ORIGINAL ARTICLE
Efficient targeted multiallelic mutagenesis in tetraploid potato
(Solanum tuberosum) by transient CRISPR-Cas9 expression
in protoplasts
Mariette Andersson
1
•
Helle Turesson
1
•
Alessandro Nicolia
2
•
Ann-Sofie Fa
¨
lt
1
•
Mathias Samuelsson
3
•
Per Hofvander
1
Received: 30 August 2016 / Accepted: 27 September 2016 / Published online: 3 October 2016
Ó The Author(s) 2016. This article is published with open access at Springerlink.com
Abstract
Key message Altered starch quality with full knockout
of GBSS gene function in potato was achieved using
CRISPR-Cas9 technology, through transient transfec-
tion and regeneration from isolated protoplasts.
Abstract Site-directed mutagenesis (SDM) has shown
great progress in introducing precisely targeted mutations.
Engineered CRISPR-Cas9 has received increased focus
compared to other SDM techniques, since the method is
easily adapted to different targets. Here, we demonstrate
that transient application of CRISPR-Cas9-mediated gen-
ome editing in protoplasts of tetraploid potato (Solanum
tuberosum) yielded mutations in all four alleles in a single
transfection, in up to 2 % of regenerated lines. Three dif-
ferent regions of the gene encoding granule-bound starch
synthase (GBSS) were targeted under different experi-
mental setups, resulting in mutations in at least one allele in
2–12 % of regenerated shoots, with multiple alleles
mutated in up to 67 % of confirmed mutated lines. Most
mutations resulted in small indels of 1–10 bp, but also
vector DNA inserts of 34–236 bp were found in 10 % of
analysed lines. No mutations were found in an allele
diverging one bp from a used guide sequence, verifying
similar result s found in other plants that high homology
between guide sequence and target region near the proto-
spacer adjacent motif (PAM) site is essential. To meet the
challenge of screening large numbers of lines, a PCR-based
high-resolution fragment analysis method (HRFA) was
used, enabling identification of multiple mutated alleles
with a resolution limit of 1 bp. Full knockout of GBSS
enzyme activity was confirmed in four-allele mutated lines
by phenotypic studies of starch. One remaining wild-type
(WT) allele was shown sufficient to maintain enough
GBSS enzyme activity to produce significant amounts of
amylose.
Keywords Genome editing New plant breeding
technique GBSS Starch Amylopectin
Introduction
Potato is ranked as the third most important food crop in
the world (Barrell et al.
2013). It has a high nutritional
value and yields a high-energy output per hectare. Potato is
not only of importanc e as a food crop, it is also one of the
major crops grown for starch production (Ellis et al.
1998).
Development of new potato cultivars using traditional
cross-breeding is complicated and slow due to tetrasomic
inheritance and high heterozygosity of cultivated varieties
(Muthoni et al.
2015). Therefore, breeding technologies
where only one or a few traits can be introduced into an
elite backgr ound is of major interest for potato. Genetic
modification (GM), by stable integration of genetic mate-
rial, has been a widely used method in potato research and
Communicated by T. Cardi.
Electronic supplementary material The online version of this
article (doi:
10.1007/s00299-016-2062-3) contains supplementary
material, which is available to authorized users.
& Per Hofvander
per.hofvander@slu.se
1
Department of Plant Breeding, Swedish University of
Agricultural Sciences, P.O. Box 101, SE-23053 Alnarp,
Sweden
2
ENEA Research Centre Casaccia, SSPT-BIOAG-BIOTEC,
Via Anguillarese, 301, 00123 Rome, Italy
3
Lyckeby Starch AB, Degebergava
¨
gen 60-20,
SE-29191 Kristianstad, Sweden
123
Plant Cell Rep (2017) 36:117–128
DOI 10.1007/s00299-016-2062-3
breeding for a long time (Barrell et al. 2013). However, a
long and expensive dere gulating process in Europe and
elsewhere has restricted commercialisation of the devel-
oped GM plants. New breeding techniques where no
recombinant DNA is introduced or maintained in the plant
chromosomes has shown promise and there is a discussion
on whether these techniques result in events that should be
regulated as GMO (Jones
2015; Waltz 2016).
