Transgenic loci obtained after Agrobacterium tumefaciens-mediated transformation can be simple, but fairly often they contain multiple T-DNA copies integrated into the plant genome. To understand the origin of complex T-DNA loci, floral-dip and root transformation experiments were carried out in Arabidopsis thaliana with mixtures of A. tumefaciens strains, each harboring one or two different T-DNA vectors. Upon floral-dip transformation, 6-30% of the transformants were co-transformed by multiple T-DNAs originating from different bacteria and 20-36% by different T-DNAs from one strain. However, these co-transformation frequencies were too low to explain the presence of on average 4-6 T-DNA copies in these transformants, suggesting that, upon floral-dip transformation, T-DNA replication frequently occurs before or during integration after the transfer of single T-DNA copies. Upon root transformation, the co-transformation frequencies of T-DNAs originating from different bacteria were similar or slightly higher (between 10 and 60%) than those obtained after floral-dip transformation, whereas the co-transformation frequencies of different T-DNAs from one strain were comparable (24-31%). Root transformants generally harbor only one to three T-DNA copies, and thus co-transformation of different T-DNAs can explain the T-DNA copy number in many transformants, but T-DNA replication is postulated to occur in most multicopy root transformants. In conclusion, the comparable co-transformation frequencies and differences in complexity of the T-DNA loci after floral-dip and root transformations indicate that the T-DNA copy number is highly determined by the transformation-competent target cells.

In view of the recent finding that different T-DNAs tend to ligate and integrate as repeats at single chromosomal positions, the frequency of transformation and cotransformation was determined during cocultivation of Arabidopsis thaliana root explants and Nicotiana tabacum protoplasts with two Agrobacterium strains. The transformation frequency of unselected A, thaliana shoots was lower than 1% whereas that of cocultivated tobacco protoplasts was approximately 18%, The cotransformation frequencies, defined as the frequencies with which cells transformed with a first T-DNA contained a second unselected T-DNA, were approximately 40% reproducible, irrespective of the selection, the transformation frequency, and the plant system used. Extrapolation of these results suggests that at least two independently transferred T-DNAs were present in 64% of the transformed plant cells, Molecular analysis of cocultivated N. tabacum shoots regenerated on nonselective medium showed that only a few transformants had a silenced (2/46) or truncated (1/46) T-DNA, Therefore, most integrated T-DNAs expressed their selectable or screenable markers in primary transgenic plants. Remarkably, 10 to 30% of the selected A. thaliana shoots or progenies lost the T-DNA marker they were selected on. As these regenerants contained the unselected T-DNA with a high frequency (17%), these selected plants might result from the expression of unstable, transiently expressed T-DNAs, In conclusion, a significant part of the T-DNAs is lost from the transformed cells.

The ability of the CRE recombinase to catalyze excision of a DNA fragment flanked by directly repeated lox sites has been exploited to modify gene expression and proved to function well in particular case studies. However, very often variability in CRE expression and differences in efficiency of CRE-mediated recombination are observed. Here, various approaches were investigated to reproducibly obtain optimal CRE activity. CRE recombination was analyzed either by transforming the CRE T-DNA into plants containing a lox-flanked fragment or by transforming a T-DNA harboring a lox-flanked fragment into plants producing the CRE recombinase. Although somatic CRE-mediated excision of a lox-flanked fragment was obtained in all transformants, a variable amount of germline-transmitted deletions was found among different independent transformants, irrespective of the orientation of transformation. Also, the efficiency of CRE-mediated excision correlated well with the CRE mRNA level. In addition, CRE-mediated fragment excision was compared after floral dip and after root tissue transformation when transforming in a CRE-expressing background. Importantly, less CRE activity was needed to excise the lox-flanked fragment from the transferred T-DNA after root tissue transformation than after floral dip transformation. We hypothesize that this is correlated with the lower T-DNA copy number inserted during root transformation as compared to floral dip transformation.

