Structure function relationships of transgenic starches with engineered phosphate substitution and starch branching

International Journal of Biological Macromolecules 36 (2005) 159–168

Structure function relationships of transgenic starches with engineeredphosphate substitution and starch branching

Andreas Blennowa,∗, Bente Wischmannb, Karen Houborga, Tina Ahmtc, Kirsten Jørgensena,d,Søren Balling Engelsene, Ole Bandsholmf, Peter Poulsend,1

a Plant Biochemistry Laboratory, Department Plant Biology, Center for Plant Molecular Physiology (PlaCe), Royal Veterinary andAgricultural University, Thorvaldsensvej 40, DK-1871 Frederiksberg C, Denmark

b FBEG, BioCentrum-DTU, Technical University of Denmark, DK-2800 Lyngby, Denmarkc Biotechnological Institute, Holbergsvej 10, DK-6000 Kolding, Denmark

d Danisco Biotechnology, Langebrogade 1, P.O. Box 17, DK 1001 Copenhagen K, Denmarke Centre for Advanced Food Studies, Food Technology, Institute of Food Science, The Royal Veterinary and

Agricultural University, Rolighedsvej 30, DK-1958 Frederiksberg C, Denmarkf KMC, Herningvej 60, DK-7330 Brande, Denmark

Received 10 January 2005; received in revised form 20 May 2005; accepted 21 May 2005Available online 15 July 2005

A

hing enzyme( structurala phate weres n (SEC-RI)i al scanningc ed starchesw elasticity andv led a highc racteristicsw rimarily thep©

K

1

c

hdm

2

dularali-eredchesver,e spe-y oreulesculeshe

0d

bstract

Potato tuber starch was genetically engineered in the plant by the simultaneous antisense suppression of the starch brancSBE) I and II isoforms. Starch prepared from 12 independent lines and three control lines were characterised with respect tond physical properties. The lengths of the amylopectin unit chains, the concentrations of amylose and monoesterified phosignificantly increased in the transgenically engineered starches. Size exclusion chromatography with refractive index detectiondicated a minor decrease in apparent molecular size of the amylose and the less branched amylopectin fractions. Differentialorimetry (DSC) revealed significantly higher peak temperatures for gelatinisation and retrogradation of the genetically engineerhereas the enthalpies of gelatinisation were lower. Aqueous gels prepared from the transgenic starches showed increased geliscosity. Principle component analysis (PCA) of the data set discriminated the control lines from the transgenic lines and reveaorrelation between phosphate concentration and amylopectin unit chain length. The PCA also indicated that the rheological chaere primarily influenced by the amylose concentration. The phosphate and the amylopectin unit chain lengths had influenced pasting and rheological properties of the starch gels.2005 Published by Elsevier B.V.

eywords:Starch branching enzyme (SBE); Genetic transformation; Antisense suppression; Potato starch; Starch structure; DSC; Rheology

. Introduction

Starch is a biomacromolecule used in a wide array of appli-ations, for example, as thickener and stabiliser, for control-

Abbreviations: DCS, differential scanning calorimetry; HPAEC/PAD,igh performance anionic exchange chromatography/pulsed amperometricetection; SBE, starch branching enzyme; SEC-RI, size exclusion chro-atography/refractive index detection∗ Corresponding author. Tel.: +45 35283334; fax: +45 35283333.E-mail address:[email protected] (A. Blennow).

1 Present address: Carlsberg Research Centre, 10 Gl. Carlsbergvej, DK-400 Valby, Denmark.

ling consistency and as texture enhancer[1]. Starches isolatefrom different sources are known to have different molecstructures resulting in a wide range of different functionties. However, there is a rising demand for new and altfunctional properties of starch. As a result modified starare widely used in foods and non-food products. Howenative starches can be selected or engineered to achievcific valuable functionalities such as freeze–thaw stabilitdecreased solubility[2]. Differences in functionality can battributed to the morphology and size of the starch granbut also to the assembly and structure of the starch molewithin the starch granules[3]. Much focus has been on t

141-8130/$ – see front matter © 2005 Published by Elsevier B.V.oi:10.1016/j.ijbiomac.2005.05.006

160 A. Blennow et al. / International Journal of Biological Macromolecules 36 (2005) 159–168

ratio between amylose and amylopectin within the starchgranules since this variable has profound effect on starchpaste rheology as shown for, e.g. amylose-free potato starch[4] and high-amylose starch[5]. Other structural objects suchas starch molecular weight distribution[6], and the degree ofamylopectin branching are also known to have specific andimportant impacts on the functional properties of starches[7].Storage starches prepared from tuberous plants are unique byhaving covalently bound phosphate esterified to a relativelylarge proportion (0.5%) of the glucose residues in the amy-lopectin. A number of studies have now shown that manythermal and rheological properties of potato starch are relatedto the degree of phosphorylation[8–11]. However, the preciseeffects of starch phosphorylation are not clearly understood.One plausible explanation is the covariation between thedegree of phosphorylation and the unit chain length distri-bution of amylopectin found for many potato starch systems[12–14] resulting in physico-chemical changes that cannotbe explained from single structural variables.

