Research Paper Abstract: Gene targeting in the moss Physcomitrella patens has created a new platform for plant functional genomics. We pro- duced a mutant collection of 73 329 Physcomitrella plants and evaluated the phenotype of each transformant in comparison to wild type Physcomitrella. Production parameters and morpho- logical changes in 16 categories, such as plant structure, colour, coverage with gametophores, cell shape, etc., were listed and all data were compiled in a database (mossDB). Our mutant col- lection consists of at least 1804 auxotrophic mutants which showed growth defects on minimal Knop medium but were res- cued on supplemented medium. 8129 haploid and 11068 poly- ploid transformants had morphological alterations. 9% of the haploid transformants had deviations in the leaf shape, 7% de- veloped less gametophores or had a different leaf cell shape. Other morphological deviations in plant structure, colour, and uniformity of leaves on a moss colony were less frequently ob- served. Preculture conditions of the plant material and the cDNA library (representing genes from either protonema, gameto- phore or sporophyte tissue) used to transform Physcomitrella had an effect on the number of transformants per transforma- tion. We found correlations between ploidy level and plant mor- phology and growth rate on Knop medium. In haploid transfor- mants correlations between the percentage of plants with spe- cific phenotypes and the cDNA library used for transformation were detected. The number of different cDNAs present during transformation had no effect on the number of transformants per transformation, but it had an effect on the overall percent- age of plants with phenotypic deviations. We conclude that by linking incoming molecular, proteome, and metabolome data of the transformants in the future, the database mossDB will be a valuable biological resource for systems biology. Key words: Bryophyte, database analysis, moss, mossDB, mu- tant collection, Physcomitrella patens. Introduction The moss Physcomitrella patens is the only land plant which hasbeenshowntointegratenuclearDNAefficientlybyhomol- ogous integration (Schaefer and Zrþd, 1997). The rate of ho- mologous recombination is more than three orders of magni- tude higher than observed in other multicellular plants (Hohe and Reski, 2003). Whereas gene targeting rates for seed plants range from 0.01 to 0.1%, for example in rice (Terada et al., 2002) or Arabidopsis (Kempin et al., 1997), this figure rises to 4% to 95% in Physcomitrella (Schaefer, 2001). Gene targeting in the moss Physcomitrella patens created a new platform for plant functional genomics (Holtorf et al., 2002) and this ap- proach is especially straightforward as the dominant phase in mosses is the haploid gametophyte (Reski,1998), making loss- of-function mutations readily screenable. In addition to a growing number of targeted gene knockouts in plant metab- olism (Girke et al., 1998), plant development (Imaizumi et al., 2002), and plastid division (Strepp et al.,1998), a large amount of Physcomitrella knock-out plants were generated by using large scale mutagenesis approaches (Nishiyama et al., 2000; Egener et al., 2002). Mutant collections and subsequent screening protocols have been very successful in helping to discover gene functions and databases have been established to compile the experi- mental data. However, the annotation of mutant phenotypes to experimental results, like sequence or protein data, is rela- tively rare. Examples of plant databases which integrate phe- notypic data, are the MTM database for maize (http://mtm. cshl.org, May et al., 2003), Gramene for cereal crops (http:// www.gramene.org., Jaiswal et al., 2002), and SeedGenesdata- base (htpp://www.seedgenes.org, Tzafrir et al., 2003) for Ara- bidopsis thaliana. These databases compile phenotypic de- scriptions from different sources that were analyzed by differ- ent researchers. The advantage of this approach is that only relevant mutant phenotypes are stored in the database, allow- ing the researcher to look for interesting phenotypes in an or- ganizedway.Therearetwodisadvantages,however.First,only a very limited part of the database input, which is mainly fo- cused on sequence data, is linked to phenotype descriptions. And second, because the database compiles phenotypic de- Large-Scale Analysis of 73329 Physcomitrella Plants Transformed with Different Gene Disruption Libraries: Production Parameters and Mutant Phenotypes G. Schween 1 ,T.Egener 1 ,D.Fritzowsky 1 ,J.Granado 1 ,M.-C.Guitton 1 ,N.Hartmann 1 ,A.Hohe 1,2 ,H.Holtorf 1,3 , D.Lang 1 ,J.M.Lucht 1,4 ,C.Reinhard 1 ,S.A.Rensing 1 ,K.Schlink 1,5 ,J.Schulte 1 ,andR.Reski 1 1 PlantBiotechnology,FacultyofBiology,UniversityofFreiburg,Schänzlestraûe1,79104Freiburg,Germany 2 Presentaddress:InstituteforVegetableandOrnamentalCrops,KühnhäuserStraûe101,99189Kühnhausen,Germany 3 Presentaddress:Albert-Schweitzer-Schule,AnderSchelmengass3,78048VS-Villingen,Germany 4 Presentaddress:InterNutrition,P.O.Box,8035Zürich,Switzerland 5 Presentaddress:SectionofForestGenetics,DepartementofPlantSciences,CenterofLifeSciencesWeihenstephan,TUMunich, AmHochanger13,85354Freising,Germany Received:December30,2004;Accepted:March11,2005 PlantBiol.7(2005):228±237  Georg Thieme Verlag KG Stuttgart ´ New York DOI 10.1055/s-2005-837692 ´ Published online May 12, 2005 ISSN 1435-8603 228