CRISPR/Cas9-mediated mutagenesis of phytoene desaturase in diploid and octoploid strawberry

Transformation and regeneration of mutant lines

We placed sgRNAs targeting exon 7 of the PDS gene under transcriptional control of the Arabidopsis U6-26 promoter (pAtU6-26 vector) [13] or the corresponding sequence of the F. vesca U6III promoter (pFvU6III vector) (Fig. 1, Additional file 1: Table S1).

Sixty-two percent of ‘Hawaii 4’ and 45% of ‘Calypso’ putative transgenic shoot lines exhibited mutant phenotypes (Table 1). Where shoot lines comprised a range of phenotypes, albino or variegated phenotypes were apparent in approximately 10% of tissues harvested. Mutant phenotypes were evident on regeneration plates and also became apparent after calli were transferred to rooting medium. Most calli regenerated a mixture of mutant phenotypes, as described within this text and in supplementary files (Additional file 2: Figure S1, Additional file 3: Figure S2, Additional file 4: Table S2, Additional file 5: Table S3). Albino shoots regenerated either directly from callus, or as axillary shoots derived from variegated shoots. The proportions of ‘Hawaii 4’ and ‘Calypso’ mutant phenotypes derived from pAtU6-26 are 25/47 lines (53%) and 19/38 lines (50%), respectively, and corresponding data for pFvU6III are 67/76 lines (88%) and 6/16 lines (38%). Native promoters are known to enhance mutation efficiency: for example, in soybean in three target genes, mutation efficiencies were increased by the use of the soybean U6-10 promoter compared to the Arabidopsis U6-26 promoter [39]. Relative mutation efficiencies in this experiment suggest that the F. vesca U6III promoter may be more effective than the Arabidopsis U6-26 consensus promoter in ‘Hawaii 4’ but not in ‘Calypso’, and that alternative U6 promoters native to the octoploid genome may enhance sgRNA expression in ‘Calypso’ and other commercial cultivars.

Types and frequency of target site mutations in transgenic lines

Amplicons of 401 bp spanning the target site were generated for 96 leaf samples (Additional file 6: Fig. S3) and used for Illumina MiSeq sequencing. Sequence data was obtained for a range of phenotypes from single and multiple shoots of 19 ‘Hawaii 4’ and 8 ‘Calypso’ shoot lines (criteria for selection were phenotype and availability of suitable material at the time of analysis) and for a single wild-type (WT) shoot of each cultivar regenerated from in vitro explants.

Transgenic shoot lines exhibit a variety of mutant sequence variants at and around the target site in a highly conserved region of the PDS gene resulting in phenotypes typical of defective PDS gene function, including pale green, variegated and albino phenotypes, which are often (although not always, as discussed below) related to the proportion of variant reads seen (Figs. 2, 3, Additional file 2: Fig. S1, Additional file 3: Fig. S2, Additional file 4: Table S2). Approximately 60–80% shoot lines included albino shoots with 100% target site sequence variant reads.

Types of allelic variant and transgenic origin of multiple shoots from individual callus lines

Leaf samples from individual shoots are either homozygous for one sequence variant type, heterozygous (having two sequence variant types), chimaeric (3 or more sequence variant types), or have a mixture of wild-type sequence and 1, 2 or more than 2 sequence variants: mono-WT, di-WT or multi-WT (Figs. 2, 3, Additional file 2: Fig. S1, Additional file 3: Fig. S2, Additional file 4: Table S2). Most homozygous and heterozygous allelic variants are of the most common sequence variant types and may have arisen from simultaneous mutation of both alleles prior to division of the cell from which the tissue is derived or at a subsequent stage in organogenesis. Multiple samples from single shoots and shoot lines often share variant sequence types and combinations of sequence types (Additional file 4: Table S2, Figs. 2, 3, Additional file 2: Fig. S1, Additional file 3: Fig. S2), including more complex patterns of chimaeric allelic variation. The timing and sequence of mutation events will determine the mixture of the allelic variants within a shoot and shoot line. The data for multiple shoots from shoot lines and leaf samples from single shoots suggest albino shoots regenerate from callus (e.g. ‘Hawaii 4’ 89, ‘Calypso’ 1, Figs. 2, 3, Additional file 5: Table S3) or by re-arrangement of existing mutations and/or continued gene editing as apical variants or axillary branches from variegated shoots (e.g. ‘Hawaii 4’ 39D Fig. 2, Additional file 5: Table S3). Analysis of leaves of some shoots provides evidence of ongoing gene editing within a shoot (e.g. ‘Hawaii 4’ 2C, 28C, 39C, D, 46Dii, H, ‘Calypso’ 5C, Figs. 2, 3, Additional file 2: Fig. S1, Additional file 4: Table S2, Additional file 5: Table S3). Four of 7 mutant ‘Calypso’ shoot lines (lines 1, 5, 7, 100) and 15 of 19 mutant ‘Hawaii 4’ shoot lines (all lines except 9, 10, 32, and 36) produced shoots which are biallelic (homozygous, heterozygous or chimaeric) mutants.

