© 2000 Nature America Inc. • http://biotech.nature.

com

RESEARCH ARTICLES

Production of very-high-amylose potato

starch by inhibition of SBE A and B
Gerhard P. Schwall1, Richard Safford1, Roger J. Westcott1, Roger Jeffcoat2, Akash Tayal2, Yong-Cheng Shi2,
Michael J. Gidley1, and Stephen A. Jobling1*
1Unilever Research Colworth, Colworth House, Sharnbrook, Bedford MK44 1LQ, UK. 2National Starch and Chemical Company, 10 Finderne Avenue, Bridgewater,
NJ 08807, USA. *Corresponding author (Steve.Jobling@Unilever.com).

Received 12 November 1999; accepted 8 March 2000

High-amylose starch is in great demand by the starch industry for its unique functional properties.
However, very few high-amylose crop varieties are commercially available. In this paper we describe the
generation of very-high-amylose potato starch by genetic modification. We achieved this by simultane-
ously inhibiting two isoforms of starch branching enzyme to below 1% of the wild-type activities. Starch
granule morphology and composition were noticeably altered. Normal, high-molecular-weight amy-
© 2000 Nature America Inc. • http://biotech.nature.com

lopectin was absent, whereas the amylose content was increased to levels comparable to the highest
commercially available maize starches. In addition, the phosphorus content of the starch was increased
more than fivefold. This unique starch, with its high amylose, low amylopectin, and high phosphorus lev-
els, offers novel properties for food and industrial applications.

Keywords: biotechnology, starch structure, amylopectin, branching enzyme

Starch is the major storage carbohydrate in plants and one of the most
important plant raw materials for both food and industrial applica-
tions. Approximately 70% of the European and US starch production
is used for industrial purposes, whereas about 30% is used as native
starch for direct human and animal consumption1. The range of starch
applications is heavily influenced by the ratio of its two major compo-
nents, essentially linear amylose and branched amylopectin, the length
and distribution of the branch chains and phosphate levels, and non-
starch components like lipids and proteins. The ratio of amylose to
amylopectin has the greatest influence on the physicochemical proper-
ties of the starch, and for many applications it is desirable to have a
pure or enriched fraction of either amylopectin or amylose. In most
crops starch contains 20–30% amylose and 70–80% amylopectin.
In the past, the search for starches with different ratios of amylose
to amylopectin was mainly achieved by extensive breeding programs
and the characterization of mutant varieties. Over the past decade,
however, attempts have been made to alter starch properties by chang-
ing levels of starch synthases (SS) and starch branching enzymes
(SBE) through genetic modification2. In potato, one granule-bound
starch synthase (GBSS I), three isoforms of soluble starch synthase,
and two isoforms of starch branching enzyme are known3,4. The suc-
cessful antisense inhibition of GBSS I led to amylose-free potato
starch2. Attempts to increase the amylose level in potatoes by inhibit-
ing amylopectin synthesis have however been much less successful.
The inhibition of SS II and SS III or the main starch branching
enzyme in tubers (SBE B; for nomenclature see refs 5,6) resulted in
novel starch characteristics but not in an increased amylose level7–9.
Recently, the antisense inhibition of a minor form of starch branching
enzyme, SBE A, resulted in a moderate increase of the apparent amy-
lose level up to 38%4. Here we report that the simultaneous inhibition
of SBE A and B results in very-high-amylose potato starch containing
insignificant levels of highly branched amylopectin.
Results
Generation of potato plants with strongly reduced levels of starch
branching enzymes A and B. Potato plants were produced by
retransformation of antisense SBE B lines9 with a 1.2 kb partial-
length SBE A antisense construct4. Plants with reduced SBE activi-
ties in leaves and tubers were regrown for two generations in the
glasshouse. A third generation was grown in the field to determine
whether the reduction of SBE activities was stable under field trial
conditions. A western blot analysis of three selected field trial lines
shows that the protein levels in tubers for SBE A and B were effec-
tively reduced to below the detection level (Fig. 1). Tuber SBE activi-
ty was reduced to less than 0.9% and leaf activity to less than 0.7% of
control plants (Table 1). During growth the transgenic lines
appeared phenotypically normal, but the reduction of SBE activity
throughout the plant had an effect on tuber yield and starch content,
which were both about half that of the wild-type (wt) control lines
(data not shown). Tuber morphology was also altered since tubers
from the high-amylose lines were more elongated than controls and
had a tendency to bud off additional tubers; a similar phenotype has
been observed before10.
SBE A + B antisense inhibition leads to high-amylose potato
starch with increased phosphate content. In total, the amylose con-
tent of 71 primary transgenic lines was determined. Amylose levels
in 19 lines were 40% or higher, as determined by a colorimetric
iodine-binding assay, and all of these showed very low levels of SBE
A by western blot analysis (data not shown). Several lines had amy-
lose contents above 60% and in the best line up to 75% (Table 1). An
alternative potentiometric amylose determination gave values of up
to 89%. This is the first description of high-amylose potato starch
that is comparable to the amylose levels of high-amylose maize
starch11 and is a considerable improvement over increased amylose
levels reported previously4,10.
Potato starch shows a naturally high degree of phosphoryla-
tion compared to starches from other crops. Changes in expres-
sion of starch biosynthetic enzymes have been shown to alter the
phosphate content of starch, although the mechanism is
unknown2,4,9. Therefore, we determined the phosphorus content
of the starch. Wild-type starch has a phosphorus content of
approximately 500 µg g-1, whereas starch from high-amylose lines

