Systemic Agrobacterium tumefaciens– mediated transfection of viral replicons for efficient transient expression in plants Sylvestre Marillonnet 1,2 , Carola Thoeringer 1,2 , Romy Kandzia 1 , Victor Klimyuk 1 & Yuri Gleba 1 Plant biotechnology relies on two approaches for delivery and expression of heterologous genes in plants: stable genetic transformation and transient expression using viral vectors. Although much faster, the transient route is limited by low infectivity of viral vectors carrying average-sized or large genes. We have developed constructs for the efficient delivery of RNA viral vectors as DNA precursors and show here that Agrobacterium–mediated delivery of these constructs results in gene amplification in all mature leaves of a plant simultaneously (systemic transfection). This process, called ‘magnifection’, can be performed on a large scale and with different plant species. This technology combines advantages of three biological systems (the transfection efficiency of A. tumefaciens, the high expression yield obtained with viral vectors, and the post-translational capabilities of a plant), does not require genetic modification of plants and is faster than other existing methods. Viral vectors designed for expression of recombinant proteins in plants hold great promise because of high absolute and relative yields, and because of the speed provided by transient expression. Most of the results of practical interest achieved so far have been obtained with vectors built on the backbones of plus-sense RNA viruses such as tobacco mosaic virus (TMV) or potato virus X 1–4 . We have recently shown that TMV-based vectors can be delivered to plant tissues using A. tumefaciens 5 (agroinfection). However, one step of this process, namely the formation of active replicons from the primary nuclear transcript, is inefficient. In a standard leaf transfec- tion experiment, this inefficiency is masked by the subsequent ability of the replicons to move to neighboring cells by cell-to-cell movement. Here we show that this bottleneck can be fully remedied by incorpora- tion of silent nucleotide substitutions into the vector and by addition of multiple introns. We demonstrate that such modifications provide for efficient processing of the DNA information into active replicons in almost all cells (as high as 94%) of Nicotiana benthamiana, an up to 1,000-fold improvement over nonoptimized TMV-based vectors, and an even higher improvement (410 6 -fold) in Nicotiana tabacum (tobacco). Finally, we show that the resulting vectors allow the development of a fully scalable and versatile whole-plant transfection protocol, that we term magnifection, for production of heterologous proteins in plants. RESULTS Viral replication following agroinfiltration of TMV-based vectors Agroinfiltration of a TMV-based viral vector containing the gene encoding green fluorescent protein (GFP) (pICH16707, Fig. 1a) into N. benthamiana leaves leads to the formation of foci of GFP fluorescence 3 d post-infiltration (d.p.i.) (shown in ref. 5 and in Supplementary Fig. 1 online). To quantify the proportion of cells initiating viral replication, a 489-bp deletion was made within the movement protein (MP) coding sequence, resulting in construct pICH14833 (Fig. 1a). Replicons derived from this construct cannot move from cell-to-cell but are able to replicate autonomously within each infected cell. Three days after agroinfiltration of pICH14833 in N. benthamiana leaf (OD 600 of the A. tumefaciens in infiltration solution was 0.7), a small number of cells expressing GFP appeared (see Supplementary Fig. 1 online), and the same pattern was still visible 2 weeks after infiltration. By counting protoplasts prepared from the infiltrated area (Figs. 1 and 2), we found that 0.6–1.6% of cells initiated viral replication. There are several reasons why RNA viral vectors might have difficulties starting the replication cycle. First, RNA viruses, such as TMV, replicate in the cytoplasm and never enter the nucleus, and have therefore evolved in an environment where they are not exposed to the nuclear pre-mRNA processing machinery. As a result, pre-mRNA transcripts made in the nucleus from viral constructs may not be re- cognized and processed properly. Second, viral vector constructs encode very large transcripts (B7.6 kb for the primary transcript of a viral vector containing a GFP gene), a size much larger than the average size (1–2 kb) of plant genes. Moreover, in nature, large eukaryotic genes often contain numerous introns that facilitate processing and export of the pre-mRNAs from the nucleus 6 . We therefore hypothesized that modifications of the constructs that would increase the efficiency of processing and export of primary transcripts from the nucleus to the cytoplasm could lead to an increase in the number of cells that would initiate viral replication. Two types of modifications were made: Published online 8 May 2005; doi:10.1038/nbt1094 1 Icon Genetics, Biozentrum Halle, Weinbergweg 22, D-06120 Halle (Saale), Germany. 2 These authors contributed equally to this work. Correspondence and requests for materials should be addressed to Y.G. (gleba@icongenetics.de). 718 VOLUME 23 NUMBER 6 JUNE 2005 NATURE BIOTECHNOLOGY ARTICLES © 2005 Nature Publishing Group http://www.nature.com/naturebiotechnology