CRISPR–Cas9-mediated editing of starch branching enzymes results in altered starch structure in Brassica napus

Abstract

Starch branching enzymes (SBEs) are one of the major classes of enzymes that catalyze starch biosynthesis in plants. Here, we utilized the clustered regularly interspaced short palindromic repeats–CRISPR associated protein 9 (CRISPR–Cas9)-mediated gene editing system to investigate the effects of SBE mutation on starch structure and turnover in the oilseed crop Brassica napus. Multiple single-guide RNA (sgRNA) expression cassettes were assembled into a binary vector and two rounds of transformation were employed to edit all six BnaSBE genes. All mutations were heterozygous monoallelic or biallelic, and no chimeric mutations were detected from a total of 216 editing events. Previously unannotated gene duplication events associated with two BnaSBE genes were characterized through analysis of DNA sequencing chromatograms, reflecting the complexity of genetic information in B. napus. Five Cas9-free homozygous mutant lines carrying two to six mutations of BnaSBE were obtained, allowing us to compare the effect of editing different BnaSBE isoforms. We also found that in the sextuple sbe mutant, although indels were introduced at the genomic DNA level, an alternate transcript of one BnaSBE2.1 gene bypassed the indel-induced frame shift and was translated to a modified full-length protein. Subsequent analyses showed that the sextuple mutant possesses much lower SBE enzyme activity and starch branching frequency, higher starch-bound phosphate content, and altered pattern of amylopectin chain length distribution relative to wild-type (WT) plants. In the sextuple mutant, irregular starch granules and a slower rate of starch degradation during darkness were observed in rosette leaves. At the pod-filling stage, the sextuple mutant was distinguishable from WT plants by its thick main stem. This work demonstrates the applicability of the CRISPR–Cas9 system for the study of multi-gene families and for investigation of gene-dosage effects in the oil crop B. napus. It also highlights the need for rigorous analysis of CRISPR–Cas9-mutated plants, particularly with higher levels of ploidy, to ensure detection of gene duplications.

CRISPR–Cas9 has been used to edit multi-gene, starch branching enzyme families in canola and study gene-dosage effects, highlighting the necessity for rigorous analysis of gene-edited plants.

Introduction

Brassica napus is one of the most important oilseed crops as global demand for edible oil and renewable industrial products are increasing with the growing worldwide population. Maximizing oil yield in B. napus is therefore of prime economic importance. Most studies aiming at oil yield improvement have, so far, been focused on exploring and exploiting mechanisms of lipid biosynthesis and turnover to achieve increased oil accumulation and/or altered quality in seeds (Weselake et al., 2008, 2009). However, the role of starch, as a major carbon reserve in plants, has received rather limited attention in B. napus. In Arabidopsis thaliana, ∼30% of carbon assimilated through photosynthesis is stored as transitory starch in source tissues such as leaves during daytime, and is degraded at night to provide a continuous supply of sugars for transport to other sink tissues via the phloem to meet plant developmental requirements (Graf et al., 2010; MacNeill et al., 2017). The diurnal cycle of starch synthesis and degradation is under concerted and intricate control, ensuring economical carbon partitioning throughout the life of the plants (Graf et al., 2010; Sulpice et al., 2014; MacNeill et al., 2017; Yu et al., 2018a). A recent study in Arabidopsis reported marked effects on oilseed yield as a consequence of manipulating starch biosynthesis (Liu et al., 2016). Thus, understanding starch metabolism and manipulating related genes in B. napus offers an alternative strategy for maximizing biomass and oil yield.

Starch consists of two types of glucan polymers that accumulate in the plastids: amylose with linear α-1,4-linked glucans and amylopectin with α-1,6-linked branches. Amylopectin accounts for typically 95% of the composition of the starch granule in Arabidopsis leaves under a normal diurnal cycle (Zeeman et al., 2002). The formation of amylopectin begins with the transfer of the glucosyl part of ADP-glucose to the glucan chain primers by the action of starch synthases, generating new α-1,4 glucosidic bonds. Starch branching enzymes (SBEs) act on the α-1,4-linked chain, resulting in the formation of α-1,6-branch points. Finally, debranching enzymes are required to hydrolyze disorganized α-1,6 glucan branches to facilitate polymer crystallization (Myers et al., 2000; Ball and Morell, 2003). Therefore, SBEs play an important role in determining the structural properties of starches (Tetlow and Emes, 2014).

Phylogenetic analysis indicates two categories of SBEs, SBEI, and SBEII, exist in maize (Zea mays), wheat (Triticum aestivum), rice (Oryza sativa), sorghum (Sorghum bicolor), and potato (Solanum tuberosum; Tetlow and Emes, 2014). The SBEI isoforms are generally endosperm specific, and the loss of SBEI activities in maize and rice has minimal effects on starch synthesis and granule morphology in leaves (Nishi et al., 2001; Blauth et al., 2002). In contrast, the disruption of SBEII in plants caused more drastic changes in starch content and structure. For example, the absence of maize SBEIIa led to significantly reduced branching in leaf starch and disturbed diurnal cycling of starch (Yandeau-Nelson et al., 2011). However, oilseed species, including Arabidopsis and B. napus, lack SBEI and SBEIIb isoforms, which are associated with storage starch synthesis in endosperm (Fisher et al., 1996; Morell et al., 1997). In Arabidopsis, two SBE isoforms are expressed, SBE2.1 and SBE2.2, that belong to the SBEII category, and were identified more than two decades ago (Fisher et al., 1996). Deficiency of either isoform in Arabidopsis has little apparent impact on starch synthesis and plant development, suggesting overlapping function, while deletion of both isoforms completely abolishes starch synthesis and impedes plant growth (Dumez et al., 2006), thus demonstrating the essential role of SBEs for starch synthesis, and in turn leaf starch for plant growth and development.

In this study, we investigated the role of SBEs in the oilseed crop B. napus, an allopolyploid (2n = 38, AACC) derived from crosses of progenitors Brassica rapa (2n = 20, AA) and Brassica oleracea (2n = 18, CC; Chalhoub et al., 2014). Generating mutants for SBEs in B. napus is the most effective way to investigate their functions, but the public availability of mutant resources, through classical approaches such as ethyl methanesulfonate mutagenesis and insertional mutagenesis, has been limited by complicated genomes and possible gene redundancy. The invention of the CRISPR–Cas9 system has effectively enabled the introduction of targeted mutations in animals, microorganisms and plants (Shalem et al., 2014; Zhang et al., 2014; Selle and Barrangou, 2015). It is noteworthy that multiple target regions can be cleaved by Cas9 directed by their corresponding sgRNAs, enabling simultaneous mutations at different regions within a single gene or different genes (Ma et al., 2015; Cong et al., 2013). This feature is extremely powerful in facilitating mutations of multiple homologs within gene families by targeting their respective regions, particularly in polyploid species. To date, the CRISPR–Cas9 system has been applied not only in the model diploid plants Arabidopsis (Feng et al., 2014; Yan et al., 2015, 2018b), Brassica oleracea (Lawrenson et al., 2015) and rice (Zhou et al., 2014; Bortesi and Fischer, 2015; Ma et al., 2015), but also in polyploid crop species such as wheat (Zhang et al., 2019a), potato (Tuncel et al., 2019), cotton (Gossypium hirsutum L.; Gao et al., 2017), Camelina sativa (Lyzenga et al., 2019), and B. napus (Zhang et al., 2019b).

The first successful use of CRISPR–Cas9 in tetraploid B. napus modifying both copies of the pod shattering regulatory gene ALCATRAZ (ALC) was reported by Braatz et al. (2017). In other studies, one or both sub-genomes from a single gene were targeted, which limits the elucidation of gene function to those families that contain two or more genes and possess functional redundancy (Okuzaki et al., 2018; Yang et al., 2018; Zhai et al., 2019). Evidence of multiplex gene editing that simultaneously targeted multiple homologs of two genes in B. napus was recently demonstrated. However, chimeras that comprised three or more different mutations occurred at a higher frequency when the Cas9 cassette was ubiquitously expressed under the 35S promoter (Zhang et al., 2019b). Despite the higher efficiency of gene editing with the ubiquitous Cas9 system, a large proportion of nonheritable mutations were created in somatic cells, which were not transmittable from generation to generation, complicating the isolation of stable homozygous mutant lines (Feng et al., 2014). Use of a promoter for reproductive cells with CRISPR–Cas9 in Arabidopsis drastically decreased the occurrence of chimeric mutations (Yan et al., 2015; Mao et al., 2016) but, to date, such an approach in B. napus has not been reported.

In this study, to increase the incidence of heritable mutations, the Cas9 protein was driven by a highly expressed embryo sac promoter termed YAO (AT4G05410), previously used in Arabidopsis (Yan et al., 2015). In the current study, two constructs assembled with four and two sgRNAs, respectively, were designed to target multiple members of BnaSBEs and applied consecutively to achieve the full knockout mutant. A range of mutants with combinations of different edited genes were used for functional analysis, providing an opportunity to dissect the role of individual genes and examine the effects of SBE gene dosage in this important oilseed crop.

Results

The B. napus genome encodes six SBE homologs

Using the sequences of two Arabidopsis SBE genes, AtSBE2.1 (At2G36390) and AtSBE2.2 (At5G03650) as queries to the Ensembl Plants database (https://plants.ensembl.org), we identified single gene homologs in both diploid Brassica species; namely Bra005269 at Chromosome (Chr.) A5 and Bra009521 at Chr. A10 in B. rapa, and Bo4g034640 at Chr. C4 and Bo9g181850 at Chr. C9 in B. oleracea, respectively. However, a similar search against the B. napus genome BLAST database (http://brassicadb.cn) found a list of six SBE homologs in cultivar ZS11 (Sun et al., 2017), consisting of four BnaSBE2.1s (ZS11A05G008670, ZS11C04G011300, ZS11A04G025480, and ZS11C04G066320) and two BnaSBE2.2s (ZS11A10G030690 and ZS11C09G072390). The cultivar DH12075, which was shown to be blackleg (Leptosphaeria maculans) resistant in Canadian field trials, was used in this study; therefore, we next examined whether this cultivar possesses the same number of SBE genes. Gene-specific primers designed for each SBE gene were used to examine genomic DNA of DH12075. Polymerase chain reaction (PCR) analyses revealed that three SBE2.1 and one SBE2.2 genes were present in DH12075. Surprisingly, one SBE2.1 gene corresponding to ZS11A04G025480 and one SBE2.2 identical to ZS11C09G072390 were missing in DH12075 (Supplemental Figure S1). In addition, alignment to a draft whole genome assembly from DH12075 confirmed the absence of the aforementioned two genes, but also found that one of the SBE2.1’s equivalent to ZS11C04G066320 was instead on an unanchored scaffold of subgenome A, which was designated as Ann. Hence, the genome of DH12075 contained four different SBE isoforms and each gene was confirmed through consecutive cloning and sequencing using genomic DNA, and subsequently designated as BnaA5_SBE2.1, BnaC4_SBE2.1, BnaAnn_SBE2.1, and BnaA10_SBE2.2, respectively.

