Identification of a common secondary mutation in the Neurospora crassa knockout collection conferring a cell fusion-defective phenotype

ABSTRACT Gene-deletion mutants represent a powerful tool to study gene function. The filamentous fungus Neurospora crassa is a well-established model organism, and features a comprehensive gene knockout strain collection. While these mutant strains have been used in numerous studies, resulting in the functional annotation of many Neurospora genes, direct confirmation of gene-phenotype relationships is often lacking, which is particularly relevant given the possibility of background mutations, sample contamination, and/or strain mislabeling. Indeed, spontaneous mutations resulting in phenotypes resembling many cell fusion mutants have long been known to occur at relatively high frequency in N. crassa, and these secondary mutations are common in the Neurospora deletion collection. The identity of these mutations, however, is largely unknown. Here, we report that the Δada-3 strain from the N. crassa knockout collection, which exhibits a cell fusion defect, harbors a secondary mutation responsible for this phenotype. Through whole-genome sequencing and genetic analyses, we found a ~30-Kb deletion in this strain affecting a known cell fusion-related gene, so/ham-1, and show that it is the absence of this gene—and not of ada-3—that underlies its cell fusion defect. We additionally found three other knockout strains harboring the same deletion, suggesting that this mutation may be common in the collection and could have impacted previous studies. Our findings provide a cautionary note and highlight the importance of proper functional validation of strains from mutant collections. We discuss our results in the context of the spread of cell fusion-defective cheater variants in N. crassa cultures. IMPORTANCE This study emphasizes the need for careful and detailed characterization of strains from mutant collections. Specifically, we found a common deletion in various strains from the Neurospora crassa gene knockout collection that results in a cell fusion-defective phenotype. This is noteworthy because this collection is known to contain background mutations—of a largely unclear nature—that produce cell fusion-defective phenotypes. Our results describe an example of such mutations, and highlight how this common genetic defect could have impacted previous studies that have used the affected strains. Furthermore, they provide a cautionary note about the use of Neurospora strains with similar phenotypes. Lastly, these findings offer additional details relevant to our understanding of the origin and spread of cell fusion-defective cheater variants in N. crassa cultures.

sequence [encoding about 10,000 protein-coding genes (3)], together with an expand ing molecular toolkit (4)(5)(6), Neurospora has contributed to key findings in several research fields and continues to represent a powerful model, particularly for animal and plant pathogens, and for agriculturally and industrially relevant species (7,8).As such, and building on decades of research, a functional genomics characterization of N. crassa has the potential to greatly contribute to our understanding of eukaryotic biology.
To maximize this potential, a consortium of laboratories working on Neurospora set out to systematically delete, via targeted gene replacement, all of the predicted protein-coding genes in this organism (9).Starting with the initial report in 2006 (10), this Neurospora Functional Genomics Project created full deletion mutants for ~9,000 of its predicted protein-coding genes [all of which are readily available from the Fungal Genetics Stock Center (11)], creating an invaluable resource to characterize gene function in this model system.Part of this project also involved phenotypic analysis of the resulting mutants, and large-scale screening studies have generated phenotypic data for nearly 1,300 N. crassa knockout (KO) strains (12)(13)(14)(15)(16), with the corresponding functional annotations.
Generation of a mutant strain followed by evaluation of the resulting phenotype is a standard and powerful approach to study gene function.A key consideration when pursuing such a strategy, however, is the existence of secondary mutations in the strain of interest that can confound the results.Indeed, it has been reported in multiple systems that laboratory strains and mutant collections exhibit considerable nucleotide variation and secondary (i.e., background) mutations (17)(18)(19)(20)(21)(22), which can impact the conclusions drawn from studies that use them.In N. crassa, it has long been observed that spontaneous mutations resulting in a phenotype characterized by altered asexual development and female sterility, reminiscent of that exhibited by many cell fusion-defective mutants (23), arise frequently (24).Indeed, the Free group has reported the existence of hundreds of strains in the Neurospora KO collection that exhibit a similar pleiotropic phenotype that appears to be unrelated to the absence of the target gene (25,26).The nature of the mutation(s) underlying this phenotype, however, is largely unknown.
A common approach to evaluating whether an observed phenotype is due to the absence of a particular gene in a KO strain in Neurospora is to perform co-segregation assays.Here, the KO strain of interest is typically back-crossed to a parental wild-type (WT) strain, and the progeny is then scored to evaluate whether the knockout cassette (i.e., the cassette used for gene replacement, harboring the selection marker) co-segre gates with the phenotype.A limitation of this assay, however, is that if a secondary mutation is responsible for the phenotype, and it is located in close proximity to the target gene locus, it may be difficult to find the recombinant strains, unless a large number of progeny is scored.A second approach to confirm whether a particular gene is responsible for an observed phenotype, is simply re-introducing a WT copy of the deleted gene and evaluating whether the phenotype is reversed (i.e., complementation assay).While several phenotypic studies have been done using the Neurospora KO collection, many of the reported phenotypes remain unvalidated, that is, no segregation and/or complementation assays have been reported to confirm that the target gene in the KO strain is indeed responsible for the observed phenotype.
In this study, we report that the Δada-3 strain from the N. crassa KO collection, which exhibits a pleiotropic phenotype commonly associated with cell fusion-defective mutants, harbors a secondary mutation responsible for this phenotype.Indeed, through whole-genome sequencing and genetic analyses, we found a ~30-Kb deletion that affects a known cell fusion-related gene, so/ham-1, and show that it is deletion of this gene-and not of ada-3-that is responsible for the cell fusion-defective phenotype in this strain.We additionally found three other knockout strains that harbor the same deletion, suggesting that this mutation may be common in the collection.Our results highlight the importance of proper functional validation of strains from the N. crassa KO collection and suggest that conclusions of studies based on the strains herein reported -or on others with similar phenotypes-may need to be reevaluated.In addition, our findings complement recent studies on fusion-defective cheater lineages and their relationship to the long-standing observation of the frequent emergence of soft-like morphological mutants in N. crassa laboratory cultures.