To aid interpretation of the patterns of allelic variation, TAIL PCR was performed to generate data relating to T-DNA integration (Additional file 4: Table S2, Additional file 8: Fig. S5). Multiple shoots from individual lines were included and putative plant chromosomal sequence (i.e. not matching vector sequence) at the left border integration site, often contiguous with partial or complete T-DNA left border sequence, was recovered from approximately half of shoot lines screened (7 of 17 ‘Hawaii 4’ and 5 of 8 ‘Calypso’ lines, Additional file 9: Fig. S6, Additional file 4: Table S2). In all cases a single unique sequence for individual lines was obtained (Additional file 10: Table S4) and where integration data was obtained for multiple shoots the data was homologous. For the remainder of shoot lines the sequence obtained beyond the T-DNA left border matches only binary vector sequence (Additional file 9: Fig. S6), indicating chromosomal integration of vector ‘backbone’ DNA, which was shown by De Buck et al. [40] to be a common occurrence in transgenic Arabidopsis and tobacco plants. Shoots of ‘Hawaii 4’ line 10 have a complex integration sequence comprising T-DNA, vector backbone sequence and sequence not matching vector DNA (Additional file 9: Fig. S6).

The combined data from both target site sequence analysis and TAIL PCR sequence analysis indicate that multiple shoots regenerated from individual callus lines are likely to share a transgenic integration origin. The only shoot line where the data indicate separate transgenic origin of individual shoots is ‘Hawaii 4’ line 79: 79C and 79E TAIL sequences do not overlap but both align to the vector backbone and may be the same (Additional file 9: Fig. S6); but 79A has a more complex pattern of mutation and does not align to vector sequence.

Correlating observed phenotypes with target site mutation data

Phenotypic observations of leaf samples of multiple shoots were compared to the mutant sequence data generated for each sample (Figs. 2, 3, Additional file 2: Fig. S1, Additional file 3: Fig. S2, Additional file 4: Table S2) and to WT sequence data. All WT shoots regenerated in these experiments on non-selective medium were observed to be green (i.e. no albino or variegated phenotypes were observed), which is consistent with observations made in previous transformation work. The clear albino and variegated phenotypes seen are quite distinct from the bleaching of untransformed tissue cultured on kanamycin, and these are therefore attributed to PDS gene mutation. All transformed shoot lines comprised a mixture of green, and/or variegated, pale green, or albino shoots, except ‘Hawaii 4’ line 89 (Fig. 2), from which only albino callus and shoots regenerated. A very small fraction of variant reads was identified in sequence data from wild-type samples: 0.6% and 0.02% of total reads in the data for ‘Hawaii 4’ and ‘Calypso’ wild types, respectively. These reads were considered background ‘noise’ as a result of index-switching between Illumina barcodes, an event that occurs at low rates during Illumina sequencing of pooled samples on a single flow cell [41].

In most cases the observed phenotype reflects the sequence data obtained. Sequence data for the majority of albino samples in both cultivars lack or have a negligible fraction of wild-type sequence reads (< 0.05–0.1% of total reads). For some shoot lines the fraction of wild-type sequence data for all shoots sequenced is negligible (‘Hawaii 4’ lines 12, 14, 79 and 89 and ‘Calypso’ line 1). The exception is ‘Calypso’ shoot 1C, which is green and probably not transformed (PCR data show the absence of a product in reactions using Nos-Kan cassette and sgRNA insert site T-DNA primers). Sequence data for ‘Calypso’ shoot line 7D (Additional file 3: Fig. S2) is anomalous as it was observed to have a variegated phenotype yet WT sequence reads were not obtained. The presence of a low fraction (< 5%) of wild-type reads in data from white phenotypes in some cases (Additional file 4: Table S2, Additional file 5: Table S3) may be attributed to difficulty in assessing phenotype (for example where the leaf samples are very small, e.g. ‘Hawaii 4’ 39B, ‘Calypso’ 5 Ci, Figs. 2, 3, Additional file 5: Table S3).

