A Simple CRISPR/Cas9 System for Efficiently Targeting Genes of Aspergillus Section Flavi Species, Aspergillus nidulans, Aspergillus fumigatus, Aspergillus terreus, and Aspergillus niger

ABSTRACT

For Aspergillus flavus, a pathogen of considerable economic and health concern, successful gene knockout work for more than a decade has relied nearly exclusively on using nonhomologous end-joining pathway (NHEJ)-deficient recipients via forced double-crossover recombination of homologous sequences. In this study, a simple CRISPR/Cas9 (clustered regularly interspaced short palindromic repeats/CRISPR-associated nuclease) genome editing system that gave extremely high (>95%) gene-targeting frequencies in A. flavus was developed. It contained a shortened Aspergillus nidulans AMA1 autonomously replicating sequence that maintained good transformation frequencies and Aspergillus oryzae ptrA as the selection marker for pyrithiamine resistance. Expression of the codon-optimized cas9 gene was driven by the A. nidulans gpdA promoter and trpC terminator. Expression of single guide RNA (sgRNA) cassettes was controlled by the A. flavus U6 promoter and terminator. The high transformation and gene-targeting frequencies of this system made generation of A. flavus gene knockouts with or without phenotypic changes effortless. Additionally, multiple-gene knockouts of A. flavus conidial pigment genes (olgA/copT/wA or olgA/yA/wA) were quickly generated by a sequential approach. Cotransforming sgRNA vectors targeting A. flavus kojA, yA, and wA gave 52%, 40%, and 8% of single-, double-, and triple-gene knockouts, respectively. The system was readily applicable to other section Flavi aspergilli (A. parasiticus, A. oryzae, A. sojae, A. nomius, A. bombycis, and A. pseudotamarii) with comparable transformation and gene-targeting efficiencies. Moreover, it gave satisfactory gene-targeting efficiencies (>90%) in A. nidulans (section Nidulantes), A. fumigatus (section Fumigati), A. terreus (section Terrei), and A. niger (section Nigri). It likely will have a broad application in aspergilli.

IMPORTANCE CRISPR/Cas9 genome editing systems have been developed for many aspergilli. Reported gene-targeting efficiencies vary greatly and are dependent on delivery methods, repair mechanisms of induced double-stranded breaks, selection markers, and genetic backgrounds of transformation recipient strains. They are also mostly strain specific or species specific. This developed system is highly efficient and allows knocking out multiple genes in A. flavus efficiently either by sequential transformation or by cotransformation of individual sgRNA vectors if desired. It is readily applicable to section Flavi species and aspergilli in other sections (“section” is a taxonomic rank between genus and species). This cross-Aspergillus section system is for wild-type isolates and does not require homologous donor DNAs to be added, NHEJ-deficient strains to be created, or forced recycling of knockout recipients to be performed for multiple-gene targeting. Hence, it simplifies and expedites the gene-targeting process significantly.

INTRODUCTION

Filamentous fungi occupy different niches in natural habitats and play a critical role in the balance of ecosystems. They are also related to health and diseases of humans and animals, important for industrial processing and agricultural production, and capable of producing a wide array of secondary metabolites with diverse biological activities that are either beneficial or detrimental. Members of the Aspergillus genus are ubiquitous in nature. There are over 500 recognized Aspergillus species described in the Catalogue of Life (https://www.catalogueoflife.org/), and they are divided into subgenera and sections (1). Aspergillus is characterized by its distinct reproductive structure called an aspergillum on which asexual conidia are borne (2). Some aspergilli also produce sclerotia, which are resting structures formed by the aggregation of hyphae into dense, pigmented structures (3). Aspergillus flavus belongs to section Flavi. It is a soilborne saprophyte and an opportunistic pathogen of humans and plants. Additionally, it can produce hepatocarcinogenic aflatoxin. Asexual conidia and sclerotia are its main propagules in the field. The A. flavus population consists of two major morphotypes. The S-type isolates produce many small sclerotia but conidiate poorly, whereas the L-type isolates produce few, large sclerotia or none but abundant conidia on most agar media (4). Extensive studies on aflatoxin production and asexual development have been conducted (3, 5, 6).

The powerful genome editing technology based on CRISPR/Cas9 nuclease has been used extensively to study and modify gene functions in various biological systems (7). In this approach, a piece of RNA called single guide RNA (sgRNA) that recognizes the target DNA sequence and directs expressed Cas9 to edit is generated. The sgRNA molecule consists of a crRNA (CRISPR RNA) protospacer sequence and a tracrRNA (transactivating CRISPR RNA). The crRNA protospacer sequence is about 20 nucleotides long but can vary in size. It is situated right before the protospacer-adjacent motif (PAM) site and complementary to the target DNA sequence that is on the opposite strand. The canonical PAM site for Streptococcus pyogenes Cas9 is NGG. However, different PAM sites are recognized by Cas9 nucleases of Streptococcus thermophilus, Neisseria meningitidis, and Treponema denticola (8). The tracrRNA serves as a binding scaffold for Cas9. While in nature crRNA and tracrRNA exist as two separate molecules, engineered fused sgRNA has become the most popular format for CRISPR/Cas9 genome editing (9). Ca9 acts as a pair of “molecular scissors” that cut double-stranded DNA at a specified sequence location, and thus, bits of DNA can be modified (7).

In the postgenomics era, elucidation of fungal gene functions by high-throughput analyses requires efficient gene-targeting systems, which reduce time and labor in generating correct gene knockouts. The availability of CRIPSR/Cas9 systems is a significant advance over the previous technology of using fungal transformation recipient strains whose component genes of the nonhomologous end-joining (NHEJ) pathway, such as those encoding DNA-dependent protein kinase subunits Ku70 and Ku80 and DNA ligase IV, have been disabled (10–14). The latter approach was designed to increase gene knockout frequencies under the NHEJ-deficient genetic background by a forced double-crossover recombinational mechanism between homologous sequences of selection marker-containing transforming DNA fragments and targeted genes of interest. However, the two technologies face the same dilemma of having limited numbers of selection markers when knocking out multiple genes is required, and recycling transformation recipient strains becomes a must.

CRISPR/Cas9 systems have been developed for many fungal genera, including Neurospora, Trichoderma, Penicillium, Fusarium, and Aspergillus (15). Current system formats consist of (i) conventional plasmid vectors containing cas9 and/or sgRNA expression cassettes along with or without homologous donor DNA substrates and (ii) the synthetic approach which directly delivers ribonucleoprotein (RNP) complexes, Cas9 and sgRNAs, into fungal protoplasts. Cross-fungal species systems have been reported but not frequently (16, 17). Transformation and gene-targeting efficiencies of already-developed multiple-species systems are usually not satisfactory. To increase the efficiencies, homologous donor DNAs (plasmids or DNA fragments) often have to be included, especially when NHEJ-deficient recipient strains are employed. A very recent review gives an excellent summary of delivery methods, DNA repair mechanisms, and targeting efficiencies on established CRISPR/Cas9 systems for aspergilli (18). The lack of a simple, efficient, and universal system means current CRISPR/Cas9 technology for aspergilli still had much room for improvement.

