Relative efficiency of split-marker versus double-crossover replacement protocols for production of deletion mutants in strain PH-1 of Fusarium graminearum

The split-marker (SM) protocol has become a popular method for production of knockout mutations in fungi. We used Southern hybridization to compare the performance and efficiency of the SM protocol with the more traditional double-crossover intact marker (IM) method for creating deletions of the mating type genes in Fusarium graminearum. Both methods successfully produced knockouts at a rate of between 24 and 75%: the SM method produced mutants more efficiently for larger constructs (>1 kb), but it was similar to IM for a smaller construct that was 865 bp. Both methods also produced strains with additional ectopic integrations at a similar rate of approximately 10%, but on average the SM produced a higher number of independent integrations in those strains. Ectopic integrations produce off-site mutations, and strains with multiple integrations are less desirable since it is more difficult to remove them by backcrossing. Southern hybridizations will be generally superior to PCR to identify strains with fewer ectopic integrations for experimental purposes.


Introduction
Inactivation of a gene by deletion or disruption is an important way to study its function (Shafran et al., 2008;Turgeon et al., 2010). Gene deletion via double-crossover replacement is usually preferred because it avoids the possibility of residual or restored gene function. Gene deletion constructs typically consist of varying amounts of sequence from the flanks upstream and downstream of the gene of interest fused to a selectable marker gene (hereafter referred to as "intact-marker" (IM) constructs). Successful gene deletion depends on homologous recombination (HR) at double-strand breaks (DSB) within each flanking sequence (Hynes, 1996;Krappman, 2007;Weld, 2006). Recombination at DSB occurs by one of two DNA repair pathways: homology-directed repair (HDR) requires the presence of DNA sequence homology, while non-homologous end joining (NHEJ) does not (Cahill et al., 2006;Kanaar et al., 1998;Krappman 2007). Saccharomyces cerevisiae uses primarily HDR for DNA repair, and the generation of gene deletion mutants by HR is consequently very efficient in this organism (Hua et al., 1997). In contrast, many filamentous fungi tend to use the NHEJ mechanism more often. This results in a reduced number of transformants in which the gene of interest has been replaced by HR, and more strains in which the marker has integrated randomly at an ectopic location in the genome (Chaveroche et al., 2000).
The split-marker (SM) protocol, first applied to yeast (Fairhead et al., 1996), is an alternative approach for gene deletion in filamentous fungi (Catlett et al. 2003;de Hoogt et al. 2000). For this procedure, fungal protoplasts are transformed with a mixture of two DNA fragments, each comprised of DNA flanking one end of the gene of interest fused to overlapping segments of a selectable marker gene. To reconstitute a functional selectable marker, the SM fragments must undergo HR at three points, and this has been reported to increase the efficiency of gene deletion (Fairhead et al., 1996;Catlett et al., 2003). However, the performance of IM versus SM protocols has been directly compared in very few cases (e.g. Fu et al. 2006), and a detailed examination of integration sites by use of Southern hybridization has been rare. The objective of the current study was to use Southern hybridization to compare SM and IM protocols and evaluate their relative efficiency for generating gene deletions versus ectopic integrations in strain PH-1 of F. graminearum. For this work we targeted the mating-type locus of F. graminearum which has a previously characterized deletion phenotype .

Materials and Methods
Fungal strains and growth conditions: Fusarium graminearum strain PH-1 (NRRL 31084) was cultured on mung bean agar (MBA) (Bai and Shaner 1996) for 7-10 days at 23°C under continuous fluorescent light to produce macroconida. Spore suspensions were prepared by adding 2 ml of sterile water to the culture and gently rubbing the surface with a sterile plastic micro-pestle. Harvested spores were filtered through glass wool, washed twice with sterile water, and adjusted to the desired concentration.
DNA extraction: Five ml of YEPD medium (20 g dextrose, 20 g bacto-peptone, 10 g yeast extract) was inoculated with an 8-mm agar plug taken from the edge of an actively growing colony. Cultures were incubated at 25°C for 5-7 days at 250 rpm. Recovered mycelia were flash-frozen in liquid nitrogen, lyophilized, and pulverized in individual 2 ml Eppendorf tubes by using a mini-pestle, or in deep 96-well plates with a 2000 GENO/Grinder® (Spex Certiprep) (500 strokes/sec for 30 sec). One ml of warm lysis buffer (0.5 M NaCl; 1% SDS; 10 mM Tris HCl, pH 7.5; 10 mM EDTA) was added per 100-200 mg fungal tissue, and samples were incubated at 65°C for 30 min, vortexing once during the incubation. After incubation the samples were transferred into individual tubes containing 660 µl PCI (25 parts phenol; 24 parts chloroform; 1 part isoamyl alcohol), mixed by inverting 4-6 times, then incubated at 65°C for an additional 30 min. The contents were mixed once during incubation. The samples were centrifuged in a tabletop centrifuge for 20 min at maximum speed to separate the phases. DNA was precipitated from the aqueous phase by using 1 volume of isopropanol, and the pellet was washed twice with 70% ethanol. The pellet was resuspended in 100 µl of TE, pH 7.9, with 2 µl of a 5 mg/ml concentration of RNase A, at 65°C for 1 h.
