5 Apr 2002 10:40 AR AR156-20.tex AR156-20.SGM LaTeX2e(2001/05/10) P1:IKH10.1146/annurev.arplant.53.100301.135202

A NEW MOSS GENETICS: TargetedMutagenesis in Physcomitrella patens

Didier G. SchaeferInstitut d’Ecologie, LaboratoiredePhytogenetiqueCellulaire, BatimentdeBiologie,Universite deLausanne, CH-1015Lausanne, Switzerland;e-mail:Didier.schaefer@ie-pc.unil.ch

KeyWords modelsystem,genetargeting,homologousrecombination,functionalgenomics

■ Abstract Thepotentialof mossasamodelsystemtostudyplantbiologyisassoci-atedwith their relatively simpledevelopmentalpatternthatneverthelessresemblesthebasicorganizationof thebodyplanof landplants,thedirectaccesstocell-lineageanal-ysis,theirsimilarresponsesto plantgrowth factorsandenvironmentalstimuli asthoseobservedin otherlandplants,andthedominanceof thegametophytein thelife cyclethatfacilitatesgeneticapproaches.Transformationstudiesin themossPhyscomitrellapatenshaverevealedatotallyuniquefeaturefor plants,i.e.,thatforeignDNA sequencesintegratein thegenomepreferentiallyat targetedlocationsby homologousrecombi-nation,enablingfor thefirst time in plantstheapplicationof thepowerful moleculargeneticapproachesusedroutinelyin bacteria,yeast,andsince1989,themouseembry-onic stemcells.This articlereviews our currentknowledgeof Physcomitrella patenstransformationandits uniquesuitability for functionalgenomicstudies.

CONTENTS

INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 477THE PHYSCOMITRELLAPATENSMODEL SYSTEM . . . . . . . . . . . . . . . . . . . . . . . 480GENETICTRANSFORMATION OF THE MOSSPHYSCOMITRELLAPATENS. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 481TransformationWith DNA WithoutHomology . . . . . . . . . . . . . . . . . . . . . . . . . . . . 481TransformationWith DNA CarryingHomologousSequences. . . . . . . . . . . . . . . . . 484Analysisof GeneFunctionby TargetedDisruption . . . . . . . . . . . . . . . . . . . . . . . . . 487AdditionalMethodologicalDevelopment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 489

CONCLUSION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 494

INTRODUCTION

A major goal of modernbiology is to understandthe relationshipbetweenbio-logical systemsandthepresenceandactivity of genes.Modelsystemsof biologyhaveprovidedboththeexperimentalmaterialandthebiologicalcontext for framing

Published in Annual Review of Plant Biology 53, 477-501, 2002which should be used for any reference to this work

1

questions that explore this relationship. The investigator’s ability to use several dif-ferent model systems allows the function of related genes to be tested in a varietyof biological and/or methodological contexts. Genomics, along with the conserva-tion of gene structure and function found among diverse organisms, represents a“golden thread” that links model systems and allows the comparison between them.

During the past 15 years, an enormous effort by the biological community hasled to the completion of nucleotide sequences of genomes from both prokaryoticand eukaryotic model systems such asEscherichia coli(9), Bacillus subtilis(62),Saccharomyces cerevisiae(20), and more recentlyCaenorhabditis elegans(15)andDrosophila melanogaster(2). The goal of these efforts has been to establishgene catalogues for particular organisms and to provide insight into the functionof the corresponding proteins. In plants, the recently completed genome sequenceof Arabidopsis thaliana(3a) will soon be followed by the completion of a genomesequence for rice (Oriza sativa) (6), and we can expect that further genomic se-quences will be produced for other plants in the future. The next challenge for thebiological community is to assess the precise function of the sequenced genes, ageneral approach that is referred to as functional genomics.

Several strategies have been applied to identify new genes and decipher proteinfunction in plants. The most basic approach is sequence comparison of a gene ora protein with existing databases. Homology to another protein whose functionhas been previously established can provide important clues as to cellular functionand biochemical properties. However, beyond this, plant functional genomics isessentially performed by the two strategies described below. The first approach,like homology comparisons, provides only an indirect indication of gene function.It makes use of global gene expression profiles and is based either on the studyof populations of cDNAs or expressed sequence tags (ESTs) deposited on DNAmicroarrays (34, 98) or on the study of populations of proteins separated by two-dimensional gel electrophoresis (proteomics) (122). Transcriptomic and proteomicapproaches identify known or new genes or proteins that are specifically up- ordownregulated and give valuable information on the expression profile of genenetworks in response to experimental conditions. Analysis of such expressiondata by high-throughput computational methods is useful for assigning genes toparticular regulatory networks and for identifying new networks and interactions.However, this is nonetheless a descriptive approach and alone does not tell theinvestigator what function a gene performs in a cell or in an organism. Similarconsiderations apply to the emerging areas of metabolomics and phenomics.

The most widely used approach for performing true, genome-wide functionalgenomics has been insertional mutagenesis. Insertional mutagenesis by transgene-sis and the phenotypic and molecular characterization of tagged mutated lines hasbeen used extensively to study plant gene function (128). Several collections ofT-DNA and transposon-tagged transformants ofArabidopsis(86) and rice (51)have been generated and successfully used to identify and characterize genesinvolved in different biological processes. Tagged mutagenesis usually gener-ates loss-of-function mutations and rarely identifies genes that act redundantly.

2

However, activation tagging, based on the random insertion of enhancers in thegenome, allows the identification of gain-of-function mutations (129), whereasgene trap screens (117) allow gene activity to be monitored in planta by gener-ating fusion between endogenous genetic elements and a reporter gene. Theseapproaches significantly complement the tools available to plant functional ge-neticists. Recent collections generated inArabidopsishave generated more than100,000 insertional lines, and this saturation of the genome essentially ensures thatat least one insertional event can be recovered for any given gene. This is particu-larly useful where fully sequenced genomes are available because the location ofan element within the genome can be determined simply by sequencing genomicsequences that flank each insertional element.

Nonetheless, insertional mutagenesis has several conceptual disadvantages:(a) It is based on random integration of tags in the genome and thus can onlygenerate stochastically distributed allelic mutations in the genome; (b) it requiresthe generation of large numbers of transformants to saturate the genome with tags;(c) it only identifies detectable viable phenotypes and does not generate conditionalmutations, which is a prerequisite for the identification and the characterizationof essential genes; (d ) the identification of the tagged mutant or gene can be verydifficult and time consuming because multiple insertion or major chromosomalrearrangements can be associated with the insertion of the tag (78); and (e) itdoes not enable the generation of specific point mutations in a gene, which is aprerequisite for detailed functional analysis and for generating an allelic series ofmutations within a specific gene that allow a full range of possible phenotypes tobe explored.

Gene targeting (GT) is the generation of specific mutations in a genome byhomologous recombination-mediated integration of foreign DNA sequences. GTcircumvents the limitations associated with stochastic mutational approaches andrepresents the ultimate genetic tool for functional genomics. GT provides themethodological core of functional genetic studies in bacteria, yeast, and severalfilamentous fungi because the integration of foreign DNA into their genome occursefficiently at targeted locations by homologous recombination (HR). It reaches itsmaximal efficiency inS. cerevisiae, and this accounts for an important part inthe extensive use of budding yeast as a model system (100, 120). Yet, GT is notroutinely used in animals and plants because the integration of foreign DNA intheir genome occurs orders of magnitude more frequently at random locations byillegitimate recombination, thus preventing the identification of the rare targetedmutant in a large population of transformants. In 1989, the observation that theratio of targeted to random integration events upon transformation of a mouseembryonic stem (ES) cell line ranges from 0.1–10% (16, 123) accounts for itsexponential development as a model system in animal biology (77). Since 1988,the feasibility of gene targeting in plants has been demonstrated, following eitherdirect gene transfer to protoplasts (87) orAgrobacterium-mediated transformation(65, 80) into either artificial (80, 87) or natural loci (65). Yet, despite numerousstudies, the ratio of targeted to random integration events observed so far in plants

3

hardlyreaches10−4, whichpreventsthegeneraluseof genetargetingapproachesfor plantfunctionalgenomics(74,89,126).

In contrastto all other plants testedso far, integration of foreign DNA se-quencesin thegenomeof themossPhyscomitrella patensoccurspredominantlyat targetedlocationsby HR (107,110).This methodologicalprogressrepresentsa true revolution in our strategiesto addressgenefunction in plantsandplacesPhyscomitrella patensin auniquepositionamongmodelsystemsin multicellulareukaryotes(108).Thisability hasbeenexploitedin anumberof studiesto directlyaddressgenefunction,andthesearediscussedherefor boththetechnicallessonsthat can be learnedand for the value of thesestudiesfor understandingbasicbiological processes.I review our currentknowledgeof genetictransformationandgenetargetingin Physcomitrella patensandemphasizehow thismosscanbeusedasamodelsystemthatadvantageouslycomplementsothertoolsavailableforfunctionalgenomicapproaches.