‘Hawaii 4’ samples, which are largely white with some green and a high fraction of wild-type reads e.g. 28 D1, 32 B1, B2 and 36 A3 (Additional file 2: Fig. S1), are possibly chimaeras with a multiple transgenic origin: for example, a mosaic of white and green cells in the L2 layer would be manifested as small green flecks at the leaf edge, and would explain the high wild-type reads in comparison with samples that show small green sectors in the central part of the leaf lamina (e.g. ‘Hawaii 4’ 2 Ci and 39 C1, with < 0.7% wild-type reads). Regeneration of chimaeric shoots comprised of transgenic and non-transgenic cells has been documented in transformed strawberry [2] and its frequency can be minimised by employing a stringent selection method (as described in this study). Chimaeric origin of shoots from more than one transformed cell is possible; however Southern blot data reported for diploid and octoploid strawberry [7, 42, 43], indicate that transformants will usually have a single copy (or few T-DNA copies with bands of similar intensity) and by inference these will be of single cell origin. Direct organogenesis (omitting a callus phase) would avoid regeneration of chimaeric shoots, but the potential for this is cultivar dependent [44].

Only one transgenic shoot 100% heterozygous for a target sequence mutation lacks the expected albino phenotype: sequence data for a green ‘Hawaii 4’ transgenic shoot, ‘Hawaii 4’ 3A (Additional file 2: Fig. S1), show that 50% of reads are deletion variants, but indicate that one allele has an amino acid substitution (threonine to asparagine) and that this encodes a functional PDS protein. Published PDS protein sequences for maize and rice also encode asparagine at this position [45] (Additional file 11: Fig. S7).

The ranges of wild-type reads commonly observed for pale green and variegated samples are 7–19% (seen in 7 samples from 7 lines of ‘Hawaii 4’) and 6–36% (seen in 15 samples from 10 lines of ‘Hawaii 4’ and 2 samples from 2 lines of ‘Calypso’), which are taken to be due to the proportion of alleles and cells expressing functional PDS protein.

In lines of both cultivars, sequence data for some phenotypes show unexpected proportions of wild-type and mutated sequences (Additional file 4: Table S2, Additional file 5: Table S3, Additional file 2: Fig. S1, Additional file 3: Fig. S2). All ‘Hawaii 4’ transgenic samples with a dark green phenotype (apart from 3A which has a sequence mutation that encodes a functional PDS protein) are single-allele sequence variants with only approximately 20% to 60% wild-type reads. In addition, there are white phenotypes with higher than expected wild-type sequence fractions (15–30%). Sequence data for ‘Calypso’ albino shoot 94 (Fig. 3) include 31% wild-type reads and a green transgenic shoot (‘Calypso’ 59, Additional file 3: Fig. S2) has only 11% wild-type reads. In apple (another diploid member of the Rosaceae), Nishitani et al. [24] also observed green transformants with partially mutated sequences of the PDS gene. It is possible that one of the PDS alleles in F. vesca is non-functional and that in the octoploid, cultivated strawberry there may be only one functional PDS gene allele in which case the fraction of wild-type sequence would reflect the proportion of non-functional PDS alleles. Albino or variegated samples with unexpectedly high wild-type reads and complex chimaeric genotypes may include mutations of both functional and non-functional alleles (Additional file 5: Table S3). Evidence for gene silencing following gene duplication and loss of redundant sequences in polyploids is well-documented [46], and gene copy loss and deletion in polyploid Fragaria has been described by Rousseau-Gueutin et al. [47]. The fact that a single wild-type amplicon was found is suggestive, but not conclusive, that the PDS1 locus is functionally diploid and that complex variants observed in amplicon profiles are due to chimaeric regeneration rather than multiple edits of homeologs.

Construction of CRISPR/Cas9 vectors

The binary vector pFGC-pcoCas9 (Addgene plasmid # 52256) was modified by insertion of pGreen plant marker constructs [52]: the ‘Nos-Kan’ cassette (comprising the nptII gene under control of nopaline synthase regulatory sequences) was cloned into the Pme1 site near the right border and the ‘35S-GFP’ cassette (comprising the modified GFP gene [53] under control of 35S regulatory sequences) was cloned into the Sma1 site in the multiple cloning site to give pCas9-K-GFP (Fig. 1). The GFP marker was not used in this study.

