An En/Spm based transposable element system for gene isolation in Arabidopsis thaliana

At the start of the research described in this thesis, the main aim was to develop, study and apply an efficient <em>En/Spm-I/dSpm</em> based transposon tagging system in <em>Arabidopsis thaliana</em> to generate tagged mutants and to provide insights in the possibilities for future applications of such a transposon tagging system in studying plant gene functions. The first step was the introduction of an active <em>En/Spm-I/dSpm</em> system into Arabidopsis. Initially a very simple T-DNA construct was transformed, containing a nearly full <em>En-1</em> element, without left and right border sequences, and with its promoter replaced by the stronger CaMV 35S promoter. As the same construct harboured a non-autonomous <em>I/dSpm</em> element, only one T-DNA transformation was needed. Transformation of this <u>'in cis</u> two- element <em>En/Spm-I/dSpm</em> system' yielded one transformant with two T-DNA insertion loci, TEn2 and TEn5, each having one, respectively, five 35S <em>-En/Spm t</em> ransposase gene copies (Chapter 1). The transposition activity of the <em>I/dSpm</em> elements turned out to be surprisingly high. Instead of a germinal excision frequency, which was generally used to express the activity of heterologous transposable element systems, the term <u>independent transposition frequency</u> or itf was coined, as a measure accounting for the entire transposition process. Although not always easy to compare for different transposable element systems, an average itf of over 10%, as was found for this <em>En/Spm-I/dSpm</em> system (Chapter 1), has not been reported for any of the <em>Ac-Ds</em> based heterologous transposon tagging systems developed for Arabidopsis (Bancroft et al., 1992; Bhatt et al., 1996; Fedoroff and Smith, 1993; Honma et al., 1993; Long et al., 1993b; Swinburne et al., 1992). Obtaining a high transposition frequency in <em>Ac-Ds</em> systems is hampered by the fact that the transposase acts as an inhibitor of <em>Ac</em> or Ds transposition when its expression exceeds a certain level. Apparently such autoregulatory mechanism is not present in the <u>'in cis</u> two-element <em>En/Spm-I/dSpm</em> system'.The need for a more sophisticated system diminished with the availability of this simple, but efficient <em>En/Spm-I/dSpm</em> transposon system and it was therefore studied in more detail to determine: 1) the ability to transpose continuously even after many plant generations; 2) the distribution of elements after transposition; 3) the ability to transpose to transcribed regions and 4) the ability to cause mutations. To start with the first issue, transposition has been studied in up to 12 generations, starting from the primary transformant. In all these generations, there was no apparent reduction in itf, demonstrating a continuous transposition of <em>I/dSpm</em> elements, irrespective of the generation number. The second issue, the distribution of elements after transposition, is another important aspect of a transposable element system. For maize transposons it is reported that the insertion site is preferentially physically and often genetically linked to the excision site (Dooner and Belachew, 1989; Peterson, 1970). This is not so remarkable considering that the proteins that perform the transposition steps will have a higher chance of encountering a nearby site on the genome instead of a distant DNA sequence. Like in maize (Peterson, 1970), the <em>I/dSpm</em> elements show a preference for insertion in genetically linked sites (Chapter 1), but the preference is not very explicit. Based on the observations of transpositions from tagged genes to sites within a few cM or only several kb away, and on the analogy to the <em>En/Spm-I/dSpm</em> elements in maize, the overall estimate is that about 30% of the elements transpose to sites genetically linked to the excision site. The mapped elements (Chapter 5) show a fairly even distribution over the genome, although there seems to be some clustering of elements (from different origin) to certain genomic regions, like the top of chromosome 4, the bottom of chromosome 1 and the lower half of chromosome 5.In accordance with the idea that DNA must be in an open confirmation to allow the access of transposase proteins before transposition (Zhang and Spradling, 1993), there are many indications that <em>I/dSpm</em> elements insert in regions of the Arabidopsis genome containing genes: a) <em>I/dSpm</em> flanking DNAs rarely contain repetitive sequences, but are mostly single copy sequences, as are genes (Chapters 1 and 5); b) about half of the <em>I/dSpm</em> elements are inserted in relatively conserved genomic regions, with no RFLPs for five restriction enzymes (Chapters 1 and 5); c) at least one third of the examined <em>I/dSpm</em> elements is inserted in close vicinity of transcriptionally active genomic sequences (Chapter 7). Insertion in unique, conserved and often transcribed DNA may not seem surprising for a plant species with little repetitive DNA and a small genome with a high gene density (Meyerowitz, 1989). However, a high frequency of insertion into genic regions of the genome offers the best chances for gene tagging.The most important aspect of a transposable element system is the possibility to generate tagged mutants. The <em>En/Spm-I/dSpm</em> system is mutagenic, with as much as 12 tagged mutants found so far. Most of these were obtained after screening for random mutant phenotypes. When screening for specific mutants, such as reduced seed dormancy, tagged alleles of the ABI3 and <em>LEC1</em> genes were found (M. Koornneef et al., unpublished results). Mutants at the <em>CER6</em> locus were obtained by targeted tagging, using the nearby <em>ap1::I/dSpm</em> allele as the <em>I/dSpm</em> element donor (A. Pereira, unpublished results). These selected and targeted transposon tagging experiments are illustrative examples of the feasibility to efficiently isolate tagged mutants of a special phenotypic or genotypic class.Ideally, a population saturated with different <em>I/dSpm</em> insertions can be made, allowing the isolation of mutants for virtually every gene. Such a 'mutation machine' can be further used for PCR based targeted gene inactivation. This novel technique, which was originally developed for <em>Drosophila melanogaster</em> (see O'Hare, 1990), exploits the abundance of transposons for the identification of insertions in genes with known DNA sequence, but no known mutant phenotype. In general, DNAs from multidimensional pools of individuals from a large population are used for a PCR using two primers. One primer is specific for the transposon terminus (directed outwards), the other is specific for the target gene. A fragment can only be amplified when a transposon insert is close enough to the target primer. This technique has been shown to work in <em>Drosophila</em> (Ballinger and Benzer, 1989; Kaiser and Goodwin, 1990), <em>Caenorhabditis elegans</em> (Zwaal et al., 1993), <em>Petunia hybrida</em> (Koes. et al., 1995) and maize (Das and Martienssen, 1995), using 'mutation machine' transposable element systems.