A phylogenetic tree using SBE homologs among Arabidopsis and B. napus including cultivars ZS11 and DH12075 is shown in Figure  1. Two large clusters were found containing SBE2.1 and SBE2.2 homologs, respectively, with the genes from ZS11 and DH12075 closely grouped. It was noted that BnaAnn_SBE2.1 represented a separate branch from BnaA5_SBE2.1 and BnaC4_SBE2.1 as well as Arabidopsis SBE2.1 probably due to their slightly divergent N-termini. Semi-quantitative reverse transcription PCR (RT-PCR) analysis was conducted to examine gene expression of four BnaSBEs in various tissues including leaf, stem, flower, and silique. All genes were ubiquitously expressed in these tissues, with a relatively lower level in stem than in leaf, flower and silique. In contrast to the other BnaSBEs, BnaAnn_SBE2.1 exhibited highest expression in siliques (Supplemental Figure S2).

Figure 1.

Phylogenetic tree of SBE2 homologs between B. napus and A. thaliana. Protein sequences of four SBE2 homologs present in the cultivar DH12075 (BnaA05_SBE2.1, BnaC04_SBE2.1, BnaAnn_SBE2.1, BnaA10_SBE2.2) and six obtained from the cultivar ZS11 (ZS11A05G008670, ZS11A04G025480, ZS11A10G030690, ZS11C04G011300, ZS11C04G066320, ZS11C09G072390) as well as two Arabidopsis isoforms (At2g36390 and At5g03650) were analyzed. Two clusters of SBE2.1 and SBE2.2 were shown. Numbers at branches indicate bootstrap confidence values out of 100. The tree is drawn to scale.

A single round of CRISPR-mediated editing generated mutations in multiple SBE genes

Two constructs containing different sgRNAs were designed for the three genes BnaA5_SBE2.1, BnaC4_SBE2.1 BnaA10_SBE2.2 simultaneously and BnaAnn_SBE2.1 alone, respectively (Figure  2). In the first instance, a total of four sgRNA expression cassettes were assembled in tandem, with two of them (t2 and t12) targeting identical regions of BnaA5_SBE2.1 and BnaC4_SBE2.1, and the other two (t5 and t6) directed towards BnaA10_SBE2.2. A second construct consisting of two BnaAnn_SBE2.1-specific sgRNAs (t16 and t17), was deployed in the genotype carrying mutations in the aforementioned three genes. For each gene, two segments of sgRNAs spaced 20–60 bp apart were designed in order to increase the editing rate and also to produce large fragment deletions (Figure  2A). The resultant sgRNA cassettes were assembled with the sequence encoding Cas9 protein driven by YAO promoter, producing two constructs, pLW08 and pLW13, respectively (Figure  2, B and C).

Figure 2.

CRISPR–Cas9-mediated gene editing of six BnaSBE isoforms. A, A list of sgRNA sequences and the intervals between two sgRNAs for a given gene as well as GC content for each sgRNA sequence. B and C, Diagrams showing the assembly of two constructs, pLW08 and pLW13. LB, left border; RB, right border; t2, t5, t6, t12, t16, and t17 indicate sgRNA sequences; pU3d, pU3b, pU6-1, and pU6-29 are Arabidopsis ubiquitin promoters that drive their respective sgRNAs.

A total of 36 positive T₀ independent lines were generated after introduction of pLW08 into 5-d-old cotyledons through Agrobacterium mediated transformation. The gene editing events were identified by direct sequencing, or by cloning and sequencing of the PCR products spanning two sgRNAs. Twenty-nine of these lines (80.5%) possessed mutations in at least one target gene, which included single base insertions (A or T), small deletions and in two cases base substitutions, but no large fragment deletions were detected (Table  1). Different sgRNAs produced remarkably different editing efficiencies. The t2 sgRNA exhibited higher rates at 93% (27/29) on BnaA5_SBE2.1 and 66% (19/29) on BnaC4_SBE2.1. The t12 sgRNA generated lower mutation efficiency at 28% (8/29) on BnaA5_SBE2.1 and 48% (14/29) on BnaC4_SBE2.1, respectively. In contrast, the efficiency of gene editing was substantially lower for t5 (24%, 7/29) and t6 (0%, 0/29), both of which target BnaA10_SBE2.2. Approximately half of the lines possessed mutations at both sgRNA-binding sites of BnaA5_SBE2.1 and BnaC4_SBE2.1, while only one line encompassing mutations at all three genes was found. As expected, the mutations occurred in close proximity to the protospacer adjacent motif (PAM) sequence (Figure  3). Our results showed that all targeted mutations were either heterozygous biallelic (two different mutations) or monoallelic (wild-type [WT]/one mutation), whereas no homozygous biallelic (two identical mutations) lines were found in the T₀ generation (Table  1 and Figure 3).

Table 1.

Mutations at three BnaSBE genes in T₀ mutated plants.

Lines	BnaA05_SBE2.1		BnaC04_SBE2.1		BnaA10_SBE2.2
	t2	t12	t2	t12	t5	t6
LW08-1	−3bp, +1bp	WT	+1bp, +1bp	WT	WT	WT
LW08-2	−1bp, −2bp	WT	−2bp, S6bp	−6bp, −8bp	WT	WT
LW08-3	−3bp, +1bp	WT	+1bp, +1bp	WT	WT	WT
LW08-7	WT	+1bp, WT	WT	WT	WT	WT
LW08-8	−2bp, WT	WT	WT	WT	WT	WT
LW08-11	−1bp, −2bp	+1bp, −14bp	−2bp, S6bp	−6bp, −8bp	+1bp, WT	WT
LW08-13	−3bp, +1bp	WT	+1bp, −2bp	WT	WT	WT
LW08-15	−2bp, −3bp	WT	WT	WT	WT	WT
LW08-16	WT	−2bp, −14bp	WT	−14bp, +1bp	WT	WT
LW08-25	−7bp, −7bp	−2bp, −14bp	WT	−14bp, +1bp	WT	WT

“−” and “+” indicate the deletion and insertion of nucleotides, respectively. “S” indicates a substitution. A total of 29 independent T₀ lines were detected to possess mutations and lines harboring the same type of mutations are not included. t2, t12, t5, t6 indicate sgRNAs.

Figure 3.

CRISPR–Cas9 induces different types of heterozygous mutant alleles. Corresponding sequencing chromatographs are shown for representative T₀ generation lines when transformed with pLW08 and pLW13, respectively. Mono-allelic mutation occurred on the sgRNA t5, but not t6, at BnaA10_SBE2.2. Bi-allelic heterozygous mutations occurred at BnaA5_SBE2.1, BnaC4_SBE2.1, and BnaAnn_SBE2.1. The three-nucleotides underlined indicate PAM sequences. The reference WT gene sequences are given as a comparison, and insertions and deletions in sequences of two alleles were labeled as “+” and “−”, respectively. All nucleotide changes are highlighted with red fonts. “S” stands for base substitution and “WT” indicates no change.

Identification of a gene duplication of BnaA10_SBE2.2

We selected the line LW08-11 which contained mutations in BnaA5_SBE2.1, BnaC4_SBE2.1, and BnaA10_SBE2.2 for further analysis. For clarification of the genotypes, capital letters are used to designate WT genotypes such as “A5, C4, A10” and lowercase for mutant alleles such as “a5, c4, a10”, respectively. By scanning sequencing chromatograms, heterozygous peaks representing biallelic mutations in both t2 and t12 sites of BnaA5_SBE2.1 and BnaC4_SBE2.1, as well as a monoallelic mutation in the t5 site of BnaA10_SBE2.2 were found (Figure  3;  Supplemental Figure S3A). After examining the gene sequences of T₁ progeny, the biallelic mutations at both BnaA5_SBE2.1 and BnaC4_SBE2.1 were stably transmitted from T₀ to the T₁ generation, with a ratio of 1:2:1 (homozygous allele1: heterozygous biallele1/2: homozygous allele2) following classic Mendelian inheritance (Table  2).

Table 2.

Genotype of three BnaSBE genes in T₁ mutated plant line LW08-11

T0 genotype	BnaA5_SBE2.1 (a51/a52)			BnaC4_SBE2.1 (c41/c42)			BnaA10_SBE2.2 (WT/a10)
T1 genotype	a51a51 (homo)	a52a52 (homo)	a51a52 (hete)	c41c41 (homo)	c42c42 (homo)	c41c42 (hete)	A10A10 (homo)	a10a10 (homo)	A10/a10 (hete)
Number of T1 progenya	43			39			67
Expected (1:1:2)	10.75	10.75	21.5	9.75	9.75	19.5	16.75	16.75	33.5
Observed	10	13	20	12	8	19	12	0	55
χ 2 test	0.62 (P = 0.73)			0.8458 (P = 0.655)			31.895 (P < 0.0001)*

The subscript of a5 or c4 means the type of editing, therefore a5₁/a5₂ and c4₁/c4₂ indicate the bi-allelic mutations of BnaA5_SBE2.1 and BnaC4_SBE2.1, respectively, and those genotypes with same subscripts such as a5₁/a5₁ indicate homozygotes. A10 and a10 indicate the WT and mutant genotypes of BnaA10_SBE2.2. “homo” means homozygote and “hete” means heterozygote.