CRISPR cassettes incorporating the sgRNA sequence and regulatory elements (Fig. 1) were synthesised by Integrated DNA Technologies (IDT) Ltd and cloned into pCR^®-Blunt II-TOPO^® (Thermo Fisher Scientific, UK). Cassettes were inserted into the EcoR1 site of pCas9-K-GFP and the orientation confirmed by sequencing. The PDS coding sequence for Malus domestica (GenBank: KU508828.1) was BLAST aligned against the genomes for F. vesca (FvAssembly 1.1) and F. × ananassa (www.rosaceae.org). The CRISPR target sequence selected (ACCTGATCGAGTAACTACTGAGG) is common to both Fragaria genomes and is located on the sense strand in exon 7 of FvAssembly 4 [9] (Fig. 1). It corresponds almost exactly to the target sequence (ex7-20 bp, ACCTGATCGAGTAACTACAGAGG) used successfully by Nishitani et al. [24]. The PAM sequences are underlined. The target guide RNA (gRNA) sequence (PDS74) comprises the first 20 base pairs of the target sequence preceded by an additional guanine base to maintain the native U6 promoter start of transcription. This is placed under transcriptional control of the consensus sequence of the Arabidopsis U6-26 promoter [13] or the corresponding sequence of the F. vesca U6III promoter and terminator sequences (Additional file 1: Table S1).

In vitro culture of Fragaria stock cultures

In vitro shoot cultures of F. vesca ‘Hawaii 4’ and F. × ananassa ‘Calypso’ were maintained in a growth room at 20 °C with a 16/8 h light/dark photoperiod provided by fluorescent lamps (colour reference 835, colour temperature 3500 K). Crowns were sub cultured at 4–5 week intervals, 5 per honey jar containing 50 ml medium. Basal culture medium was Murashige and Skoog (MS) macro and micro elements and vitamins [54], supplemented with sucrose (30 g l⁻¹) and solidified with Daishin agar (Duchefa D1004, 9 g l⁻¹). The pH was adjusted to 5.8 before autoclaving. Shoots were alternately cultured on basal medium supplemented with 6-benzylaminopurine (BAP) 0.1 mg l⁻¹ and indole-3-butyric acid (IBA) 0.1 mg l⁻¹ or BAP 0.5 mg l⁻¹, as described by Schaart [55].

Transformation of diploid and octoploid strawberry

Transformation of ‘Calypso’ was performed essentially as Schaart [55], with minor modifications. The protocol was additionally modified to achieve efficient transformation of ‘Hawaii 4’ by use of petiole explants in place of leaf explants and initial culture in liquid selection medium, as described below. Agrobacterium tumefaciens strain EHA105 [56] harbouring the binary vector was grown overnight, and pelleted at 2000×g for 10 min. The inoculum was prepared by re-suspending the overnight culture in filter-sterilised liquid MS-based medium supplemented with glucose (30 g l⁻¹) and acetosyringone (100 µM), pH 5.2, to give OD 600 nm 0.2–0.3. Petioles from apical leaves of ‘Hawaii 4’ or leaflets from young expanding leaves of ‘Calypso’ were harvested from shoots 4 weeks after subculture and submerged in the inoculum for approximately 10–15 min during explant preparation: petioles were cut into 4–5 mm pieces and leaflets were separated from each leaf before scoring transversely into 2 mm strips, leaving one leaf edge intact. Explants were then blotted to remove excess inoculum before transfer to Shoot Regeneration Medium (SRM): MS medium supplemented with α-naphthaleneacetic acid (NAA) (0.2 mg l⁻¹) and thidiazuron (TDZ) (1 mg l⁻¹). The pH was adjusted to 5.8 and Agargel ™ (A3301, Sigma) (5 g l⁻¹) was added before autoclaving. After autoclaving filter-sterilised glucose (300 g l⁻¹) was added to a final concentration of 30 g l⁻¹. After 4 days’ dark incubation at 20 °C, explants were washed in a solution of filter-sterilised ticarcillin disodium/clavulanate potassium (TCA, Duchefa T0190) (400 mg l⁻¹) in water, blotted and transferred to SRM selection medium containing filter-sterilised TCA (400 mg l⁻¹) and kanamycin sulphate monohydrate (Kanamycin, Duchefa K0126) (100 mg l⁻¹). ‘Hawaii 4’ petioles were initially cultured in Sarstedt Cell Culture Flasks (T-25) containing 15 ml liquid selection medium for 4 weeks, 30–50 explants per flask, shaking at 60 rpm at low light intensity at 20 °C, before transfer to semi-solid selection medium. Leaves were cultured throughout on semi-solid selection medium at 20 °C: approximately 10 leaves (abaxial side in contact with the medium) or 20 petiole pieces were cultured per 90 mm triple-vent Petri dish containing approximately 30 ml medium. Dishes were sealed with Parafilm^® M and incubated in the growth room as for shoot cultures. Subculture intervals were 4–6 weeks. At the second culture passage, half of the explants were transferred to SRM supplemented with gibberellic acid (GA₃) (1 mg l⁻¹) before autoclaving. Petioles were divided to separate calli developing at both petiole ends. Leaves were divided as expansion occurred to ensure contact with the selection medium, to facilitate sub culturing and to separate regenerating calli. Calli with shoots were harvested over a 7-month period, and transferred to rooting medium: MS medium 2.2 g l⁻¹, supplemented with sucrose 20 g l⁻¹, BAP 0.1 mg l⁻¹, IBA 0.1 mg l⁻¹, solidified with 9 g l⁻¹ Daishin agar, pH adjusted to 5.8 before autoclaving. Filter-sterilised kanamycin (50 mg l⁻¹) and TCA (400 mg l⁻¹) were added after autoclaving. For shoots with an obvious albino phenotype, kanamycin was omitted from the medium. Albino and variegated shoots were sub cultured onto rooting medium with TCA (400 mg l⁻¹). Control (WT) shoots were regenerated using the same method, except that the explants were not co-cultivated with Agrobacterium and antibiotics were omitted from the culture media.

