Genetic Analyses of Elys Mutations in Drosophila Show Maternal-Effect Lethality and Interactions with Nucleoporin Genes

ELYS determines the subcellular localizations of Nucleoporins (Nups) during interphase and mitosis. We made loss-of-function mutations of Elys in Drosophila melanogaster and found that ELYS is dispensable for zygotic viability and male fertility but the maternal supply is necessary for embryonic development. Subsequent to fertilization, mitotic progression of the embryos produced by the mutant females is severely disrupted at the first cleavage division, accompanied by irregular behavior of mitotic centrosomes. The Nup160 introgression from D. simulans shows close resemblance to that of the Elys mutations, suggesting a common role for those proteins in the first cleavage division. Our genetic experiments indicated critical interactions between ELYS and three Nup107–160 subcomplex components; hemizygotes of either Nup37, Nup96 or Nup160 were lethal in the genetic background of the Elys mutation. Not only Nup96 and Nup160 but also Nup37 of D. simulans behave as recessive hybrid incompatibility genes with D. melanogaster. An evolutionary analysis indicated positive natural selection in the ELYS-like domain of ELYS. Here we propose that genetic incompatibility between Elys and Nups may lead to reproductive isolation between D. melanogaster and D. simulans, although direct evidence is necessary.

Here we disrupted the X-linked CG14215 (hereafter, Elys) of D. melanogaster and analyzed the mutant phenotypes. Surprisingly, the D. melanogaster mutants exhibited an effect similar to the C. elegans mutants; homozygotes (or hemizygotes) were viable and male-fertile but female-sterile (maternal-effect lethal). Sperm penetrated the eggs produced by the mutant females, but the first mitotic division was never completed. This is one of the earliest developmental defects caused by D. melanogaster mutations (for the list of the genes, see Loppin et al. 2015) and will provide a rare opportunity to analyze Drosophila fertilization (Callaini and Riparbelli 1996;Kawamura 2001). In the present report we will describe in detail the developmental defects of the embryos in which maternally supplied ELYS is depleted.
The introgression of the Nup160 allele from D. simulans (Nup160 sim ) causes recessive female sterility in the D. melanogaster genetic background (Sawamura et al. 2010). Females homozygous or hemizygous for Nup160 sim produce eggs capable of sperm entry, but the embryos never develop (Sawamura et al. 2004). As this is similar to the maternal-effect phenotype of the Elys mutations, we wanted to compare these phenotypes in detail. We also show genetic interaction between Elys and the Nups, and discuss the possible involvement of ELYS in reproductive isolation between D. melanogaster and D. simulans.
To eliminate endosymbiotic bacteria (presumably Wolbachia) from fly stocks used for embryo immunostaining, we fed flies with medium containing 0.03% tetracycline for one generation (Hoffmann et al. 1986). This allowed us to analyze chromosomal DNA exclusively with DAPI staining, but not coexistent bacterial DNA, in the early Drosophila embryo (Lin and Wolfner 1991;Kose and Karr 1995).

Establishment of Elys mutations
No Elys mutations had been reported in D. melanogaster. Generation of Elys alleles was carried out with the CRISPR/Cas9 system described previously (Kondo and Ueda 2013). The guide RNAs (gRNAs) were selected using CRISPR Optimal Target Finder (Gratz et al., 2014; http:// tools.flycrispr.molbio.wisc.edu/targetFinder/). To generate a double gRNA construct to target the Elys locus, two pairs of oligonucleotides were annealed and cloned into the pBFv-U6.2B vector; one of the pairs of oligonucleotides is 59-CTT CGC TGC ACT CGG TCT GCT ACA-39 and 59-AAA CTG TAG CAG ACC GAG TGC AGC-39, and the other is 59-CTT CGG CCA CTG ACT CGT TGC TCG-39 and 59-AAA CCG AGC AAC GAG TCA GTG GCC-39. The Elys gRNA vector was injected into embryos of y 1 v 1 P{y +t7.7 = nos-phiC31\int.NLS}X; P{y +t7.7 = CaryP}attP40. The transgenic U6-Elys-gRNA flies were established, and mutations in the Elys locus were recovered in offspring from nos-Cas9 (y 2 cho 2 v 1 ; attP40{nos-Cas9}/CyO) and the U6-Elys-gRNA flies. Cas9-mediated targeted mutagenesis of the Elys locus was introduced on the X chromosome of y 2 cho 2 v 1 . Potential mutations of the Elys locus were identified by genomic PCR using the primers 59-AAG ACG GCC GAA TCC TGA TCT ACG-39 and 59-AGA CCA CTA GAC TGC GTT GCT TGC-39; these primers sandwich the potential deletions (the former is on exon 3 and the latter is on exon 7). Sequencing of the obtained PCR products confirmed mutations of the corresponding genomic region (Figure 1).