Alteration of specific structural motifs of the starchbiopolymers using a transgenic approach has been proven tobe superior in order to achieve extreme and/or specific func-tional alterations[15]. This principle has been demonstratedfor starches obtained from potato plants with repressed starchsynthase activity[16,17], repressed starch branching enzyme(isoforms SBEI and SBEII[14], also termed SBE B and SBEA gens yme(b rchesp s thep ed ints hos-p SBEIa syn-t s thatc genics rmr truc-t SC)r icat-i ainsi ses archc nd a6 -o witha ntra-t ation[ nulet andS calm romt ally,p ylose

content, the longer amylopectin unit chains and the increasedstarch phosphate concentration.

A thorough investigation of the structural and physico-chemical properties of starches with long-chain and highphosphate amylopectins remains to be carried out. One com-plication with this biopolymer system is the pleiotropiceffects i.e. side effects taking place that are indirectly linkedto the target mutation. These side effects tend to generatecorrelation in the data, e.g. between phosphate concentrationand amylopectin unit chain length. Hence it is a challengeto disclose the relationships between structural and physicalproperties for phosphorylated starches.

In this study starches prepared from 12 independent potatolines with suppressed SBE activity and three control lines,all grown under equal and controlled conditions, were struc-turally and physically characterised. A multivariate (chemo-metric) data approach was undertaken to enable the disclosureof the most important variables, with specific focus on findingstructure functionality relationships related to starch branch-ing and phosphate substitution.

2. Materials and methods

2.1. The generation of transgenic potato lines

cp e byt s-m -s ing theAw linesB ithp gs orien-t nator[ con-f 943o oml chinge

hootw ousec per-a genicl genee tubersw ers asd

2

int lantsw t

, respectively), ectopically expressed bacterial glycoynthase[12] and repressed starch phosphorylating enzglucan water dikinase, GWD)[11,18,19]. A main differenceetween starch in SBE deficient potato tubers and staroduced in seeds by SBE deficient maize mutants ihosphate concentration which is dramatically increas

he genetically engineered potato tuber starch[14]. Potatotarch with moderately altered unit chain lengths and phate concentrations generated by partly suppressednd SBEII activity and ectopically expressed glycogen

hase produce aqueous gels with rheological propertiean be related to the structural alterations of the transtarches[20]. Specific suppression of the SBEI isofoesulted in the generation of starch with undetectable sural differences but differential scanning calorimetry (Devealed a higher gelatinization temperature in water indng the presence of slightly elongated amylopectin unit chn these starch molecules[21]. On the contrary, antisenuppression of the SBEII isoform resulted in tuber stontaining three-fold higher phosphate concentration a–8% increased amylose concentration[22]. The simultaneus suppression of SBEI and SBEII yielded a starchlmost a complete reduction in the amylopectin conce

ion and a six-fold increase in the phosphate concentr14]. Recently, the assembly structures of starch graypes including a potato starch with suppressed SBEIBEII activity were investigated by electron and confoicroscopy[23]. The potato starch granules prepared f

he lines with suppressed SBE could not assemble normossibly due to the combined effects of the increased am

Plant transformation and growth conditions. Transgeniotato lines with reduced SBEII expression were mad

ransformation of the Dianella wild type (wt) line with plaid pSS24[24] which expresses a 1495 bp 5′ end SBEII anti

ense fragment and has a mannose selection marker usgrobacteriummediated gene transfer[24]. Two of the linesith documented changes in the starch composition (2 and B4[25]), were chosen for re-transformation wBEA3 containing a 692 bp 5′ fragment of the SBE codinequence in antisense orientation ligated in antisenseation between the patatin promoter and the 35S termi24]. Selection was achieved by kanamycin resistanceerred by the construct. The resulting transformants (Hriginating from line B2 and the H944 series originating fr

ine B4), generated tubers with suppressed starch brannzyme activity.

Nodal segments of the transgenically transformed sere transferred to soil, and grown in normal green honditions with supplementing light and controlled temture and moisture. Tubers from the generated trans

ines have simultaneously suppressed SBEI and SBEIIxpression and showed decreased SBE activity. Potatoere harvested and starch prepared from the fresh tubescribed[26].

.2. Expression analysis

The expression of SBEI, SBEII and GWD proteinuber extracts of the transgenic and in the control pere monitored by Western analysis[27] using rabbi

A. Blennow et al. / International Journal of Biological Macromolecules 36 (2005) 159–168 161

anti-SBEI, anti-SBEII and anti-GWD IgG (1:1000 dilution),respectively, and horseradish peroxidase (HRP)-conjugatedswine anti-rabbit IgG (1:5000 dilution; DAKO P0217)using a HRP substrate (BioRad). Chemiluminescence wasmonitored using the UVP AutoChemi System (UVP, Upland,CA). For each analysis 24�l potato tuber extract was used.

2.3. Enzyme activity analysis

The activity (Vmax) of soluble SBE using the iodinemethod and expressed as the decrease of absorbance (�abs) at720 nm of the amylose–iodine complex per minute was quan-tified in extracts from control and transgenic potato tubers[28].

2.4. Preparation of starch from potato tubers

Starch was prepared from freshly harvested and fullymature potato tubers as described[26]. All starches werethoroughly washed in de-ionised water.