Southern hybridizations: PCR amplicons used for probes were gel-purified and labeled with radioactive isotope P 32 using the Prime-a-gene labeling system (Promega, cat. #U1100). Between 1 and 5 µg of the genomic DNA isolated from the fungal strains was digested with 20 U of the appropriate restriction enzyme (RE) in a 50 µl reaction. DNA was precipitated and resuspended before loading the entire sample onto a 0.8% agarose gel made with 0.5X TBE buffer (20X TBE, 1L: 216g Tris base; 110g boric acid; 80 ml 0.5M EDTA, pH 8.0). The gel was electrophoresed at 35V in 0.5X TBE for 18-20 hours, followed by staining for 30 min in 0.5X TBE plus 5µg/ml EtBr for imaging, and then destaining for 30 min with fresh 0.5X TBE. DNA fragments were transferred from the gel to a charged nylon membrane (PALL Life Sciences) by electroblotting with a GENIE electroblotter (Idea Scientific) for 2 h at 12V. The DNA on the membrane was denatured in 0.4N NaOH for 10 min, then neutralized for 10 min in 2X SSC (20X SSC, 1 L; 175.3g NaCl; 88.2g Na3C6H5O7; adjusted to pH 7.0, autoclaved). DNA was fixed to the membrane by UV-crosslinking in a Spectrolinker (Spectronics Corporation). The membrane was prehybridized at 65°C for 30 min in hybridization buffer (100 mls; 25 ml 0.5 M Na2HPO4 pH 7.2; 7 g sodium dodecyl sulfate; 0.2 ml 0.5M EDTA, pH 8.0; stir to dissolve, heating if necessary) and then fresh hybridization solution was added together with the radioactively labeled probe, and the membrane was hybridized overnight at 65°C. The membrane was then washed three times with 2X SSC at 65°C for 20 min each, blotted dry, wrapped in plastic wrap, and exposed on a Storage Phosphor Screen (Molecular Dynamics) at room temperature for up to 3 d. The screen was scanned by using a Typhoon Phosphorimager (GE Healthcare). Membranes were stripped by washing for 30 min in 0.4N NaOH, followed by 15 min in 0.1X SSC, 0.1% SDS, at 45°C. Stripped membranes were rehybridized with probes for the mating-type genes and for the hyg gene. The resulting images were false-colored and overlaid to create the figures by using Adobe imaging software.
Plasmid construction: The strategy for construction of plasmid templates is shown in Figure 1. For each of the genes of interest, 0.6-1.2 kb of each flank was amplified. The PCR primers are listed in Table 1, and the PCR parameters used are shown in Table 2. The primers were designed manually from the published F. graminearum genome (Cuomo et al., 2007). Primer P2 incorporated a 3' EcoRI recognition site, and primer P3 included a 5' BamHI recognition sequence. These same RE were used to isolate the hygromycin phosphotransferase (hph) selectable marker gene from the donor plasmid pBSHyg2. To produce pBSHyg2, the hph gene with the TrpC promoter was first recovered from the plasmid pCB1636 (Sweigard et al., 1997) by digesting with SalI. Then, the Sal1 fragment was blunt-ended and ligated into the pBluescript plasmid digested with SmaI. PCR amplicons and the restricted hph gene fragment from pBSHyg2 were gel-purified with a gel extraction kit (QIAquick Gel Extraction Kit, Qiagen). Amplicons were digested with the appropriate RE and then gel-purified. The restricted fragments were ligated in a 1:2:1 molar ratio with T4 ligase (Invitrogen) at 16°C overnight. The ligase was inactivated at 70°C for 20 min, and the reactions were diluted 1000-fold and used as PCR templates with primers P1 and P4. The resulting amplicons were gel-purified and cloned into the pGEM T-easy plasmid (Promega #PR-A1360). Before cloning, "A-tails" were added by incubating 15 µl of each gel-purified amplicon with 0.4 µl Taq polymerase (Invitrogen 5U/µl #18-038-042), 2 µl 10x PCR buffer, 0.6 µl 50 mM MgCl2 and 2 µl of 10 mM dATP for 30 min at 72°C. The plasmids were introduced into electro-competent DH5α E. coli cells, and clones were confirmed by restriction digestion and sequencing, then stored as 15% glycerol stocks at -80°C.