N. crassa strains and culturing
General conditions for N. crassa growth and maintenance, as well as routine manipula tive procedures, followed those described by Davis and De Serres (27).Strains were maintained at 25°C on Vogel's minimal medium (28) with 2% wt/vol sucrose and 2% wt/vol agar (hereafter referred to as VMM).Growth tube assays to determine growth rate (29) were done on the same medium.Crosses were performed on Westergaard and Mitchell's synthetic crossing medium (30).For ascospore germination and subse quent isolation, sorbose-containing medium (FIGS [fructose-inositol-glucose-sorbose]) was used (27).Picked ascospores were then grown at 25°C on VMM, supplemented with hygromycin (200 µg/mL; Calbiochem, San Diego, CA, USA) when necessary.
Unless otherwise noted, WT and KO N. crassa strains were obtained from the Fungal Genetics Stock Center (http://www.fgsc.net/).Unless otherwise specified, strains FGSC 2489 (74-OR23-1VA) and FGSC 988 (74-OR8-1a) were used as WT for all experiments.The KO strains used were generated as part of the Neurospora Functional Genomics Project (9,10), in which each target open reading frame is replaced with a hygromycin B phosphotransferase gene (hph) cassette, which confers resistance to hygromycin.All KO strains used in this study were verified by PCR, by checking that the target gene is indeed absent and replaced by the knockout cassette.For co-segregation experiments with FGSC 11070 (ΔNCU02896, a), a standard his-3+ derivative of FGSC 9720 was used as WT for mating (see Fig. 1A) (10,31).
Images of phenotypic assays are representative of more than five independent experiments, unless otherwise noted.

Genomic DNA isolation
For genomic DNA isolation, conidia were inoculated in 4 mL of liquid Vogel's minimal medium with 2% wt/vol glucose and grown at 30°C for 24 h in constant light, with constant shaking (200 rpm).Mycelia were harvested, dried, and snap-frozen, and then ground in liquid nitrogen with a mortar and pestle.Genomic DNA was then extracted using the Puregene DNA Isolation Kit (D-7000A, Gentra Systems) according to the manufacturer's instructions.

Sequencing and analysis
For whole-genome sequencing, Illumina libraries were prepared using the Nextera DNA Library Prep Kit (Illumina), and the manufacturer's protocol was followed with one exception-we performed agarose size selection of the Nextera libraries, extracting the ~500 bp bands.Multiplexed libraries were then sequenced on an Illumina HiSeq 2,500 system, with 100-bp paired-end reads.Sequencing data (FASTQ files) were first demultiplexed, and quality was then inspected using FastQC (https://www.bioinformatics.babraham.ac.uk/projects/fastqc/).Reads were trimmed using BBDuk (version 38.18, https://sourceforge.net/projects/bbmap/) to remove adapter sequences and low-quality reads.Only paired read mates with a quality Q score ≥30 were further processed.Sequences were then mapped to the N. crassa OR74A NC12 reference genome (assem bly GCA_000182925.2) using the short-read aligner BWA-MEM (version 0.7.17-r1188), with default parameters (32).BAM alignment files were then sorted and indexed with SAMtools (33), version 1.14.Lastly, sorted files were visually inspected using the Integrative Genome Viewer (IGV) (34), version 2.11.9.
For sequencing the so/ham-1 locus in strains of interest, we amplified the relevant region from genomic DNA using a nested PCR approach (Table 1), and performed full-length amplicon sequencing using the Oxford Nanopore Technologies platform, as provided by Plasmidsaurus (Eugene, OR, USA)

Complementation assay
Complementation experiments were done with a his-3-targeting vector system, using plasmids that allow for the expression of so/ham-1 under the control of either the ccg-1 promoter or its native promoter (pCCG1-SO-GFP or pSO-SO-GFP, respectively, courtesy of N. Louise Glass).These plasmids were created using pMF272 as a backbone, which is derived from pBM60/pBM61 (35).As such, the empty his-3-targeting vector pBM61 was used as a control.For complementation, strain FGSC 11070 (ΔNCU02896, a) was crossed to FGSC 9720 (Δmus-52::bar+, his-3, A), and his-3 progeny displaying the altered asexual development phenotype characteristic of FGSC 11070 ["flat" conidiation pattern and short aerial hyphae (10)] was then used for transformation with the aforementioned plasmids.Transformants were then picked and grown on VMM.

