Testis- and ovary-expressed polo transcripts and gene duplications affect male fertility when expressed in the germline

Abstract

Polo-like kinases (Plks) are essential for spindle attachment to the kinetochore during prophase and the subsequent dissociation after anaphase in both mitosis and meiosis. There are structural differences in the spindle apparatus between mitosis, male meiosis, and female meiosis. It is therefore possible that alleles of Plk genes could improve kinetochore attachment or dissociation in spermatogenesis or oogenesis, but not both. These opposing effects could result in sexually antagonistic selection at Plk loci. In addition, Plk genes have been independently duplicated in many different evolutionary lineages within animals. This raises the possibility that Plk gene duplication may resolve sexual conflicts over mitotic and meiotic functions. We investigated this hypothesis by comparing the evolution, gene expression, and functional effects of the single Plk gene in Drosophila melanogaster (polo) and the duplicated Plks in Drosophila pseudoobscura (Dpse-polo and Dpse-polo-dup1). We found that the protein-coding sequence of Dpse-polo-dup1 is evolving significantly faster than a canonical polo gene across all functional domains, yet the essential structure of encoded protein appears to be retained. Dpse-polo-dup1 is expressed primarily in testis, while other polo genes have broader expression profiles. Furthermore, over or ectopic expression of polo or Dpse-polo in the D. melanogaster male germline results in greater male infertility than ectopic expression of Dpse-polo-dup1. Lastly, ectopic expression of Dpse-polo or an ovary-derived transcript of polo in the male germline causes males to sire female-biased broods. However, there is no sex-bias in the progeny when Dpse-polo-dup1 is ectopically expressed or a testis-derived transcript of polo is overexpressed in the D. melanogaster male germline. Our results therefore suggest that Dpse-polo-dup1 may have experienced positive selection to improve its regulation of the male meiotic spindle, resolving sexual conflict over meiotic Plk functions. Alternatively, Dpse-polo-dup1 may encode a hypomorphic Plk that has reduced deleterious effects when overexpressed in the male germline. Similarly, testis transcripts of D. melanogaster polo may be optimized for regulating the male meiotic spindle, and we provide evidence that the untranslated regions of the polo transcript may be involved in sex-specific germline functions.

Introduction

Gametogenesis in animals is sexually dimorphic. Sex differences in gametogenesis start with the establishment of the germline, continue through meiosis, and conclude with sexually dimorphic sperm and eggs (Fuller and Spradling 2007; Whitworth et al. 2012; Lehtonen et al. 2016; Cahoon and Libuda 2019). Meiosis, a central process of gametogenesis, is highly differentiated between the sexes (Hua and Liu 2021). Male meiosis starts with a single diploid cell and produces four haploid sperm; in contrast, female meiosis produces a single haploid egg and two polar bodies from a diploid precursor (Evans and Robinson 2011; McKee et al. 2012). There are additional sex differences in the meiotic spindle apparatus, meiotic chromatin, chromosomal pairing, and recombination rates (Orr-Weaver 1995; McKee 1996; Sardell and Kirkpatrick 2020).

Inter-sexual differences in gametogenesis create numerous opportunities for intragenomic and intersexual conflicts (Rice 2013; Arnqvist and Rowe 2013). For example, one allele of a gene may improve some aspect of spermatogenesis, while negatively affecting oogenesis, and vice versa for the alternative allele (VanKuren and Long 2018; Hamada et al. 2020). This type of intralocus intersexual conflict (or sexual antagonism) may be resolved by gene duplication, followed by specialization (or subfunctionalization) of one copy for spermatogenesis or gametogenesis (Tracy et al. 2010; Connallon and Clark 2011; Gallach and Betrán 2011). Such germline-specific sexual subfunctionalization may be common for genes involved sex-specific or sexually dimorphic aspects of meiosis (Reis et al. 2011).

Intersexual conflicts likely arise because of differences between the mitotic, female meiotic, and male meiotic spindle apparatus (Orr-Weaver 1995; Savoian and Glover 2014). Despite the differences across mitotic and meiotic spindles, many genes encode proteins that are required for the mitotic, female meiotic, and male meiotic spindles. For example, the Drosophila melanogaster gene mad2 encodes a protein involved in the mitotic and meiotic spindle assembly checkpoint (Li and Murray 1991; Shah and Cleveland 2000; Nicklas et al. 2001; Tsuchiya et al. 2011). In the lineage leading to Drosophila pseudoobscura, mad2 was duplicated, and each copy may have evolved a specialized meiotic function in either males or females (Meisel et al. 2010). It is possible that sex-specific subfunctionalization of each paralog resolved an intersexual conflict that arose because of sexually dimorphic meiotic spindles.

However, there has yet to be a direct test of the hypothesis that sex differences in the meiotic spindle create sexual antagonism.

Here, we use the Drosophila gene polo as a model to explore intersexual conflicts that arise as a result of the sexually dimorphic meiotic spindle aparatus. Polo-like kinases (Plks) are essential regulators of both mitosis and meiosis across eukaryotes (Archambault and Glover 2009). Specifically, Plks are required for spindle attachment to the kinetochore during prophase and the subsequent dissociation after anaphase (Sunkel and Glover 1988; Llamazares et al. 1991; Donaldson et al. 2001). The D. melanogaster genome has a single Plk gene (polo), which is necessary for chromosome segregation during meiosis in both oogenesis and spermatogenesis (Sunkel and Glover 1988; Carmena et al. 1998; Herrmann et al. 1998; Das et al. 2016). Loss of function polo mutations affect oogenesis and early embryogenesis—from oocyte determination through meiosis and into the establishment of the embryonic sperm aster (Sunkel and Glover 1988; Tavares et al. 1996; Riparbelli et al. 2000; Mirouse et al. 2006). Polo is similarly required for meiotic chromosome segregation during spermatogenesis; males with polo mutations experience high rates of nondisjunction and produce sperm with abnormal DNA content, likely because Polo is involved in the attachment of kinetochores to the spindle apparatus (Sunkel and Glover 1988; Carmena et al. 1998, 2014; Herrmann et al. 1998). Given the differences in meiotic spindles between male and female Drosophila (Orr-Weaver 1995), it is possible that polo alleles may have sexually antagonistic effects if they improve kinetochore attachment and dissolution in spermatogenesis or oogenesis, but not both.