PCR amplification of split-marker fragments, and intact-marker replacement cassettes: The plasmids were used as templates for generation of SM fragments and IM replacement cassettes (Figure 1). PCR reactions consisted of a 25 µl total reaction volume including 2.5 µl of 10X Phusion PCR buffer, 2.5 µl of 25 mM MgCl2, 2 µl of 10 mM dNTP mix, 1µl of each primer (20 nM) and 1 µl of template DNA (20-50 ng/µl). The thermocycling parameters were as follows: initial denaturation for 3 min at 94°C, followed by 35 cycles of denaturation at 94°C for 30 sec, annealing at 60°C for 20 sec, and extension at 72°C for 1 min/kb, with a final extension for 7 min at 72°C.
Preparation of fungal protoplasts: Protoplast isolation and transformation protocols were modified from methods used for transformation of Aspergillus parasiticus (Skory et al. 1990;F. Trail, personal communication). Macroconidia were harvested from 7-day-old cultures, washed, and resuspended at a concentration of 1x10 6 spores/ml. A 100 ml aliquot of YEPD was inoculated with 2 ml of the spore solution. The culture was incubated for 10-14 hours at 25°C with constant agitation (175 rpm). Mycelia were harvested by filtration, washed with sterile water, and treated with 20 ml of protoplasting buffer [500 mg Driselase D9515 (MilliporeSigma); 100 mg lysing enzyme from Trichoderma harzianum L1412 (MilliporeSigma); in 20 ml of 1.2 M KCl, filter sterilized]. Mycelium was incubated in the protoplasting buffer for 45-90 minutes at 37°C with gentle agitation (50-80 rpm). When most hyphae had released protoplasts, the solution was filtered through a 30 µm Nitex nylon membrane into sterile 50 ml centrifuge tubes. Protoplasts were pelleted by centrifugation (1729 x g) at room temperature for 5 min, then gently resuspended in STC buffer (1.2 M sorbitol; 10 mM Tris-HCl, pH 8.0; 50 mM CaCl2). Protoplasts were pelleted once more, and then resuspended in STC at a concentration of 1 x 10 8 protoplasts/ml.
Fungal transformation: For each transformation, 100 µl of the freshly prepared protoplast suspension was mixed with 100 µl STC buffer, 50 µl freshly prepared and filter-sterilized 30% PEG solution (8000 polyethylene glycol, MilliporeSigma, P2139) dissolved in a buffer (10 mM Tris-HCl, pH 8.0; 50 mM CaCl2) and 10 µl of DNA (1-3 µg). Protoplasts were transformed by IM or SM protocols, or with linearized MAT1 and MAT1-2-1 plasmids. For SM fragments, 6 µl of both fragments were combined first, and then 10 µl of the mixture was added to the transformation. The reactions were incubated at room temperature for 20 min. Two ml of 30% PEG Solution was added, and incubation continued for an additional 5 min. Four ml of STC buffer was added, and gently mixed. The transformation reaction was then added to 250 ml of cooled regeneration medium (RM) [1.0 g yeast extract; 1.0 g N-Z-Amine AS (N4517 MilliporeSigma); 7.4 g agar; and 271 g sucrose per liter]. The medium was mixed gently and aliquoted into three 100 mm Petri plates. Protoplasts were regenerated for 12-15 hours, and then plates were overlaid with 15 ml of RM amended with 150 µg/ml hygromycin B. Transformants usually appeared within 4-7 days, when they were transferred to potato dextrose agar (Difco ® ) containing 450 µg/ml hygromycin B.
Analysis of transformants: Transformants were single-spored, and genomic DNA extracted from each strain was evaluated by using Southern hybridization (Southern, 2006) to characterize integration events. The fertility phenotypes of gene deletion mutants were assessed on carrot agar (Leslie and Summerell, 2006). The colonies were incubated for 5 days at 23°C with constant fluorescent light. When the colonies covered the plate, 0.5-1.0 ml of 2.5% Tween 60 was applied to the surface of the plate, and the mycelia were flattened by rubbing gently with a sterile glass rod to induce perithecial production. Presence and appearance of perithecia, as well as presence of asci containing ascospores, were assessed 14 days post-induction.

