• Open Access

Expression of an engineered granule-bound Escherichia coli glycogen branching enzyme in potato results in severe morphological changes in starch granules

Authors

  • Xing-Feng Huang,

    1. Wageningen UR – Plant Breeding, Wageningen University and Research Center, Wageningen, the Netherlands
    2. Graduate School Experimental Plant Sciences, Wageningen University, Wageningen, the Netherlands
    Current affiliation:
    1. Center for Rhizosphere Biology, Department of Horticulture and Landscape Architecture, Colorado State University, Fort Collins, CO, USA
    Search for more papers by this author
    • These authors contributed equally to this study.

  • Farhad Nazarian-Firouzabadi,

    1. Wageningen UR – Plant Breeding, Wageningen University and Research Center, Wageningen, the Netherlands
    Current affiliation:
    1. Agronomy and plant breeding group, Faculty of Agriculture, University of Lorestan, Khorramabad, Iran
    Search for more papers by this author
    • These authors contributed equally to this study.

  • Jean-Paul Vincken,

    1. Wageningen UR – Plant Breeding, Wageningen University and Research Center, Wageningen, the Netherlands
    Current affiliation:
    1. Laboratory of Food Chemistry, Wageningen University, Wageningen, the Netherlands
    Search for more papers by this author
  • Qin Ji,

    1. Wageningen UR – Plant Breeding, Wageningen University and Research Center, Wageningen, the Netherlands
    Current affiliation:
    1. Department of Biology, Huaiyin Teachers College, Huaian, China
    Search for more papers by this author
  • Luc C. J. M. Suurs,

    1. Wageningen UR – Plant Breeding, Wageningen University and Research Center, Wageningen, the Netherlands
    Search for more papers by this author
  • Richard G. F. Visser,

    1. Wageningen UR – Plant Breeding, Wageningen University and Research Center, Wageningen, the Netherlands
    Search for more papers by this author
  • Luisa M. Trindade

    Corresponding author
    • Wageningen UR – Plant Breeding, Wageningen University and Research Center, Wageningen, the Netherlands
    Search for more papers by this author

Correspondence (Tel +31(0)317482127; fax +31(0)317483457; email luisa.trindade@wur.nl)

Summary

The Escherichia coli glycogen branching enzyme (GLGB) was fused to either the C- or N-terminus of a starch-binding domain (SBD) and expressed in two potato genetic backgrounds: the amylose-free mutant (amf) and an amylose-containing line (Kardal). Regardless of background or construct used, a large amount of GLGB/SBD fusion protein was accumulated inside the starch granules, however, without an increase in branching. The presence of GLGB/SBD fusion proteins resulted in altered morphology of the starch granules in both genetic backgrounds. In the amf genetic background, the starch granules showed both amalgamated granules and porous starch granules, whereas in Kardal background, the starch granules showed an irregular rough surface. The altered starch granules in both amf and Kardal backgrounds were visible from the initial stage of potato tuber development. High-throughput transcriptomic analysis showed that expression of GLGB/SBD fusion protein in potato tubers did not affect the expression level of most genes directly involved in the starch biosynthesis except for the up-regulation of a beta-amylase gene in Kardal background. The beta-amylase protein could be responsible for the degradation of the extra branches potentially introduced by GLGB.

Introduction

Starch is the major storage carbohydrate in plants and is present in several plant organs such as leaves, stems, fruits, roots and tubers at certain time points during plant development. This unique glucan biopolymer in the form of discrete granules is an important storage unit of energy that is captured by plants using sunlight, water and carbon dioxide. Depending on the organ and species in which it is produced, the size and morphology of starch granules can vary widely. Starch granules are approximately 0.5–110 μm in diameter and can have different shapes such as spherical, elliptical or polyhedral. The two α-glucan polymers, amylose and amylopectin, are assembled together to form a semi-crystalline starch granule (Thompson, 2000). Starch is one of the most important plant raw materials for both food and nonfood applications. It is a staple in the diet of most of the world's population and broadly used in paper, textile, adhesive and pharmacy industries (Ellis et al., 1998).

As an abundant, renewable and biodegradable polymer, starch is becoming increasingly attractive for industrial uses because of the environmental concerns about the industrial waste and greenhouse gases generated from petroleum products. Native starches are often modified by chemical or physical processing to meet the specific needs of the end uses as it does not perform ideally in its native form in a large range of industrial applications. Modification of starch biosynthesis pathway holds an enormous potential for tailoring granules or polymers with new functionalities and prevents the use of high energy consuming postharvest modifications with harsh chemicals (Kok-Jacon et al., 2003). By manipulating the expression level of genes involved in starch biosynthesis such as knocking out or overexpressing key genes, starches with altered properties have been obtained in potato (e.g. Jobling et al., 2002; Kuipers et al., 1994; Schwall et al., 2000). Most of those approaches focused on the alteration in the amylose/amylopectin ratio. Another strategy involves the expression of heterologous genes from other organisms, including bacteria, fungi and animals, which encode biosynthetic or modifying enzymes for producing starches with novel properties. For instance, expression of Escherichia coli glycogen synthase (glgA) resulted in an increased branching degree of amylopectin in potato (Shewmaker et al., 1994). The expression of an E. coli maltose acetyl transferase (mat) in potato tubers resulted in starch granules with acetyl groups (Nazarian Firouzabadi et al., 2007b).