The location of forward (F) and reverse (R) primers used to clone cDNA from testis (poloT) and ovary (poloO) transcripts are shown. B. The mRNA of the alternative polo transcripts are diagrammed. The regions cloned from testis and ovary are shown above the polo-RB transcript.

Alternative splicing of polo may be one mechanism for the resolution of sexual conflict over male and female meiotic benefits. There are two polyadenylation (pA) sites within the 3’ untranslated region (UTR) of D. melanogaster polo (Figure 1); the two polo mRNA products differ in their effects on kinetochore function, pupal metamorphosis, and female fertility, possibly because of differences in translational efficiencies between transcripts with different pA sites (Llamazares et al. 1991; Pinto et al. 2011; Oliveira et al. 2019). Mature transcripts with the proximal pA site also have a shorter 5’-UTR (Hoskins et al. 2015), suggesting differences in transcription initiation sites between the mRNAs that differ in their pA site (Figure 1A). However, functional effects of the polo 5’-UTR have not yet been identified, and both the short and long mRNA variants encode the same protein. In addition, despite the differences between the two alternative polo transcripts, it is not yet known if they have sexually antagonistic effects.

Figure 1.

D. melanogaster polo is spliced into two different isoforms that differ in their untranslated regions (UTRs). The location of a conserved upstream sequence element (USE) that affects usage of the polo-RB polyadenylation (pA) site is shown by a blue line (Oliveira et al. 2019). A cryptic pA signal, used when the USE is mutated, is shown by a red line. A. The polo gene region is diagrammed, with two alternative transcripts of polo shown (polo-RA and polo-RB). Thick black boxes show protein coding exons. Medium gray bars show the 5’-UTR (left) and 3’-UTR (right). Introns are shown as thin black lines.

Gene duplication may be another way of resolving intersexual conflicts involving Plks. Plk genes have been independently duplicated multiple times during the evolution of metazoan animals, with paralogs specialized for different functions (Habedanck et al. 2005; Bettencourt-Dias et al. 2005). While D. melanogaster has a single Plk gene (polo), the D. pseudoobscura genome harbors two duplications (three total copies) of polo (Reis et al. 2011). In addition, D. melanogaster polo is autosomal (on chromosome 3L, or Drosophila Muller element D), but element D fused to the X chromosome in the lineage leading to D. pseudoobscura. Therefore, the D. pseudoobscura ortholog of polo (Dpse-polo) is on a neo-X chromosome. An excess of genes was duplicated from the D. pseudoobscura neo-X chromosome to the autosomes (Meisel et al. 2009), including polo (Reis et al. 2011). The same autosome independently became a neo-X chromosome in D. willistoni, and mtrm (a key interactor of polo) was similarly duplicated from the neo-X onto an autosome in D. willistoni (Xiang et al. 2007; Reis et al. 2011; Whitfield et al. 2013; Bonner et al. 2020). X-to-autosome duplications may be involved in the resolution of intersexual conflict if X-linkage is unfavorable for genes with male-specific functions, possibly because of downregulated X chromosome expression in the male germline (Betrán et al. 2002; Wu and Xu 2003; Emerson et al. 2004; Meisel et al. 2010). The two duplicate copies of polo (polo-dup1 and polo-dup2) are expressed primarily in males in Drosophila persimilis (the sibling species of D. pseudoobscura), while the ancestral copy of polo is expressed in both sexes (Reis et al. 2011). The divergence in expression between polo paralogs is consistent with sex-specific subfunctionalization of a duplicated gene to resolve an intersexual conflict (Gallach and Betrán 2011). Notably, polo-dup2 has a truncated protein coding sequence (it is missing the last third of the coding region), while polo-dup1 is predicted to encode a complete Plk (Reis et al. 2011). This suggests polo-dup1 may have been retained to resolve an intersexual conflict, while polo-dup2 may be a pseudogene. Curiously, mtrm appears to have evolved under positive selection (Anderson et al. 2009), and there is evidence for divergence of Mtrm function in female meiosis across the Drosophila genus (Bonner and Hawley 2019). The evolutionary dynamics of polo and mtrm are therefore consistent with selection in response to intersexual conflicts, possibly as a result of sex differences in the meiotic spindle or kinetochore.

We evaluated if polo has sexually antagonistic effects in Drosophila, and we also explored if that conflict was subsequently resolved by testis-specific specialization of a polo gene duplication. To those ends, we examined the evolution and expression of Dpse-polo and a complete duplication of polo in the D. pseudoobscura genome (Dpse-polo-dup1). We also cloned polo transcripts into vectors for the GAL4>UAS binary expression system, and we determined the effect of driving their expression in the D. melanogaster male germline. We sampled polo transcripts from both the male and female D. melanogaster germline, in addition to Dpse-polo and Dpse-polo-dup1. We tested if expressing these different polo transgenes in the male germline affects male fertility and the sex ratio of progeny sired by these males.

Materials and Methods

Evolution of Polo protein sequences

We tested for differences in the rates of evolution of the protein sequences encoded by Dpse-polo and Dpse-polo-dup1. A previous analysis found that the nucleotide sequence of Dpse-polo-dup1 evolves faster than Dpse-polo (Reis et al. 2011), but the rate of amino acid evolution was not directly examined. To address that shortcoming, we constructed an alignment of Dpse-Polo (XM_001353282), Dpse-Polo-dup1 (XM_002132425), and D. melanogaster Polo (FBtr0074839) using MUSCLE implemented in MEGA 11 for macOS with the default parameters (Edgar 2004; Stecher et al. 2020; Tamura et al. 2021). The alignment is available as Supplemental Material. We then used Tajima’s (1993) relative rate test to compare the number of amino acid substitutions in the evolutionary lineages leading to Dpse-Polo and Dpse-Polo-dup1, treating D. melanogaster Polo as the outgroup. We also compared the number of amino acid substitutions within the N-terminal serine/threonine kinase domain, the Polo box domain (PBD), the two individual Polo boxes (PB1 and PB2), and the linker between the kinase domain and PBD.

