: EF595246] and the egg-case silk protein TuSp1 from Argiope bruennichi (A.b.) [Genbank: AB242144], showing their (1) N-terminal domains, (2) the first and last repeat in each sequence and (3) their C-terminal domains. Dashes are alignment gaps. Note the varying repeat sequence length and composition between L.h. MaSp1 and A.b. TuSp1, in comparison to the high similarity across repeat modules within a spidroin.

Figure 2

Phylogeny of species examined in this study. Tree is a composite of published phylogenies [44-47]. Major lineages and approximate divergence dates estimated by Ayoub and Hayashi [33] are indicated. Asterisks mark species from which we report new N-terminal sequences. Branch lengths are not proportional to time.

Figure 3

Phylogenetic trees from separate analyses of spidroin N- and C-terminal domain sequences. A. N-terminal domain, consensus of 4 most parsimonious trees (MPTs), B. Bayesian consensus tree for N-terminus, C. C-terminal domain, consensus of 8 MPTs, D. Bayesian consensus for C-terminus; A, C: numbers above nodes are bootstrap (BT) values, below nodes decay index, thickened branch supported by ≥ 70 BT in parsimony nucleotide analyses (Additional file 4a, 4c); B, D: numbers above nodes are posterior probability (PP) values, thickened branch supported by ≥ 0.95 PP in Bayesian nucleotide analyses (Additional file 4b, 4d).

Figure 4

Combined parsimony analysis of spidroin N- and C-termini showing consensus and conflict among domains. Matrix above branches shows character support by partition, and summed across tree in top left legend. Matrix columns from left to right: N-terminus, C-terminus, N+C termini; Rows from top to bottom: BS = branch support (decay index), PBS (partitioned branch support, and PHBS (partitioned hidden branch support). Below branches, left of the slash = bootstrap support (* = < 50%), right of the slash = node # referred to in text, thickened branches supported > 70% bootstrap replicates in parsimony nucleotide analysis (Additional file 5). Note that the K.h. MaSp1 C-terminus was coded as missing data, and the N-terminus of D.s. MaSp2 was concatenated with the D.s. MaSp2a C-terminus.

Figure 5

Bayesian consensus tree from spidroin N- and C-termini reconstructing structural motifs and gene duplications. 50% majority-rule consensus of post-burnin Bayesian trees from combined spidroin N+C terminal domains. Numbers above branches indicate PP values; thickened branches supported > 0.95 PP in Bayesian nucleotide analysis (Additional file 5b). Circles at nodes show inferred gene duplication events. Shown below branches are terminal domain amino acid synapomorphies by position (N-terminal positions from 1-168; C-terminal positions from 169-277). Gain of motif along a branch indicated as symbol in legend, and loss by an "X" of the same color. There are other equally parsimonious reconstructions for the evolution of (GA)_n and (GGX)_n motifs to those shown.