THE PHYSCOMITRELLA PATENS MODEL SYSTEM

Themajor featuresof the life cycle of Physcomitrella areindicatedin Figure1.Physcomitrella hasarelatively simpleplantarchitecturethatis neverthelesscom-posedof thesameelementsasthosepresentin otherlandplants.Its life cycle isdominatedby thehaploidgametophyte,whichpresentstwo distinctdevelopmen-talphases:theprotonemathatdisplayscharacteristicone-dimensionalfilamentousgrowth andthegametophorethatdevelopby three-dimensionalcaulinarygrowthintoatypicalsmallplant.Theformerallowsthestudyof basicbiologicalprocesseswith microbiologicalculturetechnology, whereasmorecomplex developmentalprocessessuchasorganogenesisandthe establishmentof a body plan areade-quatelyaddressedin thelatter. Mostof thebiologicalstudieshavebeenperformedon theprotonema,which is idealfor following biologicalprocessesanddevelop-mentalpatterningat thesinglecell level.

Thebiology of Physcomitrella andits potentialasa modelgeneticsystemtostudyplantbiologicalprocesseshave beendiscussedin severalexcellentreviewsand are not presentedhere(21–28,33, 56, 57, 92, 94, 111, 113, 114). Thesereviews cover mostof the work that hasbeenconductedin Physcomitrella overthe past20 yearsandhighlight the similarity of its biology with that of higherplants.Thesestudieshave shown that this mossprovidesanadequatesystemtostudyabroadrangeof biologicalprocessesincludingcelldivision,cellgrowth,cellpolarity, cellularultrastructure,photomorphogenesis,photo-andgravitropic res-ponses,hormone-mediatedresponses,signal transductionpathways,chloroplastdevelopment,filamentousgrowth, organogenesis,andplantdevelopment.

Thewild-type strainof Physcomitrella patensgrown in differentlaboratoriesderivesfrom asinglesporeisolatedin England,thusensuringgenetichomogene-ity amongstudiesconductedin differentlaboratories.Thesizeof thegenomehasbeenestimatedby flow cytometryto beapproximately480Mbp, andkaryotypicanalysishasshown that it is composedof 27 very small chromosomes(94, 95).

4

To date,nophysicalor geneticmapof thePhyscomitrella genomehasbeendevel-oped.Recently, thePhyscomitrellaESTProgram(90)conductedby theUniversityof Leeds(U.K.) and the WashingtonUniversity in St. Louis (U.S.A.) haspro-videdthescientificcommunitywith approximately16,000ESTsin thedatabase,while theGermanagrochemicalcompany BASF hasgeneratedanESTdatabaseof Physcomitrella thatcontainsapproximately120,000entriesrepresentingmorethan20,000genes.Preliminaryanalysisof genesandESTsstronglyindicatethatthesesequencesarehighlysimilarto thoseof thecorrespondinghigherplantgenes.Thesesimilaritiesextendtocodonusage(68,96),intron-exonstructure(18,61,67),and multigenefamily complexity (18, 61, 95). It is clear that the numberofPhyscomitrella genesandESTsequencesaccessiblein thedatabasewill increaserapidly, andeachentrydirectlyenablesthegenerationof thecorrespondingmossmutantby targetedtransgenesis.

Increasinginterestto studymossesalsocomesfrom thefield of evolutionarydevelopmentalgenetics.Experimentalmodel systemsin animal developmentalbiology includeorganismsfrom contrastingtaxonomicgroups,suchasworm,fly,andmammal.By contrast,modelsystemsin plantbiologyhaveessentiallyfocusedon flowering plants.The phylogeneticbasalpositionof Bryophytesamonglandplantsplacesthem in an ideal situationto addressfundamentalquestionssuchashow have plantsevolved from simpleto complex forms (29). In this respect,functionalanalysisof therecentlyidentifiedPhyscomitrella genehomologto keyplayersinplantdevelopmentsuchastheMADS(61)andhomeoboxgenes(18,102)of higherplantsmaygiveusvaluableinformationabouttheancestralmechanismsthatgovernthedevelopmentof landplants(121).Onecanpredictthatthis is onlythebeginning.

GENETIC TRANSFORMATION OF THE MOSSPHYSCOMITRELLA PATENS

Transformation With DNA Without Homology

METHODOLOGICAL CONSIDERATIONS Protoplastsof Physcomitrella provide theidealmaterialfor developingmethodsin genetictransformationowing to theeasewith which they can be isolatedand to their high physiologicaland regenera-tive capacities.Thefirst successfulgenetictransformationof Physcomitrella wasachieved by polyethyleneglycol (PEG)-mediateddirect genetransferinto pro-toplasts(112).Sincethen,the protocolhasbeenoptimized(107,109) andusedroutinelyfor transformingPhyscomitrellaandfor transientgeneexpressionassays(4,41,42,46a,47,48,52,66,79,97,101,110,119,130,131).Transientgeneex-pressionwasalsoreportedfollowingbiolisticdeliveryof DNA-coatedmicrobeadsinto protonemaltissue(58, 106),but this methodseemsto be lessefficient thanPEG-mediatedtransformationastherehasbeenonly onereportof a transgenicstrain producedthis way (58). Genetictransformationof Physcomitrella usingAgrobacterium-basedmethodshasalsobeenattemptedwithoutsuccessin several

5

laboratories (28, 94, 109). It is nevertheless possible thatPhyscomitrellamight betransformed with hyper-virulentAgrobacteriumstrains (63) or upon transforma-tion with T-DNA carrying moss genomic sequences.

To date, the neomycin resistance genenptII (7), the hygromycin resistance geneaphIV(45), and the sulfadiazine resistance genesul(52) have been used as positiveselectable markers inPhyscomitrella. The resistance genes are usually driven bythe CaMV 35S promoter (4, 41, 42, 47, 48, 58, 79, 107, 110, 112, 119) or theNOS promoter (4, 66, 97, 101, 112), although it seems that the level of resistanceachieved by NOS-driven resistance cassettes is lower than that observed with the35S promoter. Selection conditions for antibiotic resistant clones are stringent andusually use antibiotic concentrations that kill untransformed protoplasts within aweek of selection and correspond to 10 times the lethal dose 50 (LD50, i.e., 50 mg/Lgeneticin sulfate, 25 mg/L hygromycin B, or 150 mg/L sulfadiazine). Meanwhile,the cytological markers encodingβ-glucuronidase (GUS) (58, 107, 130) and greenfluorescent protein (GFP) (17, 55, 57) were shown to be suitable for transformationexperiments inPhyscomitrella.

Survival rates ofPhyscomitrellaprotoplasts after PEG-mediated transforma-tion range from 15–30%, and selection for antibiotic resistance is initiated oneweek after transformation, when protoplasts have regenerated colonies of 5–10cells. Initial relative transformation frequencies (RTF; the percentage of antibioticresistant colonies in the regenerating population) monitored 14 days after trans-fer to selective medium are extremely high, ranging from 3–30% (107). Thesevalues correspond to the percentage of protoplasts expressing a cytoplasmic GFPobserved 48 h after PEG-mediated transfection with a 35S-GFP plasmid (17). Atypical petri dish after 15 days of selection shows two classes of resistant clones(57, 107). The majority of the resistant colonies display poor growth and alteredregeneration and are episomal transformants, whereas a small percentage displaygrowth and regeneration comparable to protoplasts plated on nonselective mediumand represent integrative transformants.

EPISOMAL REPLICATIVE TRANSFORMATION Episomal antibiotic resistant cloneshave been obtained with every transforming DNA tested so far and have been rou-tinely referred to as unstable clones. Their phenotypic and molecular characteristicsare as follows: (a) Unstable clones are recovered at extremely high frequencies(RTF 10–30% and 1–5% with supercoiled and linearized DNA, respectively) thatcorrespond to up to several hundred clones perµg DNA (107). (b) They can bepropagated for years as protonemal cultures as long as constant selective pressure ismaintained but lose their resistance phenotype and the transforming DNA follow-ing a 14-day growth period on nonselective medium (4, 107, 109). (c) They displayreduced growth and altered regeneration on selective medium, both of which arepositively correlated with the antibiotic concentration (107). (d ) Unstable clonesare formed by a mosaic of transformed and untransformed cells, as observedby the presence of resistant and sensitive sectors and of GUS- or GFP-labeledand unlabeled cells in protonema regenerating on selective medium (4, 107).

6

(e) Phenotypic and molecular analyses show that maintenance of the transgene istissue specific, being restricted to the filamentous protonema, and is not transmit-ted to the gametophore (4, 107, 109). Consistent with this observation, no resistantspores have been recovered from strains regenerated under nonselective conditions(107). (f ) At the molecular level, analysis of uncut plant DNA by Southern blot-ting reveals that the transgene sequences comigrate with high molecular weightDNA and that these high molecular weight arrays are formed by direct head-to-tailrepeats of the transforming plasmid (4, 107). (g) Genomic reconstruction indicatesthat for DNA isolated from protonema the number of plasmid copies per haploidgenome ranges between 3 and 40 (4, 107). This is probably an underestimate ofthe real copy number per resistant cell given the strong mosaicism observed inprotonema. These values fall to below one copy per haploid genome when DNAis isolated from older cultures that have differentiated into gametophores (107).(h) Direct evidence that transforming sequences are effectively replicated in mosscells was provided by Southern blot analyses that showed that the bacterial Dammethylation pattern of the transfected DNA was lost in DNA isolated from theseclones (D. G. Schaefer, unpublished data).