D. pseudoobscura polo expression

We compared the expression of Dpse-polo and Dpse-polo-dup1 in males and females across seven different D. pseudoobscura tissue samples. We first obtained normalized read counts (NRC) for all D. pseudoobscura genes from an RNA-seq data set in which expression was measured in four replicates from each sex for seven different tissue samples (GSE99574; Yang et al. 2018). We calculated the median NRC for each gene across all four replicates for each tissue-by-sex combination (NRC_TS), and then we analyzed log₁₀(NRC_TS + 1). We added one to each NRC_TS value to ensure that all values were finite (because some NRC_TS values were equal to zero). We compared log₁₀(NRC_TS + 1) of Dpse-polo (FBgn0071596) and Dpse-polo-dup1 (FBgn0246554) to the genome-wide distribution of log₁₀(NRC_TS + 1) values to evaluate the relative expression of each polo gene in teach tissue-by-sex combination.

We used the same RNA-seq data to calculate the breadth of expression (τ) across tissues for Dpse-polo and Dpse-polo-dup1 (Yanai et al. 2005). In our calculation of τ, we included six non-overlapping tissue samples: 1) digestive plus excretory system; 2) gonad; 3) reproductive system without gonad; 4) thorax without digestive system; 5) abdomen without digestive or reproductive system; and 6) head. We calculated τ with the following equation: In this equation, expression of a gene in N=6 tissues is measured as log₁₀(S_i +1), where S_i is the NRC_TS in tissue i for a given sex. S_max is the maximum S_i of the gene across all six tissue samples in a given sex. Values of τ range from 0 (equal expression in all tissues, i.e., broadly expressed) to 1 (expressed in a single tissue, i.e., narrowly expressed). We calculated τ separately for male and female tissue samples.

Creating transgenic D. melanogaster carrying inducible polo transcripts

We cloned Plk transcripts from D. melanogaster testes, D. melanogaster ovaries, and whole D. pseudoobscura males. D. melanogaster testis and ovary tissues were dissected in Ringer’s solution from whole flies of the iso-1 strain (BDSC 2057). Ovaries and testes were dissolved overnight in TRI Reagent® on a rocker. Whole D. pseudoobscura males (from the MV2-25 strain) were ground in TRI Reagent® with a motorized pestle and centrifuged to remove particulates. We used the Direct-zol RNA Purification Kit (Zymo Research) to isolate RNA from each sample, following the manufacturer’s instructions.

The resultant RNA samples were used as a templates in a reverse transcription reaction (RT-PCR) with primers targeting polo (D. melanogaster testis or ovary), Dpse-polo (D. pseudoobscura males), or Dpse-polo-dup1 (D. pseudoobscura males) using SuperScript™ III reverse transcriptase (Thermo Fisher Scientific). Different primer pairs were used to amplify polo from D. melanogaster ovaries (poloO) and testes (poloT) because the primers for one tissue sample would not amplify the transcript from the other tissue sample. Each of the four cDNA products was then used as a template in a PCR with the same primers and Phusion® High Fidelity DNA Polymerase (New England Biolabs). All primer pairs were located within the 5’- and 3’-UTRs of the transcripts so that they amplified the entire protein coding sequence of the respective genes (Figure 1; Supplemental Table S1). A “CACC” adapter sequence was included at the 5’ end of each forward primer to allow the PCR products to be cloned into a Gateway™ Entry vector.

We used the Gateway™ System to clone each PCR product into a vector that could be used for germline transformation of D. melanogaster. We first used the pENTR™/D-TOPO™ Cloning Kit to create Gateway™ Entry clones for each of the four PCR products, which we transformed into One Shot™ TOP10 Chemically Competent Escherichia coli cells (Thermo Fisher Scientific). We then isolated plasmids from all four cloning products with the Invitrogen™ PureLink™ Quick Plasmid Miniprep kit. We confirmed the correct insert size using PCR with the M13 primer pair. We next used the Gateway™ LR Clonase™ II Enzyme mix to recombine each of the four PCR products into the pBID-UASC-G backbone (Addgene Plasmid #35202), which contains a φC31 integrase compatible attB sequence and UAS binding sites for the GAL4 expression system (Wang et al. 2012). We transformed One Shot™ TOP10 Chemically Competent E. coli cells with each of the four recombinant plasmids. We designed primers to amplify the inserts within the pBID-UASC-G plasmid (5’-TGCCGCTGCCTTCGTTAATA-3’ and 5’-TTCCACCACTGCTCCCATTC-3’), and we confirmed that the inserts were the correct size.

We also used Sanger sequencing of the PCR products to confirm that there were no DNA sequence errors in the resulting amplifications. We finally used the Invitrogen™ PureLink™ HiPure Plasmid Filter Midiprep Kit to isolate plasmids containing each of the four PCR products.

We created transgenic D. melanogaster that carry one of each of the four recombinant plasmids. Each of the four plasmids was injected into D. melanogaster strain VK20 (BDSC 9738), which has an attP docking site at region 99F8 of chromosome 3R. All injections were performed by GenetiVision Corporation. We confirmed successful transformation via the presence of orange eyes. We balanced the third chromosome carrying each of the transgenes over a TM3,Sb chromosome. Each of these strains has the genotype UAS-poloX/TM3,Sb, where poloX refers to the specific Plk transcript (PoloO, PoloT, Dpse-polo, or Dpse-polo-dup1).

We created at least one (and no more than three) balanced strains for each transgene, with each strain originating from a different transformed founder (Supplemental Table S2).

Assaying effects of polo transcripts on male fertility and progeny sex ratios

We tested if male germline expression of each of the four polo transcripts affects male fertility and sex chromosome transmission. Males with the UAS-poloX/TM3,Sb genotype were mated to females carrying a Gal4 driver construct that is expressed under the bag of marbles (bam) promoter (P{bam-Gal4-VP16}), which drives expression in the male germline (Chen and McKearin 2003; Sartain et al. 2011; Hart et al. 2018). After mating, all flies, eggs, and larvae were kept in 25x95mm vials containing cornmeal media in 25°C incubators with 12:12 light:dark cycles. Male progeny with the P{bam-Gal4-VP16}>UAS-poloX genotype were identified by wild-type bristles.