NATURE BIOTECHNOLOGY VOL 18 MAY 2000 http://biotech.nature.com 551

 © 2000 Nature America Inc. • http://biotech.nature.com

RESEARCH ARTICLES

A B
A

C D
Figure 1. Western blot analysis of SBE A and B in tubers. Samples
(20 µg) of soluble tuber extract from three antisense SBE A + B
lines (201, 202, 208) and one wild-type (wt) line were analyzed by
western blotting, using antiserum specific for (A) potato SBE B and
(B) SBE A. The signal at 97 kDa (*) is nonspecific and unrelated to
branching enzyme.

had levels up to 3,000 µg g-1, a sixfold increase (Table 1).

Preliminary results from butanol fractionation experiments of
© 2000 Nature America Inc. • http://biotech.nature.com

starch from line 208 show that the phosphorus is associated with
the branched-amylopectin fraction of the starch. Figure 2. Light microscopy of starch granules. (A, B) Polarized-light
High-amylose starches have an extremely altered starch struc- view of wild-type granules (A) and granules from line 208 (B). (C, D)
Heat-treated (95°C, 5 min), iodine-stained starch of (C) a wild-type
ture and a complete lack of normal amylopectin. We investigated
control and (D) line 208. Scale bar represents 50 µm.
the effect of the simultaneous inhibition of SBE A and B on the level
of granule structure as well as starch structure. The granules from
wild-type starch were oval in shape with smooth surfaces and absent in line 298 (Table 2). These characteristics are very similar to a
exhibited birefringence under polarized light, an indicator of crys- recently described very-high-amylose maize starch, which is better
talline order (Fig. 2A). The majority of starch granules from high- described as low-amylopectin starch (LAPS)11.
amylose lines showed less birefringence, indicating a reduction of After debranching with isoamylase, wild-type starch was sepa-
crystalline starch structure, and many also had irregular surfaces rated into three well-resolved fractions: an amylose peak (B1) and
with deep fissures in the center of the granules (Fig. 2B). The two amylopectin-derived peaks containing medium-length and
change in granular structure and composition also led to dramati- short glucan chains (peaks B2 and B3, respectively; Fig. 3B). These
cally altered swelling properties. Wild-type starch granules started fractions represent 23%, 31%, and 46% of the total, respectively
to swell at about 70°C and formed a viscous paste after heating to (Table 2). Debranching of starch from high-amylose lines revealed a
95°C (Fig. 2C). Starch granules from high-amylose lines showed very altered starch structure. The amylose levels (peak B1) had
increased pasting temperatures according to their amylose level increased to over 56%. Short-chain branches (peak B3) were virtu-
(data not shown), and when the amylose content was greater than ally absent, whereas medium-chain branches (peak B2) had
55% (as determined by colorimetric iodine binding), granule increased in both quantity (44%) and size such that this fraction
swelling was inhibited completely (Fig. 2D). was not well resolved from the amylose fraction (B1). These data
The molecular structure of the starch was examined by gel per- suggest that in high-amylose lines the synthesis of amylopectin with
meation chromatography (GPC) before and after debranching (Fig. short branches is severely suppressed whereas the true amylose con-
3). All high-amylose starches were examined and showed very simi- tent is significantly increased.
lar changes, but only the results from the most modified starch (line The lack of short chains in the high-amylose starch was con-
298) are shown for clarity. Normal wild-type starch was separated firmed by high-performance anion exchange chromatography
into an amylopectin peak with an apparent molecular weight of 107 (HPAEC; Fig. 4). This method allows the separation of glucan chains
Da (peak A1) and a broader peak (A2) of a lower molecular weight of debranched starch samples according to their degree of polymer-
(106 Da), which contains mostly amylose as well as a small amount ization (dp). To illustrate changes, the areas of the individual glucan
of low-molecular-weight amylopectin (Fig. 3A)4,11. In contrast, peaks were integrated and their relative peak area was expressed as a
starch from high-amylose lines eluted as a single broad peak at an percentage of the total area of all peaks (Fig. 4A) or as a difference
apparent molecular weight of 3 × 105 Da (peak
A2). Since the starch is not 100% amylose, this Table 1. Starch branching enzyme activities and tuber starch composition
peak must contain branched as well as
unbranched glucans; however, on-line intrin- Line SBE activitya Amylose contenta Phosphorus
sic viscosity measurements of the A2 peak of (U g-1 fresh weight) (% of total starch) (µg g-1 starch)
line 298 were characteristic of amylose (and Tuber Leaf Colorimetric Potentiometric
not amylopectin), indicating that the glucan
chains have long linear backbones and are not wt control 36.38 ± 1.07 10.05 ± 0.03 27.75 ± 1.9 25.59 498
densely branched as in amylopectin. The rela- 201 0.32 ± 0.13 0.07 ± 0.07 59.37 ± 5.3 77.45 2,600
tive peak areas were calculated by peak inte- 202 0.05 ± 0.02 0.00 ± 0.03 62.01 ± 1.4 80.84 3,000
gration in order to quantify the individual 208 0.13 ± 0.04 0.01 ± 0.02 64.58 ± 1.4 77.04 2,600
292 0.04 ± 0.04 0.09 ± 0.06 74.76 ± 1.2 89.14 2,300
starch fractions (Table 2). In high-amylose 298 0.08 ± 0.01 0.02 ± 0.03 72.39 ± 1.2 87.24 2,400
starches the amylopectin fraction (A1) was
approximately 5% of the total and completely aWhere given, mean and standard deviations are calculated from four samples.