It is known that the long and unbranched α-glucans such as amylose have a higher tendency to retrograde than highly branched and much shorter chain amylopectin (Jobling et al., 2002). It has also been shown that amylose-free starches have better freeze–thaw stability than amylose-containing genotypes (Zheng and Sosulski, 1998). To improve freeze–thaw stability of starch, an alternative is to manipulate the structure of starch components (amylose and amylopectin) through genetic engineering. Theoretically, it is possible to create starches with high freeze–thaw stability by (i) shortening of long chains of amylose, (ii) reducing the branch chain length of amylopectin and (iii) increasing the number of short branch chains of amylopectin. For instance, down-regulation of three starch synthase genes (GBSSI, SSII and SSIII) simultaneously by antisense in potato led to reduction in the chain length of the amylopectin branches. As a result, a significant improvement of freeze–thaw stability of potato starch was achieved in which freeze–thaw stable starches could maintain their stability through five freeze–thaw cycles (Jobling et al., 2002).

For producing starches with higher branching degree of amylopectin and novel properties in planta, the E. coli glycogen branching enzyme (GLGB) was fused to SBD at either carboxyl or amino terminus (Ji et al., 2003) and was introduced in amylose-containing (cv. Kardal) and amylose-free mutant (amf) genetic backgrounds. The SBD brings GLGB in more intimate contact with its substrate and acceptor molecules and thereby acts more effectively than GLGB alone. In this study, the influence of an engineered GLGB on structure and properties of potato starch is presented and discussed.

Results

GlgB transformants do not show a visible phenotype in plant architecture and tuber morphology

Two constructs, the pBIN19/glgB-SBD and pBIN19/SBD-glgB plasmids, were used for the expression of the GLGB-SBD and SBD-GLGB fusion proteins in potato plants (Kardal and amf), respectively (Figure 1). Transformed potato plants in this study are referred to as amfGS, amfSG, KDGS and KDSG, in which GS and SG stand for the GLGB-SBD and SBD-GLGB fusion proteins, respectively. Besides, untransformed genotypes are referred to as amf-UT and KD-UT. Thirty kanamycin-resistant, independent transformant with each of the constructs was grown to maturity in the greenhouse. None of the transgenic plants showed visible differences in the architecture of the plants or in the morphology of the tubers relatively to the untransformed controls (data not shown). The tuber yield was similar in the transgenic potato clones and in the control plants (data not shown).

Figure 1.

Schematic depiction of two different binary vector constructs used for the transformation of potato plants. Genes were in frame with GBSSI transit peptide to allow amyloplast targeting and were driven by GBSSI promoter for tuber expression. RB and LB, right and left borders, respectively. glgB, starch-binding domain (SBD), 3′NOS and linker stand for Escherichia coli glycogen branching enzyme, SBD of cyclodextrin glycosyltransferase from Bacillus circulans, NOS terminator and proline–threonine rich stretch, respectively. ▲ and ↑ signs represent kanamycin-resistant gene and cleavage site of the transit peptide.

GLGB fusion protein is incorporated into starch granules

The accumulation levels of GLGB/SBD fusion protein in starch granules of amfGS, amfSG, KDGS and KDSG transformant plant series were quantified by Western dot blot analysis according to Ji et al. (2003) in which the numbers from 0+ to 6+ were assigned, with 0+ meaning that no fusion protein could be detected and 6+ the highest amount of fusion protein. It was observed that more than 90% of the amf transformants (96% and 92% of GLGB-SBD and SBD-GLGB transformants, respectively) and over 85% of Kardal transformants (86% and 89% of GlgB-SBD and SBD-GlgB, respectively) were classified as 4+ through 6+ classes (Figure 2). Moreover, almost half of the transformants accumulated the highest amount of the fusion proteins (6+) regardless of constructs and backgrounds.

Figure 2.

Comparison of the starch-binding-domain (SBD)-containing fusion protein content of starch granules in the amf and Kardal background. (a) Classes of SBD accumulation in potato starch granules. This classification is based on the results of a Western dot blot analysis with various starch samples. 0+ represents the class with no fusion protein and 6+ the class with the largest amount of fusion protein. The 6+ category represents the transgenic granules that gave a similar signal in Western dot blot analysis as the 5+ class with half the amount of starch. (b) Distribution of the individual transformants over the seven classes of SBD accumulation in both backgrounds.

To determine whether GLGB/SBD fusion protein was present in the soluble phase of the potato tubers, potato juices of tubers from transformants were subjected to Western dot blot analysis too. As can be seen in Table 1, a relatively low (2+) amount of protein was found in the soluble part of 6+ transformants, and no fusion protein was found in the soluble fractions of any of the other classes.

Table 1. Summary of different starch characteristics in relationship with amount of fusion protein accumulation. Starch apparent amylose content (%AM), median granule size (d50) and starch gelatinization temperature are shown. Western dot blot analysis for selected starches and their corresponding juice fractions are shown. %AM, d50 and T onset data are average of three independent measurements
CloneW.B%AMT onsetd50
GranuleJuice
  1. W.B, Western dot blot; NP, not present.

amf-UTNPNP3.0 (±0.4)69.4 ± (0.0)19.3 (±0.2)
amfGS266+2+3.0 (±0.3)70.7 ± (0.2)10.7 (±0.4)
amfSG126+2+3.4 (±0.1)69.2 ± (0.1)18.9 (±0.3)
KDUTNPNP18.6 (±0.4)63.1 ± (0.1)25.2 (±0.1)
KDGS266+2+17.9 (±0.5)61.8 ± (0.1)22.8 (±0.3)
KDSG256+2+18.3 (±0.3)61.9 ± (0.1)14.2 (±0.2)

SDS-PAGE followed by immunoblotting with both anti-SBD and anti-GLGB antibodies revealed that GLGB/SBD fusion proteins were accumulated in the starch granules of the transgenic plants. The observed mass of about 100 kDa corresponded well with the predicted molecular mass of the fusion proteins that were detected with both anti-SBD and anti-GLGB (Figure 3) antibodies, indicating that the observed proteins in starch granules are indeed fusion proteins. These results showed that correctly processed GLGB/SBD fusion proteins were successfully targeted to the starch granules.