2.5. Starch analyses

Amylopectin unit chain length distribution analysis.Enzymic debranching of starch, separation of the unit chainsand distribution analysis by HPAEC/PAD was performedu

es r acidh

ea-s

da sioncT ter:2 arch(a on-l e-t ibed[

w

2

p-e tionsw from2 ins

-eD DSC( um.S ared

and 20 mg were transferred to aluminium pans. The sam-ples were pregelatinised in an oven (100◦C for 15 min) andstored at 5◦C for 8 days before the DSC analysis. The heat-ing temperature range was 20–95◦C and the heating rate was10◦C/min. DSC analyses were performed in triplicates. Fromthe DSC thermograms onset temperature (To), mid tempera-ture (Tm), completion temperature (Tc) and enthalphy change(�H) were calculated. SinceTm andTc data did not give addi-tional information, onlyTo values are presented.Rheology of starch gels. The viscoelastic properties of 5%

gels in 10 mM NaCl prepared from transgenic starches wasperformed using a Stress Tech controlled stress rheometer(Rheologica AB, Lund, Sweden). The measurement geome-try used was a 4 cm plate–plate and measurements performedin triplicates at 25◦C. Oscillatory measurements were carriedout within the linear viscoelastic region, over a frequencyrange of 0.1–7.0 Hz. From the rheological data, the storagemodulus,G′, and the loss modulus,G′′, at 1.3 Hz were cal-culated. Gel strength of the samples was quantified from theslope of the plot of logG′ versus log frequency.

Pasting profiles of starch was performed using the RapidVisco Analyser (RVA, model 4; Newport Scientific, War-riewood, Australia) using 8% starch slurries as described[6]except for that the starch was suspended in 10 mM NaClinstead of pure water to standardise the phosphate counterion during gelatinisation.

2

oe isu-a n thecr resenti ltidi-m nentsw a seti scorea theo

witho ep im,N andf Thed red byi cor-r

3

3p

sig-n o

sing a BioLC (Dionex) as described[26].The concentration of phosphatemonoesterified to th

tarch was determined as glucose-6 phosphate afteydrolysis using an enzyme-linked assay as described[9].Theamylose concentrationof the starch samples was m

ured using iodine colorimetry as described[9].Size exclusion chromatography(SEC-RI). Separation an

nalysis of amylopectin and amylose by size excluhromatography was performed as described[6] using aoyopearl TSK HW-75 F (Tosohaas) column (diame6 mm, height: 300 mm) and elution of the solubilised st80–400�g) was accomplished with 10 mM NaOH at 50◦Ct a flow rate of 0.75 ml/min. Carbohydrate was detected

ine by refractive index (Waters 410 Differential Refractomer, Millipore). Amylose content was calculated as descr6].Environmental scanning electron microscopy(ESEM)

as carried out as described[23].

.6. Differential scanning calorimetry (DSC)

Melting of native starch. The thermal gelatinisation prorties of the native starch samples under dilute condiere analysed by DSC using a Seiko DSC220 operated5 to 100◦C at 5◦C/min. All starch samples were analysedlurries of 2 mg sample and 8�l 10 mM NaCl in triplicates.Melting of retrograded starch. The thermal melting prop

rties of retrograded starch were analysed by DSC[29]. TheSC measurements were carried out using a Pyris 1

Perkin-Elmer). The instrument was calibrated with inditarch slurries of 50% starch in 10 mM NaCl were prep

.7. Data mining

Principal component analysis (PCA)[30] is a technique txtract information from a complex data structure and vlising it on a graphic interface. The method is based oonstruction of principal components (PC1, PC2,. . .) whichepresent successively smaller and smaller variances pn the original data set, e.g. chromatograms. The mu

ensional data set can be resolved in orthogonal compohose linear combinations approximate the original dat

n the least squares sense. The PCA is visualised withnd loading plots. The score and loading plot showsbjects and variables, respectively.

PCA [30] and partial least square (PLS) regressionne referencey-variable [31] were carried out using throgramme UNSCRAMBLER 7.6 (Camo A/S, Trondheorway). All PCA and PLS models were mean-centered

ull cross validated. Only validated results are reported.ata obtained for the amylose concentrations as measu

odine spectroscopy and by SEC-RI were strongly interelated, thus only the SEC-RI data were used.

. Results

.1. Tuber production and enzymes of the transgeniclants

The yield of tubers from the transgenic plants was notificantly different from the wild type line as shown for tw

162 A. Blennow et al. / International Journal of Biological Macromolecules 36 (2005) 159–168

Table 1Potato tubers of transgenic plants (average of five independent plants, S.E.in parenthesis)

Line Yield (kg) Number of tubers Tuber weight (g)

Control H943 0.76 (0.27) 22 (10) 37 (10)H944-3.1 0.81 (0.27) 41 (10) 21 (6)H944-14.5 0.85 (0.40) 37 (20) 23 (6)

of the transgenic lines (Table 1). Moreover, no phenotypicalterations of the tubers were found. However, the averagesize of each tuber was approximately half that of the wildtype plants (Table 1) indicating that the inductions of tubersin the transgenic plants were stimulated. In most of the linesthe transgenic lines showed simultaneous suppression of bothSBEI and SBEII polypeptides, respectively, as determinedby Western analysis using specific polyclonal antibodiesdirected towards the two proteins (Fig. 1). In all the lines, atleast one of the isoforms was severely suppressed. The GWDprotein encoding the starch phosphorylating enzyme glucanwater dikinase was highly expressed in all the lines exceptfor line H944-6.1. Total soluble SBE activity measurements(Fig. 2) demonstrated than the allover residual activity ranged4–57% of the control values in the transgenic lines. The activ-ity followed the expression pattern of SBEI as most clearlyexemplified by the high SBEI expression and SBE activity inline H944-9.1 and the low SBE activity in H944-16.1 appar-ently expressing wild type levels of SBEII (Figs. 1 and 2).It should be noted that the SBEII isoform is mainly starch

bound while the SBEI isoform is partly soluble. Since starchbound SBE activities are not readily quantified, these data areonly indicative for the enzyme activity status in the tuber.

