RNA editing controls meiotic drive by a Neurospora Spore killer

Neurospora Sk-2 is a complex meiotic drive element that is transmitted to offspring through sexual reproduction in a biased manner. Sk-2’s biased transmission mechanism involves spore killing, and recent evidence has demonstrated that spore killing is triggered by a gene called rfk-1. However, a second gene, rsk, is also critically important for meiotic drive by spore killing because it allows offspring with an Sk-2 genotype to survive the toxic effects of rfk-1. Here, we present evidence demonstrating that rfk-1 encodes two protein variants: a 102 amino acid RFK-1A and a 130 amino acid RFK-1B, but only RFK-1B is toxic. We also show that expression of RFK-1B requires an early stop codon in rfk-1 mRNA to undergo adenosine-to-inosine (A-to-I) mRNA editing. Finally, we demonstrate that RFK-1B is toxic when expressed within vegetative tissue of Spore killer sensitive (SkS) strains, and that this vegetative toxicity can be overcome by co-expressing Sk-2’s version of RSK. Overall, our results demonstrate that Sk-2 uses RNA editing to control when its spore killer is produced, and that the primary killing and resistance functions of Sk-2 can be conferred upon an SkS strain by the transfer of only two genes.


INTRODUCTION
Neurospora Spore killer-2 (Sk-2) is a complex meiotic drive element with the ability to be transmitted to nearly all viable offspring produced by sexual reproduction (Turner and Perkins 1979). While Sk-2's biased transmission rate is caused by spore killing, the molecular processes directing the mechanism of spore killing are unclear. We have previously demonstrated that spore killing requires at least two proteins: RFK-1 (Required For Killing-1) and RSK (Resistant to Spore Killer) (Hammond et al. 2012;Harvey et al. 2014;Rhoades et al. 2019). It is currently thought that RFK-1 provides the killing function, and that RSK provides the resistance function; however, Sk-2 is a complex meiotic drive element that spans more than 2.5 Mb of chromosome III and it includes hundreds of linked protein-coding genes (Svedberg et al. 2018). Therefore, it is possible that Sk-2 possesses other genes that are required for either killing, resistance, or both.
We sought to test this hypothesis by genetically engineering a non-Sk-2 strain (Sk S ; Spore killer sensitive) to carry copies of Sk-2's rfk-1 and rsk alleles, referred to as rfk-1 Sk-2 and rsk Sk-2 , respectively. As part of this process, we examined the role of a previously identified nucleotide within rfk-1 Sk-2 mRNAs that undergoes adenosine-to-inosine (A-to-I) mRNA editing in Sk S × Sk-2 crosses.

Strains and media
The genotypes of all strains used in this study are listed in Table 1. Vogel's Minimal Medium (VMM) or Vogel's Minimal Agar (VMA: VMM + 2% agar) (Vogel 1956) was used for vegetative propagation of all Neurospora strains unless otherwise indicated. L-histidine (Research Products International) was added to growth medium at 0.5 g per liter for propagation of strains with a his-3 mutation. Hygromycin B (Gold Biotechnology) was used at 200 μg per ml when selecting for hygromycin resistant strains. Westergaard and Mitchell's synthetic crossing agar (SCA) (Westergaard and Mitchell 1947) with 1.5% sucrose and a pH of 6.5 was used for N. crassa crosses. Brockman and de Serres agar (BDA) (Brockman and De Serres 1963) with 1% sorbose, 0.05% glucose, 0.05% fructose, 2.0% agar, and the recommended salts + minerals was used to select homokaryotic strains from heterokaryotic transformants. The mus-51 RIP70 allele is available in strain FGSC 10340 from the Fungal Genetics Stock Center (McCluskey et al. 2010).