Many studies in both animal and plant systems have shown that matrix attachment regions (MARs) can increase the expression of flanking transgenes. However, our previous studies revealed no effect of the chicken lysozyme MARs (chiMARs) on transgene expression in the first generation transgenic Arabidopsis thaliana plants transformed with a β-glucuronidase gene (uidA) unless gene silencing mutants were used as genetic background for transformation. In the present study, we investigated why chiMARs do not influence transgene expression in transgenic wild-type Arabidopsis plants. We first studied the effect of chiMARs on transgene expression in the progeny of primary transformants harboring chiMAR-flanked T-DNAs. Our data indicate that chiMARs do not affect transgene expression in consecutive generations of wild-type A. thaliana plants. Next, we examined whether these observed results in A. thaliana transformants are influenced by the applied transformation method. The results from in vitro transformed A. thaliana plants are in accordance with those from in planta transformed A. thaliana plants and again reveal no influence of chiMARs on transgene expression in A. thaliana wild-type transformants. The effect of chiMARs on transgene expression is also examined in in vitro transformed Nicotiana tabacum plants, but as for A. thaliana, the transgene expression in tobacco transformants is not altered by the presence of chiMARs. Taken together, our results show that the applied method or the plant species used for transformation does not influence whether and how chiMARs have an effect on transgene expression. Finally, we studied the effect of MARs (tabMARs) of plant origin (tobacco) on the transgene expression in A. thaliana wild-type plants and suppressed gene silencing (sgs2) mutants. Our results clearly show that similar to chiMARs, the tobacco-derived MARs do not enhance transgene expression in a wild-type background but can be used to enhance transgene expression in a mutant impaired in gene silencing.

Over the past decade, several high value proteins have been produced in different transgenic plant tissues such as leaves, tubers, and seeds 1 , 2 . Despite recent advances, many heterologous proteins accumulate to low concentrations, and the optimization of expression cassettes to make in planta production and purification economically feasible remains critical. Here, the regulatory sequences of the seed storage protein gene arcelin 5-I ( arc5-I ) of common bean ( Phaseolus vulgaris ) 3 were evaluated for producing heterologous proteins in dicotyledonous seeds. The murine single chain variable fragment (scFv) G4 (ref. 4 ) was chosen as model protein because of the current industrial interest in producing antibodies and derived fragments in crops 5 , 6 . In transgenic Arabidopsis thaliana seed stocks, the scFv under control of the 35S promoter of the cauliflower mosaic virus (CaMV) accumulated to approximately 1% of total soluble protein (TSP). However, a set of seed storage promoter constructs boosted the scFv accumulation to exceptionally high concentrations, reaching no less than 36.5% of TSP in homozygous seeds. Even at these high concentrations, the scFv proteins had antigen-binding activity and affinity similar to those produced in Escherichia coli . The feasibility of heterologous protein production under control of arc5-I regulatory sequences was also demonstrated in Phaseolus acutifolius , a promising crop for large scale production.

To use transgenic potato tubers (Solanum tuberosum cv. Désirée) for bulk production of recombinant antibodies, constructs were engineered for accumulating full-size IgGs and Fab fragments in the plant cell apoplast or endoplasmic reticulum (ER). An in-house transformation protocol was worked out for the efficient co-transformation of potato root explants. Accumulation levels in tubers of up to 0.5% of total soluble protein were found for antibodies targeted to the ER whereas five-fold lower accumulation levels were found for antibodies targeted for secretion. Additionally, different aspects important for the commercial exploitation of potato tubers as a heterologous production system were analysed. Tubers could be stored for up to 6 months without significant loss of antibody amount or activity. Minor variations in antibody accumulation levels were observed in tubers that originated from the same transformant. Most isolated IgGs and Fab fragments bound the antigen and had the correct molecular weight when compared with the hybridoma-derived standard. Processing to greenhouse or field trials, including in vitro propagation of a selected transformant, required only approximately 9 months from the start of transformation, a time frame in which hundreds of kilograms of transgenic potato tubers could easily be obtained. Small-scale purification of IgG was possible by using standard laboratory techniques. Thus, molecular farming in potato tubers can be a viable production system for economic production of clinically or industrially interesting macromolecules, such as antibodies.