These characteristics demonstrate that in episomal transformants the transform-ing sequences are concatenated to form high molecular weight extrachromosomalelements that are replicated in moss cells but poorly partitioned during mitosis(4, 107, 109). Replicative transformation inPhyscomitrellaprovides the firstexample of the successful episomal maintenance of bacterial plasmids in plantcells. Though rare, a similar pattern of behavior has also been observed for plas-mid DNA introduced intoXenopusearly embryos (70),Caenorabditis elegans(73, 118), and several other animal eukaryotes. That bacterial plasmids can beconcatenated and replicated in moss cells raises very interesting questions asto the mechanisms underlying this process. It also represents the first step to-ward the development of episomal vectors and moss artificial chromosomes inPhyscomitrella. Episomal transformation vectors provide extremely valuable toolsfor molecular genetic approaches in yeast. Their development has been achievedby the combined addition of functional elements, such as origins of DNA repli-cation (yeast ARS element), centromere (yeast CEN), and telomere from yeast orTetrahymenainto bacterial plasmids, to generate a palette of tools ranging fromARS plasmid to yeast artificial chromosome (YAC) vectors (13, 120). A simi-lar approach should be tested for vectors used to transformPhyscomitrella, andthe mitotic stability and the developmental partition of these sequences duringits life cycle should be assessed. The observation that replicative transformantsin Physcomitrellaobtained with a YAC-derived vector (11) appear to displayimproved mitotic stability and occasional transmission of the transgenes to thegametophore supports the potential for such approaches in the near future (107).Furthermore, the development of episomal vectors with improved mitotic stability[moss ARS-like plasmids (MARS)] would directly enable the general use of highfrequency replicative transformation as an approach to complement mutations inPhyscomitrella.

7

In practice,thesereplicative transformantscanbedistinguishedfrom true in-tegrative transformantsafter transformationby adoptinga selectionscreenthatalternatesperiodsof growth on selective and nonselective medium.Integrativetransgenicstrainsdisplayunrestrictedgrowthduringall stagesof selection,whereasreplicativetransformantseitherdieorshow antibioticsensitivesectorswhentrans-ferredbackto selectivemedium(56,107,109).

ILLEGITIMATE INTEGRATIVE TRANSFORMATION ForeignDNA sequencescaninte-grateby illegitimaterecombinationin thegenomeof Physcomitrella. Efficienciesarelow (58,66,97,101,107,112),andRTF rangearound0.001%(1 in 105 regen-eratingcolonies)aftertransformationwith supercoiledplasmidandareincreasedapproximatelyfivetimes(RTF 0.005%)whenlinearizedDNA is used(4,97,107,109,112).In comparisonwith thehighreplicative transformationfrequenciesob-served during the sameexperiments,thesevaluesmost likely representthe truelevelof illegitimateintegrativetransformationin Physcomitrella. Thesetransgenicstrainscancompletetheir life cycle normallyon selective mediumandmaintainantibiotic resistancewhengrown on nonselective medium.Compileddatafromexperimentsperformedwith antibioticresistanceandcytologicalmarkersdemon-stratethat thenew trait is mitotically stableandis expressedin every cell andatall stagesof development(4, 58,107,109,112).At themolecularlevel, Southernblot analysisshows thatbetween1 and50 directhead-to-tailrepeatsof thetrans-formingplasmidareintegratedin thegenome,usuallyatasinglelocus(4,97,107,112,130),althoughintegrationat two loci hasalsobeenobserved (97). Finally,geneticanalysisshows thatthetransgeneis meioticallystableandsegregatesin aMendelianwayasasinglelocuscharacterthatis independentfrom strainto strain(52,107,112).

Transformation With DNA Carrying Homologous Sequences

CONCEPTS OF TARGETED MUTAGENESIS Thebasicprinciplesandstrategiesof genetargeting(GT) andallelereplacementin eukaryoteshavebeendefinedin buddingyeast(100,120)andappliedto developtargetedmutagenesisin themouseEScellsystem(77). This methodologyhasbeenfurther improvedwhenassociatedwiththeuseof site-specificrecombinationsystems(suchasbacteriophageP1Cre/lox),which facilitatesthesubsequenteliminationof undesiredsequencesintegratedinthegenomeupontransformation.This allows very accuratetargetedtransgenesisto beperformedin theyeastandmousegenome(104,105).Theseapproachesarebasedontheuseof two maintypesof targetingvectors:insertionandreplacementvectors.Eachtypegeneratesdifferenttypesof mutations,andtheir basiccharac-teristicsandthedifferentpatternsof integrationobservedsofar in Physcomitrellaareschematizedin Figure2.

The threemain criteria usedto assessthe efficiency of both genedisruptionby targetedinsertionand point mutagenesisby targetedallele replacementare(a) the ratio of targetedto randomintegrationeventsobserved uponintegrative

8

transformation, (b) the extent of sequence homology required to efficiently targetchromosomal loci, and (c) the ratio of insertion versus replacement events moni-tored upon targeted integration of replacement vectors. Each of these parametershas now been assessed inPhyscomitrellaand is reviewed in detail below.

TRANSFORMATION WITH INSERTION VECTORS The efficiency of GT with insertionvectors has been assessed in three sets of experiments designed to target six artificialloci (107), three independent single copy genomic sequences (110), and a specificmember of the highly conserved multigene family encoding for chlorophyll a/bbinding proteins (cab) (47). Taking advantage of the fact that pUC-derived trans-formation vectors share approximately 3 kb of sequence homology, I have retrans-formed independent kanamycin- or hygromycin-resistant transformants carryingseveral repeats of either plasmid pHP23b (35S-neo) (88) or pGL2 (35S-hygro)(8) with the other plasmid to assess GT efficiency to artificial loci (107). Thesuccessful targeting of artificial loci with a 90% efficiency was supported by thefollowing experimental evidence: (a) Retransformation frequencies were on aver-age one order of magnitude higher in transgenic strains as compared to wild type,with a maximum RTF value of 0.1% (107); (b) cosegregation of the hygromycinand kanamycin resistance markers occurred in progeny of 90% of tested clones(107); and (c) PCR and Southern blot analyses provided molecular evidence forthe integration of direct repeats of the second plasmid by HR within the tandemrepeats of the previously integrated plasmid (S. Vlach & D. G. Schaefer, unpub-lished data). Independently, cosegregation of positive selectable markers used forsequential transformation was also observed in transformants generated to developtransposon-mediated mutagenesis inPhyscomitrella(25, 52). The successful tar-geting of three independent single copy genomic loci ranging in size from 2.4 to3.6 kb definitively demonstrated that targeted integration by HR was the dominantpath of integration of foreign DNA sequences in the genome ofPhyscomitrella(110). This conclusion was supported by the following experimental evidence:(a) Transformation rates with vectors carrying moss genomic sequences were onaverage one order of magnitude higher (RTF up to 0.1%) than those observed withnonhomologous control plasmids, (b) targeted integration of tandem repeats [2–30]of the vectors by HR was confirmed by Southern blot analyses in 75–100% of theplants analyzed, and (c) integration of the targeting plasmids in single loci andmeiotic stability of the targeted loci was demonstrated by segregation analyses.

The specificity of GT inPhyscomitrellawas further assessed in experimentsdesigned to disrupt theZLAB1gene (67), one of the 15 members of the cab multi-gene family (95). Using an insertion vector carrying 1 kb of the ZLAB1 genomicsequence, disruption of only the true homologous gene family member by targetedintegration of tandem repeats of the vector occurred in 30% of the transformantsanalyzed, although nucleotide sequence homology between different members wasas high as 87–93% (47).

Taken together, these data demonstrate that the efficiency of targeted insertionalmutagenesis with insertion vectors sharing more than 2.0 kb of sequence homology

9

with the moss genome is in the range of 90%, which is unique in the plant kingdom,is much more efficient than it is in mouse ES cells (32), and is comparable withefficiencies observed in budding yeast (108, 110). Targeted insertion of several[1–50] direct repeats of the transforming plasmid in the corresponding target locusoccurs by a single HR event, allowing the direct generation of insertional mutationssuch as loss-of function (see Figure 2AI ). The specificity of the recombinationprocess enables the disruption of a specific member within a gene family. Thedata also indicate that integrated transgenes provide accessible targets for thesubsequent integration of additional transforming vectors, a feature that must beconsidered in designing experiments for the sequential transformation of strains.