Figure 3.

SDS-PAGE/Immunoblot analyses of starch granule-bound proteins of amfGS26 and amfSG7. Migration of marker proteins (M) is shown to the right in each panel. SG and GS represent starch-binding domain (SBD)-GlgB and GlgB-SBD fusion proteins, respectively. Lanes designated with ‘C’ represent starches from untransformed control plants.

GLGB severely affects starch granule morphology

To investigate the effects of the accumulation of GLGB/SBD fusion proteins on the starch granule, both light microscopy (LM) and scanning electron microscopy (SEM) were used to examine alterations in the morphology of various starches from the transgenic plants. Both LM and SEM analysis revealed that the morphology of starches from transgenic plants was severely affected as a result of the expression of the fusion proteins in both genetic backgrounds (Figures 4 and 5). Starches from transgenic amylose-containing plants exhibit irregular bumpy surfaces (Figure 4b,c,e,f). Starches from amylose-free transgenic plants show both amalgamated starch granules and porous starch granules with many deep holes (Figure 5b,c,e,f), which sometimes resulted in starch granules in which cracks connect different holes (Figure 5e). With respect to the SBD position in the fusion proteins, N- as well as C-terminal fusion proteins resulted in the same phenotype in each particular genetic background. These starch phenotypes can be seen in different transformants irrespective of their level of GLGB/SBD fusion protein accumulation. Quantification of such altered starch granule phenotypes as shown in Figures 4 and 5 revealed that the higher the amount of granule-bound fusion proteins, the more altered starch granules (Figure 6). The phenotype of the starch granules remains the same after crossing with other potato lines (data not shown).

Figure 4.

SEM analyses of starch granules from Kardal-UT (a, d) in comparison with that of the different selected transformants at different magnifications. KDGS26 (6+) (b and e) and KDSG25 (6+) (c, f). Magnifications are a (×250), b and c (×500) d, e and f (×2000).

Figure 5.

SEM analyses of starch granules from amf-UT (a, d) in comparison with that of the different selected transformants at different magnifications. amfGS26 (6+) (b, e) and amfSG7 (6+) (c, f). Magnifications are a (×250), b and c (×500), d (×2000), e and f (×4500).

Figure 6.

Percentage of granules with altered morphology for the various classes of starch-binding domain-containing fusion proteins. A population of 100 starch granules in triplicate was counted. Bars indicate mean ± SD.

To investigate whether the deposition of GLGB/SBD fusion proteins affected the crystallization and the internal structure of the starch granules, sections of starch granules from two transformants where glgb is highly expressed, amfGS26 and KDGS26, as well as their corresponding controls were prepared and LM images were taken. Analysis of the starch granule section showed that both the rough surface granules and porous granules have the well-known growth rings (Figure 7). However, the growth rings in the rough and porous granules were not distributed as evenly as in control starch granules and the growth rings in the outer layer of porous granules were broken by the holes in the porous granules (compare Figure 7a with d and b with e). Moreover, the holes in the porous granules were only superficial and did not penetrate through the whole granule (Figure 7e). Furthermore, there were starch granules that did not show the concentric growth rings of normally assembled granules but contained multiple small granules inside (Figure 7c,f).

Figure 7.

Micrographs showing the section images of starch granules of Kardal (a), KDGS26 (d), amf (b) and amfGS26 (c, e, f). Starch granule sections were stained with a 10× diluted Lugol solution. The black scale bar indicates 5 μm in all images.

Altered starch granules are detectable from the initial stage of potato tuber development

The presence of GLGB/SBD fusion protein resulted in porous and amalgamated starch granules in amylose-free (amf) background and irregular bumpy surface granules in amylose-containing (Kardal) background. To investigate at which time point during potato tuber development the accumulation of GLGB/SBD fusion protein affected starch granule morphology, starches from amfGS26 and KDGS26 transformants were isolated at five tuber developmental stages according to potato tuber size, ranging from swelling stolons and four different tuber growth stages that correspond to potato tuber developmental stages 3, 4, 5, 6 and 7 in Kloosterman's study (Kloosterman et al., 2005). SEM analysis showed that starch granule morphological phenotypes caused by expression of GLGB/SBD fusion proteins were different in the Kardal and amf backgrounds at different stages of tuber development (Figure 8). Starches harvested from both amfGS26 and KDGS26 transformants showed altered phenotypes from tuber stage 3 onwards (swelling stolon) (Figure 8a–d, e–h). The starch morphology in amfGS26 appeared to follow a transient pattern. There were 50 ± 2% of starch granules that showed deep pores at tuber developmental stages 3 and 4; then, the starch morphology changed to porous (20 ± 1%) with amalgamated granules (~2%) coexisting from stage 5 onwards (Figure 8a–d). Starches from KDGS26 transformant had more than 90% altered granules with small shallow pores on the granules' surface at tuber developmental stages 3 and 4 (Figure 8e,f); then, these pores seemed to be filled with amylose and turned to irregular granules from stage 5 onwards (Figure 8g,h) in the Kardal background.

