Analysis of the influence of promoter elements and a matrix attachment region on the inter-individual variation of transgene expression in populations of Arabidopsis thaliana

Abstract

Inter-individual variation of transgene expression represents a major bottleneck for high-throughput testing of transgene constructs in plants. More specifically, relatively large populations of first generation transgenic plants are generally required to evaluate transgene constructs with respect to desired expression level and related phenotype. Aiming at a reduction of this inter-transformant variability, in this study we systematically and statistically investigated the influence of different regulatory elements on the level and variability of transgene expression in populations of Arabidopsis thaliana plants. Starting from a basic gene construct consisting of the β-glucuronidase reporter gene (uidA) controlled by the most commonly used cauliflower mosaic virus (CaMV) 35S promoter, we investigated the effect of a matrix attachment region (MAR), 5′ untranslated regions and the use of different promoter and terminator sequences on the β-glucuronidase expression (GUS expression). It was found that MARs from the chicken lysozyme gene had no influence on the level of expression or on the variability of expression in populations of A. thaliana first generation plants. Moreover, transgene expression variability was not influenced by any of four types of terminators nor by any of two different 5′ untranslated regions. Promoters, as expected, drastically influenced expression levels but also rather unexpectedly markedly affected expression variability. A set of promoters derived from the mannopine synthase genes consistently yielded reporter gene expression with higher median levels and lower variance compared to the CaMV 35S promoter. The mannopine synthase promoter derivatives can therefore be useful for a range of applications for which constitutive expression with low inter-individual variability is desired.

Introduction

Genetic transformation of plants has been widely used over the past 15 years both as a method for studying gene function and as a tool for improving particular traits of plants. It was observed in the early days of plant transformation that the expression level of the transgene can vary considerably between different plants transformed with the same transgene construct [1], [2], [3]. This high inter-transformant variability is generally undesirable because it necessitates the screening of large numbers of transformants in order to identify individuals with an acceptable expression level and an expected phenotype. The time and effort spent on screening individual transformants represents a major bottleneck for high-throughput testing of transgene constructs.

Inter-transformant variability of expression levels is thought to be attributable to a number of factors. As the integration of foreign DNA in the plant genome occurs via illegitimate recombination [4], it can take place at virtually any site. As a consequence, sequences flanking the transgene are different in independently obtained transformants. The flanking DNA can contain regulatory elements that may influence transcription of the transgene. Moreover, the chromatin structure of the flanking DNA, which determines accessibility of regulatory sequences by the transcription machinery, can be imposed on the transgene and hence influence transcriptional efficiency [5]. A second source of variation, and probably more important, is the number and configuration of copies of transgenes integrated at one or more loci in the genome [6]. Theoretically, the level of expression should increase with the copy number. However, in most cases studied thus far, expression levels do not correlate with copy number [7], [8]. Particular configurations of linked transgene copies like indirect tandem repeats [9] or intrinsic direct repeats [10] are known to cause repeat-induced transgene silencing.

In order to reduce this inter-transformant variability, several research groups have made a number of adaptations to transgene constructs. It has been proposed that, in order to reduce repeat-induced silencing effects, the duplication of sequences within the introduced construct should be avoided [11]. However, the effect of such duplications, for instance identical promoter sequences that drive both the selectable marker gene and the gene of interest, on expression variability has not been systematically analyzed. Most studies have concentrated on the effect of matrix attachment regions (MARs) introduced in vector constructs at positions flanking a gene of interest. MARs are regions in eukaryotic genomes that are attached to the proteinaceous matrix present in the nucleus [12], [13], [14]. These regions are generally A/T rich and are typically located in non-transcribed parts of the genome. MARs may act as fixed boundaries of DNA loops showing the same overall chromatin condensation and super helical density. According to the ‘loop hypothesis’, genes present on such loops are insulated by the MAR boundaries from potential influences of regulatory factors acting on neighboring loops [15]. A number of studies have investigated whether the presence of MARs flanking a transgene results in less variable transgene expression. In most of these studies, MARs significantly increased average transgene expression levels with limited or no decrease of variability in the first generation of transgenic plants [16], [17], [18], [19], [20], [21].

One outstanding example of a MAR with strong impact on transgene expression variability, even in the first generation of transgenic plants, is the A element that flanks the chicken lysozyme gene. Presence of this MAR at both ends of the T-DNA borders of an Agrobacterium tumefaciens-based transformation vector reduced variance of expression of the uidA reporter gene driven by either the potato Lhca3.St1 gene promoter or the enhanced CaMV 35S promoter in transgenic tobacco plants by 8- and 7-fold, respectively [8], [22].

To date, the effect of MARs on expression variability in Arabidopsis thaliana has not been thoroughly analyzed. A. thaliana is widely used as a model plant for the studies of gene function [23]. Systems allowing low-level variability of transgene expression would facilitate high-throughput analysis of transgenes such as is required for shotgun transformation of bacterial artificial chromosome clones during map-based cloning efforts.

Using A. thaliana as a genetic background for large-scale multi-gene delivery experiments, we aim to generate populations of transgenic A. thaliana plants that provide minimal inter-transformant expression variability. This study systematically investigates and statistically evaluates the influence of the chicken lysozyme A element MAR and regulatory elements on the level and variability of transgene expression in populations of A. thaliana plants, irrespective of T-DNA copy number or integrity of individual transgenic plants.