Efficient CRISPR/Cas9 Genome Editing of Phytoene desaturase in Cassava

1Development of superior crop varieties has relied on securing genetic gains through sexual recombination, induced random mutagenesis, and transgenic approaches. The recent emergence of targeted genome-editing technologies offers a new avenue for incorporating beneficial genetic changes in the world’s most important crop species.

2Gene-editing technologies based on zinc-finger nucleases (ZFNs), transcription activator-like effector nucleases (TALENs), and clustered regularly interspaced short palindromic repeats (CRISPR) make use of engineered nucleases to induce double strand breaks (DSB) at known DNA sequences within the genome. Subsequent repair at the target site introduces variation via error prone non-homologous end joining (NHEJ) to generate insertions and deletions (INDELs), with occasional substitution of nucleotides occurring at the repair regions. Alternatively, the error free homologous recombination (HR) repair pathway can be activated in presence of a homologous donor DNA template spanning the DSB, resulting in targeted gene replacement.

3The ZFNs and TALENs are based on protein-guided recognition mechanisms. Vector construction for targeting DNA sequences using these technologies is a relatively complex and costly process. In contrast, the CRISPR/Cas9 system depends on DNA or RNA sequence homology, in which targeting a specific DNA sequence requires only a 17–23-bp complementary nucleotide sequence called a guide RNA (gRNA). This versatility, plus high efficacy, and low cost have made CRISPR/Cas9 the most widely adopted of the three genome-editing technologies. CRISPR/Cas9 has been applied to delete, activate, and suppress targeted genes in human cells, animals, and plants. Plant species altered by CRISPR/Cas9 technology include Nicotiana benthamiana, Arabidopsis, tobacco, sorghum, and rice, tomatoes, wheat, soybean, maize, potato, sweet orange, populus, cucumber, and cotton. In several cases, CRISPR/Cas9-induced mutations have been confirmed to be heritable across sexual generations.

4Cassava (Manihot esculenta Crantz) is an essential source of calories for millions of people across the world’s tropical regions, providing storage roots for local consumption and raw materials for processed foods and industrial products. Conventional breeding for cassava improvement is constrained by its long life cycle, high level of heterozygosity, poor flowering, and seed set and inbreeding depression. CRISPR/Cas9-targeted genome editing therefore offers important new opportunities for addressing pre- and post-harvest constraints affecting cassava production and utilization. To evaluate the potential of CRISPR/Cas9 in a new species such as cassava, a reproducible system for design, construction, and delivery of Cas9/gRNA reagents must be developed and validated. We report here an efficient CRISPR/Cas9-mediated targeting of Phytoene desaturase (PDS) in two cassava cultivars. The results reported set the groundwork for future studies of gene function and trait development using the CRISPR/Cas9 gene-editing technology in this important staple food crop.

5BLAST was used to search the reference cassava genome in Phytozome 11 (version 4.1)1 for nucleotide sequences homologous to the Arabidopsis thaliana PDS gene (AtPDS: NM117498). A single copy gene, here referred to as MePDS (Manes.05G193700/cassava4.1_004359m.g), was identified. The CRISPR-P program (version 1)2 was employed to design gRNAs. Two target sequences were selected by initially identifying the NGG protospacer adjacent motif (PAM) sequence essential for Streptococcus pyogenes Cas9, followed by 20 nucleotides upstream of the PAM sequence for use as the spacer in the gRNA to target MePDS. Two gRNAs chosen were located in the 13th exon, each having a G at their 5′ end to enhance gRNA transcription in plant cells using the U6-26 promoter when transcribed as single molecule with gRNA scaffold (Figures 1A,C).

6A pair of complementary DNA oligonucleotides were synthesized (Integrated DNA Technologies3) and hybridized to form a duplex which has compatible ends with the BsaI-digested empty vector (pCAMBIA2300_CR3-EF, provided by B. Staskawicz, University of California, Berkeley, CA, United States). The empty vector construct contained Cas9 coding sequence from S. pyogenes and gRNA scaffold with a BsaI restriction site for insertion of a single gRNA. The included nptII marker gene aids in selection of transgenic cells during regeneration process. The 35S Cauliflower mosaic virus promoter was utilized to drive expression of Cas9, while the U6-26 promoter from Arabidopsis drove expression of the gRNAs (Figure 1C). The presence of the inserted gRNA and stability of the final constructs was confirmed by sequencing. The two binary vectors, MePDS-1 and MePDS-2, each carrying one gRNA, were transformed into Agrobacterium strain LBA 4404 by electroporation.