Embryo collection and immunostaining
Well-fed virgin females were mated with wild-type (Oregon-R) males and allowed to lay eggs in short vials containing fly medium on which yeast was seeded. Embryos were collected at 20-min intervals, and the following fixation was completed within an additional 10 min. After dechorionation with 50% bleach for 1.5 min, embryos were washed with water and then fixed and devitellinized by shaking in a mixture of equal volumes of heptane and methanol. Fixed embryos were stored in methanol.
To visualize sperm in the eggs, females were crossed with w; dj-GFP/ CyO males, which produce fluorescent sperm tails (dj, don juan; Figure 1 Structure of the Elys gene and its mutations. Box, exon; horizontal line, intron. 1-490 aa, seven-bladed beta propeller repeats; 714-922 aa, ELYSlike domain; 1,069-1,092 aa, coiled coil; 1,665-1,847 aa, Glu-rich. There was a 1-bp deletion (1,287T) in Elys 2 and a 3,475-bp deletion (1,512) in Elys 5 ; 59-CTC GGT CG-39 was inserted at the latter site instead. Santel et al. 1997). Egg collection, dechorionation and methanol fixation were performed as described above, followed by replacement of methanol with ethanol. Fixed eggs were stepped gradually into PBT by sequential transfers into PBT containing 75%, 50%, 25% and 0% ethanol and then were stored at 4°. For observation, eggs were incubated in 25% glycerol in PBS, mounted on glass slides with Slow-Fade Gold antifade reagent (Thermo Fisher Scientific) and then coverslipped by using a small amount of silicone grease (HIVAC-G, Shinetsu Silicone) to avoid egg-rupture.

Evolutionary analyses of Elys
By using Elys of D. melanogaster (CG14215) as a query, homologs of D. simulans (GD26978), D. sechellia (overlapping GM22978 and GM22979) and D. yakuba (GE15862) were obtained by a BLAST search (blastn in FlyBase). The sequences were aligned by using Clustal X ver. 2.1 (Larkin et al. 2007) and corrected manually. The number of nonsynonymous substitutions per nonsynonymous site (K a ) and the number of synonymous substitutions per synonymous site (K s ) were calculated, and the K a /K s ratio test (Li 1993) was conducted by using the kaks function in the seqinR package for the R environment (Charif and Lobry 2007; http://seqinr.r-forge.r-project.org). The K a /K s ratio was also calculated within a 180-bp sliding window to increase the sensitivity. PAML (Phylogenetic Analysis by Maximum Likelihood) ver. 4.9d (http://abacus.gene.ucl.ac.uk/software/paml.html ;Yang 2007) was also applied for the test.
The sequences of the common ancestors, node 1 (sechellia/simulans) and node 2 (node 1/melanogaster), were estimated, and the substitution history of the ELYS-like domain was reconstructed on the consensus unrooted phylogenetic tree: ((sechellia, simulans), melanogaster), yakuba (Lachaise and Silvain 2004). The ancestral state of node 2 was not determined unambiguously for three sites. We assumed that each replacement substitution took place with an equal probability in three branches (node 2-yakuba, node 2-melanogaster and node 2-node 1). Thus, these were in total calculated as 1/3 · 3 = 1 replacement in each branch.