<em>Two I/dSpm</em> tagged genes that have been studied in great detail are the MS2 gene (Chapters 2 and 3) and the <em>CER1</em> gene (Chapter 4). Mutants for both genes display a conditional male sterile phenotype, which is the only mutant phenotype for the <em>ms2</em> mutant, but for the <em>cer1</em> mutant it is a pleiotropic effect of a deficiency in epicuticular wax biosynthesis. The ms2 mutants are occasionally able to self-fertilize, especially in high relative humidity and late in plant development, but seed set rarely reaches more than a few percentages of wild-type seed set. This in contrast to the <em>cer1</em> mutants, which are male sterile in low relative humidity (≤50% RH) and fertile in high relative humidity (≥95% RH). Fertility cannot be completely restored by environment in the <em>ms2</em> mutants due to the drastic effect of the mutation on pollen development. The <em>MS2</em> gene is expressed in the tapetum around the time of microspore release from the microspore mother cells. The gene is needed for the development of a proper exine layer, protecting the microspore from harmful environmental influences. Consequently the few <em>ms2</em> microspores that are produced have very feeble pollen walls, which leaves only very few pollen grains intact for fertilization.The <em>CER1</em> gene acts much later in pollen development. Phenotypically <em>cer1</em> and wild-type pollen cannot be distinguished, apart from a difference in gerniination ability (Chapter 4). As <em>cer1</em> pollen germination is like wild type when applying <em>CER1</em> pollen or by pollinating in high relative humidity, there appears to be a substance missing from the pollen coat that is required for the necessary rehydration of a pollen grain. Although essential under low relative humidity conditions, this is only a minor defect, which can be easily overcome.Besides the similarity in mutant phenotypes, the <em>MS2</em> gene and the <em>CER1</em> gene share the characteristic of encoding proteins with homology to enzymes in the fatty acid biosynthesis pathway. The MS2 protein has most resemblance with the wax fatty acid reductase protein from the desert shrub jojoba, which is involved in the conversion of wax fatty acids to wax alcohols (Chapter 3). The CER1 protein shares structural features of fatty acid desaturases, and it has a proposed function as a decarbonylase, converting long carbon chain aldehydes to alkanes (Chapter 4). There are more examples of a correlation between male fertility and wax biosynthesis, as also other <em>cer</em> mutants such as <em>cer3, 6, 8</em> and 10 are known to be disturbed in male fertility. It demonstrates the general importance of fatty acid biosynthesis for male gametogenesis.The last part of this thesis has been devoted to further applications of the <em>En/Spm-I/dSpm</em> tagging system in Arabidopsis for the analysis of plant gene functions. The first description of transposable elements as controlling elements (McClintock, 1948), was based on the effect transposons had on the expression of maize genes. Especially <em>(d)Spm</em> insertions were known to cause <em>Spm</em> dependent or suppressible gene expression (Fedoroff, 1989). This effect is now also described for the <em>En/Spm- I/dSpm</em> system in Arabidopsis (Chapter 6). As in maize, an <em>En/Spm</em> suppressible allele contains an anti-parallel <em>I/dSpm</em> element insertion, which can be spliced from the mRNA. This knowledge can be further used to design an artificial gene expression system, in which an introduced gene containing an anti-parallel <em>I/dSpm</em> element, can be negatively controlled in the presence of an <em>En/Spm</em> transposase source. The reverse effect, <em>En/Spm</em> dependent gene expression, is now also described for Arabidopsis (Chapter 6), but the mechanism for dependence is not yet understood. The availability of the <em>En/Spm</em> dependent <em>lad::I/dSpm</em> mutant will be useful for further research.A more general way to study gene expression is the use of gene traps as detectors of gene activity. The pilot experiments described in chapter 7, have shown the possibility to adapt <em>I/dSpm</em> elements as gene traps. Especially their advantage in detecting genes without the need for mutation, and the possibility of studying the activity of genes which are lethal as homozygous mutants, are important additional properties of gene trap systems over "traditional" transposon tagging. In combination with its efficient transposition behaviour, an <em>En/Spm-I/dSpm</em> based gene trap tagging system seems an attractive alternative for the existing <em>Ac-Ds</em> or T-DNA based gene trap systems.Summarizing, an <em>En/Spm-I/dSpm</em> transposon tagging system has been well developed for Arabidopsis and many of its basic characteristics are studied and described. <em>I/dSpm</em> tagged mutants can be found with reasonable frequencies, either by random, selected or targeted tagging strategies. The cloning and characterization of two genes affecting male fertility has been described. Further ways to improve tagging frequencies, based on phenotypic or on genotypic selection have been discussed. In addition, the system can be exploited to study plant gene expression and gene function either by <em>En/Spm</em> controlled activity of <em>I/dSpm</em> tagged genes, or by using <em>I/dSpm</em> gene detector elements.