Blind killing of both male and female Drosophila embryos by a natural variant of the endosymbiotic bacterium Spiroplasma poulsonii.

Abstract Spiroplasma poulsonii is a vertically transmitted endosymbiont of Drosophila melanogaster that causes male‐killing, that is the death of infected male embryos during embryogenesis. Here, we report a natural variant of S. poulsonii that is efficiently vertically transmitted yet does not selectively kill males, but kills rather a subset of all embryos regardless of their sex, a phenotype we call ‘blind‐killing’. We show that the natural plasmid of S. poulsonii has an altered structure: Spaid, the gene coding for the male‐killing toxin, is deleted in the blind‐killing strain, confirming its function as a male‐killing factor. Then we further investigate several hypotheses that could explain the sex‐independent toxicity of this new strain on host embryos. As the second non‐male‐killing variant isolated from a male‐killing original population, this new strain raises questions on how male‐killing is maintained or lost in fly populations. As a natural knock‐out of Spaid, which is unachievable yet by genetic engineering approaches, this variant also represents a valuable tool for further investigations on the male‐killing mechanism.

Although the molecular mechanisms underlying these strategies are different (Harumoto & Lemaitre, 2018;LePage et al., 2017), all have as a consequence an increase in the proportion of infected individuals in host populations.
One of the most frequently reported endosymbiont that manipulates insect reproduction is Spiroplasma spp. The genus Spiroplasma is highly diverse and comprises species that are strictly pathogenic for plants and arthropods (Gasparich, 2002), but also some vertically transmitted endosymbiotic species estimated to infect 4-7% of insects (Duron et al., 2008). Spiroplasma poulsonii along with Wolbachia is one of the two endosymbionts that can naturally infect Drosophila melanogaster (Mateos et al., 2006).
Screenings of natural Drosophila populations indicate a highly variable rate of infection by Spiroplasma, ranging from complete absence to up to 60% in some populations (Watts, Haselkorn, Moran, & Markow, 2009). Some strains of S. poulsonii cause MK in Drosophila, whereby all male embryos die during early embryogenesis (Montenegro, Solferini, Klaczko, & Hurst, 2005). The recently unravelled molecular mechanism of MK involves a Spiroplasma-encoded toxin, Spaid, that potentially targets the male-specific lethal complex, a part of the X-chromosome dosage compensation system in Drosophila (Harumoto & Lemaitre, 2018). Spaid causes abnormal segregation and breakage of X chromatids in male embryos, leading to a massive DNA-damage dependent apoptosis and eventually to the death of the embryo (Harumoto, Anbutsu, Lemaitre, & Fukatsu, 2016;Harumoto & Lemaitre, 2018).
S. poulsonii also affects adult Drosophila physiology by reducing its lifespan and causing a neurodegenerative phenotype in old flies (Herren & Lemaitre, 2011). The cause of these two phenotypes is not clearly unravelled yet, although the involvement of cardiolipins or a neurotoxic protein released by Spiroplasma have been proposed as putative causes (Herren & Lemaitre, 2011;Masson, Calderon Copete, Schüpfer, Garcia-Arraez, & Lemaitre, 2018). Last, a remarkable protection is conferred to Spiroplasma-infected flies against parasitoid wasp and nematode infections (Ballinger & Perlman, 2017;Hamilton, Peng, Boulanger, & Perlman, 2016;Xie, Butler, Sanchez, & Mateos, 2014). Competition between parasites and Spiroplasma for host lipids has been proposed as a mechanism underlying this protection (Paredes, Herren, Schüpfer, & Lemaitre, 2016). Another hypothesis involves Spiroplasma toxins belonging to the Ribosome-Inactivating Protein (RIP) family, which accumulate in the hemolymph and target parasitic wasp and nematode ribosomes (Ballinger & Perlman, 2017;Hamilton et al., 2016). The effect of S. poulsonii infection on Drosophila physiology are thus highly relying on Spiroplasma secreted toxins, but the lack of genetic tools to manipulate the bacteria renders it difficult to assess precisely the role of each bacterial toxin towards each phenotype.
In this study, we describe a spontaneous mutant derived from a male-killer S. poulsonii. This strain causes an alternate phenotype in Drosophila whereby some infected offspring is killed regardless of the sex of the embryo. We show that the mutant has a clean deletion of Spaid, thus presenting a unique opportunity to study the equivalent of a Spaid full knock-out, which is technically not achievable yet by genetic engineering.