We assayed male fertility by allowing P{bam-Gal4-VP16}>UAS-poloX males to mate with wild type females from the Canton S (CanS) and Oregon R (OreR) strains. A single male and single female were combined in a 25x95mm vial with cornmeal media at 25°C, and they were observed to confirm successful copulation, as we have done previously with crosses using the same P{bam-Gal4-VP16} strain (Hart et al. 2018). After mating, the male was removed from the vial, and the female was allowed to lay eggs for 3–5 days at 25°C. The vials were stored at 25°C, and we counted the number of male and female progeny that emerged in each vial for 21 days after mating.

We tested for an effect of germline expression of each transgene on the number of progeny using mixed effect linear models. Our analysis compared the effects of UAS-PoloO, UAS-PoloT, UAS-Dpse-polo, and UAS-Dpse-polo-dup1. We analyzed all strains with the same transgene within a single model, treating strain as a random effect. For each comparison, we used the lme() function within the nlme package in R (Pinheiro and Bates 2000; Pinheiro et al. 2023) to construct a linear model with the number of progeny in a vial as a response variable, transgene as a fixed effect, and batch and strain as random effects (see Supplemental Material for R code). We tested for an effect of each transgene by separately analyzing the total number of progeny per vial, the number of male progeny, or the number of female progeny.

We also used a mixed effect logistic regression to test if the transgenes affected whether a male sires any offspring. As above, we compared the effects of UAS-PoloO, UAS-PoloT, UAS-Dpse-polo, and UAS-Dpse-polo-dup1, including all strains with the same transgene in a single model. For each comparison, we performed a logistic regression using the glmer() function in the lme4 package (Bates et al. 2015) to construct a model with whether a male sired progeny as a response variable (0=no, 1=yes), transgene as a fixed effect, and batch and strain as random effects (see Supplemental Material for R code). We tested for an effect of each transgene by separately analyzing if any progeny were sired, if male progeny were sired, or if female progeny were sired.

We additionally tested for differences in the sex ratio (relative numbers of male and female progeny) using mixed effect linear models. As above, we analyzed all strains with the same transgene within a single model. For each transgene, we used the lme() function in the nlme package (Pinheiro and Bates 2000; Pinheiro et al. 2023) to construct a linear model with the number of progeny as response variable, progeny sex (male or female) and vial as fixed effects, and batch and strain as random effects (see Supplemental Material for R code). We conclude that a transgene affects the sex ratio when progeny sex has a significant effect on the number of progeny.

Results

Accelerated evolution of Dpse-polo-dup1

We compared the number of amino acid substitutions in Dpse-polo and Dpse-polo-dup1 to test for accelerated evolution along the lineage leading to Dpse-polo-dup1 (Supplemental Table S3). There were significantly more amino acid substitutions in the lineage leading to Dpse-polo-dup1 than Dpse-polo (χ₁²=71.43, p<0.00001), consistent with the previously described faster evolution in the nucleotide sequence of Dpse-polo-dup1 (Reis et al. 2011). Of the 567 alignable amino acid positions, 80 residues (14%) were estimated to be divergent along the lineage leading to Dpse-polo-dup1. In contrast, only three amino acid substitutions were identified along the lineage leading to Dpse-polo.

We next explored amino acid divergence along the lineage leading to Dpse-polo and Dpse-polo-dup1 across the different domains of the Polo protein. Plks consist of an N-terminal serine/threonine kinase domain and a C-terminal Polo-box domain (PBD), separated by a linker. Both the kinase domain and PBD are present without any insertions or deletions in both Dpse-polo and Dpse-polo-dup1. The PBD can be further divided into Polo-box 1 (PB1) and Polo-box 2 (PB2), and there are two amino acids (histidine at position 518, and lysine at position 520) that are required to bind Polo targets (Elia et al. 2003a; b). Both residues are conserved in Dpse-polo and Dpse-polo-dup1. There were nine amino acids deleted in Dpse-polo-dup1 (out of a total of 576 codons in D. melanogaster polo), and all nine are located in the linker (Supplemental Material). One of those amino acids was also deleted in Dpse-polo. Despite the structural conservation of Dpse-polo-dup1, there were significantly more amino acid substitutions in the kinase domain, PBD, and linker of Dpse-polo-dup1, relative to Dpse-polo (Figure 2; Supplemental Table S3). Therefore, there is a consistent signal of faster amino acid evolution in Dpse-polo-dup1, yet the overall structure of Polo is conserved in both Dpse-polo and Dpse-polo-dup1.

Figure 2.

Accelerated protein coding divergence in Dpse-polo-dup1 relative to Dpse-polo. Each bar shows the percent of amino acids with a substitution along the lineage leading to either Dpse-polo (blue) or Dpse-polo-dup1 (orange), out of all alignable sites. Divergence was calculated within the N-terminal serine/threonine kinase domain (kinase), the C-terminal Polo-box domain (PBD), and the linker. Counts of amino acid substitutions are provided in Supplemental Table S3.

Dpse-polo-dup1 is highly expressed in male reproductive tissues

We tested if Dpse-polo-dup1 has male-biased expression, which would be consistent with the expression of polo-dup1 in D. persimilis (Reis et al. 2011). To those ends, we used available RNA-seq data to compare the expression of Dpse-polo and Dpse-polo-dup1 across seven different tissue samples in both males and females (Figure 3A). In each tissue sample, we observed a bimodal distribution of genome-wide expression levels, with one distribution centered close to zero (low expressed genes), and another distribution centered ∼2 orders of magnitude higher (highly expressed genes). In all sex-by-tissue combinations, Dpse-polo was expressed at a level within the distribution of highly expressed genes. In contrast, Dpse-polo-dup1 was not expressed or expressed at a low level across all female tissue samples and most male samples. The notable exceptions were male samples that included reproductive tissues (whole body, reproductive system, and gonad), in which Dpse-polo-dup1 was highly expressed, similar to Dpse-polo. The highest expression of Dpse-polo-dup1 was in testis.