^a These progeny were all from line LW08-11, and the numbers of plants used for genotyping were indicated, respectively.

^*The result is significantly different from the hypothesized expected numbers at P < 0.01.

However, the segregation of the monoallelic mutation at BnaA10_SBE2.2 did not produce the expected ratio. When assessing the sequence chromatograms, among 67 T₁ progeny, there were 12 plants with homozygous WT chromatogram peaks (A10 A10; Supplemental Figure S3B) and 55 lines carrying heterozygous peaks (A10 a10) at the t5 sgRNA region, as determined by DNA sequencing, and no progeny carrying homozygous mutant alleles (a10 a10) were identified (Table  2). It is interesting that two categories of heterozygous chromatograms were noticed from these 55 T₁ progeny. The first category, consisting of 18 plants, displayed heterozygous peaks of equal heights (Supplemental Figure S3C). The second category, detected in 37 plants, presented heterozygous peaks with one primary peak and one secondary peak (Supplemental Figure S3D), which was also observed in the T₀ plant (Supplemental Figure S3A). Since the relative peak heights are determined by the proportion of each of two sequence variants (Carr et al., 2009), we then quantified the variant numbers by cloning the PCR products and sequencing multiple colonies using the representative T₁ lines. The variant ratio of A10: a10 was 1:1 (χ² = 0.4, P = 0.527) in the first category (peaks with equal heights) and 3:1 (χ² = 0.48, P = 0.488) in the second category of chromatogram (one primary and one secondary peak). (P-value higher than 0.05 indicates that the observed values in our data are not significantly different from expected values.) Thus, three groups of genotypes were characterized within T₁ progeny: complete WT A10A10; A10: a10 with a ratio of 1:1; and A10: a10 with a ratio of 3:1, giving rise to a genotypic ratio of 1:1:2 for all three genotypes (χ² = 1.81, P = 0.4). These detailed genetic analyses therefore suggested that two identical A10 loci existed, and unbalanced heterozygous chromatograms were generated when only one of the loci was edited in line LW08-11. We therefore continued to screen for homozygous a10a10 using T₂ progeny derived from T₁ lines possessing the variant ratio of A10: a10 at 1:1 (genotyped as A10 A10 a10 a10).

Two different segregation scenarios were observed in the T₂ progeny of different lines: progeny of lines LW08-11-67 and LW08-11-11 presented the same genotypes (A10 A10 a10 a10) as the parent lines did not segregate (Supplemental Table S1), suggesting that the two loci in the above two lines are homozygous for A10 A10 and a10 a10, respectively. However, in the progeny of lines LW08-11-55 and LW08-11-61, five different genotypes (each determined by quantification of the DNA variants) were characterized: A10 A10 A10 A10, A10 A10 A10 a10, A10 A10 a10 a10, A10 a10 a10 a10, and a10 a10 a10 a10, with a ratio of 1:4:6:4:1 (Supplemental Figure S4 and Supplemental Table S1). Based on these observations, both loci in these two lines are heterozygous A10 a10 genotypes, which were most likely generated by a second editing event producing the same mutation in the unedited A10 A10 locus. These data further support the conclusion of a gene duplication of A10. We therefore designated the duplicated gene as BnaA10d_SBE2.2. Thus, one of the T₂ progeny, LW08-11-61-2, carrying the homozygous a10 a10 a10 a10 was identified, confirming successful editing at both BnaA10_SBE2.2s. In addition, this line was shown to have homozygous a5₂ a5₂ and c4₂ c4₂ alleles (subscripts denote specific alleles) at BnaA5_SBE2.1 and BnaC4_SBE2.1, respectively, and is subsequently referred to as the “quadruple mutant” (a5 c4 a10 a10d; Table  2).

Production of Cas9-free lines by segregation within a single generation

Isolation of Cas9-free mutants ensures the stable transmission of the identified mutations to the next generation since prolonged existence of the CRISPR–Cas9 construct in the mutants increases the off-target risk. To investigate whether the foreign T-DNAs could be segregated out in B. napus, we performed PCR assays using Cas9-specific primers. Among 67 T₁ progeny of line LW08-11, described above, 6 of 67 (8.9%) plants contained no Cas9 gene, hence confirming the elimination of the marker gene in these particular transgenic plants. We further used the plant selection marker hygromycin resistance gene-specific primers to conduct PCR assays and found that the plants without Cas9 were also free of the hygromycin gene. The identified quadruple mutant, homozygous for four BnaSBE genes, was also Cas9-free. Likewise, we have characterized additional Cas9-free homozygous lines (Table  3): LW08-11-67-3, a triple mutant (a5 c4 a10) consisting of homozygous mutations at BnaA5_SBE2.1, BnaC4_SBE2.1 and one of the BnaA10_SBE2.2s; a double mutant line LW08-11-61-25 (a5 c4) consisting of mutations only at BnaA5_SBE2.1 and BnaC4_SBE2.1. Furthermore, a backcross of the quadruple mutant with the WT followed by self-pollination enabled identification of the homozygous double mutant of a10 a10d (Table  3).

Table 3.

Homozygous lines containing mutations in different BnaSBE genes

Line ID	BnaA5_SBE2.1	BnaC4_SBE2.1	BnaA10_SBE2.2	BnaA10d_SBE2.2	BnaAnn_SBE2.1	BnaAnnd_SBE2.1
WT	+	+	+	+	+	+
LW08-11-61-25 (a5 c4)	−	−	+	+	+	+
F2-17-8 (a10 a10d)	+	+	−	−	+	+
LW08-11-67-3 (a5 c4 a10)	−	−	−	+	+	+
LW08-11-61-2 (a5 c4 a10 a10d)	−	−	−	−	+	+
LW13Q-59-36 (a5 c4 a10 a10d ann annd)	−	−	−	−	–	–

“+” indicated wild-type, “−” indicates gene edited mutations.

Editing of BnaAnn_SBE2.1 in the quadruple mutant background yielded plants with mutations in all six BnaSBE genes

Transformation of the vector pLW13 in the quadruple mutant background generated 36 T₀ independent lines. The editing efficiency was relatively low. In the T₀ generation only two lines, LW13Q-15 and LW13Q-27, carried mono-allelic heterozygous mutations (Table  4). However, their progeny did not display segregation patterns and all T₁ plants were heterozygotes with half of the alleles possessing mutations and the other half containing WT sequences. These observations resembled the case of LW08-11-67 indicating, again, a potential duplication of BnaAnn_SBE2.1. Interestingly, edits including insertions and deletions at the t17 sgRNA site of BnaAnn_SBE2.1 were noticed when we analyzed the T₁ progeny of one unedited T₀ line, LW13Q-22 (Table  4), suggesting that mutations were able to continuously occur in the presence of Cas9. Among 167 T₁ plants, 13 lines possessed heterozygous mutations which, in most cases, displayed the aforementioned unbalanced, heterozygous, chromatogram peaks. Three representative lines, LW13Q-22-38, LW13Q-22-59, and LW13Q-22-167, are illustrated in Supplemental Figure S5. Cloning and sequence analysis showed that the ratio of WT versus mutant variant in each line was 3:1. All these results mimicked the previously described segregation event of BnaA10_SBE2.2 and thus led to the conclusion that a second copy of BnaAnn_SBE2.1 is also present as a gene duplication, designated as BnaAnnd_SBE2.1.

Table 4.

Mutations at BnaAnn_SBE2.1 genes in transgenic lines

T0 generation			T1 generation			T2 generation
Lines	t16	t17	Lines	t16	t17	Lines	t16	t17
LW13Q-15	WT	−2bp	LW13Q-15-2	WT	−2bp	–	–	–
LW13Q-27	WT	−1bp	LW13Q-27-3	WT	−1bp	–	–	–
LW13Q-22	WT	WT	LW13Q-22-3	WT	+1bp	–	–	–
			LW13Q-22-16	WT	+1bp	–	–	–
			LW13Q-22-22	WT	−5bp	–	–	–
			LW13Q-22-41	WT	−2bp	–	–	–
			LW13Q-22-72	WT	−4bp	–	–	–
			LW13Q-22-121	WT	+1bp	–	–	–
			LW13Q-22-38	WT	−5bp	LW13Q-22-38-7	WT	−5bp, +1bp
			LW13Q-22-59	WT	+8bp	LW13Q-22-59-36	−59bp	+8bp
			LW13Q-22-167	WT	+1bp	LW13Q-22-167-8	−413bp	+1bp

These lines are the edited lines during the second round of transformation in the quadruple mutant background. LW13Q-15 and LW13Q-27 carry mutations in the T₀ generation, but no segregation occurred during the subsequent generations, indicating these are homozygous lines with only one copy of BnaAnn_SBE2.1 mutated. LW13Q-22 lines did not occur mutations at T₀ but various edits are detected from T1 generation. Among them, two lines, LW13Q-22-59-36 and LW13Q-22-167-8, were identified as null mutants for BnaAnn_SBE2.1. “−” and “+” indicate the deletion and insertion of nucleotides, respectively. t16 and t17 indicate sgRNAs.

Following the strategy used in obtaining the homozygous a10 a10d mutant, further analyses of the T₂ and T₃ progeny identified lines containing stable mutations at both BnaAnn_SBE2.1s (Table  3). The line LW13Q-22-59-36, free of the Cas9 gene, possessed two different mutated alleles and no WT alleles at either BnaAnn_SBE2.1 gene (Figure  3). Its progeny exhibited no segregation, suggesting that both mutated alleles were homozygous. This line encompassed two mutant alleles comprising a 59-bp deletion and an 8-bp insertion in BnaAnn_SBE2.1, and is designated as the “sextuple mutant” (a5 c4 a10 a10d ann annd; Figure  3 and Table 4). Notably, a large fragment deletion of 59-bp spanning the two sgRNA sequences occurred in this line, which was not observed for the sgRNAs used in the first round of transformation. Additionally, a second sextuple mutant line, LW13Q-22-167-8, was identified that contained two mutant alleles of BnaAnn_SBE2.1 and BnaAnnd_SBE2.1, consisting of a 413-bp deletion and 1-bp insertion, respectively (Table  4). Since no Cas9-free plant was characterized from this line, probably due to multiple T-DNA insertions, further investigation of this line was limited. In summary, the evidence reveals that there are six SBE isoforms in B. napus cultivar DH12075, and two rounds of CRISPR-mediated editing successfully created mutant lines homozygous for all six genes.