The absence of ada-3 is not responsible for the ADA phenotype of FGSC 11070
One of the strains generated by the Neurospora knockout project is FGSC 11070.This strain was first reported and characterized by Colot et al. (10), and was shown to exhibit an "all development altered" (ADA) phenotype, as it displayed defects in all the traits therein examined, namely basal hyphal extension, asexual development, and sexual development.The knocked out gene, which encodes for a putative Zn cluster transcription factor (36), was termed ada-3 (NCU02896).The FGSC 11070 strain exhibits a pleiotropic phenotype, featuring a "flat" conidiation pattern, short aerial hyphae, and female sterility (10,13), resembling a phenotype typically displayed by cell fusion-defec tive strains (23).Indeed, it has been shown that FGSC 11070 additionally fails to produce conidial anastomosis tubes (CATs), and ADA-3 has thus been suggested to play a role in cell fusion (25).
Upon genetic studies using the FGSC 11070 (Δada-3) strain (31), we noticed that the abnormal asexual development phenotype of this strain did not co-segregate with the hygromycin-resistance cassette and, hence, with the lack of ada-3.Indeed, when crossing 11070 with a WT strain (using WT as female, as 11070 is female sterile) and then scoring >55 single ascospore isolates, we were able to occasionally find progeny that, while it lacked the ada-3 gene (and was hygromycin resistant, HygR), it displayed a WT phenotype (Fig. 1A).This suggested that the phenotype displayed by 11070 is not a consequence of the absence of ada-3.Consistent with this, the WT-like Δada-3 progeny (i.e., the Δada-3 progeny that displays a WT phenotype) is also fully fertile, as opposed to 11070, which is female sterile (Fig. 1B).Together, these results suggest that the reported ADA phenotype exhibited by 11070 is not due to the absence of ada-3 but is, instead, a consequence of a secondary mutation.

The ADA phenotype of FGSC 11070 is due to the absence of so/ham-1
To identify the genetic alteration responsible for the pleiotropic phenotype displayed by 11070, we performed whole-genome sequencing.To do this, we analyzed the progeny from a cross between 11070 and a WT strain (Fig. 1A).Given that the hygromycin-resist ance cassette does not co-segregate with the phenotype (and, thus, the mutation of interest), we selected Δada-3 (i.e., HygR) progeny that displayed either a WT or an ADA phenotype, for comparison.We then picked four clones per group, extracted DNA, and subjected each sample individually to high-throughput Illumina DNA sequencing.
The approximately 40 Mb N. crassa genome is organized in seven chromosomes (linkage groups LG I-LG VII).Analysis of the sequencing data showed that the clones with the ADA phenotype all have a ~30-Kb (29,585 bp) deletion in LG I that physically affects three annotated genes, namely NCU10987, NCU02794, and NCU02793 (Fig. 2A).We confirmed the presence and identity of this deletion via PCR with primers flanking the predicted missing region and by cloning followed by Sanger sequencing, respectively (Table 1).In addition, and as expected, the whole-genome sequencing data confirmed the Δada-3 status of all sequenced clones.
To evaluate whether any of the genes affected by the deletion is responsible for the ADA phenotype in 11070, we characterized individual deletion strains of each of the three genes, to see if any of these single-gene mutant strains exhibits the ADA pheno type.We obtained the corresponding KO strains from the FGSC and then monitored their growth under standard conditions.We observed that while the FGSC 21640 (ΔNCU10987)  and FGSC 18264 (ΔNCU02793) strains exhibit a WT phenotype, deletion of NCU02794 (FGSC 11292) results in the ADA phenotype (Fig. 2B), which suggests that the absence of this gene is responsible for the mutant phenotype in 11070.The disrupted gene in 11292 is so/ham-1, which has been reported to play a role in cell fusion in Neurospora (37,38).Similar to many cell fusion mutants, the Δso/ham-1 strain exhibits an abnormal development, characterized by a flat, carpet-like phenotype with shortened aerial hyphae, an altered conidiation pattern, and female sterility, which is consistent with the phenotype displayed by 11070, further supporting the idea that the phenotype in 11070 is a result of the absence of so/ham-1.To directly evaluate this, we introduced a WT copy of so/ham-1 into the his-3 locus of a 11070 his-3 strain harboring the ~30-Kb deletion (see Materials and Methods).We observed that expres sion of so/ham-1 was able to rescue both the flat conidiation phenotype (Fig. 2C) and the female sterility (Fig. 2D) exhibited by the ~30-Kb deletion mutant, suggesting that the phenotype in this strain is indeed due to the lack of so/ham-1.
Together, these results confirm that 11070 harbors a secondary mutation (independ ent from the ada-3 deletion), that this secondary alteration corresponds to a ~30-Kb deletion in LG I, that this deletion affects multiple genes -including so/ham-1-, and that the flat conidiation phenotype of 11070 is not a consequence of the absence of ada-3 but of the lack of so/ham-1.In addition, these results suggest that ADA-3 is not involved in cell fusion in N. crassa as previously suggested, and that studies that have proposed a role of this putative transcription factor in cell fusion are confounded due to using a Δada-3 strain that additionally harbors a deletion in a known cell fusion protein, SO/HAM-1, whose deletion-on its own-results in cell fusion defects (37,38) (see Discussion).