Data availability
All Drosophila stocks, DNA clones and reagents are available upon request. Viability test for the Elys mutations is shown in Table S1. Sperm penetration to the eggs is shown in Table S2. Interaction between Elys and Nup37 is shown in Table S3. The lethal stage of Elys/Y; Df-Nup160/+ males was determined (Table S4). The lethal stage of Elys/Y; Df-Nup96/+ males was determined (Table S5). The cross between Elys/FM7c; Df(2L)Nup160M190/CyO females and Elys/Y males is shown in Table S6. The cross between Df(3R)/TM6C females and D. simulans Lhr males is shown in Table S6. Sperm were visualized by dj-GFP in the eggs from Elys mutant females ( Figure S1). Mating scheme to determine the lethal stage of Elys/Y; Df-Nup160/+ is shown in Figure S2. Supplemental material available at Figshare: https://doi.org/ 10.25387/g3.6279446.

RESULTS
Description of the Elys mutations X-linked CG14215 (X: 19,652,305-19,659,407 [+]) of D. melanogaster (FlyBase ID FBgn0031052) encodes a protein of 2,111 amino acids (aa) that includes an ELYS-like domain at aa 714-922 (InterPro accession number Q9VWE6; UniProtKB -X2JG50; Finn et al. 2017). We recovered two frameshift alleles (Elys 2 and Elys 5 ) that truncate the majority of the coding potential; aa 372 and 367 are predicted to be stop codons, respectively ( Figure 1). Surprisingly, the mutants were viable and malefertile (Supplemental Material, Table S1) but female-sterile in homozygotes (Table 1). Thus, the mutations can be maintained via heterozygous (Elys/FM) females and hemizygous (Elys/Y) males (or FM/Y males), where FM (first multiple) stands for a balancer X chromosome; rare FM homozygotes are also present in the stocks.

Disruption of mitotic progression of the first cleavage division by maternal effects of Elys mutations and Nup160 sim introgression
The Drosophila embryo remains a syncytium for the first two hours of development, where 13 rounds of nuclear division take place rapidly (Foe and Alberts 1983). To gain insights into the primary effect of the Elys mutations on embryonic development, we fixed embryos 10-30 min after deposition and carried out cytological analysis. Our comparative analysis of embryonic progeny produced by Elys mutant females (Elys 2 or n Elys 5 homozygotes) and the control females (Elys 2 or Elys 5 heterozygotes) revealed significant differences in the progression of the earliest cycles. Embryos from females mutant for Elys did not display mitotic progression; there was instead the accumulation of characteristics representing the first mitotic cycle (Table 2). Further investigation uncovered the maternal-effect lethality resulting from a terminal arrest in a metaphaselike state of the first cleavage division (Table 3; see below). The phenotype was essentially identical in the two Elys mutant strains. The normal mitosis of the first cleavage division in Drosophila is gonomeric (Huettner 1924;Guyénot and Naville 1929;Callaini and Riparbelli 1996;Williams et al. 1997;Loppin et al. 2015); after DNA replication in nuclei from the ovum and sperm, the haploid complements persist in separate groups on a bipolar spindle composed of two units of microtubule arrays, which we refer to as the dual spindle ( Figure 2A). The two units of microtubule arrays share the spindle poles, where the entire set of chromosomes is gathered at telophase. The Elys mutations affected the arrangement of the chromosomes and microtubule configurations of the dual spindle, because only spindles that appeared to be composed of a single unit of microtubule arrays with indiscriminately conjugated chromosomes were observed among all 102 embryos obtained from Elys 2 and Elys 5 females (Figure 2, D and E). In addition, centrosomes behaved in a peculiar manner in the embryos. An analysis of these centrosomes by Cnn immunolabeling showed that, in control embryos, the centrosome is present as a single focus at each of the spindle poles during metaphase of the first cleavage division but then splits into two adjacent foci as early as anaphase (Figure 2, A and B). In embryos of Elys mutant females in the first mitotic cycle, however, sister centrosomes were separate, giving rise to two discrete foci even when centrosomes were situated at the pole of the metaphase-like spindle ( Figure 2D). Remarkably, individualized centrosomes often detached from the spindle poles and were randomly located in the cytoplasm. We detected free asters with Cnn labeling in .70% of the embryos from both Elys 2 and Elys 5 females, whereas these were never seen in control embryos (Table 3). We observed up to four free asters within an embryo, indicative of arrest at the first cleavage division. When a spindle pole was devoid of centrosomes, the spindle appeared to be shorter in length and roundish ( Figure 2E). It is also noteworthy that, in some embryos from Elys mutant females, polar bodies anomalously formed bipolar spindles that lacked centrosomes ( Figure 2F; for control see Figure 2C), although their location within the embryo was substantively unaffected, lying near the cortex.