3.2. Structural analysis of the transgenic starches

The prime effect of suppressing the SBEI and SBEII activ-ity is expected to be a decreased number of�-1,6 branchpoints in the starch as manifested by an increased concen-tration of amylose and amylopectin with longer unit chains.As judged from HPAEC-PAD analysis carried out after enzy-matic debranching of the control starches the characteristicbimodality typical for normal starch in this chain length range[26] with a main peak at approximate degree of polymerisa-tion (DP) 14–15 and DP 45–50 is obvious (Fig. 3). All thetransgenic lines investigated showed a significant increase inpeak DP from DP 14–15 to DP 15–20 and a decrease in unitchains of DP 6–7 (Fig. 3). The bimodal appearance is also lostto a high degree. The mean DP calculated for the unit chainsbetween DP 6 and 60 shows a consistent increase in unitchain length for the transgenic starches (Table 2). Moreover,the phosphate and the amylose concentrations were increasedin the transgenic starches (Table 2).

Molecular size and structure give additional importantinformation about physico/chemical properties of starch likeviscosity and retrogradation (e.g. Shi et al.[32]). Amy-l mesa

I, SBE

Fig. 2. Activity of SBE in p

opectin and amylose, as identified by their elution volund iodine complexes in a SEC-RI system[6], were partly

II and GWD protein in the controls and in the transgenic lines.

otato tuber extracts.

A. Blennow et al. / International Journal of Biological Macromolecules 36 (2005) 159–168 163

Fig. 3. Amylopectin neutral (non-phosphorylated) chain length distributionof debranched control starch and starch from (A) Controls H935 (�), H944(�) and H943 (�), and (B) transgenic lines H944-3.1 (©) and H944-14.5(�). Single peaks between DP 6-60 in the chromatograms were integratedand analysed as described[26].

separated by the chromatography system used (Fig. 4). Amy-lopectin, being an extremely large macromolecule was almostentirely excluded from the pores in the resin while amylosewas well included in the pores (Fig. 4A and C). The mostextreme line, H944-14.5, contained approximately 50% amy-lose as opposed to the controls having approximately 25%amylose as judged from the SEC profile (Fig. 4A and C;Table 2). The concentrations calculated from the SEC profilescorrelated well with the values obtained by iodine calorimetry(Table 2). A small right shift in both amylopectin and amyloseelution profiles were also found indicating a small decreasein molecular size for the transgenic lines. By loading five-fold less sample (0.08 mg) on the column resulted in a minorshift of the elution profile (Fig. 4A and C) indicative of aonly minor aggregation of the starch molecules during separa-tion. Hence, the SEC profiles obtained chiefly reflect the truemolecular size profiles of the solubilised starch molecules.

Inspection of ESEM images of granular structures ofa starch sample (H944-14.5) with very high amylose andphosphate concentrations and long amylopectin unit chains(Fig. 5) shows the presence of multi-lobed granules withslightly more rough surface. This shows that in vivo starchcrystallisation and assembly is slightly altered during biosyn-thesis in the transgenic tubers. The structural data indi-

Fig. 4. SEC-RI profiles of starch extracted from (A) control H943 0.4 mg(solid line), control H944 0.4 mg (dotted line), control H944 0.08 mg (brokenline); (B) Line H944-3.1 0.4 mg and (C) Line H944-14.5 0.4 mg (solid line)and 0.08 mg (dotted line).

cate that the elongated amylopectin unit chains, the reducedbimodality, the high amylose content[33] and potentially thehigh phosphate concentration[23] affects the normal assem-bly of the starch molecules and results in distorted granules.

Fig. 5. ESEM images of starch prepared from normal control potatoes andtransgenic line H944-14.5.

164 A. Blennow et al. / International Journal of Biological Macromolecules 36 (2005) 159–168

Table 2Structural, melting and viscoelastic data for potato starches prepared from potato transgenic lines with reduced SBE activity

Line Amylopectinmean CL

Amylose,iodine (%)

Amylose,SEC-RI(%)

Phosphate(nmol/mgstarch)

DSC native starch DSC retrogradedstarch

Viscoelasticity

T peak(◦C)

Enthalpy(J/g starch)

Tpeak(◦C)

Enthalpy(J/gstarch)

Storagemodulus,G′ (Pa)

Lossmodulus,G′ ′ (Pa)

H944-3.1 28.1 34.3 37.7 51.2 73.2 15.8 58.7 5.2 153 28H944-3.3 28.0 38.9 46.4 53.4 73.0 12.0 70.0 5.3 140 23H944-3.5 28.3 32.5 30.4 47.3 70.4 14.9 72.0 1.2 131 29H944-6.1 27.7 37.6 39.2 52.9 71.7 16.6 72.6 3.7 133 27H944-9.1 27.7 39.3 44.9 49.5 75.6 8.7 71.5 3.4 134 19H944-14.5 27.6 43.5 50.4 63.5 75.0 8.2 72.0 2.7 124 19H944-15.1 28.1 38.7 43.2 56.1 73.1 10.4 71.4 2.7 114 23H944-24.3 28.2 31.5 34.5 43.4 71.2 14.3 71.7 2.9 144 27H944-29.4 27.8 29.0 30.2 33.4 70.5 14.3 71.7 2.8 91 26H944-35.1 28.0 39.1 43.1 48.2 73.1 14.2 71.6 2.8 153 26H943-16.1 27.8 28.6 26.7 44.8 74.5 15.6 71.6 2.8 133 28H943-16.2 28.0 34.3 38.3 52.8 70.7 15.4 68.0 4.4 98 28