Starch produced from potatoes has many uses, both in
food and technical applications, and is often chemically or
physically modified to certain specifications (Ellis et al.
1998). Some of the processes for chemically or physically
modified starches cause environmental concerns and it
would be advantageous for these processes to be replaced
by starch modified in planta. Starch is a mixture of two
components, amylose and amylopectin. Changing the ratio
of these two components greatly alters the properties of the
starch. Many applications would gain advantages by hav-
ing a change in proportion or a starch containing only one
of the two components (Zeeman et al.
2010). The so-called
waxy genotype producing essentially only amylopectin
starch was first identified in maize and the corresponding
locus was identified (Klosgen et al.
1986; Shure et al.
1983). One single enzyme was found responsible for the
synthesis of amylose, granule-bound starch synthase
(GBSS). In potato, as in most other plants, the GBSS
enzyme is encoded by a single locus (GBSSI) having four
alleles in the cultivated potato. High amylopectin potatoes
(Waxy potato) have been developed by silencing of the
GBSSI gene function through the use of antisense tech-
nology (Kuipers et al.
1994; Visser et al. 1991), RNAi
technique (Ander sson et al.
2003) or traditional mutational
breeding (Muth et al.
2008). Today, only the Waxy potato
variety Eliane
TM
, developed by radiation-induced muta-
tions (Hovenkamp-Hermelink et al.
1987; Muth et al.
2008), is grown commercially.
Gene silencing is widely used in plant research and
breeding, to study gene functions as well as to develop new
cultivars. Recently, site-directed mutagenesis (SDM) has
been applied for this purpose. Compared to chemical and
physical mutagenesis, which is random with multiple
mutations introduced throughout the genome, the SDM
techniques are designed to be target specific (Quetier
2016). The constructs for SDM, can be stably inserted into
the genome by transformation, or be transiently expresse d
to introduce in vivo mutations. Zinc finger nuclease tech-
nology (ZFN), TAL effector nucleases (TALEN) and
clustered regularly interspaced short palindromic repeat
(CRISPR) and CRISPR-associated protein 9 (Cas9) are the
main SDM techniq ues currently in use (Schiml and Puchta
2016). Development of CRISPR-Cas9 has received much
attention lately due to being a more user-friendly and cost-
efficient technique of producing target-specific constructs
compared to ZFN and TALEN. The CRISPR-Cas9 tech-
nique was first reported as successful in higher plan ts in
2013 (Li et al. 2013; Nekrasov et al. 2013; Shan et al.
2013). The method is based on a short sing le-guide RNA
(sgRNA), with a 20 bp guide sequence complementary to a
target region, a promoter and a sgRNA scaffold, which in
combination with a Cas9 nuclease (Jinek et al.
2012) can
induce mutations in a target region of choice. The resulting
double strand break (DSB) is repaired by the cell’s own
repair mechanism, either through non-homologous end
joining (NHEJ) or homologous recombination (HR) (Britt
1999). NHEJ is error prone and often leads to random-sized
inserts or deletions (indels), which may cause a knockout
of gene function.
Recently, the first studies using TALEN and CRISPR-
Cas9 to induce mutations in potato were published (Butler
et al.
2015, 2016; Clasen et al. 2016; Nicolia et al. 2015;
Sawai et al.
2014; Wang et al. 2015). In a TALEN stud y,
stable transformation was used for studying the disruption
of a sterol side chain reductase 2 (StSSR2). In that study,
two lines with confirmed indels were described in detail of
which one line was concluded to have all four alleles
mutated (Sawai et al.