2 | RESULTS 2.1 | Discovery and phenotypic characterisation of a S. poulsonii spontaneous mutant MK penetrance strongly relies on the parental genetic background (Kageyama, Anbutsu, Shimada, & Fukatsu, 2009). To investigate this effect, we crossed Oregon-R (OR R ) females infected with S. poulsonii Uganda-1, a MK strain with full penetrance (hereafter named 'MK strain'), with a hundred lines from the Drosophila Genetic Reference Panel collection (Mackay et al., 2012). Each subsequent generation was then backcrossed with the parental DGRP line for five generations to replace the OR R background by the DGRP ones. Seven of these crosses gave a progeny that contained male individuals, thus showing a defective MK activity. This defective MK phenotype was however not observed systematically upon repeating the exact same crosses several times, suggesting that it appeared randomly because of yet unidentified factors (data not shown). We then injected hemolymph from these MK defective DGRP crosses back into the original OR R background, where MK is fully penetrant. Interestingly, one newly infected OR R background line retained a reduced MK penetrance, showing that this phenotype was independent of the host genetic background but was rather caused by a change in the Spiroplasma genotype ( Figure 1a). The Drosophila OR R females infected with this new Spiroplasma variant had a normal fecundity compared to that of the original Spiroplasma strain (Figure 1b), but the viability of the embryos decreased along with female aging, while embryo viability remained stable over aging of flies infected with the original strain ( Figure 1c). The sex-ratio of the offspring was however of 50% males and 50% females, as in uninfected flies (Figure 1d,e), suggesting that the new Spiroplasma variant kills embryos of aging females regardless of their sex. We hereafter name this phenotype 'blind-killing' (BK) to oppose it to MK that selectively kills males.
In a previous report, a natural variant of S. poulsonii exhibiting weak male killing has been identified and the low MK phenotype was tied to a lower endosymbiont titre in the adult females. This suggested that a density threshold needs to be met for full MK penetrance (Anbutsu & Fukatsu, 2003). To determine whether this threshold hypothesis applies to the BK Spiroplasma strain we identified, we compared the bacterial titre in adults and eggs of MK-and BK-Spiroplasma infected OR R females. In adults (Figure 2a), no difference was observed between MKinfected females titre and BK-infected males and females. Embryos from BK-infected females had a slightly lower titre than the MK-infected ones ( Figure 2b). The difference was however significant only for embryos deriving from mothers aged of 14 days, while the BK phenotype is already detectable in one-week-old mothers, suggesting that the BK phenotype and the endosymbiont titre are not correlated. including a loss of negative geotaxis and tremors (Herren et al., 2014).
We monitored the loss of negative geotaxis in flies infected by the MK or the BK Spiroplasma strain, and found a similar intensity of decay at 3 and 4 weeks after emergence ( Figure 2d). This suggests that the BK strain is not more deleterious than the MK strain for adults but rather differentially impacts Drosophila embryos. Therefore, we focused our analysis on early embryonic stages. Furthermore, we did not observe any chromatin bridges in male embryos at stage 8-10 (data not shown), a feature associated with male killing. Altogether, our data suggest that BK-infected embryo die at earlier developmental stages than MK-infected ones, and that both male and female BK-infected surviving embryos trigger apoptosis at similar levels lower than that observed in MK-infected males. Last, two hallmarks of MK, male-specific high levels of apoptosis and chromatin bridges are not found in embryos infected by the BK strain, suggesting that killing by this strain occurs through a distinct mechanism to that of MK.  Table S1.
We first examined the expression profile of genes affected by SNPs and by the partial plasmid deletion in the BK genome. Among the 16 chromosomal SNPs, 15 were located in or close to genes that are either undetected in the RNAseq or which expression is not differential between MK and BK strains. The remaining SNP affected the terminase-coding gene SMSRO_SF011260, which had a low and variable basal expression unlikely to significantly affect Spiroplasma virulence.
As expected, Spaid, parA1 and other hypothetical protein coding genes located on the plasmid fragment missing in the BK genome were the most downregulated genes in the BK transcriptome, confirming the deletion. Intriguingly, parA2, a second plasmid-located copy of the par family, was also downregulated in BK strain.
We also found 13 genes encoding for various transporters upregulated in BK strain, including amino acids, ascorbic acid, phosphate, chromate and two glucose transporters. Genes encoding for an endoβ-N-acetylglucosaminidase and a glycosylhydrolase, which can process free oligosaccharides and release monosaccharides (Davies & Henrissat, 1995;Suzuki et al., 2002), were also upregulated in the BK strain. beside adhesins that were described in the latest genome annotation (Masson et al., 2018), only three had a differential expression level between the strains, including the gene coding for the membrane lectin Spiralin B (Killiny, Castroviejo, & Saillard, 2005), for the glycerol-3-phosphate oxidase GlpO that produces radical oxygen species (Vilei & Frey, 2001) and the protective toxin RIP2 (Hamilton et al., 2016). This notwithstanding, all of them were downregulated in BK strain compared to MK.
In conclusion, the trancriptomics of the BK strain did not reveal any overt toxicity mechanism that would explain the embryonic mortality. We however found a stronger line of evidence in favour of a metabolic shift of the bacteria that would affect the embryonic development regardless of its sex, rather than a direct pathogenic effect mediated by virulence factors.

