Effect of nuclear matrix attachment regions on transgene expression in tobacco plants

Matrix attachment regions (MARs) are thought to participate in the organization and segregation of independent chromosomal loop domains. Although there are several reports on the action of natural MARs in the context of heterologous genes in transgenic plants, in our study we tested a synthetic MAR (sMAR) with the special property of unpairing when under superhelical strain, for its effect on reporter gene expression in tobacco plants. The synthetic MAR was a multimer of a short sequence from the MAR 3¢ end of the immunoglobulin heavy chain (IgH) enhancer. This sMAR sequence was used to flank the b-glucuronidase (GUS) reporter gene within the T-DNA of the binary vector pBI121. Vectors with or without the sMARs were then used to transform tobacco plants by Agrobacterium tumefaciens. Transgenic plants containing the sMAR sequences flanking the GUS gene exhibited higher levels of transgene expression compared with transgenic plants which lacked the sMARs. This effect was observed independently of the position of the sMAR at the 5¢ side of the reporter gene. However, variation of the detected transgene expression was significant in all transformed plant populations, irrespective of the construct used.

Most genes whose expression has been studied in transgenic plants are generally expressed in appropriate patterns. However, transgene expression can vary within an extremely wide range, often showing only a very low level [1,2]. Variation in transgene expres-sion is frequently attributed to corresponding variation in the transcription potential of different chromosomal insertion sites. DNA sequences called scaffold/matrix attachment regions (S/MARs) are involved in the structural and functional organization of all eukaryotic genomes. Evolutionarily, the structures of these sequences seem to be conserved. Typically, S/MARs are located every 5 to 200 kb of sequence and are known to bind specifically to components of the nuclear scaffold, therefore suggesting that they are responsible for loop domain base formation [3,4].
Of the MAR elements reported, many do not display extensive sequence homology. It is therefore reasonable to assume that the scaffold probably recognizes some structural features of the MAR DNA rather than a specific sequence [5]. MARs appear to be functionally conserved, since animal MARs can bind to plant nuclear scaffolds and vice versa [6,7].
Most MARs have been generally characterized as AT-rich sequences. However, AT-richness per se is not a sufficient criterion for specific sequence recognition of MARs by specific binding proteins [7]. Their capacity to bind to the nuclear matrix is determined by the specific structure of DNA. A prominent structural characteristic of different MARs is their strong potential for extensive unpairing when subjected to superhelical strain [8,9]. The ability to assume a stably unpaired conformation has been described for several MARs. For example, within the MAR 3¢ end of immunoglobulin heavy chain (IgH) enhancer there is an AATATATTT motif that is a nucleation site for DNA unwinding [10]. Concatamerized oligonucleotides containing seven repeats of this sequence exhibited a strong affinity for the nuclear scaffold and increased SV40 promoter activity in stably transformed mouse cells [11].
In this paper we present results of our studies that concern the effect of a synthetic MAR on transgene expression in tobacco plants. The synthetic MAR sequences were used to flank the b-glucuronidase (GUS) gene whose transcription was under control of the 35S CaMV promoter in the binary vector pBI121. This construct was introduced into tobacco plants and the GUS reporter gene expression was monitored in stably transformed plants.

Synthetic MAR construction.
All DNA manipulations were performed using standard techniques as described by Sambrook et al. [12]. The investigated MAR was constructed similarly to that described by Bode and colleagues [11]. Complementary oligonucleotides 5¢-TCTTTAATTTCTAATATATTTAG-AAttc-3¢ and 5¢-TTCTAAATATATTAGAAAT-TAAAGAgaa-3¢ that contain the nucleation site of unwinding (underlined) were synthesized on a Gene Assembler Plus (Pharmacia), using the phosphite triester method. After deprotection and removal from the support, the synthesized oligonucleotides were desalted using NAP-25 columns (Pharmacia) and purified by gel electrophoresis. Following phosphorylation of their 5¢-ends with polynucleotide kinase T4 (Promega), the oligonucleotides were annealed into double-stranded DNA, and concatamerized by hybridization of the overlapping single-stranded ends, indicated by small letters. Multimers were separated on a 4% agarose gel, and the decamer was eluted, incubated with Klenow polymerase (Sigma) to remove single-stranded ends and cloned into the EcoRV site of pBluescript II KS+. DNA sequencing of the insert in the recombinant vector was performed using the Taq-Track Sequencing System (Promega) according to the manufacturer's instructions.