7Friable embryogenic callus (FEC) was generated from cassava cultivars 60444 and TME 204 and used as target tissue for transformation with the gene-editing constructs MePDS-1 and MePDS-2, plus empty vector (EV) and green fluorescent protein (GFP) controls. Regenerated cotyledon-stage somatic embryos and whole plantlets with altered PDS activities were visually discriminated from control tissues under a dissection microscope. Plants were established in the greenhouse by transfer to Fafard 51 soilless potting medium (Conrad Fafard, Inc., Agawam, MA, United States) before moving them to the open bench and grown at a temperature of 26°C/25°C (day/night) with 60–90% relative humidity for 3 weeks to reach approximately 8 cm in height, at which time they were transferred to a glasshouse maintained at a temperature of 32°C/27°C (day/night) and 70–95% relative humidity.

8Leaf tissue was collected from rooted in vitro plantlets. Approximately 0.1 g of leaf tissue was placed in a 2-ml screw cap tube containing ceramic beads and crushed to fine powder using a Fast-Prep 24 (MP Biomedicals, Solon, OH, United States) set at 4.0 m/s for 30 s twice. Genomic DNA was extracted using cetyltrimethylammonium bromide (CTAB) protocol and treated with RNase, DNase-free kit (Roche Diagnostics GmbH, Indianapolis, IN, United States) to remove RNA.

9Putatively edited plant lines were subjected to polymerase chain reaction (PCR) to confirm integration of the T-DNA using 200 ng genomic DNA as a template and primers specific for the Cas9 gene, U6-26 promoter, and gRNA scaffold (Table 1), under the following conditions: 98°C for 5 min; 34 cycles of 98°C for 10 s, 65°C for 20 s, 72°C for 30 s followed by final extension of 72°C for 5 min. Genomic DNA template was used for PCR amplification of a 504-bp fragment spanning the targeted MePDS sequence. Briefly, 200 ng of DNA was added to Phusion DNA Polymerase Master Mix (New England Biolabs4) in a 25 μl reaction volume and PCR performed at 98°C for 5 min; 34 cycles of 98°C for 10 s, 64°C for 20 s, 72°C for 30 s, followed by final extension at 72°C for 5 min. A single primer pair was designed to amplify both target sites based on the MePDS gene sequence from reference genotype AM560-2 in Phytozome 11, version 4.1. PCR products were resolved on a 1.5% agarose gel, excised, and amplicons purified using the QIAquick gel extraction kit (Qiagen Group), before cloning into pCR Blunt-II TOPO vector (Life Technologies5). Between 5 and 10 single colonies per plant line were subjected to Sanger sequencing using M13F and M13R primers to characterize CRISPR/Cas9 induced mutations. Sequences were aligned with the wild-type reference sequence of MePDS gene using the MUSCLE alignment program of EMBL-EBI6. The mutation frequency in transgenic plants was calculated by expression of plant lines with phenotypic alterations or mutated sequences as a percentage of total plant lines examined.

10The nucleotide sequence of A. thaliana PDS (AtPDS: NM117498) was used as bait to search for homologous sequences within the reference cassava genome in Phytozome 11 (version 4.1) and a single copy of MePDS identified. The genomic sequence of MePDS is 13,130 nucleotides in size, comprising 14 exons and 13 introns (Figure 1A), with 1,743 nucleotides of transcript sequence coding for 580 amino acids. AtPDS (NP_193157.1) and MePDS (Manes.05G193700.1) are 85% identical at the protein level, suggesting a conserved function in cassava. In order to disrupt MePDS, two distinct gRNAs were designed as predicted by the CRISPR-P program to have the highest efficacy and least off-target potential. The gRNAs were cloned into two separate binary vectors (MePDS-1 and MePDS-2) carrying the Cas9 gene (Figure 1B), Arabidopsis U6-26 promoter and gRNA scaffold (Figure 1C), and delivered by Agrobacterium into embryogenic cells of cassava.