| Protection against parasitoid wasps
Spiroplasma-mediated protection against parasitoid wasps and nematodes involves RIP toxins that are encoded by five chromosomal genes (Hamilton et al., 2016), of which only RIP1 and RIP2 are significantly expressed, RIP2 being the most expressed (Garcia-Arraez, Masson, Escobar, & Lemaitre, 2019). The down-regulation of RIP2 in the BK strain thus raised the question of the ability of the BK strain to protect the fly against natural enemies. We monitored RIP activity over the lifespan of adult females infected with the MK or the BK strain and observed a similar level for flies up to 2 weeks old, but a lower RIP activity in BK-infected flies aged 3 weeks and more ( Figure 6a,b). We observed a high mortality in pupae developing from larvae infected by the BK strain and challenged with the parasitoid wasp Leptopilina boulardii, but the surviving individuals were overall resistant as no wasp emerged from most biological replicates ( Figure 6c). We however observed two wasps emerging from BKinfected larvae on a total of 150 larvae examined while this never happened with MK-infected larvae. These two escaper wasps suggest that the protection conferred by the BK strain, although very efficient, might be weaker than that of the MK strain.

| DISCUSSION
We isolated a spontaneous variant of S. poulsonii Uganda-1 that does not cause MK in its host D. melanogaster but rather kills unselectively a subset of both male and female progeny, a phenotype we named BK.
A first key feature of the BK strain is the loss of MK, that is, the killing of male-embryos specifically. Sequencing the BK strain genome  (Carle et al., 2010;Davis et al., 2005;Saillard et al., 2008). The sequence coding for the replication protein pE, which must be borne on the plasmid to ensure its replication (Breton, Duret, Arricau-Bouvery, Béven, & Renaudin, 2008), is intact and not differentially expressed, suggesting that the plasmid replication is functional in the BK strain. However, the deletion of parA1 and down-regulation of parA2 suggests that the partition of the plasmid could be less efficient and potentially lead to its loss.
With the deleted plasmid fragment being undetectable by PCR, the BK strain can be considered as a natural knock-out of Spaid, thus confirming the function of this gene as a necessary MK factor.
Although RNAseq analysis showed that virulence factors were either not differentially expressed or downregulated in BK, with the exception of adhesin coding genes. Adhesin-like proteins have been shown to be necessary for pathogenic Spiroplasma species to adhere to and invade host cells (Béven et al., 2012;Breton et al., 2010;Hou et al., 2017;Zha, Yang, Wang, Yang, & Yu, 2018), these could thus be responsible for the BK strain increased pathogenicity. Future studies are required to address the role of these proteins in endosymbiotic Spiroplasma and determine whether they are involved in the BK phenotype.
An alternative hypothesis would be that the BK toxicity rather comes from a shift in the bacterium metabolic activity that kills the embryo in a sex-independent fashion. The RNAseq revealed two BK features related to metabolism: an up-regulation of transporters and glycosidases, and a down-regulation of translation-related genes.
tRNAs up-regulation is a mechanism that allow bacteria in co-cultures to grow better in a medium where there is competition for nutrients (Tognon, Köhler, Luscher, & van Delden, 2019). It is also a hallmark of cancer cells, where increased tRNAs can increase the translation of mRNA that bear their cognate codons (Goodarzi et al., 2016). tRNA regulation could thus be a way for Spiroplasma to modulate protein production and, in the case of BK, to decrease the activity of the whole translation machinery through the down-regulation of tRNA-Met. A decreased metabolic activity, which along with the up-regulation of transporters and glycosyl hydrolases, could indicate that BK relies more on nutrient uptake for its host than MK Spiroplasma to sustain its proliferation. Since the BK titre is not significantly different from MK in adults and eggs laid by young mothers (Figure 2), we assume that the metabolic shift of BK has little consequences on adult hosts that have unlimited access to food. In embryos, however, where resources are limited, the increased uptake capacity of the BK strain would prop up the competition for available nutrients, causing detrimental effects on Drosophila embryonic development and viability. The eggs viability would thus decrease along with Spiroplasma titre, which is consistent with our phenotypic observations (Figures 1c and 2b).