We reported previously that Nup160 sim induces maternal-effect lethality subsequent to sperm penetration in D. melanogaster (Sawamura et al. 2004), reminiscent of the above-mentioned embryonic phenotype that was due to the Elys mutations. Embryos from females hemizygous for Nup160 sim generally arrested their development in a metaphase-like state of the first mitotic cycle (Figure 2, G and H), as is the case with the embryos from Elys mutant females. Most (49/50) of the embryos had a total of two to four centrosome foci, whereas the one exception contained eight foci, which might have been attributable to another round of the centrosome cycle or the occurrence of dispermy (insemination by two sperm). Strikingly, Nup160 sim also caused abnormal centrosome behavior, which manifested as free asters in the cytoplasm in 75% (38/50) of the embryos. A noticeable difference between the effect of the Elys mutations and that of Nup160 sim could be discerned in the deformed mitotic figures that they exhibited. In the embryos of the Nup160 sim females, the union within the dual spindle was partially (12/ 49, Figure 2G) or thoroughly (24/49, Figure 2H) dissolved, resulting in two distinct spindles, each of a small size. In addition, unlike the Elys mutations, Nup160 sim did not affect microtubule configurations of the polar bodies ( Figure 2G). Taken together, both the Elys mutations and the Nup160 sim introgression commonly affected most, if not all, aspects of the first cleavage division, including mitotic centrosome behavior.

Synthetic lethality caused by Elys and Nups
Based on the phenotypic similarity between the Elys mutations and Nup160 sim introgression, we expected to find a genetic interaction between Elys and Nups. We thus made double mutants of D. melanogaster that carry an Elys mutation on the X chromosome and are hemizygous for either of nine autosomal Nup107-160 subcomplex genes. Elys/FM; +/+ females were crossed with +/Y; Df/Bal males, where Df and Bal stand for a Nup deficiency and a balancer, respectively (Table 4). Elys/Y; Bal/+ males were viable because the balancer contains the wild-type Nup + (control), but Elys/Y; Df/+ males, which carried only one dose of the n  Nup, were lethal (Nup96: viability, 0), semi-lethal (Nup160: viability, 0.01-0.04) or had low viability (Nup37: viability, 0.13-0.14). It must be stressed here that the lethality caused by the Elys mutations or the Nup160 sim introgression is maternal but the synthetic lethality caused by Elys and Nups double mutants is zygotic. Even in the last case (Nup37), most of the Elys/Y; Df/+ males died during or just after emergence: 88.9% (24/27) in Elys 2 and 82.9% (29/35) in Elys 5 . The lethality of Elys/Y; Df/+ males was confirmed by using additional Nup37 deficiencies (Table S3; viability, 0.01-0.18). An exception is Df(3R)ED10946 (viability, 1.02), but we suspect that this deficiency differs from the computational prediction and does not delete Nup37; in fact, Df(3R)ED10946 was viable, although the other deficiencies were lethal, when they were made transheterozygous against Df(3R)ED10953. Thus, three of the nine genes (Nup37, Nup96 and Nup160) exhibited haploinsufficiency (e.g., hemizygous lethal) in the genetic background of the Elys mutations. The lethal stage of the Elys/Y; Df/+ males was late pupal in Nup160 and Nup96 ( Figure S2, Table S4 and Table S5; for the staging see Bainbridge and Bownes (1981)). We also determined that the lethality of the Elys and Nup double mutants is not sex-specific. Not only Elys/Y; Df/+ males but also Elys/Elys; Df/+ females were lethal when Nup160 was made hemizygous (Table S6).