No off-target mutations were detected for all sgRNAs

To evaluate whether potential off-target effects occurred in the mutant plants, we used six sgRNAs (t2, t12, t5, t6, t16, t17) including their PAM sites as queries, respectively, for Blastn search against the genomic database of the DH12075 cultivar. Sequences with high similarity to the target sequences and a PAM site (NGG) at the 3′-end were selected for further evaluation. A total of three sequences which all contained four mismatches compared to sgRNAs are listed in Supplemental Table S2. The PCR products were amplified from a5 c4, a5 c4 a10, quadruple, and sextuple mutants, respectively, using specific primers that span the potential off-target sites, and then sequenced. We did not find any additional mutations beyond the targeted genes compared to the DH12075 WT. According to a previous investigation of off-target events in Arabidopsis (Feng et al., 2014), substantial sequence similarities for the last 12 bp of the 20-bp target sequence is characteristic of an off-target site. We noticed that in the subject ID Scaffold32939 there was only one nucleotide mismatch within the last 12-bp sequence, but no off-target effect was found in this candidate area. These analyses suggest that the CRISPR–Cas9 system employed is highly specific in B. napus.

SBE activity in gene-edited mutants

We used the above mentioned five mutants, a5 c4, a10 a10d, a5 c4 a10, “quadruple”, and “sextuple”, which carry different numbers of BnaSBE mutations to evaluate the gene editing effects on SBE enzyme activity. As shown in Figure  4A, SBE activity in a5 c4, a10 a10d, a5 c4 a10, quadruple, and sextuple mutant were 83%, 53%, 49%, 48%, and 29% of that in WT, respectively. Western blots showed that BnaSBE2.1 proteins were significantly lower in a5 c4, a5 c4 a10 and the quadruple mutant, and only at a trace level in the sextuple mutant, when compared to either WT or a10 a10d (in which all four genes for BnaSBE2.1 are functional). There was no detectable BnaSBE2.2 protein in a10 a10d, quadruple or sextuple mutants, consistent with BnaSBE2.2 null alleles, and reduced protein level in a5 c4 a10 mutant that possesses only one copy of BnaSBE2.2 (Figure  4B). We further conducted an in-gel zymogram assay of SBE activity (Figure  4C) and the corresponding BnaSBE2.1 and BnaSBE2.2 bands identified on western blots of identical nondenaturing gels (Supplemental Figure S6). In a10 a10d, a5 c4 a10, quadruple and sextuple mutants, the band of activity representing BnaSBE2.2 was not detectable, while a band associated with BnaSBE2.1 activity was still visible, even in the sextuple mutant (Figure  4C).

Figure 4.

Characterization of enzyme activities in sbe mutants. Soluble proteins were extracted from 3-week-old rosette leaves at the end of 16-h light. A, Total branching enzyme activity determined by phosphorylase a stimulation assay. Data represent mean and standard error (se) of three independent replicates. B, Western blots following sodium dodecyl sulphate–polyacrylamide gel electrophoresis of proteins extracted from a range of sbe mutants probed with SBE2.1 and SBE2.2 antibodies. The molecular weights of protein standards are indicated. C, Zymograms of SBE activities. BnaSBE2.1s and BnaSBE2.2s as well as putative PHO1 are indicated. D, Zymograms of PHO1 activities. The arrow indicates the plastidic isoform of starch phosphorylase, PHO1.

Mutations in BnaSBE2.2s lead to increased starch phosphorylase activity and altered transcription of BnaSBE2.1 genes

Less electrophoretically mobile, additional blue-stained bands were observed above SBE2.2 in SBE zymograms (Figure  4C). In-gel enzyme assays of plastidic starch phosphorylase (PHO1) corresponding to these activities were noticeably higher in a10a10d, quadruple and sextuple mutants in which both BnaSBE2.2 isoforms were absent (Figure  4D). The enhanced PHO1 activity was consistent with changes observed in chloroplast isolates (Supplemental Figure S7). Compared to the a10a10d mutant, additional mutations of BnaA5_SBE2.1 and BnaC4_SBE2.1 in the quadruple mutant, and mutations of BnaAnn_SBE2.1 in the sextuple mutant, did not significantly affect PHO1 activity. We also examined the transcriptional expressions of three BnaSBE2.1s in the Bnasbe2.2 mutant, a10a10d, and found that BnaA5_SBE2.1 and BnaAnn_SBE2.1 were upregulated (Supplemental Figure S8).

Alternate gene-splicing of BnaSBE2.1 in the sextuple mutant

To investigate why the sextuple mutant still possessed residual SBE activity, we examined the transcripts of all six genes by RT-PCR to investigate whether the CRISPR-mediated indels resulted in a reading frame shift or premature stop codons. Partial cDNA fragments covering the mutated regions of BnaA5_SBE2.1, BnaC4_SBE2.1, BnaAnn_SBE2.1, and BnaA10_SBE2.2 were amplified from the sextuple mutant, followed by cloning and sequencing. The edited BnaAnn_SBE2.1 and BnaAnnd_SBE2.1 genes generated two types of transcripts which were predicted to be truncated peptides containing 34 and 63 residues, respectively, due to the frame shift induced stop codons (Supplemental File S1). Likewise, transcripts from two BnaA10_SBE2.2s and BnaC4_SBE2.1 were predicted to produce truncated peptides consisting of 55 and 27 residues, respectively. Unexpectedly, two different types of transcripts were detected for the edited gene BnaA5_SBE2.1 in the sextuple mutant: one as a result of an induced frame shift which generated a truncated peptide (61 residues); while the other utilized an alternate splicing site “AG” to bypass the frame shift, producing a modified full length protein which differed from the WT protein at amino acids 22–47 (Figure  5; Supplemental File S1). Despite the N-terminal substitution, the presence of chloroplast transit peptides in both WT (1–78 amino acids) and modified BnaA5_SBE2.1 (1–52 amino acids) proteins were predicted using ChloroP (Emanuelsson et al., 1999).

Figure 5.

Occurrence of alternate splicing in the mutated BnaA5_SBE2.1. Different RNA splicing patterns occurred at BnaA5_SBE2.1 in the WT plant and the sextuple sbe mutants. Regular splicing of Intron 1 occurs in the WT, producing a full-length protein, while in the sextuple mutant two alternate splicing patterns were identified. In the sextuple mutant, regular splicing caused production of a truncated peptide, but alternate splicing through recognition of another “AG” site located in Exon 2 resulted in a modified full-length protein. Blue bars indicate the Exon1 and Exon2, and the open bars indicate the Intron 1 of BnaA5_SBE2.1. The unaltered sequences between WT and sextuple mutant were omitted. Splicing sites “GT” and “AG” are highlighted in red. CRISPR–Cas9-mediated edits at BnaA5_SBE2.1 including 2-bp deletion and 1-bp insertion in the Exon 1 of the sextuple mutant are indicated in blue.

Nonetheless, no WT genotypes were found for any SBE gene of this sextuple mutant. We further examined the quadruple mutant to see whether this particular example of alternate splicing occurred in these plants as well. Interestingly, the BnaA5_SBE2.1 transcripts obtained from twenty individual colonies all related to frameshift-induced stop codons. The transcripts for BnaA10_SBE2.2s and BnaC4_SBE2.1 were consistent with those in the sextuple mutant. These analyses suggested that the residual activity in the sextuple mutant was most likely derived from the modified BnaA5_SBE2.1 protein, which maintained activity despite the alterations at the N-terminus.

Mutations in BnaSBEs result in increased glucan chain length, reduced frequency of branching, and elevated phosphate content in leaf starch granules

In order to investigate the impact of suppression of multiple BnaSBEs on starch granule structure, we extracted starch from rosette leaves of the mutants carrying two to six BnaSBE mutations. Amylopectin chain length distribution (CLD) was determined by de-clustering the α-1,6 branches through isoamylase digestion, followed by separation of the side chains using anion-exchange chromatography. As illustrated in Figure  6, A and B, progressive knockout of BnaSBEs decreased the proportion of short glucan chains (degree of polymerization [DP] 6–18) and increased the proportion of long chains (DP > 18) in amylopectin. In a5 c4, a10 a10d, a5 c4 a10 and quadruple mutants, changes in CLD were detectable but moderate, whereas more pronounced changes in the proportions of short and long chains were observed in the sextuple mutant.

Figure 6.

Starch structure is altered in BnSBE mutants. Starch granules were extracted from 25-d-old leaves at the end of the 16-h-light period. A, Chain-length distribution of debranched starches. Glucan chain length (number of glucose units per chain) is shown on the x-axis, and the abundance of each chain is expressed as percentages of the total area on the y-axis. B, Difference in side-chain frequencies, shown as percentages of total area, using WT starch as reference. Three biological replicates were performed and a representative example is shown. The individual data points in (A) and (B) were listed in Supplemental File S2. C, Starch branching frequency assay expressed as nmols glucose equivalent per milligram starch. Mean and se of three independent replicates. D, Measurement of C6 phosphate content in starch granules in WT and mutant. Mean and se of three independent replicates.

The frequency of α-1,6 branch points was measured by isoamylase debranching of starch followed by quantification of reducing ends. Compared to the WT, the branching frequency in the a5 c4, a10 a10d, and a5 c4 a10 mutants was slightly decreased (13%–17%), a 30% reduction was observed in the quadruple mutant, and a decrease of 64% in the sextuple mutant was detected (Figure  6C). These changes were consistent with those of chain length distribution, revealing that reduced frequency of branching is accompanied by increased glucan chain length.