2014). Further, a transient method for
TALEN expression was developed and induc ed mutations
were verified in the target gene, ACETOLACTATE
SYNTHASE (ALS), in 2 out of 20 regenerated shoots
(Nicolia et al.
2015). In that study, no shoots with multiple
mutated alleles were described. Shortly after, another study
used transiently expressed TALEN to knockout a
VACUOLAR INVERTASE gene (Vlnv), at a mutation
frequency between 2 and 16 % and with two-thirds of the
regenerated plants having multiple alleles mutated (Clasen
et al.
2016). CRISPR-Cas9 has, in two different studies,
been shown to induce mutations in potato by Agrobac-
terium-mediated stable transformation. In the first study, a
gene encoding an Aux/IAA protein, StIAA2, was targeted
in a double haploid potato cultivar (Wang et al.
2015)
while in the second study with this method, the ALS gene
was targeted in both a diploid and tetraploid potato (Butler
et al.
2015). With confirmed stable integration, the studies
yielded mutation rates of 83 and 60 % of regenerated lines,
respectively. Furthermore, lines with mutations in multiple
alleles were detected in both cases. Most recently, TALEN
and CRISPR-Cas9 were stably introduced targeting ALS
and using a geminivirus-mediated guide, to facilitate
designed mutations (Butler et al.
2016).
In this work, we have developed full gene knockouts of
tetraploid potato using transient expression of the CRISPR-
Cas9 system designed to target a GBSS gene. This is sub-
stantiated by genetic analysis of the target gene and phe-
notypical assessment of starch qual ity. Using transient
expression of targeted CRISPR-Cas9 in tetraploid potato,
mutations in all four alleles were obtained without
118 Plant Cell Rep (2017) 36:117–128
123
stable integration of DNA and the laborious subsequent
crossing for re-establishment of a potato genotype with
desired agronomical properties.
Results and discussion
Design of CRISPR-Cas9 constructs
To determine allelic variation in parts of the GBSS gene in
the tetraploid potato variety Kuras, fragments covering
exon 8 and parts of exon 9, as well as adjacent introns
(Fig.
1a) were amplified and sequenced. The different
alleles were found to be highly similar to each other as well
as to a published GBSS gene (Supplementary Fig. S1). Out
of the four alleles, only one of them had a base pair (bp)
variation in exon 8 with an adenine (A) ? guanine
(G) shift and two alleles had an adenine (A) ? guanine
(G) shift at one position in exon 9 (Supplementary Fig. S1).
Two target regions in GBSS exon 8 were selected and
named GT1 and GT2 and one target region in exon 9 was
selected and named GT4. GT1 spanned the region in exon
8 having the allelic variation (Fig.
1a, b). Two different
promoters were chosen for driving the sgRNA expression
of the corresponding guide sequences. U6 promoters have
been the most commonly used type of promoter for driving
expression of sgRNA in plants (Bortesi and Fischer
2015).
Although Arabi dopsis thaliana has been the preferred
origin of U6 promoter, in some studies an endogenous U6
promoter has been tested with a positive influence on the
frequency of induced mutations (Sun et al. 2015). The
Arabidopsis U6 promoter was used in combination with all
three guide sequences, respectively. To study the effect of
an endogenous promoter, a U6 promoter of Solanum
tuberosum origin (StU6) (Guerineau and Waugh
1993) was
combined with the GT4 guide sequence. The use of an
StU6 promoter driving sgRNA in pota to was also reported
in the CRISPR-Cas9 study by Wang et al.
2015, although
this promoter originated from a different StU6 gene (Wang
et al.