In view of the recent finding that different T-DNAs tend to ligate and integrate as repeats at single chromosomal positions, the frequency of transformation and cotransformation was determined during cocultivation of Arabidopsis thaliana root explants and Nicotiana tabacum protoplasts with two Agrobacterium strains. The transformation frequency of unselected A. thaliana shoots was lower than 1% whereas that of cocultivated tobacco protoplasts was approximately 18%. The cotransformation frequencies, defined as the frequencies with which cells transformed with a first T-DNA contained a second unselected T-DNA, were approximately 40% reproducible, irrespective of the selection, the transformation frequency, and the plant system used. Extrapolation of these results suggests that at least two independently transferred T-DNAs were present in 64% of the transformed plant cells. Molecular analysis of cocultivated N. tabacum shoots regenerated on nonselective medium showed that only a few transformants had a silenced (2/46) or truncated (1/46) T-DNA. Therefore, most integrated T-DNAs expressed their selectable or screenable markers in primary transgenic plants. Remarkably, 10 to 30% of the selected A. thaliana shoots or progenies lost the T-DNA marker they were selected on. As these regenerants contained the unselected T-DNA with a high frequency (17%), these selected plants might result from the expression of unstable, transiently expressed T-DNAs. In conclusion, a significant part of the T-DNAs is lost from the transformed cells.

The yield and assembly of an IgG1 antibody and its derived F(ab) fragment were compared in Nicotiana and Arabidopsis. The results obtained showed a lot of interclonal variability. For 45% of the primary transgenic calluses, antigen-binding entities represented less than 0.1% of the total soluble protein (TSP). Only two of the 103 analysed transformants contained more than 1% of antigen-binding protein, with 1.26% being the highest yield. Analogous amounts of complete antibody and F(ab) accumulated in primary callus tissue. Moreover, yields were in the same range for both species as far as primary callus tissue is concerned. However, the accumulation of the F(ab) fragment in leaf tissue of regenerated plants differed significantly between Nicotiana and Arabidopsis. The F(ab) fragment accumulated to only 0.044% of TSP in Nicotiana leaves but up to 1.3% in Arabidopsis leaves. Furthermore, both species showed differences in the assembly pattern of the complete antibody. Whereas Arabidopsis contained primarily fully assembled antibodies of 150 kDa, Nicotiana showed an abundance of fragments in the 50 kDa range.

T-DNA vectors were constructed which carry a beta-glucuronidase (gusA) gene fused to the promoter of the nopaline synthase (nos) gene and the 3' end of the octopine synthase (ocs) gene. This reporter gene was cloned at different locations and orientations towards the right T-DNA border. For each construct, between 30 and 60 stably transformed calli were analysed for beta-glucuronidase activity. Depending on the T-DNA configuration, distinct populations of gusA-expressing calli were obtained. Placing the reporter gene in the middle of the T-DNA results in relatively low expression levels and a limited inter-transformant variability. Placing the gene with its promoter next to the right border led to an increase in both the mean activity and the variability level. With this construct, some of the calli expressed the gusA gene at levels four to five times higher than the mean. In all these series, at least 30% of the calli contained reporter gene activities that were less than half of the mean expression level. Separating the gusA gene from the right T-DNA border by an additional 3'-untranslated region, derived from the nos gene, resulted in an increase in the mean expression to a level almost four times higher than that of constructions carrying the reporter gene in the middle of the T-DNA. Moreover, the number of transformants with extremely low activities decreased by at least 50% and this resulted in significantly lower inter-transformant variability independently of the orientation of the reporter gene on the T-DNA.

We investigated whether the efficiency of transformation of plant cells by Agrobacterium tumefaciens during cocultivation is limited by the properties of the plant cells or by the infecting bacteria.