Starch phosphate groups tend to be associated with longer glucan chains (Blennow et al., 1998). Since ∼70% of the phosphate groups are bound at the C6 position of glucose in starch (Blennow et al., 1998), we measured the levels of glucosyl 6-phosphate esters released from digested starch. In the WT, the phosphate content of starch was 0.34 mmol·mol⁻¹ glucose residue, whereas with the increase in number of mutations of BnaSBE genes, there is a gradual increase in C6 phosphate content of amylopectin (Figure  6D). When compared to WT, the sextuple mutant possesses approximately 15-fold higher starch phosphate, consistent with the significant increase in glucan chain length and decreased branching. A dosage effect derived from incremental mutation of each BnaSBE gene was apparent. Statistical analysis showed that phosphate content was negatively correlated to starch branching frequency, with a coefficient r-value of −0.98 (derived from Figure  6, C and D).

Altered starch granule morphology in the sextuple sbe mutant

The morphology of starch granules extracted from 4-week-old rosette leaves was analyzed by scanning electron microscopy (SEM). In the WT, starch granules were of round or oval shape with smooth edges and of approximately equal sizes with an average diameter of 2.7 µm (Figure  7A). There were no discernable morphological differences in starch granules from the a5 c4, a10 a10d, a5 c4 a10 and quadruple mutants compared to that of the WT. However, an obvious alteration in starch granule morphology was observed in the sextuple sbe mutant, with uneven sized granules up to 4.4 µm, and a proportion of starch granules being distorted into irregular shapes (Figure  7B).

Figure 7.

Comparison of starch granule morphology in WT and the sextuple mutant. A, WT; (B) sextuple mutant. Starches were extracted from 4-week-old rosette leaves after a 16-h period of light. Starch granules were sputter coated with a gold/palladium blend using a Desk V Denton Vacuum, and was used to image using a FEI Quanta FEG 250 SEM. Three independently extracted starches were examined and representative images for the WT and the sextuple sbe mutants are shown. Red asterisks indicate selected irregular-shaped granules in the sextuple mutant. Bars are 10 µm.

Starch synthesis and turnover was altered in the leaves of the sextuple mutant

We next quantified starch content using rosette leaves collected during a diurnal cycle under different photoperiods. Under long-day (LD) photoperiod (16 h/8 h), starch analyzed from the first two rosette leaves at 16 d after germination accumulated at a comparable level between WT and the sextuple mutant at the end of light (EOL). At the end of dark (EOD), only 10% of starch was left in the WT, while 24% of EOL starch was detected in the sextuple mutant (Figure  8A). At 30 d after germination, the entire third rosette leaf was used for starch quantification, and the starch content in the WT at EOL was approximately three-fold that observed in leaves harvested at 16 d. Only half the amount of starch was accumulated in the sextuple mutant compared to the WT at 30 d (Figure  8B). Since starch was only partially degraded in older leaves following 8-h darkness in both WT and mutant (Figure  8B), starch turnover was further examined following extended darkness. Following an additional 9-h darkness, only 12% of EOL starch remained in WT leaves compared to 40% of EOL starch observed in the sextuple mutant (Figure  8B). When maintained in a short-day (SD) photoperiod (12 h/12 h), the sextuple mutant synthesized one-third less starch than the WT at EOL (Figure  8C). Consistently, at EOD, only 4% of starch was detected in WT, while 13% of starch remained in the sextuple mutant (Figure  8C). These results clearly demonstrated that starch synthesis in the sextuple mutant was significantly compromised as a function of developmental stages, and that the rate of starch degradation was slower in the sextuple mutant regardless of long or short photoperiods.

Figure 8.

Starch synthesis and turnover is altered in the rosette leaves of the sextuple mutant. A, Leaf starch content at the EOL and the EOD (16-h light/8-h dark) at 16 d after germination. B, Leaf starch content at the EOL, EOD (16-h light/8-h dark) and prolonged darkness of 9 h at 30 d after germination. C, Leaf starch content at the EOL and EOD (12-h light/12-h dark) at 28 d after germination. Data represent mean and se of three independent replicates in each condition.

Starch synthesis in the sextuple mutant is disrupted during embryo development, but has no impact on lipid accumulation at maturity

Given that the majority of carbon reserve during the reproductive stage is derived from assimilates synthesized in the photosynthetic leaves, we evaluated whether the disruption of starch metabolism in mutant leaves affects storage accumulation in seeds. Embryos at various stages of seed development were collected for starch and lipid measurement. As shown in Figure  9A, the starch visualized by iodine staining was initially observed in radical and hypocotyl regions (14 d after pollination [DAP]), and then accumulated throughout the entire embryo, followed by gradual degradation to a barely detectable level. Despite minimal residual starch at the maturation stage, a clearly stained spot at the root tip was persistent (Figure  9A, right lower part), which might be relevant to root gravitropism (Caspar and Pickard, 1989; Kiss et al., 1997). In WT embryos, starch content reached a maximum at 33 DAP before declining rapidly to a minimal level at 65 DAP when the seeds were at a full maturity. In contrast, starch accumulation in the embryos of the sextuple mutant was less than the WT, and reached a peak level 50% of that of the WT. Notably, the sextuple mutant embryos maintained their highest starch content over a 15-d period (25–40 DAP) before decreasing (Figure  9B). At later stages, WT and mutant embryos contained similar quantities of starch. Lipid synthesis in both WT and the sextuple mutant began after 18 DAP and then accelerated during 33-40 DAP in both WT and mutant embryos (Figure  9B), the latter time point being that at which starch content reached its maximum. There was no significant difference in the rate of oil biosynthesis, final oil content, and fatty acid composition at maturity between the WT and the sextuple mutant (Figure  9B;  Supplemental Figure S9).

Figure 9.

Starch and lipid accumulation during seed embryo development. A, Embryos at various stages (from 14 to 65 DAP) in upper part were imaged using a dissecting light microscope, and those in lower part were stained by iodine to illustrate starch distribution. Red arrowheads indicate starch in the root tip at maturation (55 and 65 DAP, respectively). Bars are 500 µm. B, Quantification of starch and lipid content throughout embryo development (mean and se of three or four independent replicates). Solid and dotted lines indicate starch and lipid contents, respectively. Black circles and squares represent WT and open circles and squares represent sextuple mutant.

Morphological differences between WT and the sextuple mutant

We compared growth of the sextuple mutant to WT plants across their life span. Under the growth conditions used (16-h light at 22°C/8-h dark at 18°C), we did not find discernable differences in rosette leaf development during the vegetative stage (leaf size and number) despite the fact that starch synthesis was compromised in the sextuple mutant (Figure  8). However, when plants were at the maturation stage, the main stem in the sextuple mutant was evidently thicker by a range of 21%–50% when measuring the perimeter of internodes (Figure  10, A and B). We also noticed that the sextuple mutant developed on average three more internodes than WT plants (Figure  10, A and C). Consistently, the additional sextuple mutant, LW13Q-22-167-8 which contains different types of mutations at BnaA10_SBE2.1s from the line LW13Q-22-59-36 (Table  4), also displayed a similar phenotype including thick stem and increased numbers of internodes. The seed number per silique in the sextuple mutant was ∼10% less than that in the WT (Figure  10D), resulting in slightly reduced total seed weight in the sextuple mutant (Figure  10E).

Figure 10.

Morphological differences between WT and the sextuple sbe mutant of B. napus. A, Representative images illustrating thick stem and more internodes on the main stem in the sextuple mutant. Arrows indicate the positions of each nodes and numbers indicate the positions of nodes from top to bottom. B, Comparison of perimeters of internodes first to sixth counted from the top and the lowest one between WT and the sextuple mutant. Data represent mean and se of six plants. C, Number of internodes on the main stem. Data represent mean and se of six plants. D, Number of seeds per silique. Data represent mean and se of a total of 60 siliques from three plants. E, Total seed weight per plant. Data represent mean and se of eight plants. Asterisks indicate statistically significant (P <0.05) according to Student’s t test.

Discussion

This study was initiated by our previous findings that expression of maize endosperm-specific isoforms SBEI or SBEIIb, in an Arabidopsis sbe null mutant, drastically altered starch metabolism and boosted biomass as well as total oil production (Liu et al., 2016). The highly conserved gene functions between B. napus and Arabidopsis provided the rationale to transfer this strategy from the model plant Arabidopsis to the canola oil crop. However, compared to the convenient availability of T-DNA insertion sbe mutant in Arabidopsis, generation of sbe null isoforms in B. napus, a tetraploid species with a more complicated genetic background, presents the initial challenge. Use of the CRISPR–Cas9-mediated gene editing system offers advantages in creating mutations at multiple genes particularly in polyploid species (Lyzenga et al., 2019; Zhang et al., 2019b). The present study demonstrated successful application of CRISPR–Cas9, editing up to six SBE genes in B. napus, enabling analysis of the roles of SBE homologs in starch metabolism and embryo lipid accumulation in B. napus.

Either subgenome A or C of B. napus usually possesses at least two homologs of the corresponding genes present in Arabidopsis. A BLAST search initially identified two homologs of BnaSBE2.1 (BnaA5_SBE2.1 and BnaC4_SBE2.1) and two of BnaSBE2.2 (BnaA10_SBE2.2 and BnaCnn_SBE2.2), respectively, from the available public database (http://www.genoscope.cns.fr/brassicanapus/). Thus, the initial design with the first construct pLW08 was to simultaneously edit three SBE genes since BnaCnn_SBE2.2 was confirmed to be absent in DH12075 (Supplemental Figure S1). The subsequent release of the genome database for the cultivar ZS11 identified two more SBE2.1 genes, ZS11C04G066320 and ZS11A04G025480 (Figure  1; Sun et al., 2017). PCR analysis confirmed the presence of one of the newly found SBE2.1, BnaAnn_SBE2.1, making a total of four SBE genes, with three BnaSBE2.1s and one BnaSBE2.2, in DH12075 cultivar. A second construct, pLW13, was thus made to further target BnaAnn_SBE2.1.

