Duplicate abalone egg coat proteins bind sperm lysin similarly, but evolve oppositely, consistent with molecular mimicry at fertilization

Abstract

Sperm and egg proteins constitute a remarkable paradigm in evolutionary biology: despite their fundamental role in mediating fertilization (suggesting stasis), some of these molecules are among the most rapidly evolving ones known, and their divergence can lead to reproductive isolation. Because of strong selection to maintain function among interbreeding individuals, interacting fertilization proteins should also exhibit a strong signal of correlated divergence among closely related species. We use evidence of such molecular co-evolution to target biochemical studies of fertilization in North Pacific abalone (Haliotis spp.), a model system of reproductive protein evolution. We test the evolutionary rates (d(N)/d(S)) of abalone sperm lysin and two duplicated egg coat proteins (VERL and VEZP14), and find a signal of co-evolution specific to ZP-N, a putative sperm binding motif previously identified by homology modeling. Positively selected residues in VERL and VEZP14 occur on the same face of the structural model, suggesting a common mode of interaction with sperm lysin. We test this computational prediction biochemically, confirming that the ZP-N motif is sufficient to bind lysin and that the affinities of VERL and VEZP14 are comparable. However, we also find that on phylogenetic lineages where lysin and VERL evolve rapidly, VEZP14 evolves slowly, and vice versa. We describe a model of sexual conflict that can recreate this pattern of anti-correlated evolution by assuming that VEZP14 acts as a VERL mimic, reducing the intensity of sexual conflict and slowing the co-evolution of lysin and VERL.

Figure 1. Evolutionary rates between sperm and egg coat proteins are correlated.

The ratio of non-synonymous to synonymous nucleotide substitutions (d _N/d _S or ω) for sperm acrosomal (lysin) or duplicated egg coat (VERL and VEZP14) proteins were estimated using the branch model of codeml with gene trees following the topology of . Values of ω for each branch of the lysin tree are plotted on the y-axis relative to x-axis values for the corresponding ZP-N motifs from (A) VERL repeats 1 and 2, or (B) VEZP14. Simple linear regression models either weighted (solid lines) or unweighted (dashed lines) by branch length as in were fit to the data , and show a significant positive or negative correlation between lysin and VERL (p = 0.06, weighted; p = 0.02, unweighted) or VEZP14 (p = 0.05, weighted; p = 0.02 unweighted), respectively. Data points with green fill correspond to the branches leading to green abalone used in binding experiments (Figure 2).

Figure 2. ZP-N from VERL and VEZP14 bind lysin with comparable affinity.

The ZP-N motif from VEZP14 and the first repeat of green abalone VERL were cloned, expressed in Drosophila S2 cells (Invitrogen, Carlsbad, CA), and affinity purified (Figure S2). (A) An in-house surface plasmon resonance (SPR) detection system was used to estimate binding of green abalone sperm lysin to surfaces coated with VERL (dark green) or VEZP14 (light green) as measured by refractive index (two replicates each). Numbered shoulders of the SPR curves indicate the intervals used to calculate the association (k_a, 1–2) and dissociation (k_d,2–3) rates via Scrubber2 (Center for Biomolecular Interaction Analysis, University of Utah). (B) Mean dissociation constants calculated from SPR data (K_D = k_d/k_a) show lysin has slightly higher, though comparable, affinity for the ZP-N motif from VERL (520 nM) and VEZP14 (580 nM; lower K_D values indicate higher affinity), demonstrating the ZP-N motif of egg coat proteins alone is sufficient to bind sperm lysin.

Figure 3. Positively selected residues occupy the exposed surface of egg coat proteins.

Positively selected sites of the (A) VERL and (B) VEZP14 ZP-N motifs were inferred among 8 North Pacific abalone taxa using the Bayes Empirical Bayes (BEB) procedure of codeml , and sites with high posterior probabilities (>95%) were mapped to the respective structural models using PyMol . The positions of positively selected sites are given with reference to the complete coding sequences of VEZP14 and VERL (repeat 1 only) from red abalone ( and , respectively). For VERL, the single site under positive selection identified with high posterior probability (V42) was inferred from the first two full VERL repeats, as statistical power from concatenated ZP-N motifs alone is insufficient to allow for predicting sites under selection via BEB (Table 1). This site occupies the same position of the structural model for ZP-N from VEZP14 (S171), with the majority of the remaining seven residues predicted to be under positive selection (N173, R187, E209, I218, L220, K228, and A233) occuring on the same exposed surface opposite the E′ extension (grey fill) thought to facilitate antiparallel pairing among intermolecular ZP-N motifs .

Figure 4. Model of sexual conflict with a female decoy.

(A and B) Hayashi et al's original diploid model of sexual conflict . (A) The mating rate (P) depends on the phenotypic distance between male and female reproductive loci (D_MF), which is determined additively from the allelic values of the male and female loci (f|f′ and m|m′). (B) The number of offspring per female (W_f) decreases from her maximum (B_max) as her mating rate deviates from the optimal female mating rate (P _opt). The scale (s_w) and exponent (x_w) parameters describe the shape of this relationship. Males maximize their reproductive output at P = 1, females at P = P_opt, and thus they are in conflict over optimal mating rate. (C) A model describing the influence of a female decoy on mating rate. A female-expressed decoy locus modifies the mating rate of a male-female pair (P_MF) by Pe, Pi and Pt. The decoy only influences mating rate when its additively determined phenotypic value is between those of the male and female (θ). The decoy's influence on mating rate depends in part on the male-decoy distance. Pe (decoy effectiveness) reduces mating rate to the female's optimal value. Pi (decoy interference) further reduces mating rate when the decoy and female are similar. Pt (the magnitude of female-decoy interference) depends on an optimal male-female distance for decoy function (D _MF = D _opt). The black line shows the effect of the decoy when D_MF = D_opt: as the distance between decoy and male decreases, the decoy causes mating rate to approach the female optima P_opt. Grey lines show how the decoy reduces mating rate (Pt) when the male and female loci are not separated by the optimal distance for decoy function: D _MF = D _opt±0.5. The scale (si, sd) and exponent (xi, xd) parameters control the shape of the Pi and Pt curves as the position of the decoy varies.

Figure 5. Population simulation of sexual conflict with a female decoy.

(A) A representative run of the simulation showing how the decoy influences mating rate and male-female co-evolution. Simulation parameters: N = 10 000, μ = 5.0×10⁻⁵ per locus, each female encounters 20 males per generation. Model parameter values are shown in Figure 4. Evolutionary dynamics: (i) Mating rate is above P _opt, and the male and female are in a co-evolutionary chase. (ii) The decoy becomes effective when it is between the male and female loci. The decoy causes the mating rate to drop below P _opt, and male-female coevolution stops. (iii) The decoy evolves toward the male to reduce interference with the female, and mating rate approaches P _opt. (iv) The decoy is no longer between the male and female loci and is thus ineffective. Mating rate increases above P _opt, and male-female co-evolution resumes. Summary. When the decoy's phenotype is not between those of the male and female, mating rate exceeds P_opt and the male and female co-evolve. When the decoy is between the male and female, it causes the mating rate to drop below P_opt. This stops antagonistic co-evolution, because in this circumstance both males and females benefit from an increased mating rate. When between the male and female, the decoy is driven toward the male locus by selection to reduce interference with the female. Thus, the decoy evolves rapidly when males and females evolve slowly and vice versa. (B) Patterns of co-evolution predicted by the model match those of natural abalone fertilization proteins. Plots of allele substitutions per 500 model generations show correlated male-female evolution and anti-correlated male-decoy evolution. Values are from the run shown in (A).