Minimum 27.6 28.6 26.7 33.4 70.4 8.2 58.7 1.2 91 19Maximum 28.3 43.5 50.4 63.5 75.6 15.8 72.6 5.3 153 29Average 27.9 35.6 38.8 49.7 72.7 13.4 70.2 3.3 129.0 25.3

Control H935 26.1 27.5 27.0 19.6 65.0 16.6 61.9 5.7 59 21Control H943 26.4 27.1 25.2 18.6 64.3 17.1 57.8 5.2 59 21Control H944 26.9 27.6 25.0 19.5 65.0 16.1 60.2 3.6 46 17

Minimum 26.1 27.1 25.0 18.6 64.3 16.1 57.8 3.6 46 17Maximum 26.9 27.6 27.0 19.6 65.0 17.1 61.9 5.7 59 21Average 26.5 27.4 25.7 19.2 64.8 16.6 60.0 4.8 54.7 19.7

Control plants were derived from in vitro cultures and grown in parallel with the transgenic plants. Control line H935 was transformed with vector without theantisense SBE sequence.

The data obtained are consistent with data for typical highamylose starches of maize and pea[5].

3.3. Thermal and rheological properties of thegenetically engineered starches

Alterations of the magnitude shown inFigs. 3 and 4forthe transgenic starches would have a marked influence on thephysical properties of starch as has already been indicatedfor genetically engineered starches with less severe struc-tural alterations[21]. Especially large and specific effectsare expected on the aqueous thermal solubilisation processfor native granular starch and re-crystallised preparations aswell as viscoelastic properties of aqueous gels prepared fromthe starches. Of specific interest for the starches in this inves-tigation is the interplay between amylose concentration andamylopectin molecular structure.

As judged from DSC experiments, the transgenic nativestarch granules showed a consistent 5–11◦ increased the peakgelatinisation temperature compared to the control starches.The gelatinisation transition interval was also significantlywider for the transgenic starches (Fig. 6), which has beendemonstrated for maize model starches[34]. In contrast,the enthalpy of gelatinisation was decreased up to 50% inthe transgenic starches. This was typically seen for starchesw nge

in enthalpy is seemingly conflicting with that observed forpotato starches with less severe suppression of the SBE activ-ity [20,21]here the enthalpy was unchanged or even slightlyincreased. However, these starches had no detectable alter-ations in the amylose content but significant variations inphosphate concentration and minor differences in the amy-lopectin chain length profile. DSC data of retrograded starch(Table 2) indicated a 3–15◦ increase in retrogradation peaktemperature for the transgenic starches indicating that crys-talline entities were formed during retrogradation that hadhigher stability and lower tendency to re-hydrate than thecontrol starches. The enthalpies for most of the retrogradedtransgenic starches were decreased, showing that similarcrystalline lattices as the native starch granule are possi-bly generated during re-crystallisation. However, the DSCtraces for the retrograded samples were considerably widerin both directions showing melting of crystalline entities atboth lower and higher temperatures as compared to the nativestarches (Fig. 6). Again, the data are in agreement with DSCdata on (non-phosphorylated) maize starches[35].

Viscoelastic analysis of gels prepared from the transgenicstarches showed some interesting features. First of all itshould be noted that the transgenic starches with very highamylose/phosphate concentrations and long amylopectinunit chains were capable to hydrate and swell much moreextensively during the gelatinisation process than the

ith high amylose concentrations. Interestingly, the cha

A. Blennow et al. / International Journal of Biological Macromolecules 36 (2005) 159–168 165

Fig. 6. DSC melting traces for gelatinisation of the native starch granules(solid lines) and retrograded starch (broken lines).

control starches as demonstrated by Rapid Visco analysis(Fig. 7), e.g. line H944-14.5 displayed a two-fold higherpeak viscosity and final viscosity compared to the controlstarch (H944). The storage modulus (G′) was generallymuch higher for the transgenic starches (Table 2) showingthat strong elastic gel networks (highG′) were formed.Moreover, the viscous modulus (G′′) was higher in thetransgenic starches showing that viscosity was preservedin the pastes. TheG′ data are in agreement with previousfindings for high amylose starches, e.g. in rice[36]. However,for the most extreme starches with combined high amyloseand phosphate concentration, lines H944-14.5 and H944-9.1there was a tendency for a drop in viscous modulus down to,or even below, the control starch values. These two lines alsoshowed the highest gel strength as concluded from the slopeof G′ (data not shown). Hence, for these extreme starches

Fig. 7. Pasting profiles of 8% starch prepared from line H944-14.5 (dottedline) and control H944 (filled line) as measured by a Rapid Visco Analyser(RVA).

extensive elastic network formation took place during gelformation at the expense of the viscosity of the gels.