Construction of binary vectors containing sMAR elements in the T-DNA.
The pBluescript II KS+ construct containing the synthetic MAR, was partially digested with EcoRI to achieve a heptamer of the sMAR with appropriate cohesive ends. After purification by gel electrophoresis, the heptamer was ligated into the EcoRI site of the binary vector pBI121, that yielded the intermediate vector pMGM0 with the sMAR at the 3¢ side of the GUS gene (Fig. 1A).
In addition, the pBluescript II KS+ construct with the sMAR was also digested with PvuII, and the fragment containing the synthetic MAR (decamer) was inserted into pBR322 (Fig. 1B). The HindIII fragment with the sMAR obtained from the recombinant pBR322 construct was then inserted into the HindIII site of the intermediate vector pMGM0. Depending upon the orientation of the insert, these generated two binary vectors which were named pMGM1 and pMGM2. In pMGM1 the decamer of the sMAR was located 27 bp upstream of the 35S CaMV promoter, while in pMGM2 the sMAR element was located 2242 bp upstream of the promoter (Fig. 1C, 2).  Cloning steps of the respective constructs are described in details in Materials and Methods. Size proportions between the plasmids are not maintained.
Plasmid integrity was verified by restriction enzyme analysis. Plants were transformed by the leaf disc transformation procedure [15]. To ensure independent transformation events, regenerated shoots were taken from opposite sites of a leaf disc. Shoots that rooted in the presence of 50 mg/ml kanamycin were considered to be transgenic. The transformation proof was obtained by amplification of a 338 bp GUS gene fragment using genomic DNA isolated from the putative transgenic plant as a template. Tobacco genomic DNA used in this reaction was extracted from leaves according to the method of Chomczynski & Sacchi [16]. PCR reactions were carried out in a Biometra cycler under the following conditions: 20 s at 94°C, 20 s at 55°C and 20 s at 72°C for 36 cycles. Each PCR reaction (10 ml) contained 250 nM of each primer (5¢-TGGAATTGATCAGCGTTGGTG-GG-3¢ and 5¢-GCCAGTTCAGTTCGTTGTTCA-CA-3¢), 0.2 mM dNTPs, 1x buffer DynaZyme (Finnzymes Oy), 0.2 U of thermostable polymerase DynaZyme (Finnzymes Oy) and 100 ng of DNA template.
Reporter gene assay. GUS activity was determined using the fluorometric assay described by Jefferson et al. [17] with methylumbelliferone glucuronide as substrate. GUS activity is expressed as pmoles of 4-methylumbelliferone (4-MU) formed per hour per microgram of total protein in the plant extract. Protein content was measured using the standard Bradford assay [18].
Statistical analysis. Statistical analysis was performed using Statistica 5 and Microsoft Excel 97.

RESULTS
Two complementary oligonucleotides containing an unwinding element (the AATA-TATTT motif) were synthesized. These oligonucleotides were hybridized to form doublestranded DNA and the obtained fragments were concatamerized. Such multimers were called synthetic MARs (sMARs). We used the sMAR concatamers to construct appropriate T-DNA vectors on the basis of the binary vector pBI121 carrying a selectable kanamycin resistance gene and the uidA reporter gene in the T-DNA. The reporter gene was flanked by synthetic MARs, as shown in Fig. 2. We constructed two types of vectors with sMAR sequences. In each case the uidA gene was flanked by the sMAR sequence in the same position at the 3¢ side of the reporter gene, but the sMAR at the 5¢ side of the uidA gene was placed either near the 35S CaMV promoter (pMGM1) or over 2 kb upstream of it (pMGM2). These two types of vectors were designated to test if the distance between the synthetic MARs or the sMAR localization relative to the promoter would change expression of the reporter gene. As a control, we used pBI121 without synthetic MARs.