The ~30-Kb deletion is present in multiple strains of the knockout collection
The identification of a secondary mutation in a strain of the N. crassa KO collection that is responsible for a flat conidiation phenotype was considered particularly relevant, given both that many deletion strains in the collection exhibit such phenotype (10,13,15,16)-in many cases due to secondary mutations (25)-and that no segregation analysis or complementation studies have been done for most of these deposited mutant strains.This suggested to us that the secondary mutation we identified might be present in multiple strains in the collection and, importantly, be responsible for their reported phenotype.
A study by Fu et al. set out to identify genes required for cell-to-cell fusion in N. crassa (25).The authors screened the KO collection for putative cell fusion mutants by initially looking at strains that exhibited a flat conidiation pattern and a defect or delay in the production of protoperithecia, a phenotype common to many cell fusion mutants (23).This was followed by subsequent segregation, complementation, and/or functional assays, which, ultimately, resulted in the identification of 24 genes that appear to be required for cell fusion between CATs (25).This gene set included ada-3, for which only co-segregation data were used as confirmation of the association between the phenotype and the lack of ada-3 (see Discussion), and which we show here to have been misclassified as a cell fusion gene.As part of their study, and in addition to the candidate cell fusion-related genes, Fu et al. also reported that while numerous KO strains exhibited a flat conidiation phenotype, they could not be rescued by introduction of the WT copy of the corresponding gene.We surmised that for at least some of these strains, the phenotype may be due to the ~30-Kb deletion.
To test this, we decided to evaluate the presence of the ~30-Kb deletion in the KO strains that exhibit a flat conidiation phenotype but for which Fu et al. (25) reported that complementation was unsuccessful, that is, that introduction of a WT copy of the target gene did not revert the mutant phenotype.These corresponded to seven strains, representing four different genes.Interestingly, we found that two of these strains, FGSC 12957 and 12958, which are the two deposited KO strains for NCU09263 (acw-4), harbor the ~30-Kb deletion (Fig. 3A) and, correspondingly, lack the genes in the deleted region in addition to the intended KO gene, acw-4.This suggested to us that the phenotype exhibited by 12957/12958 is unrelated to the absence of acw-4 and might, instead, be due to the ~30-Kb deletion (and the corresponding absence of so/ham-1).Consistent with this idea, we were able to obtain, via crossing, Δacw-4 clones that exhibit a WT phenotype (Fig. 3B).These WT-like Δacw-4 clones do not harbor the ~30-Kb deletion, while the Δacw-4 clones that exhibit an ADA phenotype do (Fig. 3C), similar to the situation with 11070 (see Fig. 1A and B).
The other five strains tested did not harbor the ~30-Kb deletion.Given, however, that previous evidence has shown that mutations in the so/ham-1 gene can arise as cheater variants during vegetative propagation [see (39) and Discussion], we surmised that these strains might nevertheless contain mutations within this gene that would be responsible for their phenotype.We found, however, that this was not the case; sequencing of the so/ham-1 locus revealed no mutations in these KO strains.
Fu et al. ( 25) also reported regulator of conidiation 1 (rco-1), ortholog of the Saccha romyces cerevisiae TUP1 gene, to be a candidate cell fusion gene.This supported an earlier study that suggested a role for rco-1 in hyphal fusion in N. crassa (40).Consis tent with this, the rco-1 mutant exhibits numerous developmental defects, including an altered conidiation pattern and female sterility (41), typical of many cell fusion mutants (23).More recently, our group reported a role for RCO-1 in modulating circadian gene expression and metabolic compensation in N. crassa (42).During the course of that study, we noticed that the two deposited KO strains for this gene, 11371 (ΔNCU06205, A) and 11372 (ΔNCU06205, a), generated as part of the Neurospora KO project, exhibit different overt phenotypes, particularly in aerial hyphae development (Fig. 4A).Furthermore, while both strains display reduced hyphal growth rates compared to the WT, a finding consistent with previous studies of the Δrco-1 strain (41), the phenotype is significantly more severe in 11372.Indeed, the observed mean growth rate (95% CI) of 11371 was 35.4% (28.8-42.03)relative to WT, while that of 11372 was 12.8% (10.35-15.17)(Fig. 4B).Given that we noticed this discrepancy at the same time as we were evaluating strains for the ~30-Kb deletion, and that Δrco-1 has been reported as a cell fusion-defective mutant, we decided to test whether any of these Δrco-1 strains harbored the ~30-Kb deletion.We found that 11372 does indeed harbor the ~30-Kb deletion (Fig. 4C), meaning that while this strain is an rco-1 deletion mutant (Fig. 4A), it also lacks the other genes in the deleted fragment, including so/ham-1.These genetic alterations in 11372 (i.e., both the lack of rco-1 and the presence of the ~30-Kb deletion) appear to have a multiplicative effect on growth rate.Indeed, the observed mean growth rate (95% CI) of 11070 was 36.4% (32.92-40) relative to WT, and the growth rate defect of the 11372 double mutant equals the product of the single-deletion effects observed in 11371 and 11070 (Fig. 4B).
Together, these results show that the ~30-Kb deletion originally found in 11070 is present in multiple strains of the Neurospora KO collection, and highlight that secon dary mutations and genomic rearrangements may be present in multiple strains in the collection and be responsible for various reported phenotypes, many of which might have been attributed to the deletion of an unrelated gene.