3.4. Multivariate data mining

As indicated by a visual inspection of the data inTable 2,the investigated transgenic starch samples showed variousdegrees of intercorrelations, and thus convincing evidenceof the impact of one specific structural factor, e.g. amy-lopectin unit chain length, on the physical properties, e.g.viscoelasticity of aqueous gels, could not be readily iso-lated. Using a multivariate approach, hidden phenomena inthe data structures could possibly be disclosed allowing for astructure–functionality model to be developed. High amyloseconcentration and long amylopectin unit chains are likely tosuppress the hydration power for the starch granules whilethe presence of covalently linked phosphate is expected toincrease the hydration capacity.

The PCA score plot using PC1 and PC2 (explaining 62and 18% of the variance, respectively) clearly clusters thetransgenic starches apart from the control lines primarilyby PC1 (Fig. 8A). The enthalpy of retrogradation (Re-enth)and the enthalpy of gelatinisation (Gel-enth) are clearlycorrelated with the control lines showing that this variableis generally reduced in the transgenic lines. Moreover, theenthalpy of gelatinisation is located diagonally througho thati osec ion)i tiono nes.T highi hainl ) andtt es oft e linesa tionsi he

rigo from the amylose (content) variable which meanst is quite strongly but inversely correlated to the amyloncentration (r2 =−0.58, pair wise least square regressndicating a direct effect of the decreased concentraf crystalline amylopectin present in the transgenic lihe PCA score plot also reveals that there exist a

ntercorrelation between the mean amylopectin unit cength (CL), the phosphate concentration (phosphatehe peak temperature of gelatinisation (GelT peak). Theransgenic lines are characterised by having higher valuhese properties. The differences between the antisensre spread out by PC2 largely influenced by the varia

n the viscous modulusG′′and the amylose content. T

166 A. Blennow et al. / International Journal of Biological Macromolecules 36 (2005) 159–168

Fig. 8. PCA for the transgenic starch system. The score plot (A) shows how the genetically engineered starches cluster away from the control lines in dependenceof the variables. The loading plot (B) indicates the correlations of the variables. (C) The relation between the phosphate concentration and the meanunit chainlength of amylopectin using PLS. Regression coefficients for thex-variables (amylopectin unit chain length, CL) in the phosphate PLS model.

starches with highest phosphate:amylose ratio i.e. lines944-3.5 and 944-16.1 are positioned in the north-eastquadrant due to a high value of the viscous modulus:G′′.This is consistent with the high viscosity expected for highlyphosphorylated starches with moderate amylose content,seemingly moderated by low amylose content rather thanextreme phosphate concentration. At the other extreme, wefind the transgenic lines with high amylose contents in thesouth–east quadrant. Hence, multiple functionalities arefound and separated within in the transgenic cluster.

Interestingly, the apparent discrepancy results between ourstudy where the enthalpy of gelatininsation was decreasedup to 50% and the potato starches with less severe suppres-sion of the SBE activity where the enthalpy of gelatinisa-tion was either unchanged or slightly higher[20,21] can beexplained when the content of amylose is considered. Therewere no detectable alterations (variations) in the amyloseconcentration but only significant variation in the phosphateconcentration and minor differences in the amylopectin unitchain length profile in the studies by Safford et al.[21] andWischmann et al.[20]. However, the PCA reveals that theenthalpy of gelatinization is strongly but inversely corre-lated to the concentration of amylose. Hence, the transgenicstarches in the present data set contribute to low gelatinisationenthalpies.

The mean unit chain length of amylopectin can predictt rrela-

tion of 0.95 using PLS regression, as illustrated inFig. 8C.The regression coefficients have a rather sharp negative peakcentered at DP 12–13 and a broad positive peak at DP 28–42,showing that the phosphate concentration negatively corre-lated to the DP 12–13 chains and is positively correlated tothe chain lengths around DP 28–42. This model is very differ-ent from the one generated for non-transgenic starches[13]showing a trimodal data structure with peaks negative cen-tered at DP 12–17, DP and DP 37–53 and a positive peakat DP 18–20, demonstrating the dramatic and fundamentalstructural difference in the transgenic starches.

4. Discussion

During the last decade, the use of biotechnology to specif-ically induce structural alterations in the starch biopolymershas been established. However, the effects on starch biosyn-thesis of a single genetic lesion have been shown to be muchmore complex than expected. In the present data set, theamylopectin fraction of the starches generated by genetictransformation of the potato plant had longer unit chains,lower molecular size and higher phosphate concentration, andthe amylose fraction was increased and had lower molecularsize. The biochemical basis for this covariation remains tobe investigated. The most important property of the potatos llow-

A. Blennow et al. / International Journal of Biological Macromolecules 36 (2005) 159–168 167

ing the decreased branching of the starch molecules. Thiseffect can be explained by the specific action on the sub-strate of the starch phosphorylating enzyme GWD whichhas recently shown a high activity for these long unit chains[37]. The minor shift in apparent molecular size of the amy-lopectin and amylose molecules as monitored by SEC is inline with the more extreme high amylose starches analysedby Schwall et al.[14] and the high amylose starch extractedfrom the lilyC. zedoaria[26]. One of these potato high amy-lose/high phosphate starches displayed almost a total reduc-tion of the amylopectin as judged by SEC[14]. However,since granular structures are formed, these low molecularsize and amylose-like molecules contain branch points asjudged from the unit chain length analysis of this starch.Hence simultaneous antisense of SBEI and SBEII generatesintermediate and highly phosphorylated starch molecules thatare assembled as distorted granular structures[23]. This inter-mediate material is known to be generally characteristic forhigh amylose starches[5]. Since starch phosphorylation isrelated to increased starch degradation in plants[38] onerational explanation for a role of hyper phosphorylation ofthese molecules in the plant can be to enable their degrada-tion by hydrolytic enzymes by distortion of double helicalformation and assembly by opening up crystalline segmentsformed by the long amylopectin unit chain. It seems, however,that the increased chain length more than compensates fort tablea pho-r plaint s ane yn-t