Figure 8.

SEM analyses of starch granules from four different tuber developmental stages (stage 3, 4, 5, 6) of KDGS26 (a–d) and amfGS26 (e–h) at different magnifications. The scale bar indicates 10 μm in all images. Magnifications are a, b, e and f (×3500), c, d, g and h (×600).

Gene expression profiles altered in GLGB/SBD transformants

Altered starch granules are detectable in glgB transformants from potato tuber developmental stage three to mature tubers regardless of the genetic backgrounds. This result suggests that expression of GLGB/SBD fusion protein in potato tubers might affect the activity of enzymes involved in starch biosynthesis and/or in the process of starch granule assembly. To better understand how GLGB/SBD fusion protein interfered with the starch biosynthesis pathway, high-throughput transcriptomic analysis of potato tubers from KDGS26 and amfGS26 and respective untransformed controls was performed using the POCI array (Kloosterman et al., 2008). As shown in Figure 9, there were 307 genes (229 up-regulated and 78 down-regulated) that had at least a twofold change in expression level with a statistic P-value below 0.05 when KDGS26 tubers were compared with KDUT tubers. Meanwhile, there were 96 genes differentially expressed in amfGS26 transformant (82 genes were induced, and 16 were repressed). Seventeen genes were shared between both Kardal and amf backgrounds (Table 2), and from these differentially expressed genes, about 50% did not have an annotation.

Table 2. Differentially expressed genes shared in both KDGS26 and amfGS26 transformant tubers relatively to the untransformed control tubers
Gene nameamfGS26KDGS26DescriptionClass
P valueFoldP valueFold
  1. ‘+’, up-regulation; ‘−’, down-regulation; NA, no annotation.

SDBN004E09u.scf0.002+20.001+6NANot assigned
MICRO.6763.C10.001+60.001+5Putative chalcone isomerase 4 [Glycine max]Secondary metabolism
bf_arrayxxx_0055d11.t7m.scf0.001+40.001+4NANot assigned
STMES26TV0.001+30.001+3Patatin A gene, patatin B gene [Solanum tuberosum]Not assigned
MICRO.2037.C60.018+40.045+3Homologue to UP|Q40151_LYCES (Q40151) Hsc70, partial (32%)Protein folding
SSBN002H22u.scf0.001+30.001+3Weakly similar to UP|Q9M621_ARATH (Q9M621), partial (17%)Protein degradation
MICRO.7062.C10.011+20.001+3Weakly similar to UP|Q2V405_ARATH (Q2V405), partial (36%)Not assigned.glycine Rich proteins
POABI58TV0.034+20.003+3S-adenosylmethionine synthetase 2 (AdoMet synthetase 2), partial (65%)Amino acid metabolism
MICRO.11846.C10.001+30.001+2NANot assigned
bf_mxlfxxxx_0054a11.t3m.scf0.001+30.002+2NANot assigned
MICRO.4281.C10.008+20.003+2S-adenosylmethionine synthetase 2 (AdoMet synthetase 2), partial (98%)Amino acid metabolism
MICRO.1805.C20.007+20.001+2Weakly similar to UP|Q60EC7_ORYSA (Q60EC7), partial (19%)Not assigned
bf_ivrootxx_0003a09.t3m.scf0.015+20.004+2NANot assigned
bf_arrayxxx_0003a04.t7m.scf0.006+20.007+2NANot assigned
BF_TUBSXXXX_0038F01_T3M.SCF0.025+20.014+2NANot assigned
MICRO.15472.C10.001+20.001+2NANot assigned
MICRO.15185.C10.001−20.001−3Similar to MADS-box transcription factor CDM36, partial (48%)Mitochondrial electron transport/ATP synthesis
Figure 9.

Differentially expressed genes in KDGS26 and amfGS26 in comparison with their untransformed control tubers. The number in the overlapping area indicates the shared number of genes in both KDGS26 and amfGS26. Numbers outside the overlapping area represent genes specifically up- or down-regulated in KDGS26 and amfGS26.

Most of the genes that code for enzymes directly involved in starch and sucrose metabolism such as ADP-glucose pyrophosphorylases (AGPase), starch synthases (SSs and GBSS), starch-branching enzymes (SBEs) and α-glucan water dikinases (PWD and GWDs) did not show any changes in KDGS26 and amfGS26 tubers in comparison with the expression in tubers of their untransformed controls. Noteworthy, there was a beta-amylase gene (Micro. 13368. C1) that showed a more than twofold up-regulation in KDGS26 transformant tubers in comparison with Kardal untransformed tubers, but this difference was not observed in amfGS26 transformant tubers.

Quantitative real-time PCR (qRT-PCR) was performed on the beta-amylase gene (Micro. 13368. C1) in both KDGS26 and amfGS26 transformants as well as in the untransformed control tubers. The results showed that the beta-amylase gene had similar expression patterns as in the microarray analysis (data not shown).