FIG 3
Strains FGSC 12957 and 12958 also harbor the ~30-Kb deletion.(A) PCR analysis to evaluate the presence of so/ham-1 or the ~30-Kb gap in the genome of the strains listed.To detect so/ham-1, primers within its coding region were used.To evaluate the presence of the deletion, primers flanking the ~30-Kb gap were used such that under normal conditions (i.e., no deletion), the size of the DNA fragment between the primers (~30 Kb) would be too large to be amplified (Continued on next page)

DISCUSSION
In this study, we report that multiple strains of the Neurospora crassa KO collection that exhibit an abnormal phenotype typical of many cell fusion-defective mutants, share a common secondary genetic alteration, namely, a ~30-Kb deletion affecting three genes.Our data suggest that it is this deletion-and not the absence of the intended target gene-that is responsible for the pleiotropic phenotype in these strains.Our results highlight the importance of proper functional validation of strains from the N. crassa KO collection-and from mutant collections in general-and suggest that conclusions of studies that have used the strains herein reported, or others with similar phenotypes, may need to be reevaluated.
A classical and powerful approach to study gene function is to create loss-of-func tion mutants and then observe the resulting phenotypes.As part of the Neurospora Functional Genomics Project (9), deletion strains for the vast majority of this organism's predicted ~10,000 protein-coding genes have been created and made available to the community.These strains represent an invaluable resource, and have been used by numerous laboratories for both focused and genome-wide screens [see, for instance, references (10,(12)(13)(14)(15)(16)(43)(44)(45)], providing key insights into gene function in this model system.
In the first study describing the methodology for high-throughput generation of loss-of-function deletion strains in N. crassa, Colot et al. (10) reported the creation of deletion strains for 103 genes that encode for predicted transcription factors.The authors then subjected the strains to phenotypic analysis, and reported that ~40% of the deletion mutants exhibit defects in basal hyphal extension, vegetative development, and/or sexual development.Nine strains were found to display defects in all three of these traits, and seven of them were thus termed "all development altered" (ada-1 to ada-7).While Southern blots were used to ensure correct and unique insertion of the knockout cassette, and to confirm the homokaryotic status of the mutant, no further analyses were done to determine whether the observed phenotypes were indeed due to the deletion of the target gene.
As part of a routine genetic study, we found that the pleiotropic phenotype exhibited by the deposited KO strain for all development altered (ada)-3 (FGSC 11070; ΔNCU02896) was not due to the deletion of ada-3.Indeed, we found that the ADA phenotype did not co-segregate with the deletion cassette-a finding consistent with a previous report (46) -which suggested that the phenotype was the result of a secondary genetic alteration (Fig. 1).To identify the mutation responsible for the phenotype in this strain, we used whole-genome sequencing, which revealed a ~30-Kb deletion in LG I that affects three genes (Fig. 2A).We observed that deletion of only one of these genes, so/ham-1, results in the same abnormal phenotype reported for FGSC 11070 (Fig. 2B).We performed a complementation assay with a WT copy of so/ham-1, and confirmed that the phenotype in 11070 was indeed a result of the deletion of so/ham-1 and is unrelated to the lack of ada-3 (Fig. 1 and 2).Given that deletion of NCU02896 does not result in a strain displaying an ADA phenotype, renaming ada-3 should be considered.