552 NATURE BIOTECHNOLOGY VOL 18 MAY 2000 http://biotech.nature.com

 © 2000 Nature America Inc. • http://biotech.nature.com
A A
B B
Figure 3. Gel chromatography of starches. Representative gel
permeation chromatogram of native (A) and debranched (B) potato
starch from a wild-type (wt) control and high-amylose line 298. The
© 2000 Nature America Inc. • http://biotech.nature.com
peaks referred to in the text are marked. The molecular weight is
based on dextran standards and is indicative only.
from control (Fig. 4B). It should be noted, however, that for a given
amount of starch, the total signal strength of the high-amylose
starches was less than 10% of the control, indicating the substantial
decrease in branched chains of all lengths in these starches. The
chain length distribution was also altered. The peak chain length had
increased from dp 13–14 in the control to dp 18–19 in the high-amy-
lose starches (Fig. 4A), and large decreases in short chains (dp 6–15)
and increases in longer chain branches (dp > 20) were observed (Fig.
4B). Taken together, these data show a lack of conventional amy-
lopectin in the high-amylose starches and are most consistent with a
population of amylose molecules, a significant minority of which
have a small number of medium-length to long branches.
Discussion
To generate potato starch with a very high amylose content, anti-
sense SBE B lines were retransformed with an antisense SBE A con-
struct. Severe reductions in both SBE A and B, as demonstrated by
enzyme assays and western blot analysis, were observed. Starch from
these lines had an apparent amylose content of 60–89% as deter-
mined by different methods based on iodine binding. It is well
known that assays based on iodine binding can overestimate the
amylose content because of interference from long amylopectin
chains. The discrepancy between different methods was reported
elsewhere11,12. Gel permeation chromatography of debranched
starch samples was therefore used to determine a true amylose con-
tent (i.e., essentially linear glucan chains) of up to 56%. In addition,
normal high-molecular-weight amylopectin was reduced to less
Table 2. Quantification of the starch fractions separated by GPCa
Peak Glucan % Peak area
fraction wt control Line 298
Potato starch A1 AP-H 65.0 0
A2 AM + AP-L 35.0 100
Debranched B1 AM 22.9 56.5
potato starch B2 MC 31.2 43.5
B3 SC 45.7 0 currently not well understood, but the described lines could help in
aQuantification of the starch fractions from the GPC profiles of Figure 3 is
based on relative peak areas. The glucan fractions separated in these peaks
are indicated: AP-H, high-molecular-weight amylopectin; AP-L, low-molecular-
weight amylopectin; AM, amylose; MC, medium chains; SC, short chains.
NATURE BIOTECHNOLOGY VOL 18 MAY 2000 http://biotech.nature.com
RESEARCH ARTICLES
Figure 4. Chain length distribution of starches. (A) Dionex HPAEC
elution profile of debranched starch samples expressed as relative
peak area in percentage of the total area, plotted against the degree of
polymerization (in glucose units). (B) Difference between the relative
peak areas. Differences were calculated by subtracting the relative
peak areas of the wild-type (wt) control from the indicated lines.
than 5% and was completely absent in the most modified starch.
This was also reflected as a strong decrease of short chains compared
to wild-type starch. This potato starch is comparable to the highest
amylose maize starch (LAPS)11, which has an apparent amylose level