11A total of 103 independent lines of regenerating somatic embryos were recovered from cv. 60444 and 85 from cv. TME 204 after selection on antibiotic-containing medium (Table 2). Visual indication of altered MePDS function was first apparent as albino or partial albino cotyledon-stage somatic embryos regenerating on selection medium (Figure 2). This phenotype was observed in 100/103 (97.1%) and 84/85 (98.9%) somatic lines of cv. 60444 and TME 204 transgenic for MePDS-1 and MePDS-2, respectively. Both constructs were equally effective for producing the mutated phenotype. A total of 47% of the somatic embryo lines from cv. 60444 and 22% from cv. TME 204 germinated to produce plantlets, generating 58 and 25 independent transgenic plant lines from 60444 and TME 204, respectively.

12A diversity of albino phenotypes was observed in the regenerated plants, ranging from full albinos with total loss of green coloration, to partial albinos displaying a mixture of green and white tissues on the same plantlet (Figures 2, 4). In all cases, non-edited wild-type, EV and GFP-expressing control plants remained fully green. Both gRNAs generated albino mutants with high frequencies ranging from 90% for MePDS-2 to 100% for MePDS-1 in cv. TME 204 (Figure 6A). Mutant plants characterized by full or partial albino phenotypes displayed diminished growth rates, shorter petioles, and reduced leaf area compared to the non-edited control equivalents (Figures 4, 5). Albino plant lines with greatly or totally diminished green tissue (Figures 4, 5A) did not survive more than two to three successive in vitro propagation cycles. A selected set of partial albino and control plants was transferred to soil, where the dwarf phenotype and albino nature were maintained when grown under greenhouse conditions. In contrast, control EV and GFP-expressing control plants were fully green and showed robust growth (Figure 5).

13Polymerase chain reaction (PCR) was performed on genomic DNA extracted from regenerated plants to detect the presence of the T-DNA and to amplify the targeted MePDS gene (Table 1). As expected, all albino and EV plant lines tested positive for presence of the Cas9 gene, U6-26 promoter, and gRNA scaffold (Figures 3B,C). A 504 bp fragment corresponding to the target region of the MePDS gene in cassava was amplified (Figure 3A), and amplicons from 38 plant lines (19 each for 60444 and TME 204) were cloned and 241 clones sequenced (Table 3). All full or partial albino plants were found to carry mutations in their MePDS gene, with the MePDS-1 and MePDS-2 constructs equally efficacious for inducing mutations in the two cultivars tested (Table 3). Both INDELs and substitutions were observed. Regardless of the construct and cultivar tested, substitutions occurred more frequently than insertions and deletions (Figure 6B). Constructs MePDS-1 and MePDS-2 directed insertion or substitution of different nucleotides. In cv. 60444, both constructs induced insertion and substitution of nucleotide ‘A’ compared to substitution of nucleotides ‘CA’ and ‘T’ in cv. TME 204. The longest deletion detected in cv. 60444 was 16 bp, compared to 101 bp observed in cv. TME 204 plant line 4 (Supplementary Figure S1). Examination of the 504 bp MePDS target sequence after Sanger sequencing revealed 1 bp nucleotide substitutions and INDELs upstream of the 5′ and or downstream of the 3′ MePDS target region for both constructs MePDS-1 and MePDS-2. We observed a single nucleotide polymorphism (SNP), which overlapped with MePDS-2 gRNA (‘C’ substituted for ‘G’) targeting the MePDS region of cv. TME 204, but this did not negatively affect CRISPR/Cas9 editing of MePDS.