FIG 3 (Continued)
with the PCR settings used.Conversely, if the deletion is present, a PCR product could be obtained under the conditions used.We used a nested PCR strategy to detect the gap (Table 1).Note that a homokaryotic strain can either be positive for so/ham-1 or harbor the deletion, but not both.As such, even though all strains were tested for both targets, only one can be detected.(B) (Left) Phenotypic analysis.WT, FGSC 12957, FGSC 12958, and two WT-like Δacw-4 progeny derived from crossing WT with 12957 (xc1478, clones 21 and 22) were imaged after culture on slants of VMM for 7 days at 25°C under constant light.(Right) Schematic diagram and genotyping of the strains on the left, to confirm the Δacw-4 status (and concomitant replacement of acw-4 with the hph cassette) of the 12957, 12958, and xc1478 strains.Note that even though both reactions were tested per homokaryotic strain, each strain can only be positive for either the WT gene or the KO cassette, but not both.(C) PCR analysis to evaluate the presence of so/ham-1 or the ~30-Kb gap in the genome of the strains shown in (B), with 11070 as reference for the deletion.Note that a homokaryotic strain can either be positive for so/ham-1 or harbor the deletion, but not both.All primer sequences are shown in Table 1.The ada-3 gene has previously been suggested to play a role in cell fusion.This was based on a study by Fu et al. (25), who set out to find cell fusion-related genes in N. crassa.For this, they screened the KO collection for strains that exhibited a flat conidia tion pattern and a defect or delay in the production of protoperithecia, a phenotype common to many cell fusion mutants (23), including so/ham-1, which is known to play a role in cell fusion in N. crassa (38).The screening was followed by segregation, comple mentation, and/or functional assays, which, ultimately, resulted in the identification of 24 genes that appeared to be required for cell fusion between conidial anastomosis tubes in Neurospora; this gene set included ada-3 (25).For ada-3, however, the authors relied solely on co-segregation data to determine that the observed phenotype was due to the absence of this gene.Indeed, given that the ada-3 locus (NCU02896) lies close to the herein identified ~30-Kb deletion in LG I, it is likely that the co-segregation data -particularly if based on scoring a relatively small number of progeny when back-cross ing FGSC 11070-might have suggested to the authors that the phenotype and the deletion cassette co-segregate, which would have led Fu et al. (25) to classify ada-3 as a cell fusion gene.As mentioned above, however, we report here that deletion of ada-3 does not result in such phenotype (Fig. 1), and that the pleiotropic phenotype reported by Fu et al. (25) for the ada-3 deletion strain (FGSC 11070) is instead due to the concomitant deletion of so/ham-1 in the Δada-3 background.Indeed, deletion of so/ham-1 on its own results in that phenotype-while deletion of ada-3 does not-and introduction of a WT copy of so/ham-1 reverts the ADA phenotype of the Δada-3 strain (Fig. 1 and 2).As such, our study shows that ada-3, on its own, appears to play no role in cell fusion, or at least not one that would result in the overt phenotype that Fu et al. (25) attributed to its deletion.