Not only Nup96 and Nup160 but also Nup37 may cause hybrid lethality In the cross between D. melanogaster females and D. simulans males, hybrid males are lethal but are rescued by the Lhr (Lethal hybrid rescue) mutation of D. simulans (Sturtevant 1920;Watanabe 1979). When Nup96 sim or Nup160 sim is made hemizygous by a deficiency chromosome of D. melanogaster or made homozygous by an introgression from D. simulans, the hybrid males cannot be rescued by D. simulans Lhr (Presgraves et al. 2003;Tang and Presgraves 2009;Sawamura et al. 2010). This is because Nup96 sim and Nup160 sim behave as recessive hybrid incompatibility genes (Strategy 2 of Sawamura 2016). In other words, a gene or genes from D. melanogaster (incompatibility partner) result in hybrid inviability in the genetic background of Nup96 sim or Nup160 sim homozygote (or hemizygote). We reported above that not only Nup96 and Nup160 but also Nup37 exhibited haploinsufficiency in the genetic background of the Elys mutations. This raises the possibility that Nup37 is also a gene for hybrid incompatibility. We thus made crosses by using deficiency chromosomes that lack Nup37. The interspecific crosses were very difficult, presumably because the deficiencies affect mating behavior; the hemizygotes exhibited the Minute phenotype resulting from the haploinsufficiency of closely linked RpS27 (Ribosome protein S27; Marygold et al. 2007). Crossing was successful only when Df(3R) ED10953 was used, and the male hybrids hemizygous for Nup37 sim were not rescued by Lhr (Table S7), although we cannot rule out the possibility that the lethality is a secondary effect of RpS27. Thus, not only Nup96 and Nup160 but also Nup37 may be hybrid incompatibility genes.

Adaptive evolution of Elys in Drosophila
Hybrid incompatibility genes generally evolve rapidly (Ting et al. 1998;Barbash et al., 2003;Presgraves et al. 2003;Brideau et al. 2006;Tang and Presgraves 2009). We thus compared the Elys gene sequences of D. melanogaster and D. simulans. Although K a /K s = 0.53 when the entire coding sequence was used, the sliding window analysis indicated positive natural selection (K a /K s . 1) around the ELYS-like domain and the Glu-rich domain of the gene ( Figure 3A). In fact, K a /K s = 1.51 and 1.10 for these two domains, respectively, even though the Glu-rich domain is 49 aa shorter in D. simulans. The sequences of D. yakuba and D. sechellia were added to the comparison of the ELYS-like domain, and amino acid replacements and synonymous substitutions were counted in each branch of the phylogenetic tree ( Figure 3B). Positive natural selection seems to have occurred on the route from node 2 (the common ancestor of D. melanogaster and D. simulans) to D. simulans, as indicated by the 26 replacements vs.. 3 synonymous substitutions. This was confirmed by the branch model of PAML; not significant for the fulllength Elys sequences but significant for the ELYS-like domain (P = 0.008 for the D. simulans branch after sprit from D. melanogaster and P = 0.048 for D. simulans branch after the sprit from D. sechellia).

DISCUSSION
ELYS function in D. melanogaster ELYS plays an important role in the NPC assembly, as noted above. Therefore, it was a surprise that Elys is dispensable for viability and male fertility in D. melanogaster (Figure 1 and Table S1). D. melanogaster might have another gene or genes, the function of which is redundant with Elys, although we have not found genes with sequence similarity. Similar to mutations in the C. elegans homolog, mel-28 (Fernandez et al. 2014;Gómez-Saldivar et al. 2016), D. melanogaster Elys exhibited a maternal effect (Table 1). Females mutant for the gene produced apparently normal eggs in which sperm can penetrate ( Figure  S1 and Table S2), but the development of the resulting embryos never progressed beyond the first mitotic division (Figure 2 and Table 2).