In this study, a modified CRISPR/Cas9 version for A. flavus based on available literature and newly acquired information about the A. flavus U6 promoter and terminator and AMA1 functional regions was developed. In short, sgRNA expression cassettes prepared by fusion PCR were cloned into a universal CRISPR/Cas9 plasmid vector. The system showed very high transformation and gene-targeting efficiencies in section Flavi aspergilli including A. flavus, A. parasiticus, A. oryzae, A. sojae, A. nomius, A. bombycis, and A. pseudotamarii. Knocking out multiple genes in A. flavus was achieved in a very short time frame. Additionally, the system was applicable to A. nidulans, A. terreus, A. fumigatus, and A. niger with satisfactory gene-targeting frequencies.

RESULTS

Localization of A. flavus U6 promoter and terminator sequences.

A search of the noncoding RNA sequence database RNAcentral (https://rnacentral.org/) for U6 spliceosomal RNA of Aspergillus species showed that U6 spliceosomal RNA is Rfam (family) 26 (RF00026). Two A. flavus NRRL3357 genomic DNA sequences that matched the U6 RNA model sequence GCCUUCGGGCAUUUGGUCAAUUUGAAACGAUACAGAGAAGAUUAGCAUGGCCCCUGCACUAAGGAUGACACGCUCAAUCAAAGAGAAGCUACCAGUUUUU were retrieved. One was on contig EQ963476 from nucleotide 1669489 to nucleotide 1669588; another was on contig EQ963483 from nucleotide 656522 to nucleotide 656621 (see Fig. S1 in the supplemental material). They shared 86% DNA identity. An 825-bp region on EQ963483 that contained the U6-encoding sequence was extracted and analyzed. The 500-bp upstream and the 119-bp downstream regions of the U6 sequence were defined as the promoter and the terminator, respectively. These sequences were used together with a 76-bp scaffold sequence for the assembly of sgRNA expression vectors (Table 1 and Fig. S1, bottom).

TABLE 1.

Sequences of universal primers and protospacers for constructing vectors expressing sgRNA cassettes

Primers	Sequence (5′→3′)
U6-F-Pa	ATACTGCAGTTCTCTTTAGAATTCAACTGTGGGT
U6-R-Ka	TATGGTACCACATATTTAAAAAAAGTCTCCTGCC
gdFd_gene	Fd_seqGTTTTAGAGCTAGAAATAGCAAGTTAA
gdRC_gene	RC_seqACTTGTTCTTCTTTACAATGATTTATATACC
	Protospacer sequencec
Target genes	Fd_seq	RC_seq
wA1 b	TGGATCTACTGGCGCGTCAC	GTGACGCGCCAGTAGATCCA
wA3 b	GAAAGATGCCTCGCAGCTTAT	ATAAGCTGCGAGGCATCTTTC
yA b	CGCCAAATGATTCTCACTAA	TTAGTGAGAATCATTTGGCG
copT	TACCTTCGCCCTCTCGACTT	CATCGGGTGCTCGTCGATCG
olgA	CGATCGACGAGCACCCGATG	CATCGGGTGCTCGTCGATCG
kojA	GGTGGAATGAGCGGCAAAGT	ACTTTGCCGCTCATTCCACC
AFLA_010700	GCTACGAACTTCAGTTGCAA	TTGCAACTGAAGTTCGTAGC
AFLA_026030	GGCGGTGTTGTCTATGATGT	ACATCATAGACAACACCGCC
AFLA_031450	TGGACGACCAGCCCAACAAC	GTTGTTGGGCTGGTCGTCCA
AFLA_064960	CGATTAAACATACCATGTAT	ATACATGGTATGTTTAATCG
AFLA_089270	AGAATCAATTCCTCCTTCAG	CTGAAGGAGGAATTGATTCT
AFLA_106990	CTTAGGTCGGACATGTCGTC	GACGACATGTCCGACCTAAG
AFLA_112760	CTAAGAAGTTTGACAGTCCCG	CGGGACTGTCAAACTTCTTAG
AFLA_127260	CCCCCGTCTCGAGGCTATGA	TCATAGCCTCGAGACGGGGG
Anid_yA d	GGCGGAGTATCATAACATCG e	CGATGTTATGATACTCCGCC
Anig_fwnA d	AGTGGGATCTCAAGAACTAC f	GTAGTTCTTGAGATCCCACT
Afum_wA d	CTCAGCGCACGCTCTTCCCG g	CGGGAAGAGCGTGCGCTGAG
Ater_melA d	CGTCTACCCTTATCACCGAC	GTCGGTGATAAGGGTAGACG

^aU6-F-P and U6-R-K were derived from A. flavus U6 promoter and terminator sequences (see Fig. S1) with tagged restriction sites of PstI (CTGCAG) and KpnI (GGTACC), respectively.

^bSee the work of Katayama et al. (40).

^cFd_seq, forward sequence; RC_seq, reverse complement sequence.

^dAnid, A. nidulans; Anig, A. niger; Afum, A. fumigatus; Ater, A. terreus. Others are A. flavus.

^eSee the work of Nødvig et al. (17).

^fSee the work of Leynaud-Kieffer et al. (49).

^gSee the work of Fuller et al. (=pksP) (45).

Functional regions of the AMA1 sequence that maintained high transformation frequencies.

The 5.2-kb HindIII AMA1 fragment of A. nidulans in pPTRII (Fig. 1A) and other derived vectors greatly enhances transformation frequencies (19, 20). It is an inverted duplication of a low-copy-number genomic repeat (21). It has been noted that using only half of the AMA1 sequence generated by PCR caused significant decreases in transformation frequencies (22, 23). In this study, the minimum AMA1 functional region in pPTRII that maintained a high transformation frequency under pyrithiamine selection was located in the 2.6-kb HindIII-PstI fragment (Fig. 1B). Therefore, the resulting plasmid, pPT70, derived from PstI digestion of pPTRII followed by self-ligation, was used as a backbone vector for CRISPR vectors.

FIG 1.

(A) Restriction map of pPTRII. The plasmid was constructed by blunt ligation of the 5.2-kb HindIII fragment (AMA1), which contains an inverted duplication of a repeat sequence, into the NaeI site of pPTRI (20). (B) Localization of a minimum functional region of AMA1 that was able to maintain high transformation frequency by restriction and transformation analyses. Fragment iii was inferred from the vector, pAsp-AMA-gpdA-ptr. Thick lines represent remaining portions of the 5.2-kb HindIII fragment. Transformation frequency: “−” indicates no pyrithiamine-resistant transformants and “+++” indicates that numbers of transformants were comparable to that of pPTRII. See Fig. S3, which contains the complete sequence of pPTRII, in the supplemental material for details. For panel A, locations of restriction sites, ptrA, and AmpR as well as the two functional regions of AMA1 shown in panel B are indicated.

Effects of genetic backgrounds and promoters driving cas9 expression.