Three constructs: pMGM1, pMGM2, pBI121 were used to transform tobacco plants using Agrobacterium tumefaciens. The independently transformed kanamycin-resistant pMGM1, pMGM2 and pBI121 plants were regenerated in vitro and then transferred to soil. There is some probability that primary transformants can be chimeras consisting of transformed and non-transformed tissues, thus resulting in irregular transgene expression patterns [19]. Therefore, we decided to examine the reporter gene expression in transgenic progeny to avoid the potential chimeric character of transformed parent plants and contamination of transformants by Agrobacterium. Following a selfing of flowering parent plants we collected seeds of six plants transformed with pBI121, eight plants transformed with pMGM1 and eight plants transformed with pMGM2. The seeding was performed on selective medium and transgenity of the obtained plants was confirmed by PCR. We obtained five progeny plants from each transformed parent plant. GUS activity was determined in leaves of all five F1 generation plants, thus in total 30 plants for the pBI121 construct and 40 plants for each of the pMGM1 and pMGM2 constructs were analyzed. Five plants from the pMGM1 population, five plants from the pMGM2 population and seven plants from the pBI121 population contained the uidA transgene integrated in the genome, but showed no GUS expression. The results concerning these individual plants were omitted from the statistical evaluations shown in Table 1. A wide range of GUS activities was observed in each population of transformed plants. Compared with the pBI121 transformed population, the pMGM1 and pMGM2 transformants showed no significant reduction in the variation of the uidA gene expression levels.
Because our raw data did not follow a normal distribution (according to Lilliefors' test and Shapiro-Wilk test), it was inappropriate to apply statistical tests dependent on a normal distribution [20]. For this reason, we used a distribution free statistical test to find if a statistical difference between GUS activity levels in different populations of transgenic plants existed. The Mann-Whitney U test supports the assumption that sMARs have an effect on transgene expression in plants (Table 1). However, the Mann-Whitney U test showed no significant difference between populations of plants transformed with pMGM1 and pMGM2. Nap et al. [20] suggested that in the case of non-normal distribution it was more appropriate to use the median rather than the mean GUS activities to assess quantitatively the effect of sMARs on transgene expression. The groups of transgenic plants containing constructs with sMARs showed the median GUS activities 2to 2.5-fold higher than the median of the control population (Table 1).
As shown in Fig. 3   transformants exhibited a wide range. The variability in GUS activity in all three groups of transformants was high, although plants with sMAR elements flanking the GUS gene had a higher level of variability compared with the control group (Table 1). In addition, the plants transformed with the sMAR constructs displayed higher levels of reporter gene expression compared with plants transformed with pBI121. The GUS level ranged from 2.5 to 1046 pmoles of 4-methyl-umbelliferone/h per mg total protein in non-sMAR transgenic plant extracts, whereas up to 8373 pmoles of 4-methyl-umbelliferone/h per mg total protein was achieved in sMAR-based transgenic plant extracts.

DISCUSSION
The interaction of chromatin with the nuclear matrix via matrix attachment regions (MARs) on DNA is considered to be of fundamental importance for higher order chromatin organization and regulation of gene expression. Matrix attachment regions (MARs) are thought to participate in the organization and segregation of independent chromosomal loop domains [21,22]. MARs as the boundary elements can define domains of independent gene regulation. Besides mediating chromatin higher-order organization, MARs may also have functional roles in gene expression. Regulatory MARs may facilitate the binding and action of regulatory proteins, perhaps providing a critical discontinuity in compacted chromatin. The location of some MARs within transcription regulatory elements suggests that MARs may serve to bring these DNA sequences into close proximity to the scaffold, thereby promoting enhancer and promoter activity by facilitating the interaction with trans-acting and/or transcription factors, which would assemble on the nuclear scaffold [23]. MARs may contain AT-rich sequences (called base-unpairing regions, BURs) that are capable of readily relieving negative superhelical strain by base unpairing or unwinding. BURs contain a core unwinding element and a cluster of sequences in which one of the strands consists exclusively of As, Ts, and Cs (ATC sequence context) [9]. BURs have been identified in animal as well as in plant MAR sequences [9,24,25]. A synthetic MAR sequence enriched in these unwinding core elements appears to increase biological activity of a linked gene in animal cells in vivo [11]. Considering the functional conservation of MAR elements and the fact that the same sequence motif was found in a scaffold-associated DNA region located downstream of the pea plastocyanin gene, the goal of our research was to investigate the usefulness of a synthetic MAR for enhancing transgene expression in tobacco plants.