To streamline the vector design for the CRISPR–Cas9 system, we modified the previously established system that carries U3d, U3b, U6-1, and U6-29 promoters from Arabidopsis, respectively, to assemble multiple sgRNAs cassettes in tandem (Ma et al., 2015). Different sgRNAs demonstrated substantial differences in editing efficiency, ranging from 0% to 93%. In line with previous results (Ma et al., 2015; Zhang et al., 2019b), the sgRNAs with higher GC content (55%–60%) achieved appreciable editing rates, while the rate for those with 45%–50% GC content was between 20% and 30% and no mutation was found for the sgRNA (t6) with 40% GC content in this study. It has been reported that GC content below 70% generated a low risk of off-target effect, while >70% could potentially increase that possibility (Tsai et al., 2015). This study has thus provided a valuable guideline for designing GC content at 55%–60% in order to obtain a desirable gene editing rate in B. napus. It is not yet clear whether differences in editing efficiency are relevant to different promoters. The majority of mutations were deletions, insertions, or base substitutions that consistently occurred at the fourth or fifth nucleotide upstream of the PAM sites. It was noted that all the mutations detected in the T₀ generation were monoallelic or biallelic heterozygotes, and no homozygous mutations were obtained. This is different from the case in rice, where most mutations were homozygous or biallelic when 35S promoter was employed to drive Cas9 (Ma et al., 2015). Notwithstanding its role in embryo development, YAO is expressed ubiquitously, with high level of expression in tissues undergoing active cell division (Arabidopsis eFP Browser, bar.utoronto.ca). We thus deduce that the YAO promoter-mediated, gene editing, occurs during the formation of shoots in the T₀ generation.

In addition to the generation of relatively small indels, another advantage of designing dual-sgRNA towards the same gene(s) is to produce large fragment deletions, which provides a more attractive strategy to disrupt the functions of gene clusters. This strategy has been utilized in a few plant species, including Arabidopsis (Feng et al., 2014), rice (Zhou et al., 2014), and Nicotiana tabacum (Gao et al., 2015). In this study, the first binary vector was assembled with two sets of dual sgRNAs at a close distance of 48–57 bp, but we did not observe large fragment deletions on any of the three targeted genes, BnaA10_SBE2.2, BnaA5_SBE2.1, and BnaC4_SBE2.1, among the 29 lines. In contrast, in those lines transformed with the second construct pLW13 where the dual sgRNAs were merely 26-bp apart, large deletions between the two sites were observed. This differs from Arabidopsis in which large fragment deletions occurred between dual sgRNAs at a distance of 150 and 229 bp in Arabidopsis (Feng et al., 2014).

Instead of using constitutive promoters, such as CaMV 35S, to drive Cas9 protein expression, in this study the embryo sac-expressed YAO promoter was used in order to minimize the occurrence of chimeric mutations (Yan et al., 2015), which are nonheritable and are usually indicated by three or more types of mutations for a given target. The existence of chimera increases the difficulty of isolating homozygous mutants in subsequent generations and has been found frequently. For instance, a high frequency of chimerism was reported when editing the BnaA7LPAT2 gene of B. napus, producing up to 13 types of mutations at one target site (Zhang et al., 2019b). By comparison, in the present study, each of the 174 editing events (29 lines, 6 target sites) identified during the first round of transformation, and 34 editing events across three generations produced during the second round of transformation, encompassed no more than two types of mutations (Tables  1 and 4), indicative of a complete lack of chimeras. Further, all biallelic mutations which occurred at T₀ were stably transmitted to the T₁ generation, following Mendelian inheritance. The Cas9-induced indels generated on the target site were very stable during the subsequent generations, suggesting that sgRNA is very specific to the target site and that Cas9 does not induce new cleavage after the target sequence was edited.

A second round of transformation was performed in the Cas9 free mutant background. The successful removal of the Cas9 cassette, together with the plant selection marker, not only eliminated the possibility of potential off-target edits, but also provided a means for designing further sgRNAs to edit more targets using the CRISPR–Cas9 system. Thus the present strategy is very applicable for use with mutants when it is difficult to edit multiple targets using a single construct. This is especially advantageous for polyploid crops, since identification of homozygous, multi-mutations from a F₂ cross-population usually takes many generations.

Two events of gene duplications of BnaA10_SBE2.2 and BnaAnn_SBE2.1 in cultivar DH12075 were characterized through analyzing sequence chromatograms (Supplemental Figures S2–S4). In most cases, heterozygous mutation is shown by a balanced peak pair, and the progeny of the heterozygote follow Mendel’s law of segregation. However, in the present study, non-Mendelian inheritance was observed when identifying homozygous mutations on BnaA10_SBE2.2 or BnaAnn_SBE2.1. Further, repeated occurrence of unbalanced chromatogram peaks, which to the best of our knowledge has not previously been reported, required an explanation. The relative height of the unbalanced peaks, observed in LW08-11 and LW13Q-22-3 lines (including their progeny) was revealed to be positively correlated with the proportion of each of two sequence variants. Hence, the progeny segregation ratio was further assessed through the genotypes derived from the proportion of different variants, which led to the conclusion that each of BnaA10_SBE2.2 and BnaAnn_SBE2.1 has two identical loci, making a total of six isoforms including four BnaSBE2.1s and two BnaSBE2.2s, in cultivar DH12075. While not being able to know whether the duplication occurs at a separate chromosome position or as a tandem duplication, the same number of SBE isoforms between cultivars DH12075 and ZS11 indicates that two duplicated genes compensated for the two missing SBE2.1 and SBE2.2 isoforms in DH12075. Our data thus suggest the necessity of analyzing sequence chromatograms when characterizing mutation variants, especially in the case of unbalanced peaks, because the secondary peak is usually ignored for “base-calling” during Sanger sequencing (Carr et al., 2009).

The CRISPR–Cas9-induced small indels or large fragment deletions are, in most cases, prone to cause frameshifts within the coding region, thereby leading to mRNA degradation due to nonsense mRNA decay or generation of truncated proteins (Cong et al., 2013). Results here showed that the same mutation in the BnaA5_SBE2.1 gene produced two different transcripts which were derived from the recognition of two splicing sites. The edited BnaA5_SBE2.1 produced a frame shift transcript in the a5 c4, a5 c4 a10 and the quadruple mutant, whereas alternate splicing, via recognition of the second “AG” site, occurred in the sextuple mutant, converting the “off-frame” sequence back to “in-frame”, resulting in a predicted full-length protein with modified amino acids in the N-terminal transit peptide region. Such findings have not previously been reported in the application of CRISPR–Cas9 in plants since the induced mutations were typically assessed only at the genomic level. A recent study in human HAP1 cells, however, demonstrated that roughly one-third of the CRISPR–Cas9-induced mutations still produce proteins (Smits et al., 2019). Similar phenomena were also found in human endothelial cell lines (Lalonde et al., 2017) and zebrafish (Danio rerio; Prykhozhij et al., 2017). Residual protein expression was, in most cases, due to alternate splicing that either skipped the edited exon or via a different translation initiation site not interfered with the mutation (Smits et al., 2019). Our data, combined with the above, suggest that future evaluation of CRISPR-induced mutation in plants should include analysis at the transcript and protein levels, in addition to genomic analysis.

In light of the modified protein produced for BnaA5_SBE2.1 derived from alternate splicing in the sextuple mutant, particularly in the transit peptide, a question raised was whether this disrupted gene function or organelle targeting. In the sextuple mutant, albeit at a much lower level, residual SBE activity was detected using the phosphorylase a stimulation assay, and in-gel zymogram analysis. Further, starch remained readily measurable in the leaves of the sextuple mutant. This is different from the Arabidopsis sbe2.1 sbe2.2 double mutant, where starch synthesis was completely inhibited (Dumez et al., 2006). Considering that all edited SBE genes, except BnaA5_SBE2.1, resulted in highly truncated sequences in the sextuple mutant, the most likely explanation is that the edited BnaA5_SBE2.1 in the sextuple mutant still retains branching enzyme activity. This is supported by the prediction of a functional chloroplast transit peptide in the modified BnaA5_SBE2.1, which would enable the edited protein to be correctly targeted to chloroplasts, and the observation that starch continues to be synthesized. Interestingly, the alternate splicing of BnaA5_SBE2.1 was only observed in the sextuple mutant, the reasons for which are unclear.

Marked increases in plastidic PHO1 activities were evident in the a10 a10d, quadruple and sextuple mutants, and were associated with editing of both BnaSBE2.2 genes. Mutations of six BnaSBEs in the sextuple mutant did not appear to result in an additional increase in PHO1 activity when compared to the quadruple mutant, indicating the induced PHO1 is associated with the loss of BnaSBE2.2, not BnaSBE2.1. The impact of mutations of only BnaSBE2.1 genes on PHO1 activity could not be evaluated since we do not have the requisite mutant. Nonetheless, this differs from the case in Arabidopsis where simultaneous loss of both SBE2.1 and SBE2.2, rather than a single isoform, stimulated PHO1 activity (Dumez et al., 2006). It has been suggested that PHO1 may be involved in starch granule initiation in Arabidopsis (Malinova et al., 2017) and in rice endosperm (Satoh et al., 2008), possibly as part of a heteromeric protein complex (Tetlow et al., 2004). The presence of multiple SBEs in B. napus could confer functional specificity via association of specific isoforms of BnaSBE2.2 within a protein complex. Alternatively, the observation that BnaA5_SBE2.1 and BnaAnn_SBE2.1 transcripts were upregulated in the a10 a10d mutant could indicate that increased PHO1 was accompanied by an increase in BnaSBE2.1s. Such coordination between the two types of enzymes is consistent with gene transcriptional profiles in Arabidopsis leaves showing that PHO1 and SBE2.1, rather than SBE2.2, exhibited the same pattern of expression during a diurnal cycle (Smith et al., 2004; Ingkasuwan et al., 2012).