Figure 3.

(A) Expression of Dpse-polo and Dpse-polo-dup1 across seven different tissue samples in males and females. The X-axis shows the log₁₀ of the median normalized expression (GSE99574; Yang et al. 2018). Each distribution shows the expression level of all genes in a given sex-by-tissue combination. The circles show the expression of Dpse-polo and the triangles show the expression level of Dpse-polo-dup1 in each sample type. Tissue samples are whole body, thorax (with digestive system removed), head, digestive system (including excretory system), abdomen (with digestive and reproductive system removed), reproductive system (without gonad), and gonad (ovary or testis). (B) The distribution of expression breadth (τ) of all genes across six unique tissue samples (excluding whole body) in females or males is plotted. The vertical line segments within each distribution show the median value. The circles show the expression breadth of Dpse-polo and the triangles show the expression breadth of Dpse-polo-dup1.

We quantified the expression breadth of Dpse-polo and Dpse-polo-dup1 using τ, which ranges from 0 (equally expressed in all tissues) to 1 (only expressed in a single tissue).

Dpse-polo had a similar expression breadth in both females (τ = 0.59) and males (τ = 0.56), which was larger than the median τ across the genome (Figure 3B). The high τ of Dpse-polo could be attributed to elevated expression in gonads relative to other tissue samples, but Dpse-polo was highly expressed across all tissues (Figure 3A). Surprisingly, Dpse-polo-dup1 had the maximal τ value of 1 when expression was measured in females (Figure 3B). This is because expression was only detected in the ovary, yet Dpse-polo-dup1 is expressed at a very low level in ovary (Figure 3A). In males, Dpse-polo-dup1 had substantially more tissue-specific expression (τ = 0.77) than Dpse-polo, and this was caused by extremely high expression of Dpse-polo-dup1 in testis (Figure 3). We therefore conclude that Dpse-polo-dup1 has almost completely male-limited expression and strong testis-biased expression.

Male germline expression of Dpse-polo-dup1 increases fertility

We used a GAL4>UAS system to express Dpse-polo and Dpse-polo-dup1 in the D. melanogaster male germline. We also expressed an ovary-derived polo transcript (PoloO) and a testis-derived polo transcript (PoloT) from D. melanogaster (Figure 1) in the D. melanogaster male germline. Each of the four transcripts were cloned into a UAS expression vector to create four separate transgenes, and we refer to each as a “UAS-PoloX” transgene. We generated 1–3 transgenic strains for each of the four UAS-PoloX transgenes. Male germline expression of each transgene was under the control of a bam-Gal4 driver (Chen and McKearin 2003). We mated individual bam-Gal4>UAS-PoloX males with single females, and we counted the number of male progeny and female progeny sired by each male (Supplemental Table S4).

We tested if expression of each Plk transgene in the D. melanogaster male germline affects the number of progeny sired. There was not a significant difference between the PoloO and PoloT transgenes on the total number of progeny, number of female progeny, or number of male progeny (all p>0.39; Figure 4). Similarly, there was not a significant difference between the Dpse-polo and Dpse-polo-dup1 transgenes on the number of total progeny, female progeny, or male progeny (all p>0.24; Figure 4).

Figure 4.

Number of progeny, number of female progeny, and number of male progeny sired by males with germline expression of different polo transcripts. Males carried a transgene with a polo transcript derived from D. melanogaster ovary mRNA [polo (ovary), i.e., PoloO], a D. melanogaster testis mRNA [polo (testis), i.e., PoloT], Dpse-polo, or Dpse-polo-dup1. Each dot shows the number of progeny sired by an individual male, and the box plots show the median and quartiles of the distribution for a given transgene.

We next tested if expressing the Plk transcripts in the D. melanogaster male germline affects if a male sires any progeny (i.e., whether a male sires 0 progeny or >0 progeny). Males that expressed Dpse-polo-dup1 in their germline sired >0 progeny more frequently than males that expressed Dpse-polo (z=1.818; p=0.0691), PoloO (z=-2.108; p=0.0350), or PoloT (z=-1.673; p=0.0943). Approximately 20–25% of males that expressed Dpse-polo, PoloT, or PoloO sired 0 progeny (Supplemental Material). In contrast, only one male (out of 22, or 4.3%) who expressed Dpse-polo-dup1 in their germline sired 0 progeny. There was not a significant difference in the number of males that sired zero progeny between those expressing D. melanogaster PoloO and PoloT in their germline (z=-1.037; p=0.300). We observed similar effects when we only counted male or female progeny (Supplemental Material).

Male germline expression of ovary derived polo transcripts causes female-biased broods

We also tested if expressing different polo transcripts in the D. melanogaster male germline affects the ratio of female:male progeny sired. More female than male progeny were sired when we expressed PoloO (F_1,46=9.35, p=0.0037) or Dpse-polo (F_1,61=3.50, p=0.066) in the male germline (Figure 5). In contrast, there was not a significant difference in female and male progeny when we expressed PoloT (F_1,49=0.175, p=0.68) or Dpse-polo-dup1 (F_1,20=0.0675, p=0.80) in the male germline.

Figure 5.

Frequency of female progeny sired by males with germline expression of different polo transcripts. Males carried a transgene with a polo transcript derived from D. melanogaster ovary mRNA [polo (ovary), i.e., PoloO], a D. melanogaster testis mRNA [polo (testis), i.e., PoloT], Dpse-polo, or Dpse-polo-dup1. Each dot shows the frequency of female progeny of progeny sired by an individual male (number of female progeny / [male + female progeny]), and the box plots show the median and quartiles of the distribution for a given transgene.

Discussion

We showed that a duplication of polo in the D. pseudoobscura genome (Dpse-polo-dup1) has the conserved structure of a canonical Plk, but its amino acid sequence is fast evolving (Figure 2). We also demonstrated that Dpse-polo-dup1 has testis-biased expression (Figure 3), suggesting specialization for male germline function. Ectopic expression of Dpse-polo-dup1 in the D. melanogaster male germline increased the probability of siring progeny relative to ectopic expression of D. melanogaster polo or Dpse-polo (Figure 4), providing additional evidence that Dpse-polo-dup1 is specialized for male germline function.