2015). Solanum tuberosum U6 small nuclear RNAs
are members of a large gene family with promoter
sequence variation (Guerineau and Waugh
1993). Naked
vector DNA (pE-GT1, pE-GT2, pE-GT4 and pE-
StU6GT4) with respective guide sequence GT1, GT2 and
500
1 500
2 000 2 500 3 000 3 500
GT4 GT2GT1
StGBSS (A23741,1)
5’-
-3’
(a)
GT4 5’- GACAAGAAGATCCCTTTGATTGG - 3’
(b)
pcoCas935SU6 (At or St)
GT (1,2,4)
sgRNA scaffold
pE-GT1, pE-GT2, pE-GT4, pE-StU6GT4
NLS
NLS
(c)
BamHI XhoI EcoRI
Poly-T NOS
pENTR
TM
11
GT1 5’- GATATTAGAATCACATAGGGTGG -3 GT2 5´- TGTTGACAAGGGTGTTGAATTGG -3’
Fig. 1 Design of CRISPR-Cas9 constructs targeting the GBSS gene.
a Illustration of Solanum tuberosum GBSS gene structure (GenBank
accession no. A23741.1). Exons 1–13 are marked with blue arrows,
outer boundaries of fragments amplified for sequence determination
of allelic variation are marked with green arrows and CRISPR-Cas9
target regions GT1, GT2 and GT4 are marked with red triangles.
b Target regions GT1, GT2 and GT4 are in red and PAM-site in
purple. Allelic variation in GT1 marked in black bold text.
c Constructs designed for CRISPR-Cas9-mediated induction of
mutations in StGBSS; from left to right: terminator (poly-T), sgRNA
scaffold, guide sequence (GT1, GT2 or GT4), U6 promoter of either
Arabidopsis thaliana or Solanum tuberosum origin, 35S promoter of
cauliflower mosaic virus origin (CaMV), nuclear localization
sequence (NLS), plant codon-optimized Cas9 gene, NLS and nopaline
synthase terminator (NOS) in vector pENTR
TM
11. The elements
shown in the illustration are not to scale in relation to each other
Plant Cell Rep (2017) 36:117–128 119
123
GT4 and plant codon-optimized Cas9 (pcoCas9) (Li et al.
2013) (Fig. 1c) was purified for protoplast transfection. To
evaluate if Cas9 and sgRNA expressed from separate
vectors had an influence on mutation efficiency, purified
vector DNA containing pcoCas9 and sg-GT1 cassettes
were co-expressed in the study.
Protoplast transfection and regeneration
DNA was transiently expressed, with the aim of inducing
mutations in the GBSS gene. Transient expression was
chosen to facilitate the generation of genotypes with
induced mutations without stable integration of DNA in the
genome. Potato could easily be transformed using
Agrobacterium tumefaciens generating a high number of
transgenic shoots in a short time period (Andersson et al.
2003; Visser et al. 1989). However, the subsequent
outcrossing of inserted T-DNA would disrupt the genetic
context of the modified potato genotype due to its
heterozygosity and tetrasomic inheritance.
Expression of the four sgRNA-pcoCas9 constructs, pE-
GT1, pE-GT2, pE-GT4 and pE-StU6GT4, as well as the
combined expression of GT1-sgRNA and pcoCas9 from
separate vectors was achieved through PEG-mediated pro-
toplast transfection. The protoplast isolation and transfection
method used was essentially as described by Nicolia et al.
2015, where a method to induce mutations in potato via
TALEN was developed (Nicolia et al.
2015). In this study,
the protocol was further developed and adapted for CRISPR-
Cas9. Besides the four different constructs used, transfection
parameters were varied, such as amount of DNA, incubation
time, concentration of protoplasts and concentration of PEG
(Table
1). After transfection, the protoplasts were embedded
in alginate until callus was formed. Approximately, 4 weeks
after transfection, the calluses were released and incubated in
liquid media for an additional 2 to 4 weeks to allow further
development. The enlarged calluses were then transferred to
solid medium for shoot development.
Shoot regeneration was similar in the 12.5 and 25 %
PEG experimental setups, where a first shoot emerged
11–15 weeks after transfection (Supplementary Fig. S2).