Plasmid construction and site directed mutagenesis
AH36 is a 1481 bp rfk-1-spanning interval of Sk-2 DNA (Rhoades et al. 2019) (GenBank KJ908288.1: positions 27900-29380). Plasmid pNR9.1 was constructed by amplifying AH36 from Sk-2 genomic DNA with primer set 353/639 and cloning the PCR product between the NotI sites of plasmid pBluescript II KS(-) (the sequences of the primers used in this study are listed in and identical orientations in pNR154.6, respectively. Plasmid pNR155.2 was constructed by transferring the 1.5 kb NotI fragment containing rfk-1 TAG>TGG from pNR127.9 to the NotI site of pTH1256.1 and selecting for the rfk-1 transcriptional direction that matched that of hph within pTH1256.1. Plasmid pNR157.6 was constructed by transferring a 1.5 kb NotI fragment containing rfk-1 TAG>TAA from pNR153.1 to the NotI site of pTH1256.1, and selecting for an rfk-1 transcriptional direction that was opposite to that of hph within pTH1256.1. DJ-PCR products were produced with Q5® High-Fidelity DNA Polymerase (New England Biolabs) via the method of Yu et al. (2004). DJ-PCR product v251a was constructed to fuse the coding region of rfk-1 TAG>TGG to the tcu-1 promoter on chromosome I of an Sk S strain while deleting the coding sequences of tcu-1 (the sequences of the tcu-1 locus and other non-Sk-2specific loci were downloaded from FungiDB [Stajich et al. 2012]). The left and right flanks for DJ-PCR were amplified from wildtype genomic DNA with primer sets 1845/1846 and 1847/1848, respectively. The center fragment for DJ-PCR was amplified from pNR155.2 with primer set 1843/1844. The fused DJ-PCR product was amplified with primer set 1849/1850 for use in N. crassa transformation. DJ-PCR product v251b was constructed to fuse the coding region of rfk-1 Sk-2 to the tcu-1 promoter on chromosome I in an Sk S strain while deleting the coding sequences of tcu-1. It was constructed similarly to v251a except that the center fragment was amplified from pNR154.6. DJ-PCR product v286 was constructed to insert hph downstream of the rsk Sk-2 gene on chromosome III in an Sk-2 strain ( Figure S1). The left and right flanks were amplified from genomic DNA of Sk-2 strain P15-53 with primer sets 1985/1986 and 1987/1988, respectively. The center fragment was amplified from plasmid pTH1256.1 with primer set 12/13. The fused DJ-PCR product was amplified with primer set 1989/1990 for use in N. crassa transformation. Strain ISU-4958 was obtained by transformation of P15-53 with DJ-PCR product v286.
Hygromycin B was used to select transformants. Strain ISU-4959 was obtained by transformation of strain FGSC 10340 with DJ-PCR product v290. Transformants were selected on medium containing hygromycin B.

Growth assays
Radial growth assays were performed in 100 mm culture dishes containing VMA or VMA + 250 μM CuSO4. A 5 μl aliquot of a 1000 conidia per μl suspension (made in sterile water) was transferred to the center of each dish. Inoculated plates were imaged after 120 h of growth at room temperature. Linear growth assays were performed in 25 ml polystyrene serological pipettes (Corning™ 4489) according to the method of White and Woodward (1995). The pipettes were inoculated with 5 μl aliquots of a 1000 conidia per μl suspension (made in sterile water) and growth front positions were recorded at 24-hour intervals.

Spore killing assays
Unidirectional crosses were performed on SCA as previously described (Samarajeewa et al. 2014). Asci (ascospore sacs) were dissected from fruiting bodies into 50% glycerol on the 14 th day after fertilization. Digital images of asci (ascospore sacs) were obtained with the aid of a Leica DMBRE microscope and imaging system.

A-to-I mRNA editing is required for spore killing
In a previous report, we demonstrated that rfk-1 Sk-2 undergoes A-to-I mRNA editing (herein referred to as RNA editing) of an early stop codon during the sexual phase (Rhoades et al. 2019).
With the sequence of rfk-1 Sk-2 , the position of the edited stop codon, and the standard rules of mRNA translation, we predicted that rfk-1 Sk-2 encodes at least two protein variants: a shorter protein of 102 amino acids and a longer protein of 130 amino acids ( Figure 1A; Rhoades et al. 2019). Here, we refer to the shorter variant as RFK-1 A and the longer variant as RFK-1 B .
We have also shown that a 1481 bp rfk-1 Sk-2 -spanning interval of Sk-2 DNA called AH36 ( Figure S2) can be placed downstream of the his-3 locus in an Sk S strain without overtly influencing vegetative growth and vegetative developmental processes (Rhoades et al. 2019).
However, if an AH36-harboring strain is crossed with an Sk S mating partner, ascus development is aborted before the formation of viable ascospores. This ascus abortion phenotype is consistent with the presence of Sk-2's killer and the concomitant absence of Sk-2's resistance protein during ascus development, and because it occurs when rfk-1 Sk-2 is present, it suggests that at least one RFK-1 variant is required for the ascus abortion phenotype.
To determine if RFK-1 A , RFK-1 B , or both are required for the ascus abortion phenotype, we inserted a modified AH36 downstream of the his-3 locus on chromosome I in an Sk S strain ( Figure 2A). The modified AH36 includes rfk-1 TAG>TAA instead of rfk-1 Sk-2 . The rfk-1 TAG>TAA sequence is identical to rfk-1 Sk-2 except that its early stop codon has been changed to TAA. We predicted that if RNA editing of the early stop codon in rfk-1 TAG>TAA were to occur, it would produce a TGA (i.e., UGA) stop codon. Thus, rfk-1 TAG>TAA should express RFK-1 A but not RFK-1 B . As a control, we reinserted an unmodified AH36 containing rfk-1 Sk-2 downstream of the his-3 locus in an Sk S strain, thereby repeating the original experiment performed by Rhoades et al. (2019).
Interestingly, we found that expression of rkf-1 TAG>TAA within AH36 during meiosis has no apparent effect on ascus development when both parents are of the Sk S genetic background, while, consistent with previous results, expression of rkf-1 Sk-2 within AH36 under the same conditions causes ascus abortion (Figure 2, B-E). These results demonstrate that RNA editing of the early stop codon in rfk-1 Sk-2 from TAG to TGG (i.e., UAG to UGG) is required for expression of RFK-1 B and that RFK-1 B expression is required for the ascus abortion phenotype.