Further consistent with male germline specialization, expression of Dpse-polo-dup1 in the D. melanogaster germline caused males to sire equal numbers of females and males, but male germline expression of Dpse-polo or a D. melanogaster polo transcript derived from ovary (PoloO) resulted in an excess of female progeny (Figure 5).

Gene duplication and testis specialization of meiotic genes

Our results suggest that the rapid evolution of Dpse-polo-dup1 may be the result of adaptive fixations of amino acid substitutions that contribute to male germline specialization. Reis et al. (2011) hypothesized that polo duplications with male-limited expression may accelerate male meiosis, which could provide a mechanism by which ectopic germline expression of Dpse-polo-dup1 increases the fertility of D. melanogaster males. It is possible that the single copy D. melanogaster polo gene is constrained from germline-specific adaptation because of the diverse functions that Polo is required to perform. Plks are required for spindle attachment to and dissociation from the kinetochore in mitosis, female meiosis, and male meiosis (Sunkel and Glover 1988; Carmena et al. 1998; Herrmann et al. 1998; Archambault and Glover 2009; Das et al. 2016). There are functional differences between mitotic, female meiotic, and male meiotic spindles (Orr-Weaver 1995; Savoian and Glover 2014), which could create pleiotropic constraints opposing the specialization of polo function across mitotic and meiotic contexts (Wagner and Zhang 2011). In other words, improvements to Plk function in male meiosis could come at a cost to mitosis or female meiosis. Similar inter-sexual fitness tradeoffs have been documented in meiotic drive systems (Fishman and Saunders 2008). Duplication of polo in D. pseudoobscura may have allowed for the resolution of those pleiotropic conflicts via male meiotic specialization of Dpse-polo-dup1 (Connallon and Clark 2011; Gallach and Betrán 2011; VanKuren and Long 2018; Hamada et al. 2020). Plk genes have been duplicated and subfunctionalized in other taxa (Habedanck et al. 2005; Bettencourt-Dias et al. 2005), suggesting that this may be a common mechanism to resolve pleiotropic constraints imposed by differences in the spindle apparatus across mitosis and meiosis.

If male-specific subfunctionalization of a polo duplication is advantageous, why does D. melanogaster not have subfunctionalized polo gene duplicates? The single D. melanogaster polo gene is autosomal (chromosome 3L, or Muller element D), but Dpse-polo became X-linked when element D fused to the X chromosome, creating a neo-X chromosome. We hypothesize that the initial retention of Dpse-polo-dup1 was favored after Dpse-polo became X-linked because X chromosome expression is reduced in the male germline (Vibranovski et al. 2009; Meiklejohn et al. 2011; Wei et al. 2022). Reduced X expression is thought to favor the retention of autosomal duplicates of X-linked genes when those genes are required for male meiosis or spermatogenesis (Betrán et al. 2002; Emerson et al. 2004; Marques et al. 2005; Potrzebowski et al. 2008; Meisel et al. 2009). Therefore, X-linkage of Dpse-polo may have favored the initial retention of Dpse-polo-dup1 in order to compensate for reduced expression in the male germline. This initial retention of Dpse-polo-dup1 may have allowed for subsequent selection for male germline specialization, which resolved the antagonistic pleiotropy over meiotic and mitotic functions. This two-step process of selective retention of male germline specific paralogs could explain why polo duplications are not observed in other Drosophila species (Reis et al. 2011). More generally, this two-step process could explain the excess gene duplication from Drosophila neo-X chromosomes and rapid (possibly adaptive) evolution of testis expressed autosomal paralogs (Meisel et al. 2009, 2010).

Sex ratio distortion and sexual conflict

We observed female-biased broods when we ectopically expressed either a D. melanogaster ovary-derived polo transcript (PoloO) or Dpse-polo in the D. melanogaster male germline (Figure 5). Female- or male-biased sex ratios can arise via meiotic drive or segregation distortion, and the mechanisms by which this occurs differ between oogenesis and spermatogenesis (Lindholm et al. 2016). In each female meiosis, only one homolog of each pair of chromosomes can segregate to the egg pole, while the remaining homolog and the sister chromatids go to polar bodies (Evans and Robinson 2011). This asymmetry in female meiosis creates an opportunity for one chromosome to outcompete its homolog to the egg pole during meiosis I, possibly because it has a “stronger” centromere (Sandler and Novitski 1957; Henikoff et al. 2001; Clark and Akera 2021). In contrast to oogenesis, segregation distortion in spermatogenesis typically arises when a chromosome carrying a “drive” allele is able to destroy sperm carrying an alternate allele (Larracuente and Presgraves 2012). When the drive locus is on the X chromosome (i.e., targeting Y-bearing sperm for destruction), this creates sex ratio distortion because males carrying the driving X sire an excess of daughters (e.g., Sturtevant and Dobzhansky 1936; Presgraves et al. 1997; Unckless et al. 2015).

Plks are essential for chromosome segregation during meiosis because they are required for kinetochore formation and dissociation. This role in spindle attachment to the centromere opens up the possibility for Plks to be involved in meiotic drive, possibly by affecting chromosome segregation. However, centromere drive is canonically associated with asymmetrical female meiosis, not symmetrical male meiosis (Lampson and Black 2017). It is therefore not clear how the kinetochore could affect sex chromosome transmission in the male germline, where all four meiotic chromatids are packaged into spermatids. Future work could use Plks to explore the possible associations between the male meiotic kinetochore and sex ratio distortion.

