Male-killing toxin in a bacterial symbiont of Drosophila

Abstract

Several lineages of symbiotic bacteria in insects selfishly manipulate host reproduction to spread in a population ¹ , often by distorting host sex ratios. Spiroplasma poulsonii^2,3 is a helical and motile, Gram-positive symbiotic bacterium that resides in a wide range of Drosophila species ⁴ . A notable feature of S. poulsonii is male killing, whereby the sons of infected female hosts are selectively killed during development^1,2. Although male killing caused by S. poulsonii has been studied since the 1950s, its underlying mechanism is unknown. Here we identify an S. poulsonii protein, designated Spaid, whose expression induces male killing. Overexpression of Spaid in D. melanogaster kills males but not females, and induces massive apoptosis and neural defects, recapitulating the pathology observed in S. poulsonii-infected male embryos^5-11. Our data suggest that Spaid targets the dosage compensation machinery on the male X chromosome to mediate its effects. Spaid contains ankyrin repeats and a deubiquitinase domain, which are required for its subcellular localization and activity. Moreover, we found a laboratory mutant strain of S. poulsonii with reduced male-killing ability and a large deletion in the spaid locus. Our study has uncovered a bacterial protein that affects host cellular machinery in a sex-specific way, which is likely to be the long-searched-for factor responsible for S. poulsonii-induced male killing.

Extended Data Figure 1. Identification and characterization of the partial male-killing Spiroplasma strain.

a, An illustration showing the origin of the Spiroplasma strains analysed in this study. Pictures show male-killing (MK) Spiroplasma of D. melanogaster. MSRO-Ug is the original male-killing strain maintained in the Oregon-R wild-type fly. Fly stocks [Df(3L)H99 and Sxl-EGFP] artificially infected with this original strain showed complete male killing (100% MK) for the first 20 generations. Afterwards, one strain (MSRO-SE) started to show the partial male-killing phenotype, while the other (MSRO-H99) kept the ability to induce complete male killing. See Methods for more detail. b, c, Sex ratio analysis of the adult progeny obtained from Oregon-R flies infected with MSRO-Ug, MSRO-H99, and MSRO-SE. We repeated experiments for three times at fourth- to sixth-generations (G4-6) after the establishment of infection. In c, the relative number of male offspring (% of females) obtained from Oregon-R female flies infected with MSRO-SE were plotted. The data point was excluded if the total count of flies was below 10. d, Relative titre of Spiroplasma within individual female fly. Adult females infected with three MSRO strains were aged for 0, 7, 14 days after eclosion and analysed by qPCR. Data were normalized with respect to 0 day females infected with MSRO-Ug. Different characters indicate statistically significant differences (P < 0.01; N.S., not significant, P > 0.05; Steel-Dwass test; see Supplementary Table 2). Please note that the titres of three strains are comparable and even the higher titre in old females (see 14 days in d) fails to induce complete male killing in MSRO-SE (c). Box and dot plots are as in Fig. 1c and sample sizes (n, number of adult flies) are shown at the bottom.

Extended Data Figure 2. Whole-genome sequencing studies of the male-killing Spiroplasma variants.

a, The comparison of genomic features of three Spiroplasma strains. MSRO-H99 and MSRO-SE are newly obtained variants isolated in this study (Extended Data Fig. 1a; see also Methods). As a control, the data of the previously reported male-killing Spiroplasma genome is also indicated (MSRO, 2015). b, Whole-genome alignment of the three Spiroplasma strains. To start the alignment from the dnaA gene, the contig #1 of MSRO-H99 was split into two fragments (contig #1.1 and #1.2). The locations of the contigs corresponding to extra chromosomes (putative plasmids; see Methods) are shown as “extra”. SpAID (gene ID: SMH99_26490) and SpAID ΔC (gene ID: SMSE_25110) are located on these extra chromosomes in MSRO-H99 (contig #4) and MSRO-SE (contig #2), respectively.

Extended Data Figure 3. Genetic alterations of the SpAID locus in the partial male-killing Spiroplasma strain.

The genome structures around the SpAID loci in the male-killing (a, MSRO-Ug and MSRO-H99) and the partial male-killing (b, MSRO-SE) Spiroplasma strains. Genes encoded on opposite strands are shown in different colours (red and blue, respectively). An 828-bp deletion and nucleotide substitutions (coloured in red; corresponding amino acid sequences are presented in one-letter code) in the 3’ region of the SpAID gene are indicated. These sequence alterations were confirmed by the Sanger method (see Methods).

Extended Data Figure 4. Neural defects of SpAID-expressing embryos.

Representative images of stage 13-14 female (a, n = 14) and male (b, n = 16) embryos maternally expressing SpAID, stained for TUNEL (green) and neural cells (Elav, magenta). Single-channel images of Elav are also shown. The boxed region in b is magnified in c with single-channel images of Elav and TUNEL. Embryos were co-stained for Elav, TUNEL, Sxl, and DNA, and selected channels are shown in a-c and Fig. 2a, b, respectively.

Extended Data Figure 5. SpAID acts through the dosage compensation machinery.

a-c, Representative images of stages 13-14 embryos ectopically expressing the MSL complex by the H83M2 transgene, stained for TUNEL (green) and Sxl (magenta). GFP-expressing control female (a, n = 15), SpAID-expressing female (b, n = 20), and male (c, n = 19) embryos are shown. d, Quantification of TUNEL signals in H83M2 embryos at stages 13-14. Different characters indicate significant differences (P < 0.001; Steel-Dwass test; see Supplementary Table 2). The box and dot plot (females, red; males, blue) is as in Fig. 1c and sample sizes (n, number of embryos) are shown at the bottom. e, f, Representative images of epithelial cells of stages 8-10 male embryos expressing GFP (e, n = 25) and SpAID (f, n = 25), stained for DNA (green) and MSL1 (magenta) from the datasets analysed in Fig. 3. All UAS transgenes were expressed maternally.