Nearly all primary transformants generated from vectors (Fig. 2) for targeting wA, yA, and copT showed known phenotypes, that is, white, gold, and yellow colonies, respectively. For olgA transformants, in addition to dull green colonies, approximately 22% of the transformants were colonies producing white conidia, identical to those of wA-targeted transformants. Although transformants of CA14 and KuPG#1 had identical colony morphologies, transformant numbers for the KuPG#1 sets (wA1 and wA3) were drastically decreased up to 200-fold using gpdA-driven cas9 constructs (Table 2). However, neither the promoters (gpdAp and GPIp) used to express the cas9 gene nor the additionally incorporated region of AMA1 (copT and olgA sets) had any bearing on the transformation frequencies for CA14 and KuPG#1. The defect in the NHEJ pathway likely was the intrinsic determinant for the observed decreases in total transformant numbers. Although KuPG#1 sets (wA1, wA3, copT, and olgA) yielded only very low numbers of primary transformants, they were all knockouts as evidenced by changes in conidial colors. For CA14, gene targeting frequencies were comparable since nearly all their primary transformants exhibited expected knockout morphologies (Table 2).

FIG 2.

(A) An AMA1-shortened and ptrA-based cas9 vector (Addgene plasmid no. 191016) for cloning PCR fragments that express sgRNA cassettes to target Aspergillus genes. KpnI and PstI are unique sites for cloning expression cassettes. (B) A CRISPR/Cas9 construct for targeting the A. flavus yA gene (Addgene plasmid no. 191015). It was also the DNA template used for generating PCR fragments, which contained the A. flavus U6 promoter and terminator for expressing sgRNA cassettes, inserted into the above-described pAsp-AMA-gpdA-ptr cloning vector. Both vectors have been deposited at Addgene, a public plasmid repository. Their complete sequences and annotated features can be found at the website (https://www.addgene.org/) by searching the plasmid numbers.

TABLE 2.

Transformation and gene-targeting efficiencies in A. flavus strains of different genetic backgrounds using cas9 expression vectors driven by different promoters

Strain	Promoter_target	Transformantc	Vector backbone
KuPG#1	gpdA_none	350wt	pAf-crispre
	gpdA_wA1a	6w/6, 4w/4	pAf-crispr
	GPIp_wA1b	2w/3	pAf-crispr
	gpdA_wA3	4w/4	pAf-crispr
	GPIp_wA3	6w/6	pAf-crispr
	gpdA_yA	NDd
	gpdA_copT	9y/9	pAsp-AMA-gpdA-ptrf
	gpdA_olgA	3olg/3	pAsp-AMA-gpdA-ptr
CA14	gpdA_none	675wt	pAsp-AMA-gpdA-ptr
	gpdA_wA1	403w/403	Af-crispr
	GPIp_wA1	308w/308	Af-crispr
	gpdA_wA3	897w/900	Af-crispr
	GPIp_wA3	154w/164	Af-crispr
	gpdA_yA	762gd/762	Af-crispr
	gpdA_copT	208y/211	pAsp-AMA-gpdA-ptr
	gpdA_olgA	113olg/144, 31w/144	pAsp-AMA-gpdA-ptr

^aThe promoter for cas9 is from the A. nidulans gpdA gene.

^bThe promoter for cas9 is from the A. flavus GPI gene (AFLA_113120).

^cApproximately 300 ng vector DNA and 2.0 × 10⁶ protoplasts were used in each transformation. wt, colonies producing wild-type yellowish green conidia; w, colonies producing white conidia; y, colonies producing yellow conidia; gd, colonies producing gold conidia when aged; olg, colonies producing dark olive-green conidia.

^dND, not done.

^eThe vector was based on pP70 and contained cas9, SV40 NLS sequence at 5′ and 3′ ends, and the 2.6-kb HindIII-PstI fragment of AMA1 (Fig. 1 and Fig. S3).

^fThe CRISPR cloning vector, pAsp_AMA_gpdA_ptr, contained an additional 292 nucleotides beyond the PstI site and including the NaeI site in AMA1 (Fig. 1 and 2 and Fig. S3).

Molecular defects associated with gene-targeting events.

A combination of PCR and sequencing analyses was performed to determine associated defects in wA, olgA, and yA genes. For CA14, gene-targeting events occurring at or around the wA1 sequence often were small deletions or an insertion, but wA1 off-target events tended to be large deletions (Fig. 3). The insertion (278 bp in size) was confirmed to be a portion of the cas9 gene. For the wA3 set, two of the 10 white-spored transformants examined had off-target large deletions. Intriguingly, the remaining eight had intact sequences from 0.5 kb upstream to 1.0 kb downstream of the target site; these apparent off-target defects were not investigated further. Likewise, five of 10 examined yellow/gold yA knockouts had small deletions (1, 13, 26, 29, and 30 bp is size). The remaining five yA knockouts not giving PCR products likely had defects extending beyond the primer sequence location(s). It appeared that genetic background also affected associated molecular defect patterns. Correct targeting of wA gene sites (wA1 and wA3) was achieved in KuPG#1, but defects were commonly large deletions. For example, one deletion was even up to 4.9 kb (Fig. 3). In the A. flavus genome, the olgA gene resides right next to the wA gene (24). Approximately 20% of the olgA gene transformants produced white (nonpigmented) conidia (Table 2). Of five white olgA transformants analyzed, three resulted from large deletions (3.3, 4.8, and 7.0 kb) that extended to the wA coding region (Fig. 3). The remaining two likely had much larger deletions, but the defects were not resolved even after extensive efforts using PCR with various wA sequence primers. Figure S6 shows sequences and locations of primers used for the wA and olgA work, including those for confirming large deletions.

FIG 3.

Graphic representation of molecular defects generated by CRISPR/Cas9 gene targeting. Squares are small deletions, and the short line with a downward arrow is an insertion. Deletions larger than 0.5 kb are drawn to scale. The lower panel shows sizes of the insertion and deletions. “k” indicates that the transformation recipient strain is A. flavus KuPG#1, which has an NHEJ-deficient (Δku70) genetic background; others are gene knockouts derived from wild-type CA14. Red and green triangles indicate locations of wA1 and wA3 targeting sites, respectively. The blue star is the location of the olgA targeting site. Large deletions extending to wA in three of five randomly selected white olgA knockouts were determined. Sizes of clustering wA and olgA and their distances are shown in base pairs.

Targeting A. flavus genes whose defects and associated phenotypic changes were not previously known.

Unlike genes involved in conidial pigment biosynthesis, not all gene knockouts will show phenotypes clearly discernible from those of the wild type and be easily selected. Therefore, a set of eight genes whose defects and resulting phenotypes were not known were tested (Table 3). Great numbers of transformants were obtained for all but AFLA_089270 under the same experimental conditions. Eight transformants from each experiment were analyzed by PCR, and products were sequenced. Only 76.5% of them gave positive PCR products from check primers (Table S1), and the lack of PCR products was not due to poor genomic DNA quality (data not shown). Table 3 summarizes their molecular defects, which included small deletions and insertions. Knockouts of the two genes, AFLA_089270 and AFLA_112760, annotated by NCBI to encode a bZIP transcription factor that probably mediated unfolded protein response and a C₂H₂ zinc finger domain protein, respectively, exhibited distinct differences in terms of conidiation on potato dextrose agar (PDA) plates in the dark (Fig. 4A). Light and V8 medium are known to promote conidiation and at the same time repress sclerotial formation of A. flavus, whereas Wickerham agar medium (WKH) is conducive to sclerotial production, but this medium effect is counteracted by light. Intriguingly, the AFLA_089270 knockout produced many sclerotia on PDA and WKH plates under constant light. The knockout on the V8 plate under constant light barely produced conidia and showed only mycelial growth with a few sclerotia, but it produced kojic acid like the wild-type CA14 (Fig. 4B).