Starch properties remain unaltered

To investigate the expected increase in branching, chain length distribution was determined by both high-performance size-exclusion chromatography (HPSEC) and high-performance anion-exchange chromatography (HPAEC). No differences were observed between starches of transgenic plants and their corresponding control plants after complete debranching by isoamylase treatment. Furthermore, the HPAEC profiles of the debranched starches from the transformants after treatment with isoamylase did not deviate from starches of untransformed controls (data not shown).

Other starch characteristics such as starch gelatinization behaviour, apparent amylose content (%AM) and starch granule size distribution were also investigated, and the results are shown in Table 1. No significant differences were found in these analyses in comparison with the control plants.

Discussion

The E. coli glgB gene fused with SBD has been targeted into both Kardal and amf potato starch granules during starch biosynthesis. The objective of expressing the glgB gene in potato was to introduce more α (1→6) linkages during amylopectin synthesis, which then would result in an altered amylopectin structure with higher branching degree in potato starch. The results showed that the chain length profiles of starches in transgenic plants were comparable to the untransformed control plants, which indicates that GLGB fused to SBD either does not introduce more α (1→6) linkages or that the number of newly introduced side chains is too small to be detected. However, our findings confirmed a previous finding that the expression of the E. coli glgB gene under control of the patatin promoter did not increase the amylopectin branching degree in amylose-containing genetic background (Krohn et al., 1994), even though the glgB gene was fused with the SBD domain. Kortstee et al. (1996) has on the other hand reported an increase in branching on 5 of the 34 transgenic plants, in amf background, containing the E. coli glgB gene.

The accumulation of GLGB/SBD fusion proteins in this study resulted in altered starch morphology (Figures 4 and 5). The porous and amalgamated starch granules were found in amf genetic background and the irregular bumpy surfaces with protrusions appeared in Kardal transformants (Figures 4 and 5). However, the physico-chemical properties of the starches from GLGB/SBD transformants are not affected (Table 1). The gelatinization behaviour was unaltered, as well as the granule size distribution and the apparent amylose content.

The GLGB/SBD fusion protein has high affinity to starch granules. In this study, regardless of the background (Kardal or amf) and construct (GLGB-SBD or SBD-GLGB) used, more than 85% of the transformants accumulated the highest amount of the fusion proteins (4 +  and higher) (Figure 2). This suggests that the affinity of GLGB/SBD fusion protein to starch is higher than for other proteins, such as GtfICAT/SBD and MAT/SBD (Nazarian Firouzabadi et al., 2007a,b) and higher than multiple SBDs (Nazarian Firouzabadi et al., 2007c). The E. coli GLGB belongs to the glycoside hydrolase (GH) family 13 and catalyses the formation of α-1,6 branch points in either glycogen or starch. Previous studies on the structure of GLGB enzyme showed that it consists of three major domains: an amino-terminal domain, a carboxyl-terminal domain and a central α/β-barrel domain containing the enzyme active site (Abad et al., 2002). GLGB on its own is able to bind to starch as has shown by Western blot analysis (Kortstee et al., 1996) of transgenic potato plants containing the glgB gene in amf background. Recent studies confirmed that the amino-terminal domain of GLGB belongs to the carbohydrate binding module family 48 (CBM48), which is a starch-binding domain (SBD) (Koay et al., 2007; Palomo et al., 2009). Although the 3D structure and binding site of CBM48 have not been characterized thoroughly, it is possible that the two SBDs in GLGB/SBD fusion protein (CBM48 and SBD) could bind starch granules similarly as the double SBD. Previous study showed that the tandem SBDs have much higher affinity to starch granules than a single SBD (Ji et al., 2004). These results support that high accumulation level of GLGB/SBD fusion proteins in the starch granules of transgenic plants might be explained bythe joint effects of the CBM48 domain of the GLGB enzyme with the extra fused SBD domain.

The alteration in starch granule morphology is unlikely caused by the activity of the GLGB enzyme on glucan branches, but it probably results from the deposition of GLGB/SBD fusion protein in starch granules, which interferes with the normal assembly of amylopectin during amylopectin biosynthesis. The starch granules' morphology is although different from the one observed in transgenic plants containing one SBD or tandem SBD domains (Nazarian-Firouzabadi et al., 2012), indicating that GLGB also plays a role. As discussed earlier, the GLGB/SBD fusion protein has three domains, a CBM48 domain, an SBD domain and the active site of GLGB enzyme, which can bind on the glucan branches during starch synthesis. The extra SBD domain fused with GLGB could make the GLGB/SBD fusion protein comparable to a double SBD during starch biosynthesis. Based on the model proposed by Nazarian Firouzabadi et al. (2007c), the exposed SBD domain of the double SBD can cross-link different granule nucleation sites and eventually result in amalgamated starch granules at a higher concentration. The GLGB/SBD fusion protein might operate similarly to a double SBD, which is supported by the comparable starch phenotypes, such as amalgamated starch granules and multiple small starch granules in the centre of the big granules in the starch granule section images (Figure 7c,f). Although the porous starch granules in amf background and the irregular starch granules in Kardal background show growth rings, the distribution of the growth rings is not as evenly as in the untransfomed control starches (Figure 7d). This indicates that the GLGB/SBD fusion protein did not affect the crystallization of amylopectin but the distribution of amylopectin structure in the granules during the starch granule formation process. The deposition of the GLGB/SBD fusion proteins on the starch surface which could be released later on and gave rise to starch granules perforated on the starch surface. These starch morphological phenotypes can be detected clearly from the tubers of transformants already at the initial stages of potato tuber development. During starch granule development, the holes are being filled with amylose later on in the amylose-containing genetic background such that amylose is protruding from these holes and some of amylose is deposited on the granule surface, which eventually gives rise to bumpy surfaces as can be seen in Figure 4, whereas the holes turn deeper in amylose-free background because of the absence of amylose.