TRANSFORMATION WITH REPLACEMENT VECTORS The minimal and optimalamounts of sequence homology required to target genomic sequences and the pat-tern of targeted integration of replacement vectors have been defined in experimentsdesigned to target thePhyscomitrellaadenine phosphoribosyl transferase gene(Ppapt) (D. G. Schaefer, M. Chakhparonian, S. Vlach, K. von Schwartzenberg,N. Houba-Herin, C. Pethe, J.-P. Zr¨yd & M. Laloue, unpublished data). Theaptgene provides an ideal model locus for GT studies in haploid organisms as its lossof function confers resistance to adenine analogues such as 2,6-diaminopurine(DAP) and can thus be directly selected (76). Using both cDNA- and gDNA-basedreplacement vectors, we have been able to establish the following parameters:(a) Targeted integration by HR is possible using stretches of continuous sequencehomology ranging from 50–200 bp (48) and reaches a maximum when targetingsequences extending over 1–2 kb are used; (b) under optimal conditions, RTFfor thePPaptlocus are in the range of 0.1%, and targeting efficiencies are above90% (i.e., more than 9/10 transformation events involve HR); (c) targeted replace-ment of genomic sequences mediated by two HR events (i.e., allele replacement)(Figures 2BII,III ) and targeted insertion mediated by a single HR coupled with anonhomologous end-joining reaction (NHEJ) (Figure 2BV) occur at similar rateswhen replacement vectors are used; (d ) replacement of the genomic sequenceby a single copy of the replacement cassette is observed in approximately halfof the replacement events (Figure 2BII ), whereas the structure of the other halfof replacement events is characterized by the presence of direct repeats of eitherthe replacement cassette (Figure 2BIII ) or the entire transforming plasmid (notillustrated); (e) targeted insertion of replacement vectors is characterized by theintegration of tandem direct repeats [1–30 copies] of either the replacement cas-sette or the entire plasmid (Figure 2BIV ); ( f ) NHEJs that occur between directrepeats of the transforming DNA or at the junctions between plasmid sequencesand the genome are identical within one set of direct repeats but vary in differenttargeted transformants. These reactions follow the main features described in otherplants for NHEJ events, i.e., deletion of plasmid sequences followed by rejoiningwithin short stretches of microhomology (43).

Taken together, these data demonstrate that targeted mutagenesis with replace-ment vectors is as efficient inPhyscomitrellaas in S. cerevisiae. Sequence

10

homologyrequirementsandtheratioof targetedto randomintegrationeventsarecomparableto thoseobserved for targetedmutagenesisin budding yeast(100)and are more favorable here than for any other multicellular eukaryoticsys-temamenableto GT (108).Yet the integrationpatternof replacementvectorsinPhyscomitrella isdifferentfromthatobservedin buddingyeast,wherereplacementvectorsintegrateessentiallyby allelereplacement,andis insteadhighly similar totargetedintegrationeventsobservedin Chinesehamsterovarycells(1). Thissug-geststhatthemechanismsaccountingfor targetedintegrationin mossaredifferentfromthemodelsproposedfor buddingyeast(85).Thesynthesis-dependent-strand-annealingmodelor the recentlyproposedbreak-inducedreplicationmodel(60)accountfor the observationsdescribedabove, but clearly further studieson themechanismsof GT in wild-type Physcomitrella andin strainsmutatedin recom-binationgeneswill beneededto clarify our understandingof themechanismsofHR in thismoss.

Analysis of Gene Function by Targeted Disruption

Efficient GT to naturalloci promptedseveralgroupsto initiate functionalstudiesby targeteddisruptionof genesinvolved in four completelyunrelatedbiologicalprocesses:thedivisionof chloroplasts,thebiosynthesisof unsaturatedfattyacids,theubiquitin-mediatedproteolyticpathway, andthepurinesalvagepathway. Usinga cDNA-basedreplacementvectorcarryingapproximately1 kb of discontinuoushomologoussequence,Reski’sgroupin Freiburg successfullydisruptedtheftsZ1geneof Physcomitrella in 14%of thetransgenicplantsanalyzed(119).In bacteria,the ftsZ proteinplaysa major role in the formationof the dividing ring duringcytokinesis,andnuclear-encodedeukaryoticftsZ homologueshavebeenfoundinplants(55,82).Theobservationthatcellsof ftsZknock-outmossstrainscontainasinglegiantchloroplastinsteadof the50lens-shapedchloroplastsusuallyfoundin wild-type cells provided the first functional evidencethat FtsZ proteinsareessentialfor chloroplastdivision. This observation wassubsequentlyconfirmedin Arabidopsis, usingantisenseapproaches(81,82).Subsequently, a secondftsZgenewasisolatedfromPhyscomitrellaandshowntobefunctionallynonredundantto the first one (55). More remarkably, Kiesslingandcoworkersusedtransientexpressionof ftsZ-GFP fusion cassettesto demonstratethat both FtsZ proteinsformapermanent,highlyorganizedfilamentousscaffold within mosschloroplasts.Thisnewly identifiedsubcellularstructureis reminiscentof cellularcytoskeleton,andtheauthorsproposedthatit mayfunctionto maintainplastidshapeandnamedit theplastoskeleton(55).Whethersimilar structuresalsoexist in chloroplastsofvascularplantsawaits further studies,but the work conductedin mossprovidednew insightsinto ourunderstandingof themechanismsof plastiddivision.

Lipids of Physcomitrella arecomposedof up to 30%arachidonicacid(44), apolyunsaturatedfatty acid(PUFA) foundonly in lower plants.Searchingfor newgenesinvolvedin thebiosynthesisof thisPUFA, thegroupof E.Heinzin Hamburgisolateda novel16-acyl-groupdesaturasegenefrom Physcomitrella (PPDES6)

11

and disrupted it with a genomic DNA-based replacement vector carrying approxi-mately 2 kb of sequence homology (41). Disruption ofPPDES6was supported byPCR and by Southern and Northern blot analyses in 5/5 randomly chosen trans-formants. These knock-out plants displayed no visible developmental phenotype,but analyses of their fatty acid composition revealed a clear biochemical pheno-type. Plants displayed reduced levels of arachidonic acid and other PUFA whosebiosynthesis involves a16-desaturation step and a concomitant accumulation ofputative precursors. ThatPPDES6encodes for a real16-desaturase was furtherconfirmed by feeding experiments in moss and expression studies in yeast. Thus,this work showed that targeted disruption inPhyscomitrellacan be used to iden-tify novel plant genes and to describe new biochemical pathways involved in thebiosynthesis of metabolites.

One major route for the selective degradation of proteins in eukaryotic cells is theubiquitin-proteasome pathway. The 26S proteasome, an ATP-dependent proteasecomplex that degrades ubiquitin-tagged proteins, is composed of a core proteinase,the 20S proteasome, associated with a pair of regulatory complexes known as the19S complex (127). The RPN10 protein is a subunit of the 19S complex that hasaffinity for multiubiquitin chains in vitro and may function in recognizing and re-cruiting ubiquitinated proteins to be degraded by the proteasome. Yet, its functionis still a matter of debate because disruption of therpn10gene inS. cerevisiaedidnot cause any obvious phenotype (such as growth defects) except for an increasedsensitivity to amino-acid homologues (124). However the same disruption didlead to an embryonic lethal phenotype in mouse (53). In our group, Girod and col-leagues isolated and disrupted therpn10gene ofPhyscomitrella(PPrpn10) with acDNA-based replacement vector carrying 1300 bp of discontinuous homologoussequence (42). Successful disruption of the gene in 2 out of 55 transgenic strainswas confirmed by PCR, Southern blot analysis, and immunological analyses. Theknock-out moss strains were viable but displayed a dramatic developmental phe-notype characterized by altered protonemal development and the impairment ofbud formation. This developmental phenotype was complemented following ex-pression ofPPrpn10gene at ectopic locations, but mutated versions of the genefailed to fully complement the phenotype, providing insight into the role of thedifferent domains within thePPrpn10gene. Thus this work provides a telling ex-ample of the need to use different model systems for the functional characterizationof complex regulatory mechanisms such as the ubiquitin/proteasome proteolyticpathway. The experiments performed inPhyscomitrellarevealed an interactionbetween this pathway and development that was not observed in yeast and thatcould not be further investigated in mouse owing to the embryonic lethality of theknock-out phenotype.

The adenine phosphoribosyl transferase gene (apt) encodes for a maintenanceenzyme of the purine salvage pathway that recycles adenine into AMP. InAra-bidopsis, aptgenes form a small family that contains at least three expressed mem-bers (35), and theaptmutant BM3 (75) carries a mutation in theAtapt1gene (35).This mutant has approximately 1% APRTase activity, is resistant to DAP, displays

12

reducedgrowth rateandmalesterility dueto abortive pollen development(91),andis unableto form callusor regeneratein vitro (64),but whethertheobservedphenotypecould be attributed to impairedpurineor cytokinin metabolismwasunclear(35,75).Physcomitrella hasonly oneaptgene(Ppapt) thatcloselyresem-blestheArabidopsisAtapt1geneat thelevel of geneticstructureandbiochemicalactivity of the encodedprotein.Physcomitrella apt null allelesdisplaya strongdevelopmentalphenotypecharacterizedby abortive gametophoredevelopment.In addition,thesemutantsdisplayincreasedsensitivity to exogenouslysuppliedadenine,which promptedusto reanalyzetheArabidopsismutantandto observethatBM3 alsodisplayedhypersensitivity to adenine,anobservationthathadnotbeenmadepreviously. Finally, we could show that expressionof the Arabidop-sis Atapt1 or Atapt2 genesat ectopiclocationsfully complementsthe apt nullphenotypein Physcomitrella, giving riseto fertile plantsthataresensitive to 2,6-diaminopurineand resistantto adenine(D. G. Schaefer, M. Chakhparonian,S.Vlach,K. vonSchwartzenberg,N. Houba-Herin,C.Pethe,J.-P. Zryd& M. Laloue,unpublisheddata).Thus,disruptingtheapt genein Physcomitrella providedtwonovel piecesof information:First,Arabidopsishomologuescanfunctionallycom-plementmutationsin Physcomitrella. Second,thedevelopmentalphenotypeob-served in mossapt null strainswasassociatedwith the inability of the plant torecycleadenine,anobservationthatwascorrelatedin Arabidopsis.

