||In the past, antisense RNA technology was used to modify the composition of potato tuber starch. Potato starch comprises amylose and amylopectin, polymers of glucose. Amylose production in potato is completely dependent on the presence of granule-bound starch synthase I (GBSSI). Inhibition of GBSSI has been achieved by transformation with antisense and sense GBSSI constructs. However, the percentages of transformants showing strong silencing were relatively low which implicated that large numbers of transformants needed to be generated in order to obtain sufficient transformants showing strong silencing. This was an undesirable situation since in practical breeding many transformants would have to be made, making the effort too time consuming and costly. Therefore, an approach that would enhance the silencing efficiency was desired.This thesis describes the application of inverted repeat constructs for the modification of potato tuber starch. Transcription of inverted repeat constructs results in the formation of double stranded RNA (dsRNA). These dsRNA molecules are cleaved in small interfering RNAs (siRNAs) by a dsRNA-specific nuclease named Dicer. Subsequently, one strand of the siRNA is incorporated into the RNA Induced Silencing Complex (RISC) which is guided to the target mRNA through conventional base-pairing interactions. The target mRNA is then cleaved opposite the centre of the guide siRNA and finally, the cleaved mRNA is degraded.The antisense strand of the siRNA can also act as a primer on target mRNA after which new dsRNA is synthesized by the action of a cellular RNA-directed RNA polymerase (RdRP). The RdRP-synthesized dsRNA will be recognized by Dicer and degraded to secondary siRNAs. These secondary siRNAs can be derived from regions upstream of the targeted regions and can thus induce RNA silencing of sequences that were not initially targeted. This phenomenon is called transitive RNA silencing.First, an extensive study on the Post-Transcriptional Gene Silencing (PTGS) effects of eight different cDNA inverted repeat constructs was performed. These cDNA inverted repeat constructs were designed in such a way that the effect of size, orientation and sequence could be determined. The orientation of the inverted repeat did not affect silencing efficiency but the size and sequence did. Small inverted repeat constructs with a repeat size of 500-600 bp were more efficient silencing inducers than the large inverted repeat constructs with a repeat size of 1.1 or 1.3 kb. The two large inverted repeat constructs comprised the5'and the3'half of the GBSSI cDNA whilst for the construction of the small inverted repeat constructs the GBSSI cDNA was divided in three regions; the5', the middle and the3'region. In both instances, the3'sequences induced the least efficient silencing implying that sequence does influence silencing efficiency. The transformants showing the highest percentages of strong silencing were obtained with two small inverted repeat constructs derived from the5'or middle region of the GBSSI cDNA. In both cases, 87 % of transformants showed strong silencing. Silencing was accompanied by a reduction in GBSSI mRNA levels but no relation between the amount of transcript and the level of silencing could be demonstrated. A similar phenomenon was observed for the accumulation of siRNAs. Although there was a relation between the presence of siRNAs and silencing, no relation between the level of siRNAs and the level of silencing was found.Another factor we tested was the influence of introns in the cDNA sequences represented in the small inverted repeat constructs. To test the effect of GBSSI intron number two, two constructs were made; one with the intron in the cDNA and one without intron number two in the cDNA. The same approach was used to address the importance of GBSSI intron number nine, which was tested in two inverted repeat constructs harbouring sequences from the middle region of the GBSSI cDNA. For both introns, no effect on silencing efficiency was found. This indicates that the introns neither enhance nor inhibit silencing.Many general silencing vectors are designed with a spliceable intron in the spacer because of a general belief that the splicing of this intron enhances gene silencing. To address whether this would apply to our experimental system, we compared constructs that contained intron number nine in a spliceable or a non-spliceable orientation in the spacer sequence. Both constructs gave rise to similar silencing efficiencies. It is likely that intron number nine was spliced in the construct in which it was in the spliceable orientation, because we demonstrated that this intron was spliced in another construct where it was located in exactly the same surrounding sequences.Since we found no differences in silencing efficiency, we postulate