Although the starch morphology has been severely altered in this study, the expression of GLGB/SBD fusion protein did not affect the expression of most of the genes that code for enzymes directly involved in starch and sucrose metabolism. The transcription level of starch synthases (SS and GBSS), starch-branching enzymes (SBE) and starch debranching enzymes (isoamylases) did not show any changes in KDGS26 and amfGS26 transformant tubers. Interestingly, a β-amylase gene was up-regulated in KDGS26 transformant tubers but not in amfGS26 transformant tubers. The β-amylase is an exohydrolase that catalyses the liberation of β-maltose from the nonreducing end of α-1,4-glucans, leaving a β-limit dextrin (Baba and Kainuma, 1987). A primary role of β-amylase is in hydrolytic transitory starch degradation (Fulton et al., 2008; Scheidig et al., 2002; Weise et al., 2005). Besides the important role in starch degradation, β-amylase is also involved in amylopectin biosynthesis by hydrolysing α-1,4-glucans before they form a crystalline structure (Delatte et al., 2005). Streb and co-workers proposed a model which suggests that the isoamylases remove miss-positioned branches to make crystallize starch granules during amylopectin synthesis and the miss-positioned branches that are cleaved off by isoamylases are degraded into maltose by beta-amylase and re-incorporated in starch biosynthesis pathway (Streb et al., 2008). The up-regulation of β-amylase and debranching enzymes (isoamylase 1 and isoamylase 2) could potentially explain the inexistence of extra branches. In this model, the GLGB could possibly bring extra branches on the amylopectin backbone structure during amylopectin biosynthesis, while the debranching enzymes and β-amylase would remove these extra positioned branches, ending up with no additional branching. In cereals, proteins have been reported to play a role in the morphology of starch granules' surface (Naguleswaran et al., 2011).

These novel starches create new opportunities for different industry. In particular, the perforated starch granules have potential interest for different industrial applications. One obvious application is in pills, where these granules could be used as vehicles of certain compounds, which are required to be released only in certain parts of the GI track. Another interesting aspect to explore in future research is the degree of substitution of these starches after derivatization, which might be higher than in wild-type starch.

Experimental procedures

Preparation of constructs

The E. coli glgB was fused in frame to the SBD of Bacillus circulans cyclodextrin glycosyltransferase (CGTase) gene at both its carboxyl (C) and amino (N) terminus. Potato GBSSI promoter segment and its transit peptide (TP) sequence were used for tuber-specific expression and amyloplast entry of the proteins (Figure 1). Two constructs, the pBIN19/glgB-SBD and pBIN19/SBD-glgB plasmids, were used for the expression of the glgB-SBD and SBD-glgB fusion proteins in potato plants (Kardal and amf), respectively.

To generate the pBIN19/glgB-SBD plasmid, a glgB-encoding fragment was obtained by PCR amplification with the primers 5′-CAACCATGGCCGATCGTATAG-3′ and 5′-TTAGATCTCTCTGCCTCCCGAAC-3′, which contained NcoI and BglII sites at their 5′ ends, respectively. This amplified glgB fragment was used to replace the first SBD fragment (also NcoI-BglII) in the pUC19/SBD2 (Ji et al., 2003), giving the pUC19/glgB-SBD plasmid. After digestion of this plasmid with HpaI and SacI, the HpaI-SacI fragment was inserted into the corresponding sites of the pBIN19/SBD2 to generate the pBIN19/glgB-SBD. The predicted molecular mass of the GLGB-SBD produced in plants is 98,424 Dalton, excluding the transit peptide. The pBNI19/SBD-glgB plasmid was assembled from below fragments: (i) GBSSI promoter, GBSSI transit peptide and SBD-linker fragment (HindIII-SalI) obtained from pUC19/SBD2 plasmid, (ii) a glgB fragment (SalI-XbaI) and (iii) a NOS terminator (BamHI-KpnI). The glgB fragment was amplified by PCR with 5′-GTGTCGACCATGGCCGATC-3′ and 5′-TCTAGAGTCATTCAGCCTCCCG-3′ as primers, which contained SalI and XbaI at their 5′ ends. The predicted molecular mass of the SBD-GLGB produced in plants is 98,089 Dalton, excluding the transit peptide. All constructs were controlled by sequencing analysis.

Potato transformation

The two different constructs were transformed into electrocompetent cells of Agrobacterium tumefaciens strain LBA4404 by electroporation. Furthermore, two different potato genotypes, namely wild-type potato (cv. Kardal, tetraploid) and the mutant amylose-free (amf, diploid), were used for Agrobacterium-mediated transformation. Potato transformation procedures were carried out according to Visser (1991). For each construct, 30 independent transformants were harvested from MS30 selection medium (Murashige and Skoog, 1962), including kanamycin (100 mg/L), and later, each was propagated to five plants to obtain enough tubers.