Thestudiesreviewedabove arethefirst examplesof theuseof efficient GT inPhyscomitrella to providerapidaccessto valuablenew informationaboutgeneticnetworksthatcontroldiversebiologicalprocesses.

Additional Methodological Development

FUNCTIONAL STUDIES OF PROMOTERS IN PHYSCOMITRELLA Thegeneralizeduseof Physcomitrella for targetedtransgenesiswill requirethe useof several well-characterizedpromotersfor theconstructionof constitutiveandinducibleexpres-sionvectors,asaresultof thefollowing featuresof Physcomitrella transformation:(a) Promotersequencesintegratedin themossgenomeprovide targetsitesfor thesubsequentintegration of an expressioncassettedriven by the samepromoter.(b) Preliminaryevidenceindicatesthat when two transgenesare driven by thesamepromoter, they maybestochasticallysilencedby amechanismthatremainsto bedetermined(D. G.Schaefer, unpublishedobservation;52).Two typesof pro-moters,endogenousmosspromotersandheterologouspromoters,arediscussedbelow. Theirusein transformationexperimentswill enableverydifferenttypesofstudiesto beperformed.

Thepresenceof anendogenousmosspromoterin a transformationvectorwillresultin thepredominanttargetedintegrationof thevectorinto thehomologouspromoterlocus.This providesa straightforward tool for monitoringthe activityof the correspondingpromoterin a native chromosomalregulatorycontext andfor characterizingpromoterfunctionalelementsafter replacementof the nativesequencewith anin vitro mutatedone.Suchefficient andaccuratepromoter-trap

13

strategies will certainly be broadly applied to characterize moss promoters and tomanipulate expression levels of endogenous genes.

Yet the use of endogenous moss promoters is clearly not ideal for functionalstudies in which investigators must be able to alter gene expression levels, and thisrequires the use of several different heterologous promoters to drive expressioncassettes that are to be introduced in the same transformed strain. So far, the riceactin-1 gene promoter (72) and the maize ubiquitin-1 gene promoter (19) have beenshown to promote GUS expression levels following transient expression in proto-plasts that were comparable to those observed in rice and maize (17, 131; D. G.Schaefer & M. Uz´e, unpublished data). In contrast, the 35S-GUS construct pBI221(50) promoted only weak GUS activity (17, 107, 131). Yet transfection with animproved version of pBI221 [pNcoGUS (12)] carrying an optimized translationinitiation context as defined by Kozak (59) resulted in a 10-fold increase in GUSactivity (17, 107, 130, 131), whereas the introduction of the 5′ untranslated intronof the rice actin-1 gene promoter between the 35S promoter and theuidA gene[pBCGA4 (72)] resulted in an additional 5-fold increase in GUS activity as com-pared to pNcoGUS, activities that were in the same range as those observed withconstructs driven by the rice actin-1 gene promoter (17). These observations indi-cate that optimizing the expression of transfected genes in moss follows the samebasic principles as those observed in other plants and that characterized hetero-logous promoters from higher plants may provide valuable tools for the futureconstruction of constitutive expression cassettes forPhyscomitrella.

HETEROLOGOUS CONDITIONAL PROMOTERS The development of a tightly regu-lated conditional gene expression system forPhyscomitrellais essential for switch-ing transgenes on and off and for the generation of conditional mutations. This isespecially important for the study of essential genes in the haploid background ofthe moss gametophyte because such experiments can then directly be performedby replacing the natural promoter with a conditional one that can be turned onduring the selection period and subsequently switched off for functional studies.Chemically inducible gene expression systems are well suited for such approachesand should ideally meet three objectives: (a) The inducible promoter should not beleaky under noninduced conditions; (b) the induction factor should reach at leasttwo orders of magnitude, and its modulation should be positively correlated withthe concentration of the chemical inducer; and (c) the chemical inducer should notinterfere with plant biological processes (38).

Knight and colleagues have studied the regulation of the abscissic acid (ABA)–and osmotic stress–induced wheatEmgene promoter (69). In transgenic strains,Em-driven GUS activity was readily detectable upon noninduced conditions (prob-ably in response to low concentrations of endogenous ABA) and was induced byfactors of 5- and 30-fold following treatment with 0.44 M mannitol or 10−4 MABA, respectively (58). The Em-GUS cassette was strongly expressed in trans-ient gene expression assay, most likely in response to the presence of mannitol inthe protoplast culture medium, and could be induced fourfold further upon ABA

14

treatment. Detectable GUS activity under noninduced conditions suggested thatsome of the molecular mechanisms involved in ABA and stress responses mightbe conserved between mosses and cereals (58). This assumption was supportedby gel retardation assay and DNAseI footprint analysis, which showed that factorspresent in moss nuclear protein extracts bind specifically to the same abscissicacid response element (ABRE) previously identified from studies with wheat nu-clear extracts (46). Thus, although the wheat Em promoter shows strong inducibleactivity in Physcomitrella, this promoter is clearly not ideal for modulating geneexpression levels in this moss as it will be continually induced by endogenousfluctuations in ABA levels in the plant.

Tetracycline-regulated gene expression inPhyscomitrellawas reported usingthe tetracycline-repressible system developed by Gatz et al. (130). In the presenceof 1–10 mg/L tetracycline in culture medium, GUS activities measured both bytransient gene expression assays and in transgenic strains were hardly detectable.In the absence of tetracycline, GUS expression was induced approximately 100-fold within 24 h to reach specific activities in the range of those observed withGUS reporter cassettes driven by the rice actin-1 gene promoter (130). Expressionof the GUS gene could subsequently be repressed upon transfer to tetracycline-supplemented medium, but the kinetics of repression could not be determinedowing to the relatively long half-life of the GUS protein in vivo. Although con-tinuous growth on tetracycline may be deleterious to moss growth (M. Laloue,personal communication), tetracycline-regulated gene expression provides a verypromising system for conditional expression of genes inPhyscomitrella.

The two-component glucocorticoid-inducible gene expression GVG system de-veloped by Aoyama & Chua (3) is composed of a constitutive cassette expressingthe glucocorticoid receptor fused to a transcriptional activator and of a reporterconstruct that carries the inducible promoter driving a reporter gene. This systemhas great potential as it should not, a priori, interfere with plant biological pro-cesses. Moreover, wild-typePhyscomitrellagrown on medium supplemented withthe synthetic glucocorticoid dexamethasone at concentrations of up to 0.5 mM didnot display any obvious alteration of growth or development. Using transient geneexpression assay in protoplasts, Chakhparonian (17) showed that GUS expressioncan be induced up to 50-fold in protoplasts grown in the presence of 30µM dex-amethasone. Yet, transfection with the reporter construct alone yielded detectableGUS activities, indicating that the inducible promoter was leaky, an observationthat was previously reported in tobacco (3). Nevertheless, glucocorticoid-regulatedgene expression represents a potentially useful system for modulating gene expres-sion inPhyscomitrella.

Conditional expression systems such as those based on the Gal4 promoter inbudding yeast (54) or thenmt1promoter in fission yeast (5, 71) have providedessential tools for the functional dissection of genetic networks in these modelorganisms. Preliminary data obtained inPhyscomitrellaindicated that both thetetracycline-regulated and the glucocorticoid-inducible systems could be used tomodulate the expression of genes in this moss, a picture that is similar to that

15

observed in higher plants (39). Additional studies are required to further improveconditional gene expression inPhyscomitrella, and the recently developed TGVsystem that combines dexamethasone induction and tetracycline repression (10)as well as the oestrogen-inducible XVE system that appears less leaky than GVG(132) remain to be tested.