TABLE 3.

Molecular defects in A. flavus genes whose resulting phenotypic changes were not previously known

AFLA_ID	Transa	Sequence defects (bp)		PCR (Y/N)b	Note
		Deletion	Insertion
010700	810	5, 21, 53, 592		4 /4
026330	456	1, 1, 1, 1, 9, 12, 39		7/1	1 bp (all identical)
031450	630	32, 403	1, 1, 1, 1	6/2	1 bp (all identical)
064960	>1,000	1, 5, 11, 72, 222		5/3
089270	132	3, 3, 3, 3, 3, 3, 3		7/1	3 bp (all identical in-frame deletion)
106990	>1,000	10, 12	113, 198, 198	5/3	113 bp from chromosome I; 198 bp (identical, from ampR gene)
112760	373	14, 14, 14, 14, 14, 14, 20		7/1	14 bp (all identical)
127260	>1,000	9, 9, 9, 9, 9, 9, 9, 9		8/0	3 types of 9-bp deletion

^aNumber of transformants; 300 ng vector DNA and 2.0 × 10⁶ protoplasts were used in each transformation.

^bEight transformants from each gene knockout experiment were randomly selected for PCR analysis. PCR products were generated by check primers, which were located about 0.5 kb up- and downstream of target (protospacer sequence, N20) sites. Y, positive PCR products, and N, no PCR products.

FIG 4.

Targeting A. flavus genes whose defects and resulting phenotypes were not previously known. (A) Representative colony morphologies of knockouts on PDA plates grown at 30°C in the dark for 4 days. The dull-green areas indicate conidia produced. Morphologies of knockouts AFLA_010700, AFLA_026030, AFLA_064960, AFLA_106990, and AFLA_127260 resembled that of wild-type CA14 (not shown here). (B) Growth of wild-type CA14 and the AFLA_089270 knockout on different media at 30°C under constant light for 7 days. The dark brown granules are mature sclerotia. Both CA14 and the AFLA_089270 knockout produced kojic acid.

The CRISPR/Cas9 system for L- and S-morphotype A. flavus isolates.

Two S-morphotype isolates, GA10-18s and SRRC1576 (=CA42), and another L-morphotype isolate, NM1-3, were further tested with the gpdA_wA1 construct to show that this system is universally applicable to other A. flavus isolates. High transformation and targeting efficiencies were obtained for all sets (Table 4). Figure 5 shows representative changes in colony morphologies of knockouts on PDA and V8 plates under dark and light conditions. PDA favored sclerotial production of the S-type isolates in the dark. Light and V8 medium promoted conidiation of the S- and L-type isolates to different extents. The combination of constant light and V8 medium had a synergistic effect on conidial production by the S-type isolates. Knockouts of the S-type isolates produced white conidia and some sclerotia on V8 medium under constant light in contrast to the abundant yellowish green conidia produced by the wild-type S-type isolates.

TABLE 4.

Gene-targeting and transformation efficiencies of various aspergilli

Species	Strain	Gene	Efficiencyc	Note
A. flavus	NM1-3a	wA	222/6.0 × 105	L-morphotype
	SRRC1576	wA	143/1.5 × 106	S-morphotype, =CA42
	GA10-8sa	wA	443/4.0 × 105	S-morphotype
A. oryzae	RIB40	wA	>597/2.0 × 106
A. parasiticus	SU1-N3	olgA	>555/2.0 × 106	nit− mutant of SRRC143
	RHN1	olgA	198/2.0 × 106	nit− mutant of SRRC2043
A. sojae	SRRC1121b	olgA	217d/2.0 × 106	=IFO 4244 (type)
A. nomius	SRRC0375	wA	>795/2.0 × 106	=NRRL13137 (type)
A. bombycis	SRRC0513	copT	90/2.0 × 106	=NRRL28900
A. pseudotamarii	SRRC2420	wA	>750/2.0 × 106	=NRRL25517 (type)
A. nidulans	SRRC73	yA	135/2.0 × 106; 24/2.0 × 106	=FGSC A4
A. fumigatus	SRRC2569	wA	32/1.2 × 106; 9/1.0 × 105	=Af293
A. terreus	SRRC2634	melA	32/1.5 × 106; 31/5.0 × 105	=NIH 2624
A. niger	SRRC2379	fwnA	4/2.0 × 106; 16/1.0 × 106	=ATCC 16888 (type)

^aUSDA, ARS, National Peanut Research Laboratory (Dawson, GA, USA) culture collection.

^bSRRC, Southern Regional Research Center culture collection.

^cNumbers of transformants generated from protoplasts used in transformation. For each transformation, approximately 300 ng of vector DNA was used. When protoplasts generated were fewer than 2.0 × 10⁶, all were used in the experiments.

^dApproximately 20% transformants were white colonies (Fig. 7A and Fig. S9).

FIG 5.

Morphologies of wild types and wA knockouts of A. flavus L- and S-type isolates. The L-type isolate is NM1-3 (left). S-type isolates are GA10-18s (top) and CA42 (right). The S-type wild types and knockouts produced copious sclerotia (dark brown granules) but sparse conidia on PDA plates in the dark; constant light promoted their conidiation somewhat on PDA plates. They produced abundant conidia on V8 medium plates when grown under constant light. The L-type isolate produced abundant conidia regardless of growth conditions and medium types. All three wA knockouts produced white conidia, whereas the respective wild types produced yellowish green conidia.

Strategies for targeting multiple genes in A. flavus.

Strains containing AMA1- and ptrA-based plasmids easily lost their pyrithiamine resistance under nonselective growth conditions. For CA14, the rate of spontaneous mutation to pyrithiamine resistance was found be less than 10⁻⁶ when knockout spores harvested from V8 plates were spread onto Czapek solution agar plates containing pyrithiamine and checked for growth (data not shown). Therefore, an olgA knockout grown on V8 plates was used for production of pyrithiamine-sensitive spores. For both second (targeting copT and yA) and third (targeting wA) rounds, high transformation (>800 transformants) and gene-targeting (>95%) frequencies were maintained. Triple gene knockouts (olgA, copT, and wA; olgA, yA, and wA) were readily generated in a month based on visual selection of knockouts with changed conidial colors (Fig. 6). The consistent high transformation and gene-targeting efficiencies suggested that simultaneously knocking out multiple genes by cotransforming different sgDNA constructs was feasible. As a proof of concept, three single-gene constructs for targeting yA, wA, and kojA, which were located on chromosomes 7, 4, and 5, respectively, were used to cotransform the olgA knockout. From a single trial, over 300 primary transformants on six regeneration plates were obtained. The majority were white (nonpigmented), and a few were yellow or dull green (Fig. 6B). Fifty of them, which included five yellow and two dull green colonies, were used for determining percentages of single-, double-, and triple-gene knockouts. For genes involved in the biosynthesis of conidial pigments, their order is wA→copT→yA→olgA (25). A white knockout derived from the olgA recipient thus could genetically be either a wA mutant or a wA/yA double mutant, but defects in yA should result in only yellow mutants. Hence, yA sequences in white knockouts were examined by PCR amplification followed by sequencing to distinguish wA mutants from wA/yA double mutants. The kojA gene encodes an oxidoreductase directly responsible for kojic acid production (26). Transformants that had a functional kojA gene were evidenced by the production of an orange red chelate compound of kojic acid and iron(III) on kojic acid medium (KAM) plates (Fig. 6B). For example, yellow knockouts of #2 and #50 had an intact and a defective kojA, respectively. The overall percentages for single- (wA, yA, or kojA), double- (wA/yA, wA/kojA, or yA/kojA), and triple-gene knockouts were 52%, 40%, and 8%, respectively (Fig. 6C).