Analysis of transgenic lines

DNA was extracted from leaf material using the DNeasy Plant Kit (Qiagen, Manchester, UK). Primers used in PCR and sequencing reactions (Additional file 12: Table S5, Fig. 1) were synthesised by IDT Ltd. PCR screening of putative transgenic lines for Nos-Kan cassette and sgRNA insertions was performed using PCRBIO Taq Mix Red (PCR Biosystems) and reaction conditions as detailed in Additional file 13: Table S6. PCR amplicons spanning the target region or Left Border T-DNA-genomic DNA junction sites were sequenced by Illumina MiSeq: primers incorporated target-specific sequences at the 3’ ends, and Illumina adapter overhang sequences (P5 forward overhang, P7 reverse overhang) at the 5’ ends. A 401 bp genomic region was amplified using primer pair P5-300 F and P7-700F, binding 186 -> 161 bp upstream and 195 -> 170 downstream of the target region, respectively. Amplicons were prepared for sequencing following Illumina guide 16S Metagenomic Sequencing Library Preparation (Part # 15044223 Rev. B). Template was amplified using Q5^® Hot Start High-Fidelity 2X Master Mix (M0494, New England Biolabs Ltd.) and purified using Mag-Bind^® TotalPure NGS (M1378, Omega Bio-tek Inc.). Dual indices and Illumina sequencing adapters were attached using Nextera XT Index Kit (15055290, Illumina) and 2x KAPA HiFi HotStart ReadyMix (KK2605, Kapa Biosystems). PhiX control V3 (15017660, Illumina) and the amplicon library were diluted to 4 pM concentration and the sample was spiked with 30% PhiX before loading onto a MiSeq Reagent Nano Kit v2 500 cartridge (MS-103-1003, Illumina).

TAIL PCR was used to amplify T-DNA-genomic DNA junctions using PCRBIO Taq Mix Red (PCR Biosystems Ltd): first round PCR was performed using arbitrary degenerate primer AD3 [57] and T-DNA-specific primer TAIL R1 (binding 257 -> 228 downstream of the Left Border). Nested primer TAIL R2 (binding 209 -> 188 downstream of the left border) and AD3 were used for the second round PCR. PCR reactions were performed using a Verity 96-Well Thermal Cycler (#4375786, maximum block ramp of 3.9 °C/Sec and a maximum sample ramp of 3.35 °C/Sec). The PCR cycle was essentially as Liu et al. [57] with minor modifications (Additional file 13: Table S6). Amplicons were purified using Mag-Bind^® TotalPure NGS and Sanger-sequenced by Eurofins Genetic Services Ltd. The sequencing primer (TAIL SEQ) binds 177 -> 146 downstream of the T-DNA left border.

Sequencing analysis and software

Illumina sequencing reads were trimmed to remove low quality data and Illumina adapters with fastq-mcf (http://code.google.com/p/ea-utils) before being aligned to the reference F. vesca ‘Hawaii 4’ genome v4.0.a1 [9], using BWA v0.7.15 [58]. Variants in respect to the reference genome were predicted and quantified from aligned reads using the CrispRVariant package v1.9.2 [59]. Nucleotide and protein alignments were performed using Geneious version 10.0.2 (http://www.geneious.com) [60]. Schematic maps were prepared using IBS software [61].