There are some similarities between the effect of polo on sex ratios and other known segregation distortion systems that may help us understand how polo expression affects sex ratios. First, most genes that cause segregation distortion in Drosophila are recent gene duplications that acquired germline-specific expression (Merrill et al. 1999; Montchamp-Moreau et al. 2006; Tao et al. 2007a; b; Helleu et al. 2016; Lin et al. 2018). For example, the D. melanogaster Segregation Distorter (SD) chromosome is preferentially transmitted relative to wild-type second chromosomes in SD/+ heterozygous males (Temin et al. 1991; Larracuente and Presgraves 2012). The Sd locus that is responsible for SD drive is a truncated duplication of a gene encoding the Ran GTPase-activating protein (RanGAP), and the Sd gene is sufficient to create the driving effect of the SD chromosome (Merrill et al. 1999). In addition, simply overexpressing RanGAP in the male germline causes segregation distortion in a way that mimics the effect of the SD locus (Kusano et al. 2001). This driving effect of overexpression is reminiscent of the sex ratio distortion we observe when ectopically expressing PoloO or Dpse-polo in the male germline (Figure 5).

The mechanism by which RanGAP and Sd affect segregation distortion could be informative of how male germline polo expression affects progeny sex ratios. RanGAP is an integral component of nucleocytoplasmic transport (Stewart 2007), but RanGTP also affects microtubule function. For example, organization of the bipolar mitotic spindle is coordinated by Ran GTPase gradients that determine microtubule nucleation and stabilization around chromosomes (Caudron et al. 2005). In Saccharomyces cerevisiae, microtubule extension from spindle poles depends on the Ran GDP/GTP exchange factor, which functions in the opposite enzymatic direction of RanGAP (Tanaka et al. 2005). In addition, manipulating RanGTP in vertebrate oocytes affects meiosis II spindle assembly (Dumont et al. 2007), demonstrating a sex-specific meiotic effect. Genes encoding nuclear transport proteins, such as RanGAP, are frequently duplicated, adaptively evolving, and thought to be important loci of genetic conflict (Betrán and Long 2003; Presgraves and Stephan 2007; Presgraves 2007; Tang and Presgraves 2009; Tracy et al. 2010; Phadnis et al. 2012; Mirsalehi et al. 2021). The mechanisms by which RanGTP and other nuclear transport molecules affect male meiosis could be similar to how male germline expression of Plks affect sex chromosome transmission.

Second, there are multiple documented sex ratio distortion systems in D. pseudoobscura. One of these systems is caused by a “driving” X chromosome (known as Sex Ratio, or SR) that eliminates all Y-bearing sperm (Sturtevant and Dobzhansky 1936; Policansky and Ellison 1970). Males carrying the SR X chromosome therefore produce nearly all female offspring. In addition, male hybrids between two D. pseudoobscura subspecies (pseudoobscura and bogotana) are weakly fertile and sire only female offspring (Orr and Irving 2005). Decreased fertility and segregation distortion of the sex chromosomes in hybrids are both caused by incompatibilities between alleles at an X-linked locus and at least seven other interacting loci throughout the genome (Phadnis and Orr 2009; Phadnis 2011). The shared genetic architecture underlying decreased fertility and female-biased sex ratios is similar to our observation that ectopic expression of polo transcripts in the D. melanogaster male germline can affect both fertility and sex ratios. Exploring links between male fertility and sex chromosome transmission could provide further insights into how male germline expression of polo affects sex ratios.

Sex ratio distortion and meiotic drive are often framed as intragenomic conflicts, which are often studied independently of intralocus sexual antagonism (Lindholm et al. 2016). Our results provide evidence that expression of polo variants can create intragenomic conflict, and, more specifically, sexually antagonistic effects (Rowe et al. 2018). We hypothesize that polo alleles that optimize function in mitotic or female meiotic chromosome segregation can have deleterious effects when expressed during male meiosis. We observe these effects when we express polo or Dpse-polo in the D. melanogaster male germline and female-biased broods are sired. In contrast, we hypothesize that selection to optimize Dpse-polo-dup1 for male germline function ameliorates those deleterious effects. This hypothesis explains why ectopic expression of Dpse-polo-dup1 in the D. melanogaster male germline increases fertility (relative to polo and Dpse-polo) and does not skew sex ratios. Our hypothesis is also consistent with a model in which gene duplication has resolved a sexual conflict (Connallon and Clark 2011; Gallach and Betrán 2011). While our results provide a link between intralocus sexual antagonism and meiotic drive, it is not clear if sexual conflicts over meiotic functions respond to or cause meiotic drive or segregation distortion.

Mechanisms by which polo could affect male meiosis and sex chromosome transmission

Our results are suggestive of mechanisms by which Polo transcripts could affect sex chromosome transmission and male fertility. First, we observe that ectopic expression of PoloO in the D. melanogaster male germline causes female-biased broods, while expression of PoloT does not (Figure 5). The PoloO and PoloT transgenes in our experiments had the same protein sequence, but they differed slightly in the UTRs they contained. PoloT has a 5’-UTR that is 45 bp longer PoloO, while PoloO has a 3’-UTR that is 17 bp longer than PoloT (Figure 1; Supplemental Table S1). It is therefore possible that a region of the 5’-UTR promotes found in PoloT equal transmission of the X and Y chromosomes, or a region of the 3’-UTR found in PoloO causes preferential transmission of the X chromosome (resulting in female-biased broods).

UTRs are known to affect both mitotic and meiotic functions of Plks. For example, the two polo transcripts in D. melanogaster differ in the lengths of their 3’-UTRs (Figure 1), which affects translational efficiency and possibly kinetochore function, metamorphosis, and female fertility (Llamazares et al. 1991; Pinto et al. 2011; Oliveira et al. 2019). In addition, an allele in the human PLK1 3’-UTR affects mRNA secondary structure and stability (Akdeli et al. 2014), and shorter 3’-UTRs in many genes are associated with enhanced cell proliferation (Sandberg et al. 2008; Mayr and Bartel 2009). Some of these effects are caused by different pA sites, which should not differ between PoloO and PoloT—they share the same pA site that was engineered into their cloning backbone (Wang et al. 2012). However, the transcription rate of polo affects pA site selection, possibly via auto-regulatory feedback (Pinto et al. 2011). The GAL4>UAS system that we used may therefore have affected expression levels of polo transcripts in a way that shifted the relative usage of the pA site in the cloning backbone and a cryptic pA site in the 3’-UTR (Figure 1). It is also possible that the additional sequence in the PoloO 3’-UTR may affect the testis function of polo via effects on transcript stability or translational efficiency.