In the present study, we carefully examined the maternal effect of the Elys mutations (Table 3) and Nup160 sim introgression in early Drosophila embryos and showed that they share the embryonic phenotype of developmental arrest in a metaphase-like state of the first cleavage division. Therefore, the Nup160 sim introgression in D. melanogaster appears to behave like a loss-of-function allele of Elys. The prior steps of fertilization, such as the establishment of the sperm aster and pronuclear apposition, were unaffected, and no figures showing anaphase of the first cleavage division or later were observed. In these embryos, abnormally individualized centrosomes and their dissociation from the spindle poles were obvious, implying ELYS and Nup160 in mitotic centrosome behavior. Consistently, a proteomic analysis of Drosophila embryonic centrosomes shows that ELYS is actually a centrosome component (see Table S1 of Müller et al. 2010), although its function has not yet been established. Centrosomal localization of Nup160 is unknown in Drosophila, but the protein has been detected in spindle poles and proximal spindle fibers of HeLa cells .
The developmental arrest could be accounted for by failure in structural changes of the nuclear envelope during the semi-open mitosis of early Drosophila embryos and/or disrupted interactions between the kinetochore and microtubules (Güttinger et al. 2009). Both ELYS and the Nup107-160 subcomplex can be detected in an interdependent manner at spindle poles and kinetochores (Zierhut and Funabiki 2015). Also, the halting of mitotic progression could reflect the abnormal persistence of spindle-associated Cyclin B owing primarily to the dissociation of centrosomes from spindle poles, as the polar localization of centrosomes is required to initiate local destruction of Cyclin B in mitotic spindles of the Drosophila syncytium (Huang and Raff 1999;Wakefield et al. 2000). The fact that the Elys mutations and the Nup160 sim introgression result in very different outcomes with respect to the deformed morphology of the first mitotic spindle suggests that the ELYS and Nup160 proteins may have both common and distinct roles in the spindle assembly characteristic of the first cleavage division.
The present cytological study clearly demonstrates that ELYS and Nup160 are commonly involved, at a minimum, in centrosome behavior during the first cleavage division. Studies on subcellular localization of n  a Males were crossed with Elys/FM7c females. The replicates that produced maternal nondisjunctional flies at high frequency were not included in the data, because some of the mothers must have been XXY. b Full genotypes are available upon request. Df(3R)ED10953 exhibits a slight Minute phenotype, because the Nup37 locus is close to RpS27.
c Presumably produced by the break in FM7c (see Hutter 1990). d The presence of the chromosome 2 balancer could not be determined.
e Calculated as (class II · class III)/(class I · class IV).
g One was a gynandromorph.
h One was apparently paternal nondisjunctional XO.
i Two were apparently paternal nondisjunctional XO.
the ELYS and Nup160 proteins and their protein-protein interactions are needed to further elucidate their functions. Because ELYS determines the subcellular localization of the Nup107-160 subcomplex (Belgareh et al. 2001;Boehmer et al. 2003;Harel et al. 2003;Walther et al. 2003;Loïodice et al. 2004;Franz et al. 2007;Gillespie et al. 2007;Rasala et al. 2006Rasala et al. , 2008Doucet et al. 2010;Bilokapic and Schwartz 2013;Inoue and Zhang 2014), we expected genetic interaction between Elys and Nups. Among the nine Nup107-160 subcomplex components examined, Nup37, Nup96 and Nup160 indeed exhibited haploinsufficiency in the genetic background of the Elys mutations (Table  4, Table S3 and Table S6); Elys/Y; Df/+ males were lethal at the pupal stage ( Figure S2, Table S4 and Table S5). Interestingly, those three Nups are located in close proximity in the NPC (see Figure 1 of Hurt and Beck 2015). Furthermore, Bilokapic and Schwartz (2012) have suggested that ELYS binds near an interface of the subcomplex consisting of Nup120 (the yeast homolog of Nup160) and Nup37 in Schizosaccharomyces pombe. This might cause the epistatic interaction detected in the present analysis. Notably, the effect of Elys mutations and Nup160 sim introgression is different than that of double mutations of Elys and Nups; the former survived to adulthood on their own and the lethality was only revealed as maternal effect while the latter exhibited a strong zygotic phenotype. These results suggest that ELYS and Nups may act at the same component of the mitotic machinery, or at another unidentified biological process, resulting in more severe synthetic lethal interactions.