RFK-1 B but not RFK-1 A is toxic to vegetative tissue
While the above results demonstrate that RFK-1 B is required for the ascus abortion phenotype, they do not address the role of RFK-1 A in ascus abortion. For example, perhaps RFK-1 A and RFK-1 B must both be present within an ascus for ascus abortion to occur. We tried to test this hypothesis by inserting an rfk-1 TAG>TGG -containing AH36 downstream of the his-3 locus in an Sk S strain. The rfk-1 TAG>TGG sequence is identical to rfk-1 Sk-2 except that the early TAG stop codon has been changed to TGG. As a result, rfk-1 TAG>TGG should express RFK-1 B but not RFK-1 A . However, our attempts to insert an rfk-1 TAG>TGG -containing AH36 downstream of the his-3 locus in an Sk S strain failed to produce a successful transformant (data not shown).
Our inability to recover a transformant carrying an rfk-1 TAG>TGG -containing AH36 downstream of the his-3 locus in an Sk S genetic background suggested that rfk-1 TAG>TGG is toxic to Sk S strains. To test this hypothesis, we inserted the coding region of rfk-1 TAG>TGG immediately downstream of the copper repressible tcu-1 promoter on chromosome I while suppressing expression of rfk-1 TAG>TGG with 250 μM of CuSO4 ( Figure 3A; Lamb et al. 2013). As a control, we also inserted the coding region of rfk-1 Sk-2 immediately downstream of the tcu-1 promoter.
Interestingly, while the tcu-1(P)-rfk-1 Sk-2 strain grew similarly on low and high copper media, the tcu-1(P)-rfk-1 TAG>TGG strain grew poorly under low copper conditions and normally under high copper conditions (Figure 3, B and C; Table 2). These results demonstrate that RFK-1 B can exert its toxic effects on vegetative tissue. They also demonstrate that RFK-1 A is not required for the toxicity of RFK-1 B in vegetative tissue.

RSK Sk-2 neutralizes RFK-1 B toxicity in vegetative tissue
To determine if RSK Sk-2 can neutralize RFK-1 B 's toxicity in vegetative tissue, we inserted the coding region of rsk Sk-2 downstream of the high expression ccg-1 promoter on chromosome V in an Sk S strain ( Figure 4A). We then combined the ccg-1(P)-rsk Sk-2 and tcu-1(P)-rfk-1 TAG>TGG alleles in a single strain through a sexual cross, and analyzed the resulting strain's growth characteristics on low and high copper media. Interestingly, we found there to be little difference in the strain's vegetative morphology or linear growth rate under low and high copper conditions ( Figure 4, B and C; Table 2). These observations contrast with those made of strain ISU-4956, which possesses tcu-1(P)-rfk-1 TAG>TGG but not ccg-1(P)-rsk   (Figure 3, B and C; Table 2).
These findings demonstrate that RSK Sk-2 neutralizes the toxicity of RFK-1 B in vegetative tissue.