CRE/LOX-MEDIATED SITE-SPECIFIC RECOMBINATION IN PHYSCOMITRELLA The site-specific recombination Cre/lox system of bacteriophage P1 is used to introducesite-specific mutations in a genome. It is composed of the Cre recombinase and asmall DNA-recognition target site, theloxPsite, that consists of two 13-bp invertedrepeats flanking an 8-bp asymmetric spacer that defines the orientation of the site.Recombination events mediated by the Cre recombinase lead to the excision or theinversion of DNA sequences located between two direct or inverted repeats ofloxPsites, respectively. In plants, the Cre/lox system was successfully used to generatemutations as diverse as translocations, deletions, or inversions and to regulateconditional gene expression or site-specific integration (83, 84, 126). One of itsdirect applications in plant biotechnology is the excision of selectable markers andplasmid sequences in genetically modified crops (31, 133). Yet, it is in combinationwith efficient GT that this system realizes its full potential because the latter allowsloxP sites to be introduced at specific locations within the genome. The utility ofthis approach is well illustrated by the extensive use of the Cre/lox system tomanipulate the genomes of yeast (103) and mouse (105). In this respect, setting upmethodologies for accurate targeted mutagenesis inPhyscomitrellanecessitatesthe development of an efficient site-specific recombination system.

The possibility of using the Cre/lox system to excise plasmid repeats inserted inthe genome ofPhyscomitrella patenswas recently demonstrated by Chakhparonian(17) in two independent strains carrying one or five copies of a plasmid containinga 35S nptII cassette flanked by twoloxP sites in direct orientation. Followingtransient expression of a 35S Cre cassette [pMM23 (30)], protoplasts of thesestrains were screened for plasmid excision, based on loss of the resistance marker.Loss of plasmid repeats by Cre-mediated site-specific recombination was observedin 3–10% of protoplast-derived colonies and was confirmed by PCR, Southernblot analyses, and sequence analysis. Such an efficient excision is in the range ofthat observed in other plant systems following transient expression of Cre. Thismethodology permits the rapid and easy identification of recombined clones in asmall population. This finding has one immediate consequence for strategies oftargeted mutagenesis inPhyscomitrella: loxPsites should be built into the designof every vector to be used for transformation. Their presence allows the gen-eration of point mutations by gene conversion (as illustrated in Figure 2B), therecycling of selectable markers for sequential transformation, or the eliminationof integrated plasmid repeats. Ultimately the combination of efficient GT withsite-specific recombination will enable the study of plant biological processes inPhyscomitrellawith the type of accurate targeted mutagenesis strategies that havebeen developed in budding yeast and mouse (77, 103).

16

TAGGED MUTAGENESIS AND GENE- AND ENHANCER-TRAP SCREENS Thus far, I havediscussed the potential of targeted transgenesis inPhyscomitrellafor the functionalanalysis of plant genes. The unique opportunity that this offers in the plant kingdomcontrasts singularly with the limited genetic knowledge we have of the biologyof Physcomitrella. This prompted Hasebe’s group to use large-scale insertionalmutagenesis approaches in combination with efficient GT to generate collectionsof tagged and trapped lines ofPhyscomitrella.In these collections, new phe-notypes and gene expression patterns can be identified and immediately furthercharacterized (46a, 79). The first report from Nishiyama et al. (79) described theconstruction of two moss genomic libraries mutagenized by shuttle mutagenesis(116) with a mini-transposon designed either to tag moss genes with anptII markeror to trap them with auidA marker. Following transformation ofPhyscomitrellawith these libraries, tagged lines displaying a visible developmental phenotypeand trapped lines identified by GUS histochemical staining were recovered atfrequencies of 4% and 3%, respectively (79). A preliminary analysis of thesetransformants indicated that such an approach was suitable for generating mutantsthat displayed a broad range of developmental phenotypes and for identifyinggene expression patterns that cover many of the different developmental stages orcell types found inPhyscomitrella. Using the same overall experimental strategy,Hiwatashi et al. (46a) developed furtheruidA-tagged gene-trap and enhancer-trapelements and used them to transformPhyscomitrellaunder both homologous andnonhomologous conditions. Transformation rates were one order of magnitudehigher under homologous conditions, suggesting that the use of mutagenized geno-mic libraries enhanced the yield of tagged and trapped lines. GUS-positive gene-trap and enhancer-trap lines were recovered at frequencies of 4% (235/5637) and29% (1073/3726), respectively, which represent very high frequencies for theapplication of such strategies. More importantly, this work demonstrates that thetrapped gene can be efficiently identified using RACE PCR withuidA-specificprimers even though multiple copies of the mini-transposon were inserted in a trapline (46a). To account for the presence of multiple copies of the mini-transposonin tagged and trapped lines, the authors proposed that mini-transposon-taggedsequences concatenate extrachromosomally to generate large tandem arrays ofmutagenized genomic sequences that subsequently integrate by HR in a singlelocus within the terminal genomic sequence of the array (46a), a hypothesis thatfits well with the structural feature of replicative and integrative transformationof Physcomitrellareviewed here. This enables the study of moss genes or en-hancers trapped upon shuttle mutagenesis without the simultaneous disruption oftheir original locus in the plant, a situation that presents advantages in the haploidgametophyte ofPhyscomitrella.

The two studies reviewed above represent major progress in the developmentof Physcomitrellafor plant functional genomics. They provide the first large-scalecollection of tagged insertional mutants inPhyscomitrellaavailable to the public inwhich a broad range of new phenotypes and new gene expression patterns in vivocan be identified. They also show that subsequent isolation of the mutated gene is

17

facilitatedby thepresenceof themini-transposontagandprovidethefirstexampleof a completereversegeneticanalysissuccessfullycompletedin Physcomitrella.Finally, they show that, in a mannersimilar to its applicationin buddingyeast(14), shuttlemutagenesiscanbe usedto undertake large-scaleanalysisof geneexpression,proteinlocalization,andgenedisruptionin Physcomitrella.

CONCLUSION

ThedatareviewedabovedemonstratethatGTefficiency in Physcomitrella isamaz-ingly high andcanonly becomparedwith thatobserved in buddingyeast(108).Underoptimalconditions,suchefficiency enableswithin fiveweekstheisolationof dozensof targetedmutantsfrom a singletransformationexperiment(41, 107,110). The efficiency of GT in Physcomitrella, combinedwith the possibility ofusingthesite-specificCre/loxsystem,allows theapplicationof thesophisticatedstrategiesof targetedmutagenesis,which so far have beenappliedonly in yeastandmouseEScells.Thus,at themethodologicallevel, Physcomitrella fulfills theultimaterequirementfor itsdevelopmentasamodelsystembecausethegenerationof predeterminedpointmutationsatany locationof its genomeis directlyfeasible.

Prospectsfor improving genetargetingfrequenciesin animals(125)andplants(74) (126)have beenreviewedrecently. The ideal situationwould be to achievea useful rateof GT in a wild-type background,as this would allow unambigu-ous interpretationof the data.This requirementis completelyfulfilled by thePhyscomitrella system.Thestrategy basedon thegenerationof a recombinationsubstratein vivo asdescribedby Rong& Golic in Drosophila(99) hasyet to betestedin higherplants.Theuseof promotertrappositive-negative screensasap-pliedin mammaliancells(115)alsolookspromising.Althoughit will notincreasetheyieldof targetedintegrationevents,it will mainlystrengthenthescreenfor suchevents.Currentstrategiesalsoincludeattemptsto improve theratioof targetedtorandomintegrationeventsby modulatingtheactivity eitherof endogenousgenesinvolved in DNA recombinationand repair or following the overexpressionofheterologousgenesinvolved in thesepathways(74, 126).Yet, asemphasizedbyVasquezetal. (125),suchstrategiesshouldbeadoptedwith greatcautionandide-ally shouldbeperformedonlybytransientexpressionof advantageousgenesratherthanby their inactivationor constitutive expression,assuchmutationsmayseri-ouslyimpair themaintenanceof genomeintegrity. This is dramaticallyillustratedby theobservationthat lossof functionof theRad50genein Arabidopsisstimu-latesintrachromosomalrecombinationin planta(40) but simultaneouslyleadstoplantsterility (36) andacceleratedsenescencedueto telomereinstability (37). Inthis respectagain,Physcomitrella patensprovidesanidealmodelsystem.

Why is genetargetingsoefficient in Physcomitrella? We (108,110,111)pre-viously proposedthat it could be correlatedwith the dominanceof the haploidgametophytein its life cycle, whereasReski (93) proposedthat efficient GTcould be correlatedwith the fact that protoplastsare highly synchronizedandblockedinG2duringtransformation.Bothhypothesescanbetestedexperimentally.

18

In buddingyeast,extensivegeneticandbiochemicalstudieshaveshown thateffi-cientGT is tightly associatedwith therepairof DNA double-strandbreaks(85).Yetamutationthatwouldspecificallyaltertheratioof targetedto randomintegra-tion eventsobserveduponintegrative transformationhasnot yet beenidentified,indicatingthatthemechanismsthatcontrolthisratioarenoteasytounravel (85).Inthisrespect,Physcomitrella providesavaluablealternatesystemto studythispro-cess.The identificationandfunctionalanalysisof mossgenesinvolved in DNArepair and recombinationmay provide insight into thesemechanismsand ulti-matelyfacilitatethedevelopmentof strategiesto targetgenesefficiently in highereukaryotes.