Extended Data Figure 6. Expression of SpAID by using a weak GAL4 driver.

The number of adult progeny (females, red; males, blue) obtained from crosses between the armadillo-GAL4 driver line (weak and ubiquitous expression) and four UAS transgenic lines (GFP, SpAID, ΔANK, and ΔOTU; n = 6 independent crosses for GFP and SpAID, n = 8 independent crosses for ΔANK and ΔOTU). The UAS-GFP line was used as a negative control. With this weak GAL4 driver, SpAID still eliminated all male progeny, while both ΔANK and ΔOTU had no impact on male viability. An asterisk indicates the statistically significant difference (P < 0.01; N.S., not significant, P > 0.05; two-tailed Mann-Whitney U test; see Supplementary Table 2). Box and dot plots are as in Fig. 1c. The total numbers of adult counts for each genotype and sex are shown at the bottom.

Extended Data Figure 7. A proposed model for SpAID-induced male-killing phenotypes.

SpAID utilizes the OTU domain and ankyrin repeats (ANK) to target the host nucleus and the male X chromosome. “MSL” and “Ac” indicate the dosage compensation complex and resultant histone acetylation, respectively. See text for other explanations.

Figure 1. Expression of SpAID selectively eliminates male offspring.

a, Spiroplasma-induced male killing in Drosophila. Infected females (top) produce only female offspring (bottom). A picture shows a male-killing Spiroplasma of D. melanogaster detected by DNA staining. b, Protein structure of SpAID, which contains ankyrin repeats (ANK, red), the OTU deubiquitinase domain (blue), an N-terminal signal peptide (SP, black), and a C-terminal hydrophobic region (HR, green). SpAID ΔC of the partial male-killing strain encodes a protein with an amino acid substitution (Q787C) and C-terminal truncation. The structures of two deletion constructs of SpAID (ΔANK and ΔOTU) are also indicated. The numbers represent amino acid (aa) residues. c, The number of adult progeny obtained from crosses between the Actin-GAL4 line and four UAS transgenic lines (GFP, SpAID, ΔANK, and ΔOTU; n = 10 independent crosses for each transgene). The UAS-GFP line was used as a negative control. We counted the number of resultant offspring (females, red; males, blue) expressing the transgenes (+, having both Actin-GAL4 and UAS transgenes) and siblings not expressing the transgenes (-, having only UAS transgenes) as internal controls. Different characters indicate statistically significant differences (P < 0.0001, P < 0.05 for ΔANK; N.S., not significant, P > 0.05; Steel-Dwass test; see Supplementary Table 2). Box plots indicate the median (bold line), the 25th and 75th percentiles (box edges), and the range (whiskers). Dot plots show all data points individually. The total numbers of adult counts for each genotype and sex are shown at the bottom.

Figure 2. Expression of SpAID reproduces male-killing phenotypes during embryogenesis.

a, b, Representative images ofstages 13-14 female (a, n = 14) and male (b, n = 16) embryos maternally expressing SpAID, stained for apoptosis (TUNEL; green), Sxl (magenta), and DNA (blue). Single-channel images of TUNEL and Sxl are also shown. c, Quantification of TUNEL signals in stages 11-12 and 13-14 embryos (females, red; males, blue). Different characters indicate statistically significant differences (P < 0.0001; Steel-Dwass test; see Supplementary Table 2). Box and dot plots are as in Fig. 1c. Sample sizes (n, number of embryos) are shown at the bottom. Embryos were co-stained for Elav, TUNEL, Sxl, and DNA, and selected channels are shown in a, b and Extended Data Figure 4.

Figure 3. SpAID acts through the MSL complex.

a, b, Epithelial cells in stage 9 male embryos expressing GFP (a) and SpAID (b), stained for pH2Av (green), MSL1 (magenta), and DNA (blue). c, Quantification of pH2Av foci in embryos expressing GFP and SpAID. d, Percentage of pH2Av foci overlapping with MSL1 signals in male embryos expressing GFP [median (interquartile range): 6.3% (0-19.3%)] and SpAID [52.9% (46.2-64.2%)]. e, f, Dividing cells in stage 9 male embryos expressing GFP (e, proper segregation) and SpAID (f, a broken bridge) stained for DNA (green) and MSL1 (magenta). g, The number of chromatin bridges containing (black; numbers on the left) or not containing (grey; numbers on the right) MSL1 signals in embryos expressing GFP and SpAID. h, The number of MSL1 focal signals in male embryos expressing GFP (black) and SpAID (green). The same datasets of stages 8-10 embryos were analysed in a-h (n = 50 or 48 images per condition). Different characters or asterisks indicate significant differences in c (P < 0.001; Steel-Dwass test), d (P < 0.0001; χ² test), and h (P < 0.0001; two-tailed Mann-Whitney U test) (see Supplementary Table 2). Box and dot plots in c, d, h are as in Fig. 1c and sample sizes are shown at the bottom in c. All UAS transgenes were expressed maternally.

Figure 4. Subcellular localization of SpAID.

a-f, Larval salivary glands expressing SpAID-GFP (a, female, n = 13; b, male, n = 17), ΔANK-GFP (c, female, n = 9; d, male, n = 16), and ΔOTU-GFP (e, female, n = 15; f, male, n = 12) stained for MSL1 (magenta) and DNA (blue). For GFP (green), raw fluorescent signals were detected. Magnified views of nuclei are shown. Dark spots inside nuclei in GFP images represent the nucleolus. Arrowheads indicate GFP signals associated with plasma membranes.