Research
As part of their screening, Fu et al. ( 25) also reported seven KO strains for which re-introduction of a WT copy of the target gene could not rescue the mutant phenotype of interest (i.e., a flat conidiation pattern and a defect or delay in the production of protoperithecia).Given that this phenotype is consistent with the ADA phenotype, we explored whether at least some of these strains harbored the ~30-Kb deletion as a secondary mutation, such that the lack of so/ham-1 would be responsible for their observed phenotype.We indeed found two such strains, FGSC 12957/12958, which, while they both lack the target gene NCU09263/acw-4, also harbor the ~30-Kb deletion.Via crossing, we were able to obtain Δacw-4 strains that either had the ~30-Kb deletion or did not feature such deletion; the former displayed the ADA phenotype, while the latter did not, suggesting that the ADA phenotype in these strains is due to the ~30-Kb deletion and further confirming that it is unrelated to the absence of acw-4.Maddi et al. (47) reported that while they did observe co-segregation of the mutant phenotype with the deletion cassette in the Δacw-4 strain, the phenotype was not reverted by introduction of a WT copy of acw-4, which supports the idea that there is a secondary mutation responsible for the phenotype (38).Our data suggest that this secondary mutation corresponds to the ~30-Kb deletion, specifically to the absence of so/ham-1.Importantly, the ~30-Kb deletion in the Δacw-4 strain would explain the results reported by Maddi et al. (47): given that the deletion lies close to acw-4 locus in LG I, one would expect to see moderate co-segregation of the phenotype with the deletion cassette but no complementation when transforming with a WT copy of acw-4.The other five strains analyzed based on the study by Fu et al (25) do not harbor the ~30-Kb deletion.Given that a previous study reported that mutations in so/ham-1 can arise as cheater variants during vegetative propagation, resulting in the same pleiotropic phenotype exhibited by these strains (39), we tested whether they contained mutations in this gene. of variance, F (3, 18) = 257.5,P < 0.0001.Different letters indicate statistically significant differences between groups (Tukey's Honest Significant Difference , P < 0.05).All statistical analyses were performed using GraphPad Prism v. 9.5.1 (733).(C) PCR analysis to evaluate the presence of so/ham-1 or the ~30-Kb gap in the genome of the strains shown in (B).Note that a homokaryotic strain can either be positive for so/ham-1 or harbor the deletion, but not both.All primer sequences are shown in Table 1.
No mutations, however, were found.Additional studies of these strains are needed to uncover the molecular mechanisms underlying their phenotype.
Among the 24 genes that Fu et al. (25) reported as involved in cell fusion, was regulator of conidiation 1 (rco-1) (41).This gene had previously been reported to play a role in cell fusion in Neurospora (40), which lent support to the screening.During a study in our laboratory on the role of RCO-1 in circadian gene expression and metabolic compensation in N. crassa (42), we noticed that the two deposited KO strains for this gene, FGSC 11371 and 11372, exhibit different phenotypes (Fig. 4A).We found that, in addition to lacking rco-1, 11372 harbors the ~30-Kb deletion.Interestingly, while it has previously been reported that deletion of rco-1 results in a reduced growth rate (41), we noticed that the defect is even more dramatic in 11372 (Fig. 4B).Mutations in so/ham-1 are known to affect growth rate (37), and the phenotype exhibited by 11372 appears to be the result of the compounded effect of the lack of both rco-1 and so/ham-1.This multiplicative effect suggests that these loci may have an independent effect on growth rate in N. crassa (48).
Given that FGSC 11372 has a secondary mutation that affects its overt phenotype, it would thus be inappropriate for functional studies of rco-1: any result would be confounded by the fact that, in addition to lacking rco-1, this strain is also a so/ham-1 mutant.As such, the conclusions of functional studies of rco-1 based on 11372 may need to be reevaluated [see (13,16,25,40,(49)(50)(51)].Of particular notice among these studies is the report by Carrillo et al. (13), who characterized the growth and developmental phenotypes of loss-of-function deletion mutants of a large set of predicted transcription factor genes in N. crassa, including the transcriptional co-repressor RCO-1.In this work, it was reported that ~40% of viable mutants had a growth rate significantly slower than WT, and that deletion of rco-1 resulted in the lowest growth rate observed in the study, at ~10% of the WT rate.This is significantly different from the rate observed for Δrco-1 in previous studies (41) and in our laboratory (Fig. 4B) (~30% of WT), but is similar to the rate we observed for 11372 (Fig. 4B), which, as reported here, lacks both rco-1 and so/ham-1; indeed, the authors used 11372 for their analysis of rco-1 and, thus, the classification of Δrco-1 as the slowest-growing mutant in this data set is likely unwarranted.Similarly, this study also identified 92 mutant strains with a defect in aerial hyphae height and/or conidia production, and reported that the Δrco-1 strain exhibited the shortest aerial hyphae among all tested mutants.This, again, is likely a consequence of the additional mutation in this strain, as mutations in so/ham-1 are also known to affect aerial hyphae development (38).
Secondary mutations exist in multiple strains of the N. crassa KO collection (25,26), and many of these appear to result in an ADA phenotype.Interestingly, it has long been known that spontaneous mutations affecting growth rate and morphologyresulting in phenotypes similar to that of so/ham-1-are common and may arise in any strain, even in WT back-crosses (24).Indeed, such mutants are frequently found in all Neurospora laboratories.The reasons behind this are unclear, but some recent data shed light on how these may be propagated, and allow us to speculate about how they might have contaminated the N. crassa KO collection.In two studies, involving experimental evolution, whole-genome sequencing, and various competition assays, the Aanen group (39,52) proposed a model to explain how cheater genotypes, which appear to involve cell fusion-defective mutants -including so/ham-1 mutants-may be propagated during clonal growth.Maintaining live cultures by serial transfers may favor the appearance of such cheater lineages and, ultimately, reduce overall performance (conidial spore yield).Briefly, the authors proposed that upon a de novo cell fusion cheater mutation, the mutant nuclei will have increased representation in the