Remarkably, this is the second strain described in the literature within a 2-years period as a natural variant that lost MK ability (Harumoto & Lemaitre, 2018). In lab stocks, MK is used as a proxy to assess the infection status. If males emerge from the progeny of infected flies, the whole progeny is considered as uninfected and discarded, while potentially being well-infected by a natural non-malekiller variant. The frequency of such variants is thus likely to be underestimated in lab conditions. This also suggests that MK could be quite unstable in the absence of positive selective pressure, notably because of the plasmid location of Spaid (while other phenotypes such as protection against natural enemies are mediated by chromosomal, thus more stable genes).
The prevalence of endosymbionts capable of MK is highly variable between species, ranging from 1% for S. poulsonii in Drosophila willistoni (Williamson & Poulson, 1979) up to 99% for Wolbachia infecting some populations of the butterfly Hypolimnas bolina (Dyson & Hurst, 2004), and also within populations of the same species (Kageyama et al., 2009). Most associations with MK endosymbionts are recent and thus likely transient, indicating that fixation of the symbiont in the host population leads in most cases to the extinction of the host species because of the paucity of males (Dyer & Jaenike, 2004;Hatcher, Taneyhill, Dunn, & Tofts, 1999). Some examples of long-lasting associations have been discovered but their evolutionary stability despite of MK remains unexplained (Dyer & Jaenike, 2004). A current hypothesis is that hosts undergo pressure to evolve resistance mechanisms against MK, although experimental evidence of the emergence of such resistances is scarce (Dyson & Hurst, 2004;Hayashi, Nomura, & Kageyama, 2018;Jiggins, Hurst, & Yang, 2002 infection was performed by injecting them with 9 nl of undiluted hemolymph extracted from infected females using a Nanoject II (Drummond) as previously described (Herren & Lemaitre, 2011). The DGRP line from which we isolated the BK strain was #25191. in 2019 with no significant difference between replicates. Titre measurements have been performed by qPCR as previously described (Herren & Lemaitre, 2011) using the ΔΔCT method to quantify the abundance of Spiroplasma dnaK copy number relative to host rps17 copy number. Titre measurements have been performed three independent times in adults except BK-M (two times), and two times in embryos. Survivals and negative geotaxis assays (climbing assays) as well as DAPI and Terminal deoxynucleotidyl-transferase dUTP nick end labeling staining and quantification have been performed as described, along with msl-1 staining used to sex the embryos (Harumoto et al., 2014;Herren et al., 2014;Herren & Lemaitre, 2011). These experiments were repeated three independent times, except for DAPI staining that has been performed once.

| Phenotypic characterisation
Wasp challenges have been performed using the progeny of 1-week-old mothers as previously described (Paredes et al., 2016) and were repeated five independent times. RIP assays were performed by reverse transcription quantitative PCR (RT-qPCR) as previously described (Hamilton et al., 2016) and were repeated three independent times.

| Plasmid detection by PCR
DNA from pools of 5 flies was extracted as previously described (Herren & Lemaitre, 2011). PCRs were carried out on total DNA using GoTaq G2 with the following cycling protocol:

| Genome sequencing, assembly and annotation
S. poulsonii DNA was extracted from fly hemolymph as previously described (Masson et al., 2018). Processing of the samples was performed in the University of Lausanne Genomic Technologies Facility.