FIG 6.

Targeting multiple genes in A. flavus using a sequential approach (A) or a simultaneous approach (B). (A) An olgA knockout from the first round was used as the recipient strain for the second rounds of targeting copT and yA, respectively. One knockout from each was randomly selected for spore production on V8 plates; they were used as recipient strains for the third rounds that targeted wA. Selection of knockouts was based on known association of conidial gene defects with formation of distinct conidial pigments. (B) Cotransformation of three individual constructs for targeting yA, wA, and kojA. Fifty primary transformants (top left) were transferred onto KAM agar plates for the examination of conidial colors and production of kojic acid (bottom left). (C) The Venn diagram shows numbers of single-, double-, and triple-gene knockouts.

Gene targeting in section Flavi species, A. nidulans, A. fumigatus, A. terreus, and A. niger.

Aspergillus flavus is in the genus section Flavi, which includes more than 30 species (27). Their close genetic relatedness suggests that the developed CRISPR/Cas9 system is applicable to them. Therefore, six other species, A. oryzae, A. parasiticus, A. sojae, A. nomius, A. bombycis, and A. pseudotamarii, for which constructs contained identical protospacer sequences and were available, were selected for experiments (Table S2). As for A. flavus isolates, great numbers of transformants and high gene-targeting frequencies up to nearly 100% were obtained (Table 4), and they displayed expected conidial colors (Fig. S7 to S9). For A. sojae knockouts, approximately 20% of colonies, like those A. flavus olgA knockouts that had large deletions extending to wA, were white colonies (Fig. 7A and Fig. S8). A further evaluation of this system on four aspergilli, A. nidulans, A. fumigatus, A. terreus, and A. niger, which are genetically distant from A. flavus, from two independent trials showed that lower numbers of primary transformants than those from section Flavi species were produced (Table 4 and Fig. S10). Nonetheless, gene-targeting frequencies for selected conidial genes among these primary transformants, conservatively estimated, were greater than 90% as a majority of all exhibited expected changes in colony phenotypes when examined on PDA plates, that is, yellow, white (nonpigmented), or light fawn for yA, wA and melA, and fwnA knockouts, respectively (Fig. 7B).

FIG 7.

Targeting conidial genes of section Flavi species and other aspergilli. (A) Morphologies of four wild-type section Flavi aspergilli on PDA plates (top) and their primary (gene-targeted) transformants on CZ regeneration medium plates (bottom). (B) Morphologies of wild-type A. nidulans, A. fumigatus, A. terreus, and A. niger on PDA plates (top) and their primary transformants transferred onto PDA plates (bottom). The gene identifiers (IDs) for A. nidulans yA, A. fumigatus wA (=pksP), A. terreus melA, and A. niger fwnA are AN6635, ATEG_03563, AFUA_2G17600, and An09g05730, respectively.

DISCUSSION

Using donor DNA, a short fragment homologous to a gene-targeting site, was thought to be required to increase targeting frequencies on different genetic backgrounds of A. fumigatus (28–30) and other aspergilli such as A. nidulans, A. niger, and A. aculeatus (17, 31). This technique is referred to as microhomology-mediated end joining (MMEJ) (32) and has been adopted for other fungi (33). Zhang et al. (29) reported that MMEJ-based CRISPR mutagenesis is independent of the NHEJ pathway. In this study, without using homologous donor DNA fragments A. flavus with either a wild-type or an NHEJ-deficient genetic background had nearly 100% gene-targeting frequencies. However, transformation frequencies on the NHEJ-deficient background were extremely low (Table 2). These low transformation frequencies imply that majority double-stranded breaks induced by CRISPR/Cas9 were not repaired and are lethal. Consistently, previous studies reported that, using only Cas9-sgRNA plasmids without including donor DNAs, no transformants were obtained from an A. niger kusA-deficient strain (34, 35). Incorporation of donor DNA in both cases likely would significantly increase transformation frequencies. In more than a decade, over 100 A. flavus genes have been disrupted by several research groups (3, 5). These gene function studies were carried out nearly exclusively using NHEJ-deficient strains (24, 36). A concern about using NHEJ-deficient strains is that it could have unexpected consequences in DNA repair capacity and promotes error-prone NHEJ (37, 38). For example, a slight increase in the mutation rate for the A. flavus NRRL3357 Δku70 strain was reported, although no aberrant morphological changes were found (39).

Sequence defects caused by CRISPR-mediated gene-targeting events are complex. On the wild-type A. flavus genetic background, most induced mutations in the knockouts were small deletions (Tables 1 and 3), which is often found in other aspergilli (17, 23, 40). However, large deletions not frequently reported for fungi are not uncommon, for example, those of wA and olgA knockouts (Fig. 1). Most recently, deletions larger than 1.0 kb and near 3.0 kb also were reported for A. sojae knockouts (16). Therefore, unresolved gene defects in some of the A. flavus knockouts likely contained large deletions (Table 3, no PCR products). It was estimated that overall, approximately 20% of A. flavus knockouts contained large deletions; some will be off-target types. A binary logistic regression analysis of published CRISPR/Cas9 articles indicates that number and position of mismatches between target and off-target sites are the two major factors affecting the occurrence of off-target effects (41). Numerous approaches have been developed to reduce off-target effects (42). It is not known whether the off-target incidence observed in this study has anything to do with the supposedly high levels of the Cas9 nuclease expressed in fungal cells. Literature regarding this aspect is lacking. The Cas9 levels apparently are related to copy numbers of the transforming vectors and the A. nidulans gpdA promoter used to drive cas9 expression. AMA1 is an inverted duplication of a low-copy-number genomic repeat (21), and AMA1-containing vectors exist as multicopy extrachromosomal elements (plasmid) in cells. The gpdA promoter is a constitutive promoter of the gene that encodes glyceraldehyde-3-phosphate dehydrogenase (GPD), and GPD has been reported to constitute up to 5% of the soluble cellular protein in A. nidulans (43, 44). Therefore, the gpdA promoter is the main determinant for high levels of Cas9 production, and the multicopy vectors in part contribute indirectly to the production. In addition to gpdA, other strong promoters such as amyB (amylase gene) and tef1 (translation elongation factor 1α gene) are frequently employed to construct cas9 expression vectors (17, 22), but off-target events are seldom reported by those studies. In the future, when high-throughput whole-genome sequencing becomes a routine technique, it may be able to provide a definite answer.