that it is not a general rule that inclusion of a spliceable intron in the spacer of an inverted repeat construct enhances silencing efficiency. We think that the extent to which the presence of an intron in the spacer sequence enhances silencing efficiency might be different for each intron.In plants and C.elegans examples of transitive silencing, i.e. silencing of sequences that were not initially targeted, have been described. Since it was shown that potato has an RdRP homolog, we investigated whether transitive silencing occurred in potato. Two approaches were used to induce transitive silencing. A construct harbouringa NOSterminator inverted repeat preceded by the middle region of the GBSSI cDNA and transcribed by a GBSSI promoter, was tested for its ability to induce silencing of GBSSI. Silencing of GBSSI could be induced indicating that transitivity occurs in potato in a3'to5'direction. To address whether the endogenous GBSSI mRNA can function as a template for the synthesis of new dsRNA, transformants harbouring GBSSI cDNA IR constructs were analysed for the accumulation of secondary siRNAs derived from regions adjacent to the targeted regions. Secondary siRNAs upstream of the targeted region were detected indicating that, in potato, the endogenous GBSSI transcript can function as a template for the synthesis of new dsRNA. No transitivity in the5'to3'direction could be detected. In plants, transitive silencing has been observed in several transgenic systems but very few examples of transitive silencing of endogenous genes exist. In these examples, it was not investigated whether the endogenous or transgene mRNA was used as a template. To our knowledge, we are first to show that endogenous mRNA is used as a template in transitive silencing of endogenous genes in plants. We also showed that the transitivity could extend over a distance of 761 bp, which is the largest distance reported until now.When an inverted repeat construct contains promoter sequences, transcriptional gene silencing (TGS) can be induced. In this case, alterations at the DNA or chromatin level prevent transcription of the targeted genes. An alteration at the DNA level is methylation of cytosine residues. This can be induced by RNA signals and is described as RNA-directed DNA methylation. Even though transformants obtained through the use of GBSSI cDNA inverted repeat constructs show strong silencing of GBSSI, there is still production of GBSSI mRNA. To obtain transformants in which the production of GBSSI mRNA was completely inhibited, we induced TGS of GBSSI by constructing GBSSI promoter inverted repeats. Three different regions of the GBSSI promoter were tested in inverted repeat constructs.In potato, four classes of GBSSI alleles are known. These four classes of alleles are highly homologous in the coding region, but vary in promoter sequences. The variation in promoter sequences was exploited to make an 'allele-specific' promoter inverted repeat construct. This construct contained the sequences from -531 to -330 bp relative to the transcription initiation site (TIS) in the GBSSI A2 allele. It was hypothesized that this construct would inhibit the expression of the A2, A3 and A4 GBSSI alleles, but not of the A1 GBSSI allele. However, the percentages of weakly silenced transformants obtained with this construct were too low to distinguish the effect between two potato cultivars that varied in allele composition. Therefore, this approach could not be used to induce allele-specific silencing.Weak silencing effects were also observed in 57-60 % of the transformants harbouring the inverted repeat construct with the promoter sequences from -766 to -168 bp relative to the TIS. We demonstrated that within this sequence, CG methylation had occurred, but apparently this was not sufficient to obtain strong silencing.The most efficient silencing efficiency was induced with the inverted repeat construct containing the promoter sequences from -766 to +194 bp, relative to the TIS. Very strongly silenced transformants were obtained in which the transcription of GBSSI was completely or almost completely inhibited. This was accompanied by the accumulation of promoter-specific siRNAs and CG and CNN methylation.Here, we showed that transcriptional silencing of the endogenous GBSSI promoter can be induced through the use of promoter inverted repeats. However, in order to obtain strong silencing, it is important to include sequences in the vicinity of the TIS. As compared to the PTGS approach, the TGS approach could be a good alternative to obtain very strongly silenced transformants that show a complete or almost complete inhibition of GBSSI transcription. Therefore, we think it is worthwhile to study whether the silencing effect in TGS transformants is stable during storage of tubers and in field experiments.