nics ela-t f ther stu-l rhe-o uchc

witht amy-l thet e thev h gel,w ersee ydra-t ins int , thep esulti tur-b ctingt f theD cen-t pressh em-p dly be

opposed by destabilising forces exerted by phosphate substi-tution on the amylopectin unit chains. The results indicatea dominating role of the decreased glucan branching for thephysical properties of these starches. However, the data alsodemonstrate that the increased phosphate concentration in thetransgenic starches can give new functionality, most notablyincreased hydration capacity, not present in high amylose andnon-phosphorylated starches prepared from, e.g. maize.

The high throughput analysis of the plant metabolomeand the evaluation of the dataset using multivariate analysiswere recently pioneered[39]. Albeit the high concentrationsand major impact on primary metabolism of biopolymers inplant systems, these are generally not included in metaboliteprofiling strategies. Structural and physical profiling of starchbiopolymers engineered by specific alterations of the genomewould provide the tools for biopolymers to be included inthe rapidly developing functional genomics models. Mul-tivariate analysis has recently been employed to study andpredict structural and functional properties of complex starchsystems[7,13,40]. By collecting biopolymer structural andphysical data for mutant and transgenic plant systems, pre-dictive biotechnological strategies can be employed to cre-ate novel functionalities in between, or extending, from themodel data set. In the present study, major dependencies in thepotato starch system such as the one between phosphate andamylopectin unit chain length, as well as more subtle vari-a tedu cisem paredf alter-a ionso s andm ns int

5

s foru n ofp om-b rateda ins,h sizea lowerm singP log-i icity,t nde

A

theE

he increased phosphorylation generating much more smylopectin crystallites. The relationship between phosylation and starch degradation in the plant may also exhe lower molecular size of the amylopectin molecules affect of hydrolytic activity taking place during starch bios

hesis. These hypotheses remain to be investigated.All the structural alterations found in the transge

tarches would potentially have a major effect on the ginisation of the starch and the viscoelastic properties oesulting aqueous gels. However, it is not trivial to poate clear relationships between structural, thermal andlogical properties for a granular polymer system with somplexity.

The combined decreased molecular size combinedhe increased amylose concentration and elongatedopectin unit chains of the starch biopolymers found inransgenic potato starches would in principle decreasiscous modulus increase the elastic modulus of a starchile higher phosphate concentration would have the invffect of the starch gel as effects of phosphate induced h

ion and disordered entanglement of the glucan unit chahe gel. For the thermal properties as monitored by DSCresence longer unit chains in the starch granule would r

n more stable or larger crystalline units. However, disances may occur as an effect of amylose chains obstru

he crystalline clusters as indicated by the broadening oSC traces in the transgenic lines. The high amylose con

rations generated in the transgenic starches would supydration and swelling resulting in higher gelatinisation teratures. However, these effects would again suppose

tions within the transgenic cluster were readily indicasing a multivariate data mining approach. More preodels can be generated only by including starches pre

orm mutants generating more specific and independenttions in the structural motifs including, e.g. concentratf phosphate and amylose as well as unit chain structureolecular sizes of the amylose and amylopectin fractio

he starch.

. Conclusion

The present work has demonstrated the possibilitiesing a transgenic approach for functional modificatiootato tuber starch. Starch biopolymers with a complex cination of molecular and physical properties were generising from amylopectin fractions with longer unit chaigher phosphate concentration and lower molecularnd increased amylose concentrations with apparentolecular size. Some of the complexity was clarified uCA, by which it was possible to disclose unique rheo

cal effects such as combined high viscosity and elastypical for gels with combined high hydration capacity antangled non-crystalline networks.

cknowledgements

We would like to thank James Sanderson for providingSEM images.

168 A. Blennow et al. / International Journal of Biological Macromolecules 36 (2005) 159–168

Per Lassen Nielsen, Helle Kildal Mogensen and LisByrsting Møller are thanked for excellent technical support.This project was supported by The National STVF FrameProgramme Exploring the Biosynthetic Potential of Potato,The Danish National Research Foundation, The DanishBiotechnology Programme, The Danish Directorate forDevelopment (Centre for Development of Improved FoodStarches) and The Committee for Research and Developmentof theOresund Region (Oforsk).

References

[1] R.P. Ellis, M.P. Cochrane, M.F.D. Dale, C.M. Duffus, S. Lynn, I.M.Morrison, R.D.M. Prentice, J.S. Swanston, S.A.J. Tiller, Sci. FoodAgric. 77 (1998) 289.

[2] A. Blennow, in: A.-C. Eliasson (Ed.), Starch in Food: Structure Func-tion Applications, CRC Press Woodhead Publishing Ltd., Cambridge,UK, 2004, p. 97, ISBN 0-8493-2555-2.