of 90% (potentiometric iodine binding) and amylose levels and
short-chain glucan levels of 53% and 4.5%, respectively (as deter-
mined by GPC of debranched starch).
Previously the antisense inhibition of potato SBE B and SBE A as
single constructs was reported. The reduction of SBE B (the main
tuber branching enzyme) had no effect on the amylose content9, but
the inhibition of a minor branching enzyme (SBE A) led to an
increased apparent amylose content of up to 38%, although the
amylose content as determined by GPC of debranched starch was
not significantly altered4. The SBE A isoform is clearly required to
synthesize amylopectin with a normal branching structure, but the
role of the SBE B isoform is less clear. It would appear that SBE A can
complement the activity of SBE B but that the reverse is only partial-
ly true, implying that the specificities (substrate preference, chain
length transfer, etc.) of the enzymes differ. The two SBEs do interact,
however, as inhibition of both SBE A and B to below 1% of the wild-
type activity seems to be necessary to significantly increase the
apparent and true amylose content in potato to the levels described
above. More work, such as the biochemical characterization of
recombinant enzymes or the inhibition of SBE B in a low-SBE A line,
is required to clarify the role of each isoform.
In contrast to potato, a deficiency in only SBE A seems to be suf-
ficient to significantly increase the amylose level in ae and r mutants
of maize and pea. However, in the ae-derived low-amylopectin
(LAPS) mutant in maize it was shown that, in addition to an almost
complete reduction of SBE A, the SBE B isoform was also decreased
to one third of its wild-type level13.
In addition to altered amylose:amylopectin ratios of the starch,
we observed a five- to sixfold increase in the phosphorus content,
which is associated with the branched starch fraction. This effect was
already shown but to a lesser extent in lines with inhibition of a sin-
gle SBE4,9. The mechanism leading to this increase in phosphorus is
a better understanding of the mechanisms involved.
To our knowledge, this publication is the first report of a true
high-amylose starch generated by genetic modification. This
unique starch with its very high amylose, low amylopectin, and
553
 © 2000 Nature America Inc. • http://biotech.nature.com
RESEARCH ARTICLES
high phosphorus levels offers novel properties for food and indus-
trial applications.
Experimental protocol
Generation of transgenic potato lines. Transgenic potato lines were generated
by retransformation of antisense SBE I (SBE B) lines 12, 15, 17, and 18
described9 with a 1.2 kb antisense SBE A construct4. Note that the SBE A/B
nomenclature is used in this paper rather than SBE I/II to be consistent with a
previous publication4. Both antisense constructs were under the control of the
double 35S promoter from cauliflower mosaic virus. Transformation with
Agrobacterium tumefaciens was essentially carried out as described by Safford
and coworkers9. A field trial was carried out at Plant Breeding International
Cambridge under licence 96/R2/7. The tubers were planted on July 11 and har-
vested on October 2. The lines 201, 202, and 208 were derived from antisense
SBE B line 15, whereas lines 292 and 298 came from antisense SBE B line 12.
Enzyme assays and western blot analysis. Enzyme assays and western blot
analysis were carried out as described4. For the enzyme assays a phosphorylase a
stimulation assay was used, and one unit of SBE activity is defined as the incor-