14The approach developed by in populus was used to classify plant lines as putatively homozygous or heterozygous for the induced mutations with slight modification. Sequences derived from albino mutants were aligned and compared with the reference wild-type sequence of MePDS. Mutant lines in which all clones examined carried one identical type of mutation were classified as putatively homozygous mono-allelic, while mutants whose clones displayed two identical types or patterns of mutations were considered to be putatively homozygous bi-allelic. Mutants categorized as heterozygous were chimeric, having at least three types or patterns of mutations and/or wild-type sequence present in the same plant. Based on this scheme, 1/19 plant lines (5.23%, line 9) of cv. 60444 was putatively homozygous mono-allelic since all five clones carried two substitutions (Figure 7). Further scrutiny revealed that 1/19 plant lines (5.23%, line 8) of cv. 60444 and 4/19 plant lines (21.1%, lines 2, 6, 7, and 9) of cv. TME 204 were putatively homozygous bi-allelic (Table 3); 17/19 (89.5%) of cv. 60444 and 15/19 (78.9%) of cv. TME 204 mutant plant lines were categorized as heterozygous (Table 3). Assessment of sequence and phenotypic data further indicated that the majority of full albino plant lines edited with MePDS-1 were putatively homozygous for mutations at MePDS (Table 3). Plant line 5 of cv. TME 204 which remained fully green showed sequence alteration in only one of the two MePDS alleles (Figure 7). The altered sequence (putatively homozygous or heterozygous) and albino (full or partial albino) frequency data agree with over 90% of CRISPR/Cas9 directed mutation frequency (Figure 6A). When the 504 bp MePDS target sequence was examined for the presence of off targets, Sanger sequencing revealed 1 bp nucleotide substitutions and INDELs upstream of the 5′ and or downstream of the 3′ target region. In most cases, such off-target effects occurred next to a random PAM site NGG or NCC within the MePDS gene.

15CRISPR/Cas9 genome-editing technology has become an effective tool for basic research and trait development in crop plants. To date, however, no such capacity has been reported for the tropical crop cassava. Efficient gene editing in this important staple food crop would present new opportunities to address biotic and abiotic constraints in cassava production and post-harvest utilization. Toward this end, CRISPR/Cas9 was employed to target the PDS gene in cassava. PDS codes for key enzymes in the carotenoid biosynthesis pathway. Defects in PDS gene function cause dwarfism and albino phenotypes in Arabidopsis due to impaired chlorophyll, carotenoid, and gibberellin biosynthesis. Reports of CRISPR/Cas9 targeting of PDS in rice, apple, sweet orange, and populus resulted in similar phenotypic aberrations. Deliberate alteration of the PDS gene was therefore chosen as a visual marker to reveal efficacious application of CRISPR/Cas9-targeted genome editing in cassava. PDS offers an attractive target for developing gene-editing techniques due to its monogenic nature and the easily assessed albino phenotype resulting from its mutation.

16In order to maximize our chance of success, two gRNAs were designed to interrupt MePDS function by targeting exon 13 in cassava cultivars 60444 and TME 204. The 13th exon was chosen because CRISPR-P output indicated that it provided optimal gRNAs starting with ‘G’ for use with the U6 promoter for achieving successful gene editing. In addition, previous studies report that mutations to the PDS gene at any location translate into phenotypic alteration. Existing genetic transformation systems were employed to integrate the CRISPR/Cas9 tools into embryogenic cells. Albino phenotypes were recovered from both cultivars after transformation with two gRNA constructs (Figures 2, 4, 5 and Table 2). Albino somatic embryos were visualized as early as 8 weeks after transformation and could be regenerated to plantlets displaying full and partial albino characteristics (Figures 2, 4, 5), in a manner similar to that reported in other plant species. Recovery of the albino mutants occurred at unexpectedly high frequencies of 93–95% of the regenerated plant lines of cvs. 60444 and TME 204, respectively. The albino phenotypes observed in the current study resulted from loss of MePDS function generated by action of the CRISPR/Cas9 reagents. The observed diversity and degree of the albino phenotype recorded here could have resulted from differential timing of the targeted mutation. Due to the integrated nature of the editing tools, editing will continue from CRISPR/Cas9 expressing cassettes within a given cell until the target site is mutated. This can result in chimeric plants consisting of cells with non-modified genomes, and populations of cells carrying different gene edited events. The exact pattern of these would depend on timing of CRISPR/Cas9-mediated editing in relation to the plant regeneration process. In rice, for example, the occurrence of full albino mutants totally lacking chlorophyll and partial albinos characterized by a mixture of albino and green tissues was attributed to induction of mutations before and after first cell division.