It is notable that a transcript that appears to affect X chromosome transmission in spermatogenesis was cloned from the ovary (PoloO), whereas a testis-derived transcript (PoloT) has no such effects (Figure 5). We cloned different transcripts from ovary and testis because the PCR primers that amplified polo transcripts in one tissue sample did not work in the other tissue sample. Upon examination of RNA-seq reads mapped to the polo locus in D. melanogaster, we discovered that the testis and ovary transcripts of polo may have atypical UTR configurations (Supplemental Figure S1). Specifically, testis transcripts appear to have the longer 5’-UTR (similar to polo-RA) and the shorter 3’-UTR (similar to polo-RB). In contrast, ovary transcripts appear to have the shorter 5’-UTR (similar to polo-RB) and the longer 3’-UTR (similar to polo-RA). These different UTR configurations may explain why we could amplify a longer 5’-UTR in PoloT and a longer 3’-UTR in PoloO (Figure 1). These differences are also consistent with our hypothesis that a sequence in the 5’-UTR of polo has testis-beneficial effects or a sequence in the 3’-UTR is ovary-beneficial. These testis- and/or ovary-specific effects may provide a mechanism for sexual conflict over transcript expression or splicing, possibly via transcript stability or translational efficiency.

A second important observation is that expression of polo or Dpse-polo in the D. melanogaster male germline decreases male fertility relative to Dpse-polo-dup1 (Figure 4). We hypothesized that the higher relative fertility of males expressing Dpse-polo-dup1 is caused by amino acid substitutions that optimize the protein for testis function. An alternative hypothesis is that high testis expression of Plks in the male germline decreases fertility, and Dpse-polo-dup1 encodes a Plk with a mild loss of function with a lower fertility cost. In this hypothesis, ectopic expression of Dpse-polo-dup1 would be less costly than expression of the fully functional Plks encoded by polo and Dpse-polo. Negative effects of high polo expression have been shown in D. melanogaster intestinal stem cells, where constitutively active Polo suppresses intestinal stem cell proliferation, induces abnormal accumulation of β-tubulin in cells, and drives stem cell loss via apoptosis (Zhang et al. 2023). However, other experiments have shown Polo overexpression by 2.5-fold using GAL4>UAS does not affect its physiological function in mitosis (Martins et al. 2009). It therefore remains to be determined Dpse-polo-dup1 has fewer negative effects when ectopic expressed in the male germline, or if it has beneficial effects because of selection for testis specialization.

We hypothesize that changes to polo transcript stability, translational efficiency, or protein coding sequence affect sex chromosome segregation in male meiosis. This hypothesis is motivated by the observation that mutations to polo cause high rates of nondisjunction and sperm with abnormal DNA content (Sunkel and Glover 1988; Carmena et al. 1998, 2014; Herrmann et al. 1998), and we observed that ectopic expression of transcripts in the D. melanogaster male germline affected the sex ratio of broods in our experiments (Figure 5). Polo may affect chromosomal transmission through its interactions with Mei-S332. Mei-S332 associates with centromeres in prometaphase of meiosis I, and phosphorylation by Polo is required for removal of Mei-S332 during segregation of sister chromatids in anaphase II (Goldstein 1981; Kerrebrock et al. 1992; Tang et al. 1998; Clarke et al. 2005). Mutation of mei-S332 causes nondisjunction during meiosis II because of defective sister chromosome cohesion after metaphase I, which affects orientation going into meiosis II (Davis 1971; Goldstein 1980). Nondisjunction of autosomes could decrease fertility by increasing the frequency of autosomal aneuploids. Another outcome of elevated meiosis II nondisjunction is that mei-S332 mutant males produce an excess of XX sperm (i.e., co-inheritance of sister chromatids), relative to XY sperm (co-inheritance of homologous chromatids), in addition to an excess of nullo-XY sperm (Kerrebrock et al. 1992). These two predictions could provide a functional link between sex ratio distortion and male fertility. However, while an excess of XX sperm would cause males to sire female-biased broods, an excess of nullo-XY sperm would lead to male-biased broods, effectively canceling each other out. If polo expression affects Mei-S332 function, it is therefore not clear how this would result in female-biased broods.

Conclusions

We showed that a fast evolving, testis-expressed duplication of polo in the D. pseudoobscura genome (Dpse-polo-dup1) does not impose fertility costs nor does it skew progeny sex ratios when expressed in the D. melanogaster male germline. In contrast, ectopic testis expression of ovary derived polo genes and transcripts causes males to sire female-biased broods. These results are consistent with adaptive specialization of Dpse-polo-dup1 for male germline-specific function, possibly related to unique requirements associated with the male meiotic spindle apparatus. Alternatively, Dpse-polo-dup1 may be a hypomorphic Plk variant that does not have deleterious effects when over-expressed in the male germline, in contrast to other Plks. The initial duplication of polo may have been selectively retained because neo-X-linkage caused decreased male germline expression of the ancestral polo locus, favoring an autosomal paralog to compensate. This could explain why D. pseudoobscura has a testis-expressed paralog of polo but D. melanogaster does not. These results provide evidence for divergent selection pressures on spindle assembly genes in mitosis, female meiosis, and male meiosis. We hypothesize that these divergent selection pressures create pleiotropic conflicts or sexual antagonism, which can then be resolved by duplication and germline-specific specialization of a paralog.

Supplemental Figures

Supplemental Figure S1.

RNA-seq read mapping across the polo gene region. Transcript structures of polo-RA and polo-RB are shown, with UTRs in gray and protein coding sequence in tan. RNA-seq reads mapped per nucleotide position are shown for samples from testes, ovaries (virgin females), and ovaries (mated females). Average coverage across all samples is shown as a line graph, with predicted exons indicated by red brackets. Figure downloaded and modified from FlyBase JBrowse. Original RNA-seq data from modENCODE (Brown et al. 2014).

Acknowledgements

We thank Samantha Pacheco and Taylor Nunley for assistance with experiments, and other members of the Meisel lab for valuable discussions. This work was supported by startup funds from the University of Houston to RPM and a University of Houston Summer Undergraduate Research Fellowship to RV.