Plant materials

In vitro plantlets of all different transformants, KDGS, KDSG, amfGS and amfSG, and their untransformed controls (Kardal and amf) were grown in MS medium for 2 weeks and transferred to the greenhouse and maintained in a climate chamber for tuber development. Tuberization in potato is a highly unsynchronized process, meaning that in the same plant tubers from different developmental stages, from stolons to mature tubers, can be found simultaneously (Trindade et al., 2003). To access different tuber developmental stages, tubers of selected transformants (KDGS26 and amfGS26) and untransformed controls were collected and classified into five developing tuber stages according to their tuber (and starch granule) size (stage 3, tuber initiation about 0.2 cm in diameter; stage 4, tuber size 0.2 cm; stage 5, tuber size about 0.5 cm; stage 6, tuber size about 1 cm; stage 7, tuber size around 2 cm) as in Kloosterman et al. (2005) and frozen in liquid nitrogen immediately. Tubers from all stages were harvested 3–4 h into the light period. Three biological replicates for each stage sample and each specific sample were harvested from at least five individual plants.

Isolation of tuber starches

All tubers form each individual transformant (five clones) were peeled, mixed and ground using a Sanamat rotor (Spangenberg, Oosterhout, the Netherlands) to which 0.01% of sodium metabisulphite (Na2S2O5) was added. Collected juices were taken to 4 °C to settle for at least 3 h. A part of the juice was kept and frozen at −20 °C for further analysis. The settled starch was washed at least three times and passed through a Whatman filter paper on a Buchner funnel. All starch samples were later air-dried at room temperature and powdered with a sieve shaker (Retsch, Haan, Germany).

For the potato tuber developmental series, tubers at each stage from five plants were pooled and cut in pieces and putted in small centrifuge tubes. 1.0 mL 3% Rapidase in 0.2 m NaAc with 0.5 g/L Na2SO3 (pH = 5.5) was added to tubes and incubated overnight at 30 °C. Starches were collected by centrifuging 2 min (800 g) and removing the cell wall rests using tweezers. The starch sediment was washed three times with 30 mm phosphate buffer (pH = 7.0), one time with demi-water, one time with acetone and finally air-dried at room temperature.

The tubers used for starch analysis were all produced in the green house starting from in vitro plantlets. For the potato tuber developmental series, plants have been generated both from in vitro plantlets and from tubers. No differences were observed between the material produced from in vitro plantlets and from tubers.

Western dot blot analysis

Western dot blot analyses, using gelatinized starch, were performed for SBD-containing fusion proteins. The amount of accumulated protein was quantified as described by Ji et al. (2003). To visualize the fusion proteins, an aliquot of 1 : 500 dilution of anti-SBD (Ji et al., 2003) was used as primary antibody. The soluble fractions were also subjected to Western dot blot analyses according to Ji et al. (2003).

Immunoblot analysis of granule-bound fusion proteins

Sodium dodecyl sulphate–polyacrylamide gel electrophoresis (SDS-PAGE) was carried out as described by Nazarian Firouzabadi et al. (2007a,b). Fifty milligrams of starch sample from selected transformants was solubilized in 1 mL SDS sample buffer [final concentrations: 1 m Tris-HCl, pH 6.8, 2%(w/v) SDS, 10% (v/v) glycerol, 3% (v/v) β-mercaptoethanol] and incubated at boiling temperature for 5 min before being analysed on a gel. Samples were centrifuged, and initially, equal amounts of control and transformants (25-μL supernatant) were loaded onto lanes of a 12% polyacrylamide separating gels (145 mm × 95 mm × 3 mm; BioRad, Leicester, UK). The Precision Plus Protein (Bio-Rad) ladder was used as size standard. Proteins were then transferred onto a Hybond ECL (Amersham, Buckinghamshire, UK) membrane and immunoblotted using either anti-SBD or anti-GLGB antibodies as described by Nazarian Firouzabadi et al. (2007a,b).

Analysis of physico-chemical properties of different transformants

Average granule size and granule size distribution of the starches were determined with a Coulter Multisizer II, equipped with an orifice tube of 200 μm (Beckman-Coulter, High Wycombe, UK). Approximately 10 mg of starch was dispersed in 160 mL of Isoton II. The granule size distribution was recorded by counting approximately 50 500 (±500) particles. The coincidence (the frequency of two granules entering the tube at the same time and consequently being counted as one) was set at 10%.

Starch granule morphology was investigated by light microscopy (LM; Axiophot, Zeiss, Oberkochen, West Germany) and scanning electron microscopy (SEM, JEOL 6300F; SEM Tech Solutions, Tokyo, Japan). For LM, starch granules were stained with a 20× diluted Lugol's solution (1% I2/KI).

The starch granules were mounted onto brass holders with double-sided sticky carbon tape (EMS, Washington, DC). The sample holder was placed in a dedicated preparation chamber (Oxford Instruments CT 1500 HF; Eynsham, UK) and sputter coated with 10 nm platinum. Specimens were analysed with a field emission scanning electron microscope (JEOL 6300 F) at room temperature at a working distance of 15 mm, with SE detection at 2.5 kV. All images were recorded digitally (Orion, 6 E.L.I. sprl., Labuissière, Belgium) at different scan rates depending on the sample charging. The images were optimized and resized with Adobe Photoshop CS.