In myopinion,Physcomitrella ispoisedtobecomeanextremelyvaluablemodelsystemin biology. Theefficiency of GT in Physcomitrella canandshouldbeusedimmediately, asit fills a majormethodologicalgapandadvantageouslycomple-mentsthetoolsavailablefor thestudyof plantfunctionalgenomics.Thesuitabilityof thePhyscomitrella systemto studyplantbiology is supportedby thestructuraland functional similarities observed betweenmossesand higher plantsboth inbiologicalprocessesandatthegeneticlevel.Theplantscientificcommunitycouldadvantageouslybenefit from a whole genomesequencingprogramto gain therequiredinformationaboutPhyscomitrella genesfor future functionalgenomicstudies.The field is moving fast,andwe have all the tools in our handsto de-velopthePhyscomitrella systemasa“greenyeast”for addressingplantbiologicalquestions.Furthermore,theuseof GT technologyin Physcomitrella allowsfor thestudyof complex developmentalprocessessuchasorganogenesisor theestablish-mentof a bodyplan,processesthatcannotbestudiedin buddingyeast.As such,this plant may prove to be in a uniquepositionto unravel the geneticnetworkscontrollingcomplex developmentalprocessesin multicellularEukaryotes.

ACKNOWLEDGMENTS

I would like to thankJean-PierreZryd for his constantsupportandthevaluablediscussionswehadoverthepast15years.I acknowledgemy formercollaboratorsSusanVlach andMikhail Chakhparonianfor their contribution andenthusiasmandMichel Laloueandhiscollaboratorsin Versaillesfor thefruitful collaborationwe have hadover thepastsevenyears.I amvery gratefulto MichaelLawton forhiscritical review andvaluablesuggestionson themanuscript.

LITERATURE CITED

1. Adair GM, ScheererJB, BrothermanA,McConville S,WilsonJH,etal.1998.Tar-getedrecombinationat theChineseham-ster APRT locus using insertion versus

replacementvectors. Somat. Cell Mol.Genet.24:91–105

2. AdamsMD, CelnikerSE,Holt RA, EvansCA, GocayneJD,etal.2000.Thegenome

19

sequence ofDrosophila melanogaster.Science287:2185–95

3. Aoyama T, Chua N-H. 1997. A gluco-corticoid-mediated transcriptional induc-tion system in transgenic plants.Plant J.11:605–11

3a. Arabidopsis Genome Initiative. 2000. An-alysis of the genome sequence of the flow-ering plantArabidopsis thaliana. Nature408:796–815

4. Ashton NW, Champagne CEM, WeilerT, Verkoczy LK. 2000. The bryophytePhyscomitrella patensreplicates extra-chromosomal transgenic elements.NewPhytol.146:391–402

5. Basi G, Schmidt E, Maundrell K. 1993.TATA box mutations in theSchizosaccha-romyces pombenmt1 promoter affect tran-scription efficiency but not transcriptionstart point or thiamine repressibility.Gene123:131–36

6. Bennetzen JL. 1999. Plant genomics takesroot, branches out.Trends Genet.15:85–87

7. Bevan MW, Flavell RB, Chilton M-D.1983. A chimaeric antibiotic resistancegene as a selectable marker for plant celltransformation.Nature304:184–87

8. Bilang R, Iida S, Peterhans A, PotrykusI, Paszkowski J. 1991. The 3′ terminalregion of the hygromycin-B-resistancegene is important for its activity inEsche-richia coli andNicotiana tabacum. Gene100:247–50

9. Blattner FR, Plunkett GI, Bloch CA, PernaNT, Burland V, et al. 1997. The completegenome sequence ofEscherichia coliK-12. Science277:1453–74

10. Bohner S, Lenk I, Rieping M, Herold M,Gatz C. 1999. Transcriptional activatorTGV mediates dexamethasone-inducibleand tetracycline-inactivable gene expres-sion.Plant J.19:87–95

11. Bonnema AB, Peytavi R, Van DaelenRAJ, Zabel P, Grimsley N. 1993. Develop-ment of an in vivo complementation sys-tem for identification of plant genes us-ing yeast artificial chromosomes (YACS).

In Bacterial Wilt, ed. GL Hartmann, ACHayward, pp. 177–82. Canberra: Aust.Cent. Int. Agric. Res.

12. Bonneville J-M, Sanfa¸con H, Futterer J,Hohn T. 1989. Posttranscriptionaltrans-activation in cauliflower mosaic virus.Cell 59:1135–43

13. Burke DT, Carle GF, Olson MV. 1987.Cloning of large segments of exogenousDNA into yeast by means of artificialchromosome vectors.Science236:806–12

14. Burns N, Grimwade B, RossmacdonaldPB, Choi EY, Finberg K, et al. 1994.Large-scale analysis of gene expression,protein localization and gene disruptionin Saccharomyces cerevisiae. Genes Dev.8:1087–105

15. Caenorhabditis eleganssequencing con-sortium. 1998. Genome sequence of thenematodeC. elegans: a platform for in-vestigating biology.Science282:2012–19

16. Capecchi MR. 1989. Altering the genomeby homologous recombination.Science244:1288–92

17. Chakhparonian M. 2001.Developpementd’outils de la mutagenese ciblee par re-combinaison homologue chez Physcomi-trella patens. PhD thesis. Univ. Lausanne,Switz. www.unil.ch/lpc/docs/

18. Champagne C, Ashton N. 2001. Ances-try of KNOX genes revealed by bryophyte(Physcomitrella patens) homologs.NewPhytol.150:23–36

19. Christensen AH, Quail PH. 1996. Ubiqui-tin promoter-based vectors for high-levelexpression of selectable and/or screen-able marker genes in monocotyledonousplants.Trans. Res.5:213–18

20. Clayton RA, White O, Ketchum KA, Ven-ter JC. 1997. The first genome from thethird domain of life.Nature387:459–62

21. Cove DJ. 1983. Genetics of Bryophyta. InNew Manual of Bryology, ed. RM Schus-ter, pp. 222–31. Tokyo, Jpn.: Hattori Bot.Lab.

22. Cove DJ. 1992. Regulation of develop-ment in the moss,Physcomitrella patens.

20

In Developmental Biology. A MolecularGenetic Approach, ed. S Brody, DJ Cove,S Ottolenghi, VEA Russo, pp. 179–93.Heidelberg: Springer-Verlag

23. Cove DJ. 2000. The moss,Physcomitrellapatens. J. Plant Growth Regul.19:275–83

24. Cove DJ, Ashton NW. 1988. Growth regu-lation and development inPhyscomitrellapatens: an insight into growth regulationand development of bryophytes.Bot. J.Linn. Soc.98:247–52

25. Cove DJ, Kammerer W, Knight CD,Leech MJ, Martin CR, et al. 1991. De-velopmental genetic studies of the mossPhyscomitrella patens. In Molecular Bio-logy of Plant Development, ed. GI Jenkins,W Schuch, pp. 31–43. Cambridge, UK:Co. Biol.

26. Cove DJ, Knight CD. 1987. Gravitrop-ism and phototropism in the mossPhys-comitrella patens. In Developmental Mu-tants in Higher Plants, ed. H Thomas,D Grierson, pp. 181–96. Cambridge, UK:Cambridge Univ. Press

27. Cove DJ, Knight CD. 1993. The mossPhyscomitrella patens, a model systemwith potential for the study of plant re-production.Plant Cell5:1483–88

28. Cove DJ, Knight CD, Lamparter T. 1997.Mosses as model systems.Trends PlantSci.2:99–105

29. Cronk QCB. 2001. Plant evolution anddevelopment in a post-genomic context.Nat. Rev. Genet.2:607–19

30. Dale EC, Ow DW. 1990. Intra- and inter-molecular site-specific recombination inplants mediated by bacteriophage P1 re-combinase.Gene91:79–85

31. Dale EC, Ow DW. 1991. Gene transferwith subsequent removal of the selectiongene from the host genome.Proc. Natl.Acad. Sci. USA88:558–62

32. Deng C, Capecchi MR. 1992. Reexamina-tion of gene targeting frequency as a func-tion of the extent of homology between thetargeting vector and the target locus.Mol.Cell. Biol.12:3365–71

33. Doonan JH, Duckett JD. 1988. The bryo-

phyte cytoskeleton: experimental and im-munofluorescence studies of morphogen-esis.Adv. Bryol.3:1–31

34. Eisen MB, Brown PO. 1999. DNA arraysfor analysis of gene expression.MethodsEnzymol.303:179–205

35. Gaillard C, Moffat BA, Blacker M, LaloueM. 1998. Male sterility associated withAPRT deficiency isArabidopsis thalianaresults from a mutation in the gene APT1.Mol. Gen. Genet.257:348–53

36. Gallego ME, Jeanneau M, Granier F,Bouchez D, Bechtold N, et al. 2001. Dis-ruption of the Arabidopsis RAD50 geneleads to plant sterility and MMS sensitiv-ity. Plant J.25:31–41

37. Gallego ME, White CI. 2001. RAD50function is essential for telomere mainte-nance in Arabidopsis.Proc. Natl. Acad.Sci. USA98:1711–16

