Didier G. Schaefer(1), Pierre-François Perroud(2), Remko Offringa(3), Michel Laloue(4) and Jean-Pierre Zryd(1).

(1) Institut d'écologie, laboratoire de phytogénétique cellulaire, UNIL Lausanne, (2) Washington University, St Louis, USA, (3) University of Leiden, Holland, (4)INRA Versailles, France

Agrobacterium-mediated transformation of the moss Physcomitrella patens occurs predominantly at illegitimate location

In the moss Physcomitrella patens, integration of foreign DNA sequences in the genome following PEG-mediated direct gene transfer occurs essentially at targeted location by homologous recombination (Schaefer 2001, Cur Op. Plant Biol. 4, 143-150). We present here phenotypic, genetic and molecular evidence demonstrating that Physcomitrella patens can be transformed with hypervirulent Agrobacterium strain. Agrobacterium-mediated Physcomitrella patens has the following characteristics

(1) Transformation is acetosyringone dependent and can only be achieved with hypervirulent strains such as AGL-1 (Lazo et al. 1991, Bio/Technology 9, 963-968)

(1) Transformation rates are similar with T-DNA sharing sequence homology with the moss genome and with non homologous T-DNA. This contrast with PEG-mediated naked DNA transformation where the presence of sequence homology increases transformation rates by at least one order of magnitude.

(1) T-DNAs integrate frequently at several locations of the genome and either direct and inverted repeats of either/and/or the T-DNA or the entire binary vector as well as some abortive T-DNA transfer events are detected. This corresponds to integration patterns observed in Arabidopsis with the same Agrobacterium strain, but differs radically from PEG-mediated naked DNA transfer that is characterised by single locus insertion and tandem direct repeats integration patterns.

(1) The ratio of targeted to random integration events is below 1% after Agrobacterium-mediated transformation whereas it is around 90% following PEG-mediated transformation with the same binary vectors.

These data provide the first example of the successful transformation of a Bryophyte with Agrobacterium. Our observations suggest that the mechanisms involved in the integration of foreign DNA sequences in the moss genome are different for protein coated T-DNA and naked DNA. More interestingly these results provide the first example of a complete reversion of the ratio of targeted to random integration events monitored in the same organism upon genetic transformation. This situation is ideal for studying in detail both the mechanisms of Agrobacterium-mediated gene transfer (in mutated moss and/or Agrobacterium genetic background) and the factors that affect gene targeting efficiencies (in moss strains affected in DNA recombination-repair-replication genes). Previous studies of Agrobacterium-mediated yeast transformation have not provided such strong discrepancy between targeting efficiencies and transformation methods. The moss Physcomitrella patens may thus provide an extremely suitable model system to address these questions.