Noticeable decreases in the proportion of short glucan chains and increased long chains, in the mutant, indicate the class II SBE enzymes tend to enrich the short chain proportions in leaf starches and support previous data regarding the function of class II SBEs. Starch branching analyses clearly demonstrated that deficiency of BnaSBEs produced amylopectin with reduced branching frequency. The most striking changes in starch structural properties were observed in the sextuple sbe mutant and were only marginally evident in sbe double and triple mutants, consistent with functional redundancy between BnaSBE2.1 and BnaSBE2.2 isoforms. Even so, starch-bound phosphate was elevated with an increase in the number of mutated BnaSBEs and appeared to be proportional. Thus, starch phosphorylation in B. napus is subject to a SBE gene dosage effect, and inversely correlated to starch branching frequency. This is consistent with previous studies that reduced amylopectin branching resulted in elevated phosphate content in potato tuberous starch (Wischmann et al., 2005). Furthermore, the gene-dosage effect of starch branching frequency and phosphate content was also obvious when comparing the a5 c4 a10 and quadruple mutants (Figures  6, C and D), implying that the BnaA10_SBE2.2 duplicate was functionally active. In leaf starch, the amount of phosphate varies between species. Unlike in tubers, phosphate was barely detectable in potato leaf starch (Santacruz et al., 2005), whereas the phosphate content in Arabidopsis leaf starch (Liu et al., 2016) was of a similar order of magnitude to that in B. napus cultivar DH12075.

Further experiments focused on the sextuple mutant in which defects in starch structure were most apparent. Starch content at EOL was compromised in leaves of SD (12/12) plants, and in leaves of LD (16/8) plants at 30 d after germination but not in young leaves (16 d after germination), indicating that BnaSBEs employ higher control over starch synthesis at middle stage of rosette leaf development when starch is required to release more carbon reserves for leaf growth. Prolonged darkness resulted in a two-fold increase in starch in the sextuple mutant at EOL, compared to the WT, indicative of a slower rate of starch degradation.

The simplest explanation for decreased degradation of starch in the sextuple mutant is that it is a consequence of its altered structure. In agreement with this, a starch-excess phenotype during the dark phase has also been observed in a maize sbe2a mutant, in which an overall reduction in branching and the loss of granular uniformity were evident in leaves (Dinges et al., 2003; Yandeau-Nelson et al., 2011). In this regard, SBEII class enzymes are therefore also essential for proper starch degradation in addition to their direct roles in starch synthesis. Thus activities of enzymes involved in starch degradation are also likely to be perturbed as a consequence of the modification of starch structure. It is not yet clear whether slower mobilization of starch reserves in the sextuple mutant was associated with the thick stem phenotype, but increased diameter and vascular bundle areas of the stem has been found to be beneficial for stem lodging resistance (Wu et al., 2021). Given that the present study also demonstrated that the Bnasbe sextuple mutant maintained WT levels of seed oil synthesis, it may be worthwhile to evaluate the role of thickened B. napus stems during abiotic stress as a strategy to minimize yield loss under adverse environmental conditions.

Conclusion

The present study has exploited CRISPR technology to generate a range of sbe mutants and to characterize the effects of reduced SBE activities on starch structure and starch mobilization in an important oilseed crop. Importantly, the identification of gene duplication events, by monitoring sequencing chromatograms, and characterization of alternate gene-splicing patterns emphasize the need for rigorous genetic analysis when evaluating CRISPR–Cas9-induced mutations. To the best of our knowledge, these approaches have not previously been reported when applying CRISPR technology to plants. Such analyses, especially in species with higher ploidy levels, are likely to be critical for accurate interpretation of the effects of editing multi-gene families.

Materials and methods

SBE gene selection and sgRNA design

SBE gene sequences for B. rapa, B. oleracea, and B. napus were obtained from Ensembl Plants database (https://plants.ensembl.org) and national center for biotechnology information (NCBI) database (https://www.ncbi.nlm.nih.gov). The equivalent gene sequences in the cultivar DH12075 (developed by G. Seguin Schwartz and G. Rakow at the Agriculture and Agri-Food Canada, Saskatoon Research & Development Centre), used in this study, were confirmed through PCR cloning and sequencing using primers (Supplemental Table S3): 87694/87696 for BnaA10_SBE2.2, 87697/87698 for BnaCnn_SBE2.2, 87699/87701 for BnaA5_SBE2.1, 87702/87703 for BnaC4_SBE2.1, 94299/94300 for BnaAnn_SBE2.1 and 94298/94297 for the isoform homologous to ZS11A04G025480. A phylogenetic tree aligning SBE homologs was constructed using MEGA-X and the evolutionary history was inferred by using the Maximum Likelihood method and JTT matrix-based model (Tamura et al., 2007). Two sgRNA sequences, separated by 20–60 bp, were selected for targeting each individual BnaSBE gene, and a total of six sgRNAs, which were located within the first two exons (Figure  2, Supplemental Diagram), were therefore designed to target all BnaSBEs. All sgRNAs follow the PAM (N₂₀)GG on sense or anti-sense strand, and a Blast search (https://blast.ncbi.nlm.nih.gov) against the B. napus database was performed to evaluate their specificities in the genome. Target sequences having a difference of at least two bases compared with off-target sequences at the PAM-proximal region, and those having more than five-base mismatches within the 20-bp sequence were considered as an effective sgRNA. Potential off-target genes were selected with sequences containing less than four nucleic acid mismatches when compared to targeted sgRNAs.

Plasmid construction

The sgRNA primers were synthesized and the promoters of AtU3d (Addgene plasmid #66200), AtU3b (Addgene plasmid #66198), AtU6-1 (Addgene plasmid #66202), and AtU6-29 (Addgene plasmid #66203) were pre-assembled to individual plasmids, respectively (Ma et al., 2015). The sgRNA primers were ligated to the above plasmids in a reaction of 1x BsaI buffer, 1x ligase buffer, 5U of BsaI (NEB), 200U of T4 DNA ligase (NEB), 10-ng plasmid, and 1-μM primer pairs, and incubated in a thermo-cycler for five cycles (37°C, 5 min; 16°C, 5 min). The resultant products (1 μL) were amplified in a volume of 25 μL including 0.2-μM U-F and gRNA-R primers (Supplemental Table S3) using iProof High-fidelity DNA polymerase (Bio-Rad) for 27 cycles (98°C, 10 s; 56°C, 15 s; 72°C, 20 s). The PCR products were diluted 10 times and used for templates for a second round of PCR reaction for 27 cycles (98°C, 10 s; 56°C, 15 s; 72°C, 20 s), together with combinations of the site-specific primer pairs harboring BsaI restriction sites. The PCR fragments were gel purified and used for Golden Gate ligation in a 10-μL volume of 1x BsaI buffer, 1x ligase buffer, 5U of BsaI (NEB), 200U of T4 DNA ligase (NEB), 25-ng fragments, and incubated for 15 cycles (37°C, 2 min; 10°C, 3 min; 20°C, 5 min). The resultant, assembled, tandem sgRNA cassette was then digested with KpnI and SpeI restriction enzymes and ligated with the binary vector containing the Cas9 element driven by an Arabidopsis YAO promoter (Yan et al., 2015), to generate two plasmids pLW09 and pLW13, respectively.

Plant transformation

Cotyledons cut from 5-d-old seedlings were inoculated with Agrobacterium MP90 harboring the above binary vector, pLW09 or pLW13, and then kept in darkness at 22°C for 2 d followed by 4°C for 3 d. The explants were moved to MS medium containing 300 mg·L⁻¹ timentin and 5 mg·L⁻¹ hygromycin for 3–5 weeks. New, emerged shoots were transferred to fresh media for shoot elongation and rooting. Rooted shoots were transferred to peat pots or soil directly for growth. Each plant was bagged after entering the reproductive stage to prevent cross-pollination.

Detection of mutations and identification of Cas9-free plants

Genomic DNAs were extracted from leaves of T₀ generation transgenic plants. Each targeting region was amplified by gene specific primers (Supplemental Table S3): 87699/87701 for BnaA5_SBE2.1, 87702/89426 for BnaC4_SBE2.1, 89201/89200 for BnaA10_SBE2.2; 94313/94300 for BnaAnn_SBE2.1, respectively. The PCR fragments were purified and used for sequencing. Bi-allelic and monoallelic heterozygotes were represented by two overlapping chromatograph peaks (Supplemental Figures S2–S4). Two sequence variants were determined through cloning the PCR products into a plasmid and sequencing colonies. When determining the ratio of 2 sequence variants, approximately 20 randomly selected colonies were used for sequencing of each. The primers 92926/92927 for Cas9 and 88425/88426 for hygromycin were used to identify transgene-free lines.

RT-PCR analysis

Total RNAs were extracted from leaf tissues of WT and sbe mutants (RNeasy mini plant kit, Qiagen), and cDNAs synthesized with a QuantiTect Reverse Transcription Kit (Qiagen). To detect their transcripts, primers 94756/94607 for BnaA5_SBE2.1, 94757/94607 for BnaC4_SBE2.1, 99986/99958 for BnaAnn_SBE2.1, and 94733/94734 for BnaA10_SBE2.2 (Supplemental Table S3), were used for RT-PCR. The B. napus 18S rRNA was amplified using primers 93483/93484 as internal positive controls. The amplified BnaSBE transcripts were cloned into a plasmid, and at least 20 colonies were randomly selected for sequencing.

Protein extraction and measurement

Total soluble protein was extracted by mixing ground powders of rosette leaves and extraction buffer (50-mM Tris–HCl [pH 8.0], 1-mM Na₂-EDTA, 150-mM NaCl, 1-mM DTT, 1% protease arrest, G Biosciences) followed by centrifuge at 16,000g for 5 min. Protein content was determined using the Bio-Rad Quick Start Bradford reagent.

Chloroplast protein isolation

For isolating chloroplasts, two rosette leaves at 4-week-old collected at the end of dark were homogenized in grinding buffer (100-mM HEPES [pH 7.5], 660-mM sorbitol, 4-mM Na₂-EDTA, 2-mM MgCl₂, 2-mM MnCl₂). After filtering the homogenate through two layers of miracloth, the filtrate was centrifuged at 1,000g for 8 min at 4°C. The pellet was re-suspended gently with grinding buffer. An aliquot of suspended material was layered on a Percoll gradient (7 mL 85% [v/v] and 8 mL 40% [v/v]) and centrifuged at 3,000g for 15 min. Chloroplasts were collected at the interface of the two Percoll gradients and diluted with four times volume of HEPES-sorbitol buffer. After centrifugation at 1,000g for 6 min, the resultant pellet was resuspended in ruptured buffer (100-mM Tricine [pH 7.5], 7.5-mM MgCl₂), frozen in liquid nitrogen and thawed to rupture organelles. After centrifuging at 16,000g for 10 min, the supernatant was used to measure protein concentration.