[3] J.J.M. Swinkels, Starch/Starke 37 (1985) 1.[4] R.G.F. Visser, L.C.J.M. Suurs, P.A.M. Steeneken, E. Jacobsen,

Starch/Starke 49 (11) (1997) 443.[5] W. Banks, C.T. Greenwood, D.D. Muir, Starch/Starke 26 (9) (1974)

289.[6] A. Blennow, A.M. Bay-Smidt, R. Bauer, Int. J. Biol. Macromol. 28

(2001) 409.[7] H. Fredriksson, J. Silverio, R. Andersson, A.-C. Eliasson, P.Aman,

Carbohydr. Polym. 35 (1998) 119.60

s.

[ .[ , A.

[ M.R.159.

[ dr.

[ .-C.

[

[16] J.R. Lloyd, V. Landschutze, J. Kossmann, Biochem. J. 338 (1999)515.

[17] S.A. Jobling, R.J. Westcott, A. Tayal, R. Jeffcoat, G.P. Schwall, Nat.Biotechnol. 20 (2002) 295.

[18] R. Lorberth, G. Ritte, L. Willmitzer, J. Kossmann, Nat. Biotechnol.16 (1998) 473.

[19] G. Ritte, J.R. Lloyd, N. Eckermann, A. Rottmann, J. Kossmann, M.Steup, Proc. Natl. Acad. Sci. 99 (2002) 7166.

[20] B. Wischmann, A. Blennow, F. Madsen, K. Jørgensen, P. Poulsen,O. Bandsholm, Food Hydrocolloids 19 (6) (2005) 1016–1024.

[21] R. Safford, S.A. Jobling, C.M. Sidebottom, R.J.M. Westcott, D.Cooke, K.J. Tober, B.H. Strongitharm, A.L. Russel, M.J. Gidley,Carbohydr. Res. 35 (1998) 155.

[22] S.A. Jobling, G.P. Schwall, R.J. Westcott, C. Sidebottom, M. Debet,M.J. Gidley, B. Jeffcoat, R. Safford, Plant J. 18 (1999) 163.

[23] A. Blennow, M. Hansen, A. Schulz, K. Jørgensen, A.M. Donald,Sanderson, J. Struct. Biol. 143 (2003) 229.

[24] P. Poulsen, I.S. Sørensen, Genetic Modification of Starch in PlantsGB Patent Application GB2,360,521 (2001).

[25] M. Joersbo, I. Donaldson, J. Kreiberg, S.G. Petersen, J. Brunstedt,F.T. Okkels, Mol. Breeding 4 (1998) 111.

[26] A. Blennow, A.M. Bay-Smidt, B. Wischmann, C.E. Olsen, B.L.Møller, Carbohydr. Res. 307 (1998) 45.

[27] W.N. Burnette, Anal. Biochem. 112 (1981) 195.[28] A. Blennow, G. Johansson, Phytochemistry 30 (1991) 437–444.[29] F.E. Ortega-Ojeda, A.-C. Eliasson, Starch/Starke 53 (2001) 520.[30] H. Hotelling, J. Edu. Psychol. 24 (1933) 417.[31] S. Wold, H. Martens, H. Wold, Lect. Notes Math. 973 (1983) 286.[32] Y.-C. Shi, T. Capitani, P. Trzasko, R. Jeffcoat, J. Cereal Sci. 27

(1998) 289.[33] P.J. Jenkins, A.M. Donald, Int. J. Biol. Macromol. 17 (6) (1995)

[

[

[ )

[ m.

[ ends

[[ )

[8] Y.P. Kim, D.P. Wiesenborn, P.H. Orr, L.A. Grant, J. Food Sci.(5) (1995) 1060.

[9] A.M. Bay-Smidt, A. Blennow, M. Bojko, B.L. Møller, Ann. TranNord. Rheol. Soc. 7 (1999) 31.

10] S. Hopkins, R. Gormley, Food Sci. Technol. 33 (5) (2000) 38811] A. Viksø-Nielsen, A. Blennow, K. Jørgensen, K.H. Kristensen

Jensen, B.L. Møller, Biomacromolecules 2 (2001) 836.12] K. Shewmaker, C.D. Boyer, D.P. Wiesenborn, D.B. Thompson,

Boersig, J.V. Oakes, D.M. Stalker, Plant Physiol. 104 (1994) 113] A. Blennow, S.B. Engelsen, L. Munck, B.L. Møller, Carbohy

Polym. 41 (2000) 163.14] G.P. Schwall, R. Safford, R.J. Westcott, B. Jeffcoat, A. Tayal, Y

Shi, M.J. Gidley, S.A. Jobling, Nat. Biotechnol. 18 (2000) 551.15] J. Kossman, J. Lloyd, Crit. Rev. Plant Sci. 19 (2000) 171.

315.34] N. Inouchi, D.V. Glover, Y. Sugimoto, H. Fuwa, Starch/Starke 43

(12) (1991) 473.35] N. Inouchi, D.V. Glover, Y. Sugimoto, H. Fuwa, Starch/Starke 43

(12) (1991) 468.36] T. Sasaki, T. Yasui, J. Matsuki, T. Satake, Starch/Starke 54 (9) (2002

410.37] R. Mikkelsen, L. Baunsgaard, B.L. Møller, A. Blennow, Bioche

J. 377 (2) (2003) 525.38] A. Blennow, S.B. Engelsen, T.H. Nielsen, L. Baunsgaard, Tr

Plant Sci. 7 (2002) 445.39] O. Fiehn, Plant Mol. Biol. 48 (1–2) (2002) 155.40] L.G. Thygesen, A. Blennow, S.B. Engelsen, Starch/Starke 55 (2003

241.