aerial hyphae (and the resulting conidia) by virtue of the reduced chance of cheaters to participate in hyphal fusion.Upon clonal propagation and germination, a fraction of these mutants will fuse with WT or heterokaryotic mycelia, thus forming new heterokary ons.Importantly, while the mutants are unable to initiate fusion, WT hyphae can fuse to them (39).Fusion will give cheaters access to a well-connected mycelia, with efficient resource distribution and utilization, and will allow them to propagate effectively.When cheaters reach higher frequencies, however, the culture becomes increasingly fragmen ted, and cheater mycelia will remain largely unconnected from WT hyphae, which will ultimately reduce the total spore yield of the culture.The spread of such female-sterile cheaters is consistent with the early observation made by Westergaard and Hirsch that cultures of Neurospora commonly become female sterile after extended periods of vegetative propagation (53).Culture degeneration, with loss of female fertility, has also been reported in Fusarium species (54).
As such, the results of the Aanen group suggest that many of the spontaneous morphological mutants frequently found in Neurospora labs may be the result of unintended selection of cell fusion-defective cheater variants during routine propaga tion (39).In this context, it is conceivable that among the conidia in a spore suspension from a seemingly WT serially propagated culture, there could be multiple cheater mutant nuclei.Indeed, early reports suggested that cultures of strains typically used as WT in Neurospora (of the Oak Ridge background) may commonly become heterokaryotic for mutations that result in female sterility and soft-like phenotypes (24,55).It is thus possible that these nuclei, in the conidial suspension, are transformed and selected for when generating, for instance, KO strains, such as the ones described in our study.In this scenario, it could be proposed that strains FGSC 11070, 12957/12958, and 11372 might have been derived from transformation of mutant conidia from the same parental strain.Efforts to track down this information, however, were unsuccessful, as the Neurospora KO project was completed several years ago.While all of these mutants appear to have been generated in the same laboratory (Dunlap lab, Dartmouth University), the group confirmed they were done on different transformation plates/different batches (J.C. Dunlap, personal communication), but the source of the conidia (i.e., the exact batch used) is unknown.Alternatively, although intuitively less likely, the existence of this mutation in these different KO strains might have been due to independent events; as suggested by Grum-Grzhimaylo et al. (39), different types of mutations in so/ham-1-of which the ~30-Kb deletion could be an example-might be common as cheaters, and could have appeared independently in different clonally propagated WT cultures used as a source of conidia for transformation experiments for the Neurospora KO project.Indeed, Grum-Grzhimaylo et al. (39) found that 75% (6/8) of the cheater lines analyzed had acquired mutations in the so/ham-1 gene.In addition, ~50 years ago, a spontane ous female sterility mutant, female sterile-n (fs-n), was obtained via conidial isolation of an established WT laboratory strain (74-OR23-1 A, FGSC 987), where it occurred in a heterokaryotic condition (55).The fs-n strain was more recently subjected to wholegenome sequencing and found to contain a + 1:A frameshift mutation in so/ham-1 (56).
In summary, we report here that multiple strains of the N. crassa KO collection harbor a ~30-Kb deletion in LG I, and that this deletion might have affected the conclusions of various studies.We found four strains in this collection with such a secondary mutation affecting so/ham-1, but given that our analysis was neither systematic nor comprehen sive, the extent of the issue is currently unknown but likely affects many more strains, particularly if other types of mutations in so/ham-1 are considered.Our data should serve as a cautionary note regarding the use of strains from the N. crassa KO collection-and mutant collections in general-and highlight that standard validation assays should always be performed to confirm that a phenotype in a Neurospora KO strain is indeed due to the mutation of the target gene.These include not only strain validation by PCR and segregation assays but, ideally, also complementation.In addition, it is important to consider proper strain storage and handling (57).Users should place primary stocks into suspended animation and minimize their vegetative transfer, to avoid the acquisition of mutations.Furthermore, users would be best served by occasionally replacing laboratory stocks with those from the FGSC and by purifying strains of interest (e.g., by crossing and ascospore isolation).While whole-genome sequencing could be proposed for routine strain and sample verification, the associated costs and the paucity of suitable and easily accessible bioinformatics tools create a barrier for such an approach (17), further supporting the implementation of the simple assays and procedures described above.
Secondary mutations are widespread in the N. crassa KO collection, and many of these result in cell fusion-defective phenotypes, so researchers should consider the so/ ham-1 status of KO strains that exhibit such a phenotype before performing any assays.With careful validation of the results, mutant collections, including the N. crassa KO collection, will continue to serve as incredibly useful resources for the community.

FIG 4
FIG 4 FGSC 11372, an rco-1 KO strain, harbors the ~30-Kb deletion.(A) (Left) Phenotypic analysis of WT and the two rco-1 KO strains generated as part of Neurospora Functional Genomics Project, FGSC 11371 and 11372.The strains were imaged after culture on slants of VMM for 7 days at 25°C under constant light.(Right) Schematic diagram and genotyping of the strains on the left, to confirm the Δrco-1 status (and concomitant replacement of rco-1 with the hph cassette)of the 11371 and 11372 strains.Note that even though both reactions were tested per homokaryotic strain, each strain can only be positive for either the WT gene or the KO cassette, but not both.(B) Growth rate assay.Conidia from the strains listed were inoculated on race tubes containing VMM and were then placed under constant light conditions at 25°C.The growth front was marked on the tubes every 24 h, and the distance between the marks was then used to calculate the linear growth rate per day.Dots represent data from four to six independent biological replicates.Shown are the mean ± 95% confidence intervals.Analysis (Continued on next page)

TABLE 1
Primers used in this study