Although ELYS sequences are well conserved in metazoans (Rasala et al. 2006), our present analysis detected positive natural selection in the ELYS-like domain of the protein in the branch leading to D. simulans (Figure 3 and Table S8). This might be the consequence of coevolution between ELYS and Nups. Indeed, recurrent adaptive evolution has been detected in five Nup107-160 subcomplex components (Nup75, Nup96, Nup107, Nup133 and Nup160) and two mobile Nups (Nup98 and Nup153) in D. melanogaster and D. simulans (Presgraves et al. 2003;Presgraves and Stephan 2007;Tang and Presgraves 2009).

Possible involvement of ELYS in reproductive isolation
Several genes responsible for hybrid lethality between D. melanogaster and D. simulans have been identified (for recent reviews, see Sawamura 2016; Castillo and Barbash 2017). Lhr and Hmr (Hybrid male rescue), which encode chromatin binding proteins, are one such incompatibility pair (Watanabe 1979;Hutter and Ashburner 1987;Barbash et al. 2003;Brideau et al. 2006;Thomae et al. 2013;Blum et al. 2017), and gfzf (GST-containing FLYWCH zinc-finger protein) is an upstream gene in this incompatibility (Phadnis et al. 2015).
Nup96 and Nup160 are also involved in reproductive isolation (Presgraves et al. 2003;Tang and Presgraves 2009;Sawamura et al. 2010). Nup96 sim and Nup160 sim synergistically cause hybrid incompatibility (Sawamura et al. 2014), but the D. melanogaster alleles of Nup160 and Nup96 are not the dominant autosomal incompatibility partner of Nup96 sim and Nup160 sim , respectively (Tang and Presgraves 2015). Then, what is (are) the incompatibility partner(s) of Nup96 sim and Nup160 sim ? One can envision that at least one recessive gene must be located on the X chromosome of D. melanogaster (X mel ), because the hybrid inviability is revealed in X mel Y sim but not in X mel X sim , where Y sim and X sim stand for the Y and X chromosomes of D. simulans, respectively (Strategy 2 of Sawamura 2016). We here propose that the X-linked Elys of D. melanogaster may be the incompatibility partner of Nup96 sim and Nup160 sim .
Our proposal is based on three observations. (1) Elys mutations mimic the maternal Nup160 sim introgression phenotype in D. melanogaster (Figure 2), which suggests that Elys affects the same cascade as the Nup160 sim incompatibility. (2) Epistatic interaction was detected between Elys and Nup37, Nup96 or Nup160 in D. melanogaster (Table 4). (3) Male hybrids between D. melanogaster and D. simulans cannot be rescued by the Lhr mutation if Nup37, Nup96 or Nup160 of D. melanogaster is deficient (Table S7; Presgraves et al. 2003;Tang and Presgraves 2009;Sawamura et al. 2010).
In this model we presume that D. melanogaster ELYS does not function properly-and thus NPC formation and mitotic centrosome behavior are compromised-if Nup37, Nup96 or Nup160 is from D. simulans. We must also note that the incompatible D. simulans allele of the Nup107-160 subcomplex genes is recessive; the presence of the D. melanogaster allele is enough to avoid incompatibility. Thus, hemizygous Nup160 sim introgression causes female sterility (maternal-effect lethality) with a phenotype that is similar to the Elys mutations of D. melanogaster (Figure 2). But Nup96 sim introgression does not cause female sterility (Sawamura et al. 2014) and Nup37 sim has not been tested.
Recently, rhi (rhino) and del (deadlock), which encode piRNA pathway proteins, were shown to be another incompatibility pair (Parhad et al. 2017). This pathway might have been adapted to suppress the species-specific transposable element mobilization (Kelleher et al. 2012;Parhad et al. 2017). ELYS plays an important role in the piRNA pathway; PIWI is released from messenger ribonucleoprotein particles by binding to NPCs via Xmas-2, ELYS and other NPC components (Ilyin et al. 2017). The piRNA pathway evolution might result in the incompatibility between Elys and Nups.
Thus, Elys is a candidate for a gene of reproductive isolation between D. melanogaster and D. simulans, but direct evidence is necessary. We are going to test the viability and female fertility of flies (D. melanogaster or the D. melanogaster/D. simulans hybrid) that carry various combinations of Elys and Nup alleles.