Increasing loop domain size does not diminish effects of matrix attachment regions on transgene expression in tobacco cells in culture

It is now widely held that the chromatin of eukaryotic organisms is organized into loop domains in which chromatin ¢bers are attached to the nuclear matrix by speci¢c interactions of nuclear matrix proteins with DNA sequences called matrix attachment regions (MARs) [1,2]. Several MARs have been isolated and incorporated into constructs used for transformation. Work with animal cell culture systems has consistently shown much higher average levels of reporter gene expression in cell lines in which reporter genes are £anked with cloned MARs than in control lines without £anking MARs [1]. MARs have similar eiects in plant cell systems. For example, in a tobacco cell culture system, we have shown that a heterologous MAR (a yeast MAR that binds weakly to the tobacco nuclear matrix) increases reporter gene expression by 12-fold, and a strong matrix-binding tobacco MAR increases reporter gene expression by 60-fold [3]. The mechanisms by which MARs increase reporter gene expression are unknown. MARs do not appear to act as typical enhancers, as they have little eiect in transient expression assays [1,2]. Most proposed mechanisms involve chromatin structure [1,2]. According to one model, MARs aiect transgene expression by creating independent, topologically isolated domains. The transgenes in these independent domains are insulated from chromatin structure of the native domains (either transcriptionally active or repressed) into which they become incorporated. If transgenic MARs do act by creating independent domains containing the transgene, the question remains of why the independent domains have a transcriptionally active chromatin structure. We have previously speculated that independent domains created by cloned MARs may have transcriptionally active chromatin structures because they are too small to form stable, transcriptionally repressed, condensed chromatin ¢bers in vivo [2,3]. In our previous experiments, the putative domain formed by cloned MARs contains 3 kb of DNA, enough to form 16 plant nucleosomes. Even fewer nucleosomes may form on the transgenes in vivo, as close proximity to the nuclear matrix may sterically inhibit nucleosome formation. As the structure of the 30-nm chromatin ¢ber is not well understood, the number of nucleosomes necessary to form a stable, transcriptionally repressed structure is unknown. Formation of folded chromatin ¢bers presumably is dependent upon nucleosome^nucleosome interactions. If we consider the solenoid model of the 30-nm ¢ber (six nucleosomes per turn of the solenoid), it is logical to assume that a minimum of 12 nucleosomes would be required to form a stable, folded structure. This would allow each nucleosome to interact with at least one nucleosome other than its linear neighbors. Further turns of the solenoid would be expected to further stabilize the structure. For example, in a structure containing 18 nucleosomes (three turns of the solenoid), the nucleosomes of the middle turn could interact with nucleosomes of the outside two turns of the solenoid. Carruthers et al. [4] have provided evidence that a reconstituted linear DNA fragment containing 12 nucleosomes can form a stable, folded structure, but the relationship of this structure to the 30-nm ¢ber is unclear. The work of Butler and Thomas [5] indicates that more nucleosomes may be required. These workers observed a change in the hydrodynamic properties of native, rat liver nucleosome oligomers above the size of 50 nucleosomes. They attributed the change to higher-order folding. The considerations mentioned above suggest that in our previous work [2,3] the 16 nucleosomes that would be expected to form on 3 kb of DNA bounded by MARs may be below or near the minimum required to form stable, folded chromatin ¢bers in vivo. Based on these ideas concerning the stability of higher-order structure in chromatin, we have asked if the enhancement of transgene expression by MARs can be counteracted by increasing the amount of DNA in the putative loop domain to a point at which a stable condensed chromatin ¢ber could be formed. In order to answer this question we have transformed plant cells in culture by microprojectile bombardment as we have previously described [3]. A co-transformation procedure is used in which a selectable marker (NPTII conferring kanamycin resistance) is carried on a separate plasmid from the reporter gene. As in our previous experiments we have used the reporter plasmids, pGHNC11 and pGHNC12 (Fig. 1A). These plasmids contain the GUS reporter gene cassette £anked by the RB7-6 tobacco MAR and a control of the GUS cassette without £anking MARs. We also used a plasmid (pNMCS1 in Fig. 1A) in which V DNA has been inserted between the GUS reporter cassette and the MAR. This `spacer' DNA would increase the size of the putative MARbounded loop domain to 52 nucleosomes. The V DNA (a 6.6kb HindIII fragment, nucleotides 37586^44141) has an AT content of 51% and does not bind to the tobacco nuclear matrix (data not shown). A control plasmid (pNMCS2) with the V DNA and the GUS reporter cassette but no MARs was also used. Kanamycin-resistant transformed cells were grown in liquid culture for 2 months with transfers every 7 days. At this time, cells were harvested and protein extracts were made in order to measure GUS speci¢c activity [3]. Fig. 1B shows the results of the GUS speci¢c activity measurements. Increasing the amount of DNA in the putative independent domain to a size that would support 8.6 turns of a nucleosome solenoid does not diminish the eiect of MARs on enhancing reporter gene expression. The speci¢c activity of the MAR-SPACERGUS-MAR transformed cell lines is slightly higher than that in the MAR-GUS-MAR lines, but the diierence is not statistically signi¢cant. In the lines transformed with constructs lacking MARs, the GUS activity was slightly lower in the