Enzyme assays

SBE activity was assayed using a phosphorylase a stimulation assay according to Smith (1988). Briefly, the assay was conducted in a volume of 1,400 µL containing 100-mM sodium citrate (pH 7.0), 0.2 U rabbit muscle phosphorylase a (Sigma, P-1261), 1-mM EDTA, 1-mM DTT, 2.5-mM adenosine monophosphate, 0.3% (w/v) α-amylase inhibitor acarbose (Bayer), and 100-µg protein extracts. To start the reaction, [U-¹⁴C] glucose 1-phosphate (specific activity 74 MBq mol⁻¹, PerkinElmer Inc.) was added to a final concentration of 50 mM at 28°C. Each aliquot of 200 µL at time points of 0, 15, 30 45, 60, 90, 120 min, respectively, were removed and boiled for 1 min. The glucans were precipitated by adding 20 µL of 80 mg·mL^–1 glycogen (from oyster, Sigma, G8751) followed by 1 mL of methanol:KCl (75%:1% (v/w)) followed by centrifugation at 16,000g for 5 min. The pellets were washed in water and re-precipitated before transferring to scintillation vials for determining ¹⁴C counts.

In-gel zymograms were conducted to visualize SBE activities. Samples of 100-μg soluble protein were electrophoresed at 90 V for 3 h on nondenaturing 6% (w/v) acrylamide gels containing 0.3% (w/v) acarbose at 4°C for 3 h. Following electrophoresis, gels were incubated in buffer containing 20-mM glucose 1-phosphate, 2.5-mM adenosine monophosphate, 0.04% (w/v) maltoheptoase, 1-mM Na₂-EDTA, 1-mM DTT, and 2.4U phosphorylase a at 28°C overnight. Gels were stained with Lugol’s solution and photographed. SBE isoforms were identified by blotting the proteins onto nitrocellulose membranes and probing with Arabidopsis anti-SBE2.1 and anti-SBE2.2 antibodies (Liu et al., 2016).

To detect starch phosphorylase activity, 100-μg soluble proteins from whole-cell extracts or isolated chloroplast were electrophoresed on a nondenaturing 7% (w/v) acrylamide gel containing 0.1% (w/v) glycogen (from oyster, Sigma, G8751) and incubated in buffer containing 20-mM glucose 1-phosphate, 0.1% (w/v) glycogen at 28°C for overnight. Phosphorylase was visualized after staining gels with Lugol’s solution.

Leaf starch quantification

Starch quantification was determined according to Smith and Zeeman (2006). Intact vegetative leaves collected at various time points, or embryos at various stages, were boiled in 80% (v/v) ethanol for 30 min, five times to remove soluble sugars. The plant material was transferred to a mortar, homogenized thoroughly, and suspended in water. An aliquot of 500 μL of homogenate was removed to an Eppendorf tube and boiled for 10 min and incubated overnight with an equal volume of 200-mM sodium acetate (pH 5.5), 6 U of α-amyloglucosidase (Megazyme) and 1 U of α-amylase (Megazyme). Following overnight incubation with shaking at 37°C, the sample was centrifuged at 10,000g for 5 min at room temperature. An aliquot of 10–20 μL from each supernatant was used to determine glucose content (Smith and Zeeman, 2006).

Leaf starch purification

Rosette leaves harvested at the end of light were homogenized in pre-cold isolation buffer (100-mM MOPS [pH 7.2], 2.6-mM sodium thiosulfate, 0.05% [v/v] Triton X-100, 5-mM DTT, 5-mM Na₂-EDTA) using a Polytron. Leaf lysate was filtered through miracloth and then centrifuged at 3,900g for 20 min. The pellet was suspended in 20-mL isolation buffer on ice and was layered onto 20-mL 90% (v/v) Percoll, and centrifuged at 3,900g for 10 min at 4°C. The pellet was re-suspended in 2-mL isolation buffer and re-layered onto 2-mL 90% (v/v) Percoll. The resulting pellet was then washed with water for three times and acetone for one time, and left to airdry in fume hood.

Amylopectin chain length distribution

Approximately 2-mg starch was gelatinized in 50-μL dimethyl sulfoxide by boiling, and then mixed with 1-mL pre-heated 10-mM sodium acetate (pH 3.5) and fully solubilized. After cooling, an aliquot of 6-μL sodium acetate solution containing 2400U isoamylase (Sigma, Cat. 08124) was added. The mixture was incubated at 1,200 rpm (Thermomixer, Eppendorf) for 24 h in the darkness, and then filtered through 0.45-μm nylon filter membrane. A volume of filtered 20 μL was loaded and analyzed by a Dionex ICS 3000 High-Performance Anion-Exchange Chromatography system equipped with a PA-100 column (Thermo Fisher Scientific). The percentage of each DP chain was calculated based on peak area (Bertoft et al., 2008; Annor et al., 2014).

Starch branching frequency assay

The quantitative branch-linkage assay was used to determine glucan/glucose reducing ends according to Waffenschmidt and Jaenicke (1987). Five mg starch was debranched with isoamylase as above, followed by boiling for 10 min. Debranched starch sample (200 μL) was mixed with 100 μL of solution A (97.1 mg of disodium bicinchoninate, 3.2 g of sodium carbonate monohydrate, and 1.2 g of sodium bicarbonate in a total volume of 50 mL) and 100 μL solution B (62 mg of copper sulphate pentahydrate and 63 mg of L-serine in a total volume of 50 mL), and incubated at 100°C for 15 min. After cooling, 600 μL of water was added to the sample. Absorbance was measured at 560 nm using a spectrophotometer. The number of reducing ends was calculated based on the linear line which was calibrated by a concentration gradient of glucose.

Starch-bound phosphate measurement

Glucosyl 6-phosphate esters were determined in starch hydrolysates by an enzymatic cycling assay described by Nitschke et al. (2013).

SEM

Starch granules were sputter coated with a gold/palladium blend using a Desk V Denton Vacuum, and was used to image using a FEI Quanta FEG 250 SEM.

Fatty acid composition and lipid quantification

Fatty acid profile and oil content measurement were conducted according to Wang et al. (2012). Total lipid was extracted from 10 to 20 embryos according to the protocol from the Kansas Lipidomics Research Center. Briefly, embryos together with 50-μg tripentadecanoin as internal standard were heated in 2 mL isopropanol with 0.01% butylated hydroxytoluene at 85°C for 10 min and then homogenized. Chloroform and methanol were added for extraction. Samples from five rounds of extraction were combined and dried under nitrogen stream. The dried lipids were heated with 2 mL of 3 N methanolic HCl at 80°C for 2h. Fatty acid methyl esters were extracted with hexane and analyzed by an Agilent 6890-series gas chromatograph (Agilent Technologies, Inc., Wilmington, DE, USA). A 60 m × 0.22 mm internal diameter with a 0.25 μm film thickness GC column BPX70 (SGE Inc., Austin, TX, USA) was used. The fatty acid composition was represented as percentages of individual fatty acids and oil content was determined according to the amount of internal standard.

Accession numbers

AtSBE2.1, At2G36390; AtSBE2.2, At5G03650; YAO, At4G05410; PHO1, At3G29320.

In B. napus cultivar ZS11, BnaSBE2.1s are: ZS11A05G008670, ZS11C04G011300, ZS11A04G025480, ZS11C04G066320; BnaSBE2.2s are ZS11A10G030690, ZS11C09G072390.

In B. napus cultivar DH12075, sequence data for BnaA5_SBE2.1, BnaC4_SBE2.1, BnaAnn_SBE2.1 and BnaA10_SBE2.2 genes in can be found in the GenBank/EMBL data libraries under accession numbers OL416042, OL416041, OL416043 and OL416040, respectively.

Supplemental data

The following materials are available in the online version of this article.

Supplemental Figure S1. Characterization of SBE isoforms in the DH12075 cultivar through PCR analysis.

Supplemental Figure S2. RT-PCR analysis of BnaSBE gene expressions in different tissues.

Supplemental Figure S3. DNA sequencing chromatograms for heterozygous mutations of BnaA10_SBE2.2.

Supplemental Figure S4. DNA sequencing chromatograms of BnaA10_SBE2.2 in T₂ progeny of the lines LW08-11-55 and LW08-11-61.

Supplemental Figure S5. DNA sequencing chromatograms for heterozygous mutations of BnaAnn_SBE2.1.

Supplemental Figure S6. Identification of SBE2.1 and SBE2.2 following nondenaturing PAGE.

Supplemental Figure S7. Zymograms of PHO1.

Supplemental Figure S8. RT-PCR analysis.

Supplemental Figure S9. Fatty acid composition in mature seeds of the WT and sextuple mutant.

Supplemental Table S1. Genotype of BnaA10_SBE2.2 gene from T₂ progeny of the line LW08-11.

Supplemental Table S2. Characterization of potential off-targets.

Supplemental Table S3. Primers used in this study.

Supplemental File S1. DNA and protein sequences illustrating edits in BnaSBE genes in the sextuple sbe mutant.

Supplemental File S2. Individual data points represented in Figure  6, A and B.

 

L.W. designed and performed most of the experiments; Y.W. analyzed leaf starch content and helped genotyping; A.M. conducted starch branching enzyme assay; F.N. measured phosphate content in leaf starch granules; all the authors analyzed the data; M.J.E. and I.J.T. conceived the study; L.W. and M.J.E. managed and supervised the project; L.W., M.J.E., and I.J.T. wrote the manuscript.

The author responsible for distribution of materials integral to the findings presented in this article in accordance with the policy described in the Instructions for Authors (https://academic.oup.com/plphys/pages/general-instructions) is: Michael J. Emes (memes@uoguelph.ca).