Insertion appears to occur less frequently than deletion. Fuller et al. (45) showed that inserted sequences originated from either a portion of or a whole transforming DNA construct. In this work, the inserts in a wA (273 bp, Fig. 1) and two AFLA_106990 (198 bp, Table 2) knockouts indeed came from similar origins. A 2.5-kb insert, which encompassed U6 and partial AMA1 sequences, also was found in an olgA knockout (data not shown). For Trichoderma reesei, chromosomal fragments were found inserted into disrupted loci in addition to a cotransformed plasmid (46). Nødvig et al. (17) also reported that short 60- and 84-bp inserts that caused mutations in A. nidulans yA came from different loci on chromosome V. Similarly, the 113-bp insert in the AFLA_106990 knockout originated from chromosome I (100% nucleotide sequence identity), whereas AFLA_106990 resides on chromosome III. No sequences homologous to respective protospacer sequences were found around integrated sites of these inserts (data not shown). How an NHEJ pathway of wild-type A. flavus integrates or recruits a DNA replication mechanism to repair double-stranded breaks induced by CRISPR/Cas9 remains unclear.

The original A. nidulans autonomously replicating AMA1 fragment has a large size of 5.2 kb (19). In this study, a reduced size of at least 2.6 kb was able to maintain high transformation frequencies and facilitated subsequent vector handling. The derived vectors could consistently give 1,000 to 2,000 transformants per microgram of vector DNA (Tables 2 and 3). For A. niger, shortening AMA1 to 2.5 kb compared to the full-length AMA1 decreased the transformation frequency to about a quarter and gave 25 transformants per microgram of DNA (47); that number at most is about 1/10 of that found for the overall A. flavus transformation frequencies (Tables 2, 3, and 4). Intriguingly, using only half of the AMA1 generated by PCR in A. oryzae and Aspergillus kawachii drastically reduced transformation frequencies (22, 23). In this study, combining a shortened AMA1 with ptrA in vectors expedited sequential deletion of multiple A. flavus genes because the vectors were easily lost under nonselective (without pyrithiamine) growth conditions. It thus circumvents the required recycling steps for transformation recipients and selection markers by counterselection. For example, using 5-fluoroorotic acid and 5-fluoroacetamide for pyrG- and amdS-based NHEJ-deficient systems, respectively (48, 49), both A. flavus S- and L-types are present in the fields. The S-type isolates consistently produce greater amounts of aflatoxin than do L-type isolates. S-type A. flavus has been implicated as the primary causal agent in aflatoxin outbreaks (50). Currently available A. flavus NHEJ-deficient strains are exclusively derived from two L-type isolates, CA14 and NRRL3357 (24, 39). Hence, studies in S-type A. flavus, particularly in aspects related to sclerotial biogenesis, biosynthesis of secondary metabolites, and its tenacious aflatoxigenicity, have been hampered. The developed system should promote and facilitate gene function studies in S-type A. flavus. Its gene-targeting frequencies in A. flavus, conservatively, reached over 95%, and roughly 80% of knockouts would contain small deletions that can be easily determined (Table 3); these targeting frequencies were better than those reported for NHEJ-deficient systems that rely on homology-directed repair (18). This improved CRISPR/Cas9 technology thus could make the dependence of over a decade on A. flavus NHEJ-deficient recipient strains and auxotrophic selection markers a thing of the past.

Using the same backbone CRISPR/Cas9 vectors containing AMA1, double- and triple-gene knockouts were generated in A. niger with good targeting efficiencies (89% and 38%, respectively) from a single cotransformation. However, these frequencies were achieved only when both NHEJ-deficient recipient and homologous donor DNAs were used (35). Likewise, donor DNAs and a tandem sgRNA expression cassette containing different (A. oryzae U6 and A. fumigatus U6-2) promoters were used for another A. niger NHEJ-deficient recipient. But targeting frequencies for single-, double-, and triple-gene knockouts were 31.8%, 54.5%, and 13.6%, respectively (51). These frequencies are comparable to those for A. flavus, but vector design for the former is more complicated. Bundling all sgRNA-encoding DNAs in a single vector lacks flexibility, as reported in another A. niger study that used various tRNA-sgRNA arrays. To access multiple-gene-targeting frequencies, five different tRNA-based sgRNA expression cassettes with significant redundancy had to be constructed because of the concern that using identical copies of tRNA sequences could render sgRNA cassettes unstable (52). For A. oryzae, Katayama et al. (22) showed that only the placement of two sgRNA cassettes into a single vector, not cotransformation of individual sgRNA vectors, could generate double-gene knockouts. A reason for this result likely is due to the authors using the PCR-generated half-AMA1, which lost its autonomously replicating capacity significantly and gave very low numbers of transformants. Cotransformation with individual sgRNA vectors instead of bundling of all sgRNAs in a single cassette further allows combining various vectors for studying gene interactions and saves a tremendous amount of time in vector design and construction because of the simplicity. The cotransformation experiment using three individual gene-targeting constructs to generate triple-gene knockouts did not give a satisfactory result. Nonetheless, nearly half of the primary transformants were double- (40%) or triple-gene (8%) knockouts (Fig. 6B), which means that, if needed, quadruple-gene knockouts can be readily generated from two rounds of transformation by screening a few primary transformants from each round.

Gene knockout using the same backbone CRISPR/Cas9 vector (pFC332), which contains AMA1 and the hygB selection marker, has been shown in section Nigri species (A. aculeatus, A. niger, Aspergillus carbonarius, Aspergillus brasiliensis, and Aspergillus luchuensis). However, targeting frequencies varied greatly, and it failed to yield transformants from Aspergillus tubingensis, a Nigri species (17). In this study, the developed CRISPR system for A. flavus was proven applicable to other section Flavi species (A. parasiticus, A. nomius, A. bombycis, A. pseudotamarii, A. oryzae, and A. sojae) and achieved extremely high gene-targeting frequencies. Similar CRISPR/Cas9 systems for A. oryzae and A. sojae have been reported (16, 22, 40). Nonetheless, the modified version serves as an alternative to them in terms of applicability and simplicity, that is, no forced recycling step or donor DNAs required and better transformation/targeting frequencies. Aspergillus section Flavi includes 33 species that are split into eight clades (27). In addition to A. oryzae and A. sojae, which are widely used in Asian food fermentation, many of them are natural producers of aflatoxins. The current system worked well for aspergilli in the A. flavus clade (A. flavus, A. parasiticus, A. oryzae, and A. sojae), Aspergillus tamarii clade (A. pseudotamarii), and A. nomius clade (A. nomius and A. bombycis). This system should work efficiently in the majority of, if not all other, Flavi aspergilli.