poration of 1 µmol of glucose into a methanol-insoluble polymer per minute.
Starch analysis. Starch extraction and the determination of phosphorus
levels were described by Safford and colleagues9. Starch analysis by GPC and
Dionex HPAEC was performed as described11. Intrinsic viscosity measure-
© 2000 Nature America Inc. • http://biotech.nature.com
ments of GPC eluants were made using a Model H 502 Viscometer (Viscotek,
Houston, TX). The apparent amylose content was determined by a colori-
metric iodine-binding assay14 and by potentiometric iodine titration11. The
true amylose content was determined by area integration of the amylose peak
in the GPC profile of debranched starch.
Acknowledgments
The authors wish to thank Chris Sidebottom and Martine Debet for their contri-
butions in the early phase of the project, Tina Sanders, Alice Belton, Alison
Burrows, and Bob Cowper for their technical assistance.
554
1. Lillford, P.J. & Morrison, A. Structure/function relationship of starches in food.
In Starch structure and functionality. (eds Frazier, P.J., Richmond, P. &
Donald, A.M.) 1–8 (The Royal Society of Chemistry, Cambridge, UK; 1997).
2. Heyer, A.G., Lloyd, J.R. & Kossmann, J. Production of modified polymeric
carbohydrates. Curr. Opin. Biotechnol. 10, 169–174 (1999).
3. Smith, A.M. Making starch. Curr. Opin. Plant Biol. 2, 223–229 (1999).
4. Jobling, S.A. et al. A minor form of starch branching enzyme in potato
(Solanum tuberosum L.) tubers has a major effect on starch structure: cloning
and characterisation of multiple forms of SBE A. Plant J. 18, 163–171 (1999).
5. Burton, R.A. et al. Starch branching enzymes belonging to distinct enzyme
families are differentially expressed during pea embryo development. Plant J.
7, 3–15 (1995).
6. Commission on Plant Gene Nomenclature. Nomenclature of sequenced plant
genes. Plant Mol. Biol. Rep. 12, S1–S109 (1994).
7. Edwards, A. et al. A combined reduction in activity of starch synthases II and
III of potato has novel effects on the starch of tubers. Plant J. 17, 251–261
(1999).
8. Lloyd, J.R., Landschütze, V. & Kossmann, J. Simultaneous antisense inhibi-
tion of two starch synthase isoforms in potato tubers leads to accumulation
of grossly modified amylopectin. Biochem. J. 338, 515–521 (1999).
9. Safford, R. et al. Consequences of antisense RNA inhibition of starch branch-
ing enzyme activity on properties of potato starch. Carbohydr. Polym. 35,
155–168 (1998).
10. Tjaden, J., Möhlmann, T., Kampfenkel, K., Henrichs, G. & Neuhaus, H.E.
Altered plastidic ATP/ADP-transporter activity influences potato (Solanum
tuberosum L.) tuber morphology, yield and composition of tuber starch. Plant
J. 16, 531–540 (1998).
11. Shi, Y.-C., Capitani, T., Trzasko, P. & Jeffcoat, R. Molecular structure of a low-
amylopectin starch and other high-amylose maize starches. J. Cereal Sci. 27,
289–299 (1998).
12. Cheetham, N.W.H. & Tao, L. The effects of amylose content on the molecular
size of amylose, and on the distribution of amylopectin chain length in maize
starches. Carbohydr. Polym. 33, 251–261 (1998).
13. Sidebottom, C., Kirkland, M., Strongitharm, B. & Jeffcoat, R.
Characterization of the difference of starch branching enzyme activities in
normal and low-amylopectin maize during kernel development. J. Cereal Sci.
27, 279–287 (1998).
14. Morrison, W.R. & Laignelet, B. An improved colourimetric procedure for
determining apparent total amylose in cereal and other starches. J. Cereal
Sci. 1, 9–20 (1983).
NATURE BIOTECHNOLOGY VOL 18 MAY 2000 http://biotech.nature.com