After 6 months, 34–43 % of all developed calluses from
the pE-GT2 and pE-GT4 expressions at the 25 % PEG
experimental setup had produced a shoot. Shoot develop-
ment was peaking between months 4 and 7, but shoots
continuously emerged up to one year after transfection
(Supplementary Fig. S2). In two out of three experiments
with the highest PEG concentration of 40 % and with the
lower density of 50,000 protoplasts/mL culture medium,
the shoot regeneration was delayed and depressed (Sup-
plementary Fig. S2). High density of protoplasts or use of
feeder cells is generally believed to stimulate cell division,
by releasing stimulating substances into the medium (Xu
and Xue
1999). However, this needs to be balanced since a
high density of viable protoplasts will hamper distinct
separation of individually developing calluses.
While the majority of the viable shoots developed into
plantlets undistinguishable from the parental phenotype,
some of the regenerated shoots failed in elongation and
hence development into plantlets. Based on all regenera ted
shoots in this study, an average of 25 % were stunted in
growth and eventually died. In previous studies where
silencing of GBSS in potato has been the objective, no
impact on plant growth has been reported (Hofvander
2004; Kuipers et al. 1994). Therefore, it is not likely that
the cause for a stunted growth is a complete knockout of
Table 1 Protoplast transfection experimental setup and results thereof
Guide
sequence
Promoter PEG
(%)
Transfection
time (min)
DNA
(lg)
Protoplasts
(/mL)
Number regenerated mutated
lines (number analysed)
Frequency mutated
lines (%)
GT1 A.t.U6 12.5 3 10 80 000 8 (161) 5.0
GT1 A.t.U6 25 3 5 80 000 7 (85) 8.2
GT1 ? Cas9
a
A.t.U6 25 3 5 ? 5 80 000 4 (130) 3.1
GT1 A.t.U6 40 30 15 50 000 5 (43) 11.6
GT2 A.t.U6 12.5 3 10 80 000 12 (286) 4.2
GT2 A.tU6 25 3 5 80 000 5 (108) 4.6
GT2 A.t.U6 40 30 15 50 000 2 (79) 2.5
GT4 S.t.U6 12.5 3 10 80 000 18 (188) 9.6
GT4 A.tU6 25 3 5 80 000 21 (407) 5.2
GT4 S.tU6 25 3 5 80 000 44 (426) 10.3
GT4 S.t.U6 40 30 15 50 000 3 (138) 2.2
Mutation frequency (%) calculated for each experiment based on mutated lines detected using high-resolution fragment analysis by the total
number of regenerated lines analysed
a
GT1 and Cas9 expressed from separate vectors
120 Plant Cell Rep (2017) 36:117–128
123
the GBSS enzyme function. Plants regenerated from potato
leaf protoplasts can lead to somaclonal variation, which
might be a plausible explanation for the observed negative
effect on plant development in this study (Larkin and
Scowcroft
1981; Shepard et al. 1980). Although a sub-
stantial share of shoots was discarded due to failed devel-
opment, the regeneration rate of the method ensures a vast
number of shoots to select from and where the subsequent
selection process does not differ from other plant breeding
methods.
High-resolution fragm ent analysis (HRFA)
for screening of mutations
Efficient screening methods for induced mutations are
crucial to analyse a large number of genome-edited
regenerated plants. Here, we used a high-resolution frag-
ment analysis, HRFA, with 96-format DNA extraction,
PCR amplification and capi llary electrophoresis. A leaf
sample of each regenerated shoot was subjected to DNA
isolation followed by PCR amplification covering the
respective target sites using a fluorescently labelled for-
ward primer. Capillary electrophoresis was applied to the
amplified reactions for separation with high resolutio n and
to fluorescently detect PCR products differing in size. The
method was highly sensitive and indels as small as 1 bp
were detected (Fig.
2a). In addition, multiple mutated
alleles, with indels of differing size, were distinguished
(Fig.