CRISPR/Cas9 vectors for genetically distant cross-species gene targeting were reported by Nødvig et al. (17). However, the sgRNA vectors designed for section Nigri aspergilli differ from that designed for A. nidulans; the latter contains an auxotrophic selection argB marker. Also, their design placed individual sgRNAs in the middle of a large transcript whose expression was controlled by the A. nidulans gpdA promoter and trpC terminator. The resulting sgRNA then was released by activities of two ribozyme sequences flanking the sgRNA, a hammerhead-type ribozyme and a hepatitis delta virus ribozyme at the 5′ end and the 3′ end, respectively (53). Not only was the vector design complicated but the reported targeting frequency for A. nidulans yA was low also, and the majority of primary transformants remained wild-type green. The simple system developed for A. flavus worked well for wild-type A. nidulans, which would make the creation of auxotrophic transformation recipients an obsolete practice. Additionally, it achieved nearly 100% targeting frequencies (Table 4 and Fig. 7B). Without relying on an NHEJ-deficient A. fumigatus recipient strain and donor DNAs, it gave a result comparable to those reported by other gene-targeting studies that used either vector or RNP complex (28, 29). Al Abdallah et al. (28) reported that only the A. fumigatus ΔakuB strain gave high gene-targeting frequencies (97%) regardless of the homologous sequence length of donor DNAs, whereas using the wild-type strain resulted in much lower gene-targeting efficiencies that were homology dependent. For A. terreus, the developed system achieved better gene-targeting frequencies than a recent study (>90% versus 71%) (54). That study still employed an auxotrophic A. terreus NHEJ-deficient recipient strain and two selection markers, an auxotrophy curing pyrG for the vector and the hph gene (hygromycin resistance) for homologous donor DNA. It lacks simplicity and flexibility somewhat and thus cannot be readily applied to other A. terreus isolates because individual NHEJ-deficient recipients must be generated first. In this study, the numbers of primary transformants produced from A. fumigatus, A. terreus, and A. niger appeared somewhat low, but they did not markedly affect gene-targeting frequencies. These low numbers probably were due to the adopted procedures used, which were established specifically for A. flavus. Optimizing conditions of protoplast preparation (medium, mycelial growth stage, enzyme recipe, digestion time, etc.), transformation, and regeneration for these species by expert groups may improve the outcomes. Since section Nidulantes, section Fumigati, section Terrei, and section Nigri contain 65, 60, 16, and 26 species (55–58), respectively, the developed CRISPR/Cas9 system likely will have a broad application in aspergilli.

MATERIALS AND METHODS

Strains and media.

The initial test of the CRISPR/Cas9 system was performed using wild-type L-morphotype A. flavus CA14 and its NHEJ-deficient strain, KuPG#1 (59). Three more A. flavus isolates, one L-morphotype strain, NM1-3, and two S-morphotype strains, SRRC1576 (=CA42) and GA10-8s, were further used (60, 61). Additional section Flavi species, which included A. parasiticus, A. oryzae, A. sojae, A. nomius, A. bombycis, and A. pseudotamarii (27), and aspergilli from other sections such as A. nidulans, A. fumigatus, A. terreus, and A. niger (culture collection strains of the Southern Regional Research Center, New Orleans, LA, USA) were also tested. Kojic acid medium (KAM) used to examine kojic acid production (26) was modified slightly and contained 0.25% yeast extract, 0.1% K₂HPO₄·3H₂O, 0.05% MgSO₄·7H₂O, 2% glucose, and 1.5% agar (pH 6.0). Production of kojic acid on KAM was indicated by formation of diffusive bright orange red substance, a chelate of kojic acid and iron ion. Strains were maintained on potato dextrose agar (PDA; EMD, Darmstadt, Germany) plates. V8 juice agar medium, which consists of 50 mL V8 juice and 20 g agar per L, adjusted to pH 5.2 prior to autoclaving, was used for spore production. Wickerham agar medium (WKH) (62) was used for sclerotial production.

Determination of A. flavus U6 promoter and terminator sequences.

The Rfam database (https://rfam.org/) is a collection of RNA families, and each family is represented by multiple sequence alignments, consensus secondary structures, and covariance models. To identify sequences of A. flavus U6 promoter and terminator, A. flavus genomic DNA sequences that matched the U6 RNA model sequence were retrieved from the NCBI (National Center for Biotechnology Information) genome database for analysis (see Fig. S1 in the supplemental material).

Identification of a minimum region required for AMA1’s autonomous replication function in pPTRII.

The autonomously replicating plasmid pPTRII (Fig. 1A), which contains a 5.2-kb HindIII fragment of AMA1 (19, 20), is derived from the integrative plasmid pPTRI (20). Restriction sites on pPTRII were analyzed using the DNAMAN software (Lynnon Soft, Vandreuil, QC, Canada). Various regions of pPTRII were removed by restriction digestion and self-ligation to give test constructs (Fig. 1B). For example, PstI digestion alone and self-ligation removed 3.0 kb, including half of AMA1, from pPTRII, and the resulting vector was called pP70. These vectors were individually transformed into A. flavus to assess their abilities to maintain the autonomous replication capacity based on numbers of pyrithiamine-resistant transformants produced. Transformation frequencies of these constructs were compared to that of the pPTRII control.

Synthesis of cas9 sequence and expression in an AMA1-containing vector.

A codon-optimized Streptococcus pyogenes cas9 sequence whose 3′ end was tagged with the simian virus 40 (SV40) nuclear localization signal (NLS) sequence, PKKKRKV, was synthesized by Twist Bioscience (South San Francisco, CA, USA) and cloned into NotI and NheI sites of the pTwist cytomegalovirus (CMV) hygromycin (Hygro) vector. The cas9 sequence was derived from an A. oryzae study (40) with minor modification, which included removal of four HindIII and three KpnI sites to allow subsequent recombinant DNA manipulation. The cas9-containing pTwist vector was cut first with NheI, blunt ended at 72°C for 10 min by AccuPrime Pfx DNA polymerase (Invitrogen, Carlsbad, CA, USA), and then cut with NotI. The gel-purified fragment was cloned into NotI and SmaI sites of pTR1-GPD-TRPC (deposited as Addgene plasmid no. 190103), a robust expression vector that uses the constitutive gpdA promoter and trpC terminator of A. nidulans (63). The resulting expression cassette, gpdAp-cas9-trpC, was digested with HindIII and KpnI and cloned into the pP70 vector and gave vector pP70-gpdAp-cas9-trpC.

Addition of SV40 NLS sequence to the 5′ end of cas9.

The originally synthesized cas9 gene fragment contained only a single copy of the SV40 NLS sequence at its 3′ end. However, it did not give satisfactory gene-targeting frequencies from preliminary tests (unpublished results). Hence, an additional copy of the SV40 NLS sequence was cloned into the 5′ end of a cas9-containing vector, pP70-gpdAp-cas9-trpC. Two complementary 90-nucleotide oligonucleotides (Fig. S2A), which contained the SV40 NLS sequence and each side flanked by a 16-bp overlap recombinational sequence designed for In-Fusion cloning (TaKaRa Bio USA), were synthesized by Integrated DNA Technologies (IDT; Coralville, IA, USA). Mixed oligonucleotides were placed in an annealing buffer (10 mM Tris, pH 7.5, 50 mM NaCl, 1 mM EDTA) in a PCR tube, heated at 95°C for 2 min, and allowed to cool gradually to room temperature. The DNA fragment was cloned into the NotI site of the above-described expression vector. The resulting plasmid was named pAf-crispr.

Inclusion of an additional 292 bp beyond the PstI site of pAf-crispr.