DISCUSSION
In this study, we have provided evidence that RNA editing (specifically, A-to-I mRNA editing) of an early stop codon in rfk-1 Sk-2 is necessary for spore killing. We have also shown that RFK-1 B , one of two RFK-1 variants, is toxic to vegetative tissue. Finally, we have shown that RSK Sk-2 neutralizes RFK-1 B in vegetative tissue. These results are consistent with the killer neutralization (KN) model of Sk-2 meiotic drive (Hammond et al. 2012).
The KN model holds that Sk-2's killer protein is neutralized by Sk-2's resistance protein during early stages of ascus development. At later stages of development, likely after ascospore delimitation, the resistance protein becomes restricted to ascospores that inherit Sk-2 (i.e., Sk-2 ascospores). Ascospores that do not inherit Sk-2, such as the Sk S ascospores of an Sk-2 × Sk S cross, succumb to the killer's toxic effects because they cannot produce the resistance protein. If the KN model is accurate, the killer should have a half-life that is longer than that of the resistance protein, or the killer should be able to move from Sk-2 ascospores to Sk S ascospores after ascospore delimitation.
Although it had previously been shown that the early stop codon in rfk-1 Sk-2 mRNAs is edited from UAG to UGG during the sexual phase, the role of this edit in spore killing was not investigated (Rhoades et al. 2019). Here, we have shown that editing of the early stop codon in rfk-1 Sk-2 mRNAs allows for expression of RFK-1 B . We have also shown that alleles which cannot express RFK-1 B , such as rfk-1 TAG>TAA , do not trigger the ascus abortion phenotype ( Figure 2E), which occurs when the Sk-2 killer is expressed without co-expression of a compatible resistance protein (Rhoades et al. 2019). The most likely explanation for why rfk-1 TAG>TAA does not trigger ascus abortion is that its TAG to TAA mutation prevents RNA editing from changing the resulting early UAA stop codon to UGG, a tryptophan codon. If RNA editing does occur, we expect that it would change the early UAA stop codon in rfk-1 TAG>TAA mRNAs to UGA, which is also a stop codon. Therefore, it is likely that rfk-1 TAG>TAA mRNAs can only be used by ribosomes to produce RFK-1 A .
Our study of rfk-1 Sk-2 suggests that it encodes at least two protein variants: the 102 aa RFK-1 A and the 130 aa RFK-1 B . Furthermore, our results support a model in which RFK-1 A is non-toxic and RFK-1 B is toxic. We have presented two lines of evidence for the non-toxicity of RFK-1 A : first, the RFK-1 A -specific rfk-1 TAG>TAA allele does not trigger ascus abortion in Sk S sad-2 ∆ × rfk-1 TAG>TAA crosses ( Figure 2E), demonstrating that expression of RFK-1 A alone during ascus development is not harmful; and second, expression of rfk-1 Sk-2 during the vegetative phase does not reduce growth rate or overtly influence vegetative morphology. This latter finding is relevant to the non-toxicity of RFK-1 A because A-to-I mRNA editing does not occur during the vegetative phase in N. crassa (Liu et al. 2017), and thus translation of rfk-1 Sk-2 mRNAs during the vegetative phase should produce RFK-1 A but not RFK-1 B . We have also presented two lines of evidence for the toxicity of RFK-1 B . First, ascus abortion occurs during Sk S sad-2 ∆ × rfk-1 Sk-2 crosses but not during Sk S sad-2 ∆ × rfk-1 TAG>TAA crosses (Figure 2, D and E), and RFK-1 B should only be expressed in the former cross type; and second, expression of the RFK-1 B -specific rfk-1 TAG>TGG allele during the vegetative phase significantly restricts vegetative growth ( Figure   3, B and C; Table 2).
The KN model holds that the Sk-2 killer is neutralized by the Sk-2 resistance protein. Our results strongly suggest that the Sk-2 killer is RFK-1 B and that RFK-1 B is neutralized by RSK Sk-2 .
Furthermore, our results suggest that a second Sk-2 specific gene product is not required for either the toxicity of RFK-1 B or the resistance properties of RSK Sk-2 . For example, although expression of RFK-1 B in vegetative tissue is toxic in an Sk S genetic background (Figure 3, B and C; Table 2), this toxicity can be overcome by co-expression of RSK Sk-2 (Figure 4, B and C; Table   2). Still, Sk-2 is a complex meiotic drive element containing hundreds of linked protein coding genes (Svedberg et al. 2018), and it remains possible that some of these Sk-2-linked genes modify the efficiency of the spore killing and/or resistance processes. Testing this hypothesis will likely require construction of a functional two-gene meiotic drive element containing only   Complete growth assay results are provided in Table 2. are identical to those shown in Figure 3B. (C) Linear growth on VMA was tracked over time for strains ISU-4956 (rfk-1 TAG>TGG ), ISU-4962 (rfk-1 TAG>TGG rsk Sk-2 ), and ISU-3866 (wild type).