17Molecular analysis was performed on albino and green plant lines, with multiple clones sequenced per plant line to identify mutations within one or both MePDS alleles. Sequence data indicated the presence of INDELs and substitutions in all (100%) of the 38 (19 from each cultivar) regenerated plant lines examined at the DNA level. Irrespective of cultivar and construct used, INDELs mostly occurred 1–3 bp upstream from the PAM site. Predominantly single nucleotide insertions and substitutions (Figure 7) were detected in MePDS resulting from both gRNA constructs. However, larger deletions, including 1 of 101 bp in TME 204, were also observed (Supplementary Figure S1). The INDELs recorded here could have resulted from NHEJ repair following integration of our constructs MePDS-1 and MePDS-2 as previously reported in apple and populus following CRISPR/Cas9-targeted mutation of the PDS gene. Observation of single nucleotide substitutions upstream or downstream of the target region (Figures 6B, 7 and Supplementary Figure S1) implies activation of HR pathway to repair DSBs created in MePDS. Simultaneous triggering of HR and NHEJ repair pathways proceeding integration of CRISPR/Cas9 reagents was documented earlier in apple, N. benthamiana, and cotton. We also recovered wild-type sequences from some colonies derived from plants that clearly showed altered phenotype (Figure 7). This finding is similar to data reported for CRISPR/Cas9 gene targeting in N. benthamiana and rice. Gene edited TME 204 plant line #4, transgenic for MePDS-1, returned three clones with double nucleotide substitutions (CA) (Supplementary Figure S1). This rare occurrence seems to indicate that CRISPR/Cas9 can direct differing frequency and types of mutations. Induction of high mutation rates by MePDS-2 gRNA that carried a single bp mismatch with the cv. TME 204 target sequence was common. This finding is not surprising in light of a similar recent report in which a target sequence with a single nucleotide difference was deliberately chosen to successfully induce CRISPR/Cas9 directed mutation in homologous Chloroplastos alterados 1 genes of cotton (GhCLA1). Four transgenic plant lines out of the 69 (4/69, 5.8%) examined displayed unexpected visible green phenotype like WT and EV controls. However, sequence data showed presence of nucleotide substitutions in one allele of MePDS in these plants (Figure 7). This phenomenon is consistent with results obtained in citrus, Arabidopsis, and N. benthamiana, and could result from presence of very low populations of photo-bleached PDS cells among non-modified tissues. Alternatively, the presence of a single, fully functional allele may be sufficient to ensure functionality of MePDS. Possibly because of the high mutation efficiency achieved, we simultaneously induced putatively homozygous (up to 21.1% in TME 204) and heterozygous (up to 89.5% in 60444) mutations in MePDS gene in these T0 events. All putatively homozygous, mono-allelic plants showed full albino phenotypes while putatively homozygous, bi-allelic plants were either full or partial albinos, with putatively heterozygous counterparts displaying partial albino phenotypes.

18The very high efficiency of mono and bi-allelic gene editing achieved from two constructs across both cultivars studied equates with similar reports in potato (83%) and cotton (98–100%). The surviving edited mutants showed persistence of altered phenotype across micropropagation cycles and under greenhouse conditions. These results indicate significant potential for successful application of gene-editing technologies for the development of cassava with improved farmer and consumer traits. There is need, however, to further develop CRISPR/Cas9 gene-editing capacity for this crop. It is considered that the high efficiencies achieved here result, at least in part, from integration of the CRISPR/Cas9 and gRNA cassette. Continuous expression of the editing tools maximizes opportunity for production of mutations at the target site. Although a powerful tool for basic research, such an approach is not optimal for developing enhanced germplasm with value for cassava farmers. Segregating away the integrated T-DNA by sexual crossing in cassava is possible, but is not desirable due to the crop’s heterozygous nature and associated in-breeding depression. It is important, therefore, to develop genome-editing capacity for cassava that does not require integration of the CRISPR tools. Such techniques include direct delivery into totipotent cells and use of ribonucleoprotein (RNP) as alternatives to Agrobacterium-mediated transgenesis. The expanding range of alternatives to Cas, such as Cpf1, also brings increased flexibility for targeting genes, transcription factors, and their control elements. The MePDS knockout system described here generates easily detectable mutated events in a relatively short time frame of 6–8 weeks, and due to direct visualization of the phenotype, does not require sequencing to confirm editing at the target. It therefore provides a valuable method to facilitate rapid assessment and optimization of CRISPR/Cas9 and other genome-editing technologies in cassava.

19JO designed the experiments, produced, and analyzed the gene edited plants and wrote the manuscript. TA and II interpreted the data and edited the manuscript. RB designed the analysis and interpreted sequencing data, wrote and edited the manuscript. DN designed the experiments and edited the manuscript. NT supervised the design of experiments and analysis of edited plants, wrote and edited the manuscript.

20The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.