2b–d). It was not possible to detect mutations yield-
ing a nucleotide substitution with the method. However,
since the majority of the nucleotide exchanges only result
in an amino acid substitution in the corresponding prot ein,
these mutations were of less interest since a complete
knockout of enzyme activity was intended. A limitation of
the method was that alleles having same sized indels did
co-elute and could not be resolved. A corresponding wild-
type fragment labelled with a different fluorescent dye was
included in the analysis (Fig. 2f). Loss of wild-type size
fragment signal in the regenerated lines was accordingly a
confirmation that all four alleles had receive d mutations.
The applied HRFA method can easily be adapted to mul-
tiplex analysis of samples with more than one gene targeted
by designing PCR amplicons differing in size and labelled
with different fluorophores.
A range of different techniques are in use for screening
plants for induced mutations. Detection of indels using
PCR and capillary electrophoresis is used here and else-
where (Ramle e et al.
2015; Yang et al. 2015); in compar-
ison to the more generally used DNA sequencing method
(Bortesi and Fischer
2015) it has a much shorter time span
from sampling to identification of mutants. Furthermore,
screening polyploid plants by sequencing, and verifying
coverage of all alleles might lead to analysis of many
replicates. A 454 sequencing approach or similar might
give a more reliable result, but is still costly in comparison
to HRFA. Other commonly used methods for detection of
mutation are high-resolution melt analysis (HRM) (Wit-
twer et al.
2003) as well as cleaved amplified polymorphic
sequences (CAPS) (Konieczny and Ausubel
1993), a
method based on the loss of function of a restriction site
located in the predicted cleave site in the target region.
CAPS are efficient and easy to use, but limit the target
guide design and the number of target sites that can be
identified in a gene. The use of a heteroduplex mobility
assay is yet another alternative, but the allelic specificity
and sensitivity of the method is lower, having a published
resolution limit of 3 bp indels (Delwart et al.
1993; Ito
et al.
2015).
Identification of mutated lines
To validate CRISPR-Cas9 functionality when transiently
expressed in protoplasts of a tetraploid potato cultivar and
to estimate the mutation efficiency of eleven different
experimental setups, a total of 2051 regenerated shoots
were analysed by HRFA. As shown in Table
1, directed
mutations were induced to a frequency of 2.2–11.6 %. No
striking difference could be found between the three GBSS
sites targeted, GT1, GT2 and GT4, yielding mutation fre-
quencies of 8.2, 4.6 and 5.2 %, respectively, in the 25 %
PEG experimental setup. Hence, the CRISPR-Cas9 tech-
nique was not very sensitive to target region and corre-
sponding guide sequence chosen in this study. Similar
findings have previously been published; for example, in a
study where CRISPR-Cas9 was transiently expressed for
targeted genome editing in rice (Xie and Yang
2013).
Even though the mutation frequency using pE-GT1
increased with a higher PEG concentration, DNA amount
and incubation time used (5.0, 8.2 and 11.6 % at 12.5, 25
and 40 % PEG experimental setups, respectively, Table
1),
the tendency was not consistent among the different guide
sequences studied. pE-GT2 expressed at 12.5, 25 and 40 %
PEG experimental conditions, resulted in mutation fre-
quencies of 4.2, 4.6 and 2.5 %, respectively. In the
experiment where Cas9 and the GT1-sgRNA were
expressed from separate vectors (DNA ratio 1:1 (w/w)), a
frequency of 3.1 % was found which was somewhat lower
(2.6-fold) than when expressed from a single construct at
the same transfection conditions (Table
1).
Previous studies have shown that the choice of promoter
driving the guide sequence could have an impact on the
mutation frequency (Sun et al.
2015). It has also been
shown that the strength of the promoter can have an effect
on target specificity, and a high conce ntration of sgRNA-
Cas9 transcripts can increase off-target mutations (Hsu
et al.
2013; Pattanayak et al. 2013). The U6 promoters used
Plant Cell Rep (2017) 36:117–128 121
123