A vector containing an additional 292-bp sequence beyond the PstI site of AMA1 (Fig. 1 and Fig. S2 and S3) was also constructed to investigate whether it would increase transformation frequencies on the NHEJ-deficient genetic background of A. flavus KuPG#1. A 338-bp DNA fragment synthesized by IDT (Fig. S2B), which contained two 18-bp overlap sequences at both flanking regions, was cloned into the PstI site of pAf-crispr by In-Fusion cloning. The resulting plasmid was pAsp-AMA-gpdA-ptr (deposited as Addgene plasmid no. 191016, Fig. 2A).

Construction of pAf-CRISPR-yA and generation of gene-specific CRISPR vectors.

A DNA fragment for the sgRNA cassette to target the A. flavus yA gene was synthesized by IDT. It contained the A. flavus U6 promoter, a short crRNA (yA) sequence fused to the scaffold tracrRNA sequence, and the A. flavus U6 terminator (Fig. S2C). The fragment was cloned into the pAf-crispr vector to give pAf-CRISPR-yA (deposited as Addgene plasmid no. 191015, Fig. 2B). pAf-CRISPR-yA also was the PCR template for the preparation of other DNA constructs expressing sgRNAs. In this protocol, two protospacer-specific DNA fragments were first generated by PCR with primer sets of U6-F-P/gdRC_gene and gdFd_gene/U6-R-K, respectively (Table 1). To this end, 40 pmol of each primer and 2 ng of pAf-CRISPR-yA as the template were added to 20 μL AccuPrime SuperMix (Invitrogen, Carlsbad, CA, USA) and subjected to 30 cycles of PCR. The amplification conditions consisted of denaturation at 94°C for 30 s, annealing at 55°C for 30 s, and extension at 72°C for 1.0 min. The two PCR fragments without further purification were then directly fused and amplified by another round of PCR with primers U6-F-P and U6-R-K. The PCR conditions were similar except that annealing time was set for 2 min. The resulting fragment, after being cut with PstI and KpnI, was cloned into the same sites of either pAf-crispr or pAsp-AMA-gpdA-ptr. A concise protocol for the steps is summarized in Fig. S4.

Replacement of A. nidulans gpdA promoter with A. flavus GPI gene promoter.

A glycosylphosphatidylinositol-anchored protein (GPI-AP)-encoding gene, AFLA_113120, was previously found to be highly expressed under various growth conditions (64). Its promoter contained an intron at the 5′ nontranslated region (Fig. S5). The A. nidulans gpdA promoter used to drive cas9 expression in vectors pAf-CRISPR-wA1 and pAf-CRISPR-wA3 was replaced by the A. flavus GPI promoter to examine whether this indigenous promoter could increase transformation frequencies on the NHEJ-deficient genetic background of A. flavus KuPG#1. It was done by digesting the vectors first with HindIII and NotI followed by cloning the tagged PCR product of the 1.4-kb GPI promoter into the same restriction sites.

Preparation of protoplasts and fungal transformation.

For A. flavus, approximately 5 × 10⁷ conidia harvested from V8 agar plates were inoculated into 100 mL Czapek-Dox broth (CZ; Becton Dickinson, Sparks, MD, USA) supplemented with 0.5% Casamino Acids. The culture was shaken at 150 rpm for 11 to 12 h at 30°C. Mycelia were collected on a 100-μm nylon cell strainer, transferred to a 50-mL tube, and resuspended in 20 mL of filter-sterilized enzyme solution that contained 1.0 g VinoTaste Pro (Novozymes, Bagsvaerd, Denmark) and 200 mg lysing enzymes (Sigma, St. Louis, MO, USA) in 0.55 M KCl, 0.05 M citric acid, pH 5.8. The digestion was for 2 to 3 h at 30°C with shaking (60 rpm). Protoplasts were collected by filtering through a 40-μm nylon cell strainer and pelleted using a table centrifuge. The protoplasts were washed twice with a solution of 0.6 M KCl, 50 mM CaC1₂ and 10 mM Tris-HCl, pH 8.0. Fungal transformation was performed as described (65). Polyethylene glycol (PEG) solution consisted of 30% (wt/vol) PEG3350 (Rigaku, Japan), 0.6 M KCl, 50 mM CaCl₂, and 10 mM Tris-HCl, pH 7.5. In a typical A. flavus transformation experiment, 2 × 10⁶ protoplasts and 0.3 μg of vector DNA were used. The transformation mixture was added to 150 mL molten CZ regeneration agar medium supplemented with pyrithiamine (0.1 μg/mL), and six plates were poured. Small pyrithiamine-resistant colonies usually started forming in 2 to 3 days at 30°C. The number of transformants was estimated in the early stage since growing colonies in/on poured plates tended to overlap or merge when they gradually increased in size. Conditions of culture growth, protoplast preparation, and transformation optimized for A. flavus CA14 were used for other section Flavi aspergilli, A. nidulans (section Nidulantes), A. fumigatus (section Fumigati), A. terreus (section Terrei), and A. niger (section Nigri).

Direct PCR and sequencing analyses for gene defects of A. flavus transformants.

Primary transformants were transferred onto PDA plates and grown at 30°C for 2 to 3 days. A Phire Plant Direct PCR master mix (Thermo Scientific, Waltham, MA, USA) was used for amplification of DNA targets directly from fungal mycelia without DNA purification (66). A pin-size young mycelium was sampled with a sterile toothpick and transferred to a 0.5-mL microcentrifuge tube containing 20 μL of the dilution buffer provided with the kit. The content was smashed and dispersed with the toothpick using an action of up-and-down strokes. An 0.5-μL amount of the resulting solution was used directly for PCR amplification in a final volume of 20 μL. A protocol that consisted of an initial denaturation at 98°C for 5.0 min, followed by 40 cycles of denaturation at 98°C for 5 s, annealing at 60°C for 5 s, and extension at 72°C for 20 s, was used. PCR products were purified by a DNA Clean & Concentrator-5 kit (Zymo Research) and sequenced at the Genomics and Bioinformatics Research Unit of the Agricultural Research Service, U.S. Department of Agriculture (Stoneville, MS, USA). For those genomic DNA preparations that did not yield PCR products, quality checking was carried out by a new round of PCR with kojR primers (Table S1) whose PCR product was 1.0 kb.

Sequential and simultaneous approaches for multiple-gene targeting.

An A. flavus olgA knockout was inoculated on a V8 plate and grown at 30°C for 3 days. Freshly harvested spores were the inoculum for protoplast preparation. For sequential gene targeting, 2 × 10⁶ protoplasts and 300 ng vector were used for each round of transformation. The copT and yA genes were separate targets for second rounds of knockout experiments. Selection of resulting knockouts was based on known changes in conidial colors and the association of defects of conidial genes with accumulation of conidial pigments. A copT knockout and a yA knockout randomly picked from regeneration plates were directly transferred onto V8 plates and grown at 30°C for 3 days for spore production. The target gene for two third rounds of knockout experiments was wA. For simultaneous multiple-gene targeting, the olgA knockout was the transformation recipient. A total of 1 × 10⁶ protoplasts and combined vectors, 150 ng each for targeting kojA, yA, and wA, were used in cotransformation. Determination of single-, double-, and triple-gene knockout events was based on three criteria: (i) association of conidial gene defects with known conidial colors, (ii) loss of kojic acid production or not on KAM medium, and (iii) PCR and sequencing results of yA targeting regions in selected white (nonpigmented) wA knockouts with location-specific primers.