To test the chain length distribution, 5 mg of starch from transgenic potatoes was suspended in 250 μL of DMSO and gelatinized for 10 min at boiling temperature. After cooling down to the ambient temperature, 700 μl of 50 mm NaAc buffer pH 4.0 was added. A sufficient amount of isoamylase (Hayashibara Biochemical laboratories, Okayama, Japan; 59 000 U/mg protein) to debranch the starch polymers completely was added to the mixture, which was incubated for 2 h at 40 °C. After inactivation of the enzyme for 10 min at boiling temperature, 1 mL of 25% DMSO was added. For HPSEC, the samples were used as such (HPSEC). The HPSEC was performed on a p680HPLC pump system (Dionex, Sunnyvale, CA) equipped with three TSKgel SWXL columns in series (one G3000 and two G2000: 300 mm × 7.5 mm; Montgomeryville, PA) in combination with a TSKgel SWXl guard column (40 mm × 6 mm) at 35 °C. Aliquots of 100 μL were injected using a dionex ASI-100 Automated Sample Injector and subsequently eluted with 10 mm NaAc buffer (pH 5.0) at a flow rate of 0.35 mL/min (3-h run). The effluent was monitored using a RID-6A refractometer (Shimadzu, Kyoto, Japan). The system was calibrated using dextran standards (10, 40, 70, 500 kDa; Pharmacia, Uppsala, Sweden). Dionex Chromelon software version 6.50 SP4 Build 1000 was used for controlling the HPLC system and data processing.

High-performance anion-exchange chromatography was used to obtain a better separation of the smaller amylopectin side chains (in the range of 2–45 glucose residues). HPAEC was performed on a GP40 gradient pump system (Dionex) equipped with a carboPac PA 100 column (4 mm × 250 mm; Dionex) at 35 °C. The flow rate was 1.0 mL/min, and 20-μL samples was injected with Dionex AS3500 automated sampler. Two eluents were used, eluent A (100 mm NaOH) and eluent B (1 m NaAc in 100 mm NaOH), for mixing the following gradient: 0→5 min, 100% eluent B (rinsing phase); 5→20 min, 100% eluent A (conditioning phase); 20→25 min, linear gradient from 0→20% eluent B (100→80% eluent A); 25→50 min, linear gradient from 20→35% eluent B (80→65% eluent A); 50→55 min, linear gradient from 35→50% eluent B (65→50% eluent A); 55→60 min, 50% eluent B (50% eluent A). The sample was injected at 20 min. The eluent was monitored by ED40 electrochemical detector in the pulsed amperometric mode (Dionex).

The temperature at which starch granules start to gelatinize was determined by differential scanning calorimetry (DSC) using a Perkin-Elmer Pyris 1 equipped with a Neslab RTE-140 glyco-cooler. The instrument was calibrated with indium (melting point 156.6 °C) and zinc (melting point 419.47 °C). Before DSC analysis, the moisture content of starch samples was determined by drying them overnight in an oven at 105 °C and weighing them before and after drying. Precisely, 10 mg of starch was transferred to a stainless-steel pan, and the starch content of the pan was adjusted to 20% by adding an appropriate amount of water. The pan was sealed and allowed to equilibrate for at least 1 h. The samples were heated from 40 to 100 °C at a scanning rate of 10 °C/min. An empty sample pan was used as a reference. For each endotherm, the onset temperature of gelatinization (To), as well as the difference in enthalpy (ΔH), was automatically calculated by the software.

Analysis of starch structure

Five mg of starch from selected transformants (Figure 4) was incubated in two different ways. Assay (a): Starches were incubated in 100 mm phosphate buffer pH = 7.5 and incubated for 5 days at 40 and 55 °C. Phosphate buffer was removed, and isoamylase treatment for HPSEC and HPAEC analyses was performed as mentioned in the previous section. Assay (b): To create enough space for the granule-bound GLGB, five milligrams of starch was solubilized in 250 μL of DMSO, 800 μL of 10 mm phosphate buffer was added and the mixture was incubated for 48 h at 30 °C. Hundred micro liter of 1 m NaAC was added followed with sufficient amount of isoamylase (59 000 U/mg protein; Hayashibara Biochemical laboratories). To keep the DMSO at 25% final concentration, 550 μL of distilled water and 250 μL of DMSO were added. The HPSEC and HPAEC analyses were performed as mentioned earlier.

Sample preparation and hybridization for microarray and data analysis

RNA extraction and POCI microarray hybridization were performed as described by Kloosterman et al. (2008). Furthermore, analysis of differential expressed genes using the MapMan tool (Thimm et al., 2004) (http://gabi.rzpd.de/projects/MapMan/) was conducted to understand biological processes that were affected by expression of GLGB/SBD fusion protein. Differences between averages and P-values for the t-tests were performed in Microsoft Excel. Log2 ratios for all hybridizations were exported into Genstat® v11 (VSN International, Hemel Hempstead, UK) for statistical analysis and calculation of expression estimates and standard errors for each feature. Calculated estimates and standard errors of each feature were finally imported into the GenemathsXT 1.6 (Applied Maths, St-Martens-Latem, Belgium) software package for additional analysis.

Genes with a log2 expression ratio ≥1 (expression ratio twofold) or less than or equal to −1 (expression ratio 0.5-fold) in combination with a P-value < 0.05 were considered significantly different.

Acknowledgements

This work was partly financed by the Ministry of Science, Research & Technology of Iran. We are grateful to Mrs. Isolde Pereira and Mr. Dirkjan Huigen for the tissue culture and greenhouse assistance, respectively. We also wish to thank Dr. Vic Morris and Dr. Mary Parker of the Institute of Food Research (UK) for starch section imaging. We are thankful to Dr. Bjorn Kloosterman who kindly helped with microarray experiments and Dr. Chris Maliepaard for his advice on statistic data analysis.

Ancillary