cifB- transcript levels largely explain cytoplasmic incompatibility variation across divergent Wolbachia

Abstract

Divergent hosts often associate with intracellular microbes that influence their fitness. Maternally transmitted Wolbachia bacteria are the most common of these endosymbionts, due largely to cytoplasmic incompatibility (CI) that kills uninfected embryos fertilized by Wolbachia-infected males. Closely related infections in females rescue CI, providing a relative fitness advantage that drives Wolbachia to high frequencies. One prophage-associated gene (cifA) governs rescue, and two contribute to CI (cifA and cifB), but CI strength ranges from very strong to very weak for unknown reasons. Here, we investigate CI-strength variation and its mechanistic underpinnings in a phylogenetic context across 20 million years (MY) of Wolbachia evolution in Drosophila hosts diverged up to 50 MY. These Wolbachia encode diverse Cif proteins (100% to 7.4% pairwise similarity), and AlphaFold structural analyses suggest that CifB sequence similarities do not predict structural similarities. We demonstrate that cifB-transcript levels in testes explain CI strength across all but two focal systems. Despite phylogenetic discordance among cifs and the bulk of the Wolbachia genome, closely related Wolbachia tend to cause similar CI strengths and transcribe cifB at similar levels. This indicates that other non-cif regions of the Wolbachia genome modulate cif-transcript levels. CI strength also increases with the length of the host's larval life stage, presumably due to prolonged cif action. Our findings reveal that cifB-transcript levels largely explain CI strength, while highlighting other covariates. Elucidating CI's mechanism contributes to our understanding of Wolbachia spread in natural systems and to improving the efficacy of CI-based biocontrol of arboviruses and agricultural pests globally.

Keywords: Drosophila; Endosymbiosis; Wolbachia; host–microbe interactions; reproductive parasitism.

Fig. 1.

CI phenotype and Wolbachia and host phylogenies. (A) CI kills a proportion of eggs when uninfected females mate with infected males. All other crosses are compatible. (B) Bayesian Wolbachia chronogram with absolute divergence estimates in thousands of years (95% credible interval of the median) for 10 Group-A Wolbachia, with Group-B wMau that infects D. mauritiana as an outgroup (37). The chronogram was estimated using 156 full-length and single-copy genes (130,359 bp) of equal length and based on prior calibration of rates of Wolbachia divergence (38). All nodes have posterior probabilities >0.95. wMel-like and wRi-like clades are highlighted. (C) Phylogram of Drosophila host species used in this study, based on 20 conserved and single-copy genes (38). All nodes have posterior probabilities >0.95. Dashed lines pair Wolbachia strains with their Drosophila host species and indicate topological discordance. Female D. teissieri is displayed to the bottom right. (D) Six Wolbachia induced strong CI, two yielded weak CI, and two caused nonsignificant reductions in egg hatch. “Compatible crosses” include the three compatible crosses in Fig. 1A. The wAur compatible cross includes only the infected female x infected male cross since uninfected wAur males were largely inviable. CI strength was calculated as 1—(CI-cross hatch rate/mean hatch of compatible crosses) (10) and displayed as a percentage. 0% CI represents no deviation from average compatibility, and 100% CI represents no eggs hatching. BCa confidence intervals of CI strength are displayed to the right of the plots. Significant differences are based on Mann–Whitney U tests between compatible and CI crosses for each strain. Names of strains are displayed in orange text if they do not cause significant CI, blue text if they cause weak CI, and black text if they cause strong CI. Significant differences are *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001. Nonsignificant results are denoted with ns. Exact P-values are reported in Supplementary Table S1, and Supplementary Fig. S1 displays all cross types individually.

Fig. 2.

CifA and CifB variation. (A) Similarity matrix displaying pairwise amino acid similarity for CifA proteins ranging from 100% to 15.4%, displaying considerable genetic diversity. (B) CifA (left) and CifB (right) protein architecture with HHpred annotations with P > 80% are displayed as large colored boxes. CifA had no confident annotations. The PD-(D/E)XK pair was the only feature represented in all CifB structures. (C) AlphaFold structures of CifA and CifB’s PD-(D/E)XK pair from four divergent Cif Types. Proteins are colored by pLDDT—a metric of structure confidence. pLDDT scores range from 0 to 100 where pLDDT > 90 is high confidence, 90 > pLDDT > 70 is confident, 70 > pLDDT > 50 is low confidence, and pLDDT < 50 is very low confidence.

Fig. 3.

Wolbachia densities and cif-transcript levels in testes vary for Drosophila-associated Wolbachia. (A) Wolbachia-density fold change (FC) from testes DNA extracts was calculated as 2^–∆∆CT of the Wolbachia gene ftsZ to the host gene mid1 relative to a randomly selected wMel-infected sample. Notably, wAno and wBai are strong-CI Wolbachia that have significantly lower Wolbachia densities than weak-CI wMel. Wolbachia relationships presented in Fig. 1B and mean BCa CI-strength estimates in parentheses are displayed for reference. (B) cifA- and (C) cifB-transcript levels in testes of infected flies of each strain. cif-transcript level FC was calculated as 2^–∆∆CT of cif-transcript levels to an exogenous spike-in control relative to the average transcript levels of cif_wMel[T1], cif_wRi[T2], cif_wTei[T4], and cif_wTri[T5] samples. Names of strains are displayed in orange text if they do not cause significant CI, blue text if they cause weak CI and black text if they cause strong CI. Significant differences are based on FDR-adjusted pairwise t tests relative to wMel abundance in A and cif_wMel[T1] transcript levels in B and C. *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001. Exact P-values are reported in Supplementary Table S1.