Results and Discussion
The mating-type locus of F. graminearum, known as MAT1, is a complex locus that contains four individual open reading frames (Yun et al., 2000). In this study, three DNA intervals from the MAT1 locus were targeted for deletion. These included the entire MAT1 locus, the MAT1-1-1 gene, and the MAT1-2-1 gene. SM and IM constructs were produced for each interval and used in transformation experiments. Each experiment generated between 50 and 70 hygromycin-resistant transformants within 4-7 days, from which 20-25 individual transformants were arbitrarily chosen for further analysis. There were no noticeable differences among the constructs in the number or appearance of the transformants they produced on regeneration medium.
Southern hybridization analysis revealed the presence of three different types of integrations among the recovered transformants; these included double-point HR resulting from HDR and producing a gene deletion (D-HR), single-point HR producing an integration of the transformation construct adjacent to the original gene (S-HR), and ectopic integration of the transformation construct by NHEJ (Figure 2).
Both types of construct (SM and IM) generated gene deletion (D-HR) strains at rates ranging from 24-75%. The SM protocol produced more deletion strains than IM for the two longest sequences, MAT1 (6070 bp) and MAT1-1-1 (1091 bp). However, deletion efficiencies for the smallest MAT1-2-1 sequence (865 bp) were similar for SM versus IM (Table 3). In yeast, increase in GC content of the targeted DNA, comprising the ORF to be deleted and the flanking DNA included in the disruption construct, resulted in increased representation of HR transformants (Gray and Honigberg, 2001). We noticed a similar trend in our study. Overall, MAT1 (48.6% GC) resulted in 32% D-HR, MAT1-1-1 (51.2% GC) produced 56% D-HR, while MAT1-2-1 (52.7% GC) yielded 67% D-HR.
Among the strains with D-HR events resulting in gene deletions, approximately one in ten had additional ectopic (NHEJ) integrations of the DNA (Table 3, Figure 2). Such strains may be less desirable since additional integrations cause off-site mutations that could independently affect the phenotypes of the transformants (Weld et al., 2006). The SM and IM protocols both generated a similar proportion of strains with additional ectopic integrations (13% for SM, 12% for IM).
SM produced more strains with S-HR events than IM (Table 3). Separate hybridizations with each flank of the SM construct pairs (not shown) suggested that SM also produced more independent NHEJ integrations overall than IM. Of the NHEJ integrations involving SM, 18% comprised just one of the SM flanks. Of those single flank integrations, 74% were of the upstream flank. One possible explanation for the larger relative frequency of NHEJ events with SM is the increased number of recombinogenic ends in the transforming DNA. Transformation experiments in which the template plasmids were linearized within the plasmid sequence, thus avoiding creation of homologous recombinogenic ends, produced fewer integrations, supporting this hypothesis (Table 3). Single ectopic insertions can be more easily removed by backcrossing than multiple integrations, making them less problematic for research purposes. However, we do not know whether transformation with either method also produces unmarked off-site mutations. This question will require further study.
Deletion of any of the three MAT sequences by either SM or IM produced transformants that were completely infertile and could reproduce only asexually, as expected. There were no apparent differences related to the deletion approach used. MAT1 and MAT1-1-1 deletion strains produced numerous small perithecial initials containing no ascospores after induction (Figure 3). Most MAT1-2-1 deletion strains produced few or no perithecial initials and had a thick layer of aerial mycelium that was easily removed from the agar (Figure 3). These phenotypes were consistent with other descriptions of similar knockouts in F. graminearum . Strains without HR events, whether with single or multiple ectopic integrations, had wild type homothallic fertility phenotypes. The perithecia were of normal size and contained ascospores (not shown).
The SM and IM methods both efficiently produced gene deletion mutants in F. graminearum strain PH-1. SM produced deletions with a higher efficiency for the two largest sequences but was similar to IM for the smallest. Fusarium graminearum has a relatively high rate of HR compared to other fungi (Maier et al., 2005;Trail, 2009). In cases where HR is less efficient, SM could provide more of an advantage. The SM protocol facilitates high throughput gene deletion studies, and that convenience provides an additional compelling reason to use SM versus IM. However, we observed that SM produces higher numbers of independent ectopic integrations, including integrations that involved only one of the two flanks, and this may result in more variability among individual transformants due to off-site mutations, as well as complicating their removal by backcrossing. Although PCR is more convenient for screening transformants for strains in which the gene of interest has been successfully deleted, Southern hybridizations are helpful for the identification of strains with fewer ectopic integrations for experimental purposes.