Our studies complement the expanding amount of data that illustrate that MARs are useful tools for enhancing transgene expression in plants and animals. In order to evaluate the synthetic MAR for functional properties in gene expression, two chimaeric gene constructs were prepared. Figure 2 shows structures of these two constructs containing the sMAR regions. We assumed that if the sMARs are responsible for loop formation, then the base pair sequence between the sMARs should be of an appropriate length. Therefore, we tested two distances between the sMAR sequences flanking the GUS gene: 5 kb and 2 kb in the pMGM2 and pMGM1 constructs, respectively. Because sMARs, independent of their arrangement in the T-DNA, showed no reduction of transgene expression variation, these sMAR elements probably did not play an insulating role for the boundaries of the loop domain. Independently of the localization of the sMAR in the T-DNA, the level of b-glucuronidase expression was increased in comparison with the control construct. This confirms that BURs are important for gene expression in plants, but perhaps more interestingly suggests that the unwinding property of MARs is not an exclusive determinant for creating independent domains. It could be that the sMAR acted as an enhancer in our experiments. This is supported by research conducted elsewhere, the results of which have shown that MARs can often stimulate transgene expression, but are not sufficient to completely over-ride genomic position effects in plants (reviewed in [26]). The enhancement of transgene expression by synthetic MARs used in our study seems to have an additive character. One could hypothesize that this sMAR augmented the positive effect of endogenous cis-elements on transgene transcrip-tion. Van der Geest et al. [27] reported that the b-phaseolin 5¢ MAR placed between an enhancer and promoter of a reporter gene acted as an enhancer facilitator and not an enhancer blocker. Furthermore, other studies revealed that insertion of the T-DNA of Agrobacterium was usually within or near transcriptionally active loci [28] and that natural MARs located in these regions were somewhat effective in promoting high and stable transgene expression [29,30]. Our data and results that have been reported previously by others, suggest that MARs, having the ability to unwind, have a common positive effect on transcription. It is plausible that the synthetic MAR is able to function as a topological switch by storing and releasing supercoil-generated energy during transcription. The transcriptional process generates local negative supercoiling 5¢ of a transcribed gene and positive supercoiling 3¢ of it.
Other studies on MAR function in plant transgene expression have provided interesting but unequivocal conclusions. The effect of any particular MAR on transgene expression is observed only after integration of the transgene with the plant genome. According to the loop model, MARs play a role in increasing transgene expression and reducing variability of that expression between independent transformants. Enhancement of transgene expression in stably transformed plant cell lines or whole plants has been reported for the soybean Gmhsp17.6.L There are several possible explanations for the reported differences in the magnitude of the MAR-mediated augmentation of transgene ex-pression. These include: the transformation method, vector design, inheritance features of the MAR and the type of tissue that is used for transformation. The second effect of MARs as boundary elements that has been reported is the normalization of transgene expression. Boundary elements placed in a gene-flanking arrangement should insulate the gene from regulatory elements, such as enhancers or silencers, that reside outside the MAR-defined domain. This should result in the MAR being able to decrease the variability of transgene expression and even lead to a copy number-dependent expression. In the majority of studies to date, the decrease of the variability of transgene expression was rather moderate. Furthermore, the expression was rarely dependent on the copy number of the transgene. Only the presence of the chicken lysozyme A element MAR in transgene constructs significantly, up to 7-fold, reduced variability between transformation events by eliminating low-level expressing plants, but without affecting maximal transgene expression levels [38,40,41]. Although a normalization in reporter gene expression was obtained, variation was still evident among independent plants containing transgenes flanked by MARs. This was attributed mostly to environmental or developmental influences. On the other hand, any MAR element may play various functions. A lack of reduction in variation that has been reported for some MARs illustrates that not all of these elements have boundary functions or the ability to eliminate transgene silencing.
In summary, the results from different studies are not unequivocal. Some MARs have both general and specific properties. This heterogeneity can complicate functional analysis, because results obtained with different MAR sequences may not be comparable. In this study a synthetic MAR element was used that exhibited only one of the structural features characteristic for this group of DNA elements. Therefore, our results can only be associated with the potential of MARs for unpairing and the effects that this feature of MARs may have on plant transgene expression.
The authors wish to especially thank Jason A. Able, Agata Rapiñska and Artur Rogowski for the helpful and critical review of the manuscript.