38. Gatz C. 1996. Chemically inducible pro-moters in transgenic plants.Curr. Opin.Biotech.7:168–72

39. Gatz C. 1997. Chemical control of geneexpression.Annu. Rev. Plant Physiol.Plant Mol. Biol.48:89–108

40. Gherbi H, Gallego ME, Jalut N, Lucht J,Hohn B, et al. 2000. Homologous recom-bination in planta is stimulated in the ab-sence of Rad50.EMBO J.2:287–91

41. Girke T, Schmidt H, Z¨ahringer U, ReskiR, Heinz E. 1998. Identification of a novel16-acyl-group desaturase by targeted dis-ruption inPhyscomitrella patens. Plant J.15:39–48

42. Girod PA, Fu HY, Zryd JP, Vierstra RD.1999. Multiubiquitin chain-binding sub-unit MCB1 (RPN10) of the 26S protea-some is essential for developmental pro-gression inPhyscomitrella patens. PlantCell 11:1457–71

43. Gorbunova V, Levy AA. 1999. How plantsmake ends meet: DNA double-strandbreak repair.Trends Plant Sci.4:263–69

44. Grimsley NH, Grimsley JM, Hartmann E.1980. Fatty acid composition of mutantsof the mossPhyscomitrella patens. Phy-tochemistry20:1519–24

21

45. Gritz L, Davies J. 1983. Plasmid-encodedhygromycin B resistance: the sequence ofhygromycin B phosphotransferase geneand its expression inEscherichia coliandSaccharomyces cerevisiae. Gene25:179–88

46. Guiltinan MJ, Marcotte WR, QuatranoRS. 1990. A plant leucin zipper proteinthat recognizes an abscissic acid responseelement.Science250:267–71

46a. Hiwatashi Y, Nishiyama T, Fujita T,Hasebe M. 2001. Establishment of gene-trap and enhancer-trap systems in themossPhyscomitrella patens. Plant J.28:105–16

47. Hofmann AH, Codon AT, Ivascu C, RussoVEA, Knight C, et al. 1999. A specificmember of the cab multigene family canbe efficiently targeted and disrupted in themossPhyscomitrella patens. Mol. Gen.Genet.261:92–99

48. Houba-H´erin N, Reynolds S, Schaefer D,von Schwartzenberg K, Laloue M. 1997.Molecular characterization of homolo-gous recombination events in the mossPhyscomitrella patens. Plant Sci., Coll.Gen. Soc. Fr. Physiol. Veg., ed. JC Petch,A Latche, M Bouzayen, 3:22–23. Toulo-use: SFPV-INRA

49. Deleted in proof50. Jefferson RA, Kavanagh TA, Bevan MW.

1987. GUS fusions:β-glucuronidase as asensitive and versatile gene fusion markerin higher plants.EMBO J.6:3901–7

51. Jeon JS, Lee S, Jung KH, Jun SH, JeongDH, et al. 2000. T-DNA insertional mu-tagenesis for functional genomics in rice.Plant J.22:561–70

52. Kammerer W, Cove DJ. 1996. Geneticanalysis of the result of re-transformationof transgenic lines of the moss,Physco-mitrella patens. Mol. Gen. Genet.250:380–82

53. Kawahara H, Kasahara M, Nishiyama A,Ohsumi K, Goto T, et al. 2000. Develop-mentally regulated, alternative splicing oftheRpn10gene generates multiple formsof 26S proteasome.EMBO J.19:4144–53

54. Keegan L, Gill G, Ptashne M. 1986. Sep-aration of DNA binding from the trans-cription-activating function of eukaryoticregulatory protein.Science231:699–704

55. Kiessling J, Kruse S, Rensing SA, Har-ter K, Decker EL, et al. 2000. Visualiza-tion of a cytoskeleton-like FtsZ networkin chloroplasts.J. Cell Biol.151:945–50

56. Knight CD. 1994. Studying plant devel-opment in mosses: the transgenic route.Plant Cell Environ.17:669–74

57. Knight CD. 2000. Moss, molecular toolsfor phenotypic analysis. InEncyclopediaof Cell Technology, ed. E Spier, pp. 936–44. New York: Wiley

58. Knight CD, Sehgal A, Atwal K, WallaceJC, Cove DJ, et al. 1995. Molecular res-ponses to abscissic acid and stress are con-served between moss and cereals.PlantCell 7:499–506

59. Kozak M. 1991. Structural features in eu-karyotic mRNAs that modulate the ini-tiation of translation.J. Biol. Chem.266:19867–70

60. Kraus E, Leung W-Y, Haber JE. 2001.Break-induced replication: a review andan example in budding yeast.Proc. Natl.Acad. Sci. USA98:8255–62

61. Krogan N, Ashton N. 2000. Ancestry ofplant MADS-box genes revealed by bryo-phyte (Physcomitrella patens) homolog-ues.New Phytol.147:505–17

62. Kunst F, Ogasawara N, Moszer I, Alber-tini AM, Alloni G, et al. 1997. Thecomplete genome sequence of the gram-positive bacteriumBacillus subtilis. Na-ture390:249–56

63. Lazo GR, Stein PA, Ludwig RA. 1991.A DNA transformation-competentAra-bidopsisgenomic library inAgrobacte-rium. BioTechnology9:963–68

64. Lee D, Moffat BA. 1994. Adenine salvageactivity during callus induction and plantgrowth.Physiol. Plant87:483–92

65. Lee KY, Lund P, Lowe K, Dunsmuir P.1990. Homologous recombination in plantcells afterAgrobacterium-mediated trans-formation.Plant Cell2:415–25

22

66. Leech MJ, Kammerer W, Cove DJ, Mar-tin C, Wang TL. 1993. Expression ofmyb-related genes in the mossPhyscomitrellapatens. Plant J.3:51–61

67. Long Z, Wang S-H, Nelson N. 1989. Clo-ning and nucleotide sequence analysis ofgenes coding for the major chlorophyll-binding protein of the mossPhyscomit-rella patensand of the halotolerant algaDunaliella salin. Gene76:299–312

68. Machuka J, Bashiardes S, Ruben E, Spoo-ner K, Cuming A, et al. 1999. Sequenceanalysis of expressed sequence tags froman ABA-treated cDNA library identifiesstress response genes in the mossPhysco-mitrella patens. Plant Cell Physiol.40:378–87

69. Marcotte WR, Bayley CC, Quatrano RS.1988. Regulation of a wheat promoter byabscissic acid in rice protoplasts.Nature335:454–57

70. Marini NJ, Etkin LD, Benbow RM. 1988.Persistence and replication of plasmidDNA microinjected into early embryos ofXenopus laevis. Dev. Biol.127:421–34

71. Maundrell K. 1993. Thiamine-repressibleexpression vectors pREP and pRIP for fis-sion yeast.Gene123:127–30

72. McElroy D, Blowers AD, Jenes B, Wu R.1991. Construction of expression vectorsbased on the rice actin 1 (Act1) 5′ regionfor the use in monocot transformation.Mol. Gen. Genet.231:150–60

73. Mello CC, Kramer JM, Stinchcomb D,Ambros V. 1991. Efficient gene transferin C. elegans: extrachromosomal mainte-nance and integration of transforming se-quences.EMBO J.10:3959–70

74. Mengiste T, Paszkowski J. 1999. Pros-pects for the precise engineering of plantgenomes by homologous recombination.Biol. Chem.380:749–58

75. Moffat B, Pethe C, Laloue M. 1991. Meta-bolism of benzyladenine is impaired ina mutant ofArabidopsis thalianalackingadenine phosphoribosyl transferase activ-ity. Plant Physiol.95:900–8

76. Moffat BA, Sommerville C. 1988. Posi-

tive selection for male-sterile mutants ofArabidopsislacking adenine phosphori-bosyl transferase activity.Plant Physiol.86:1150–54

77. Muller U. 1999. Ten years of gene target-ing: targeted mouse mutants, from vectordesign to phenotype analysis.Mech. Dev.82:3–21

78. Nacry P, Camilleri C, Courtial B, CabocheM, Bouchez D. 1998. Major chromo-somal rearrangements induced by T-DNAtransformation in Arabidopsis.Genetics149:641–50

79. Nishiyama T, Hiwatashi Y, Sakakibara K,Kato M, Hasebe M. 2000. Tagged muta-genesis and gene trap in the moss,Phy-scomitrella patensby shuttle mutagene-sis.DNA Res.7:9–17

80. Offringa R, de Groot JA, Haagsman HJ,Does MP, van den Elzen PJM, et al. 1990.Extrachromosomal homologous recombi-nation and gene targeting in plant cellsafter Agrobacterium-mediated transfor-mation.EMBO J.9:3077–84

81. Osteryoung KW, Pyke KA. 1998. Plastiddivision: evidence for a prokaryoticallyderived mechanism.Curr. Opin. PlantBiol. 1:475–79

82. Osteryoung KW, Stokes KD, RutherfordSM, Percival AL, Lee WY. 1998. Chloro-plast division in higher plants requiresmembers of two functionally divergentgene families with homology to bacterialFtsZ.Plant Cell10:1991–2004

83. Ow DW. 1996. Recombinase-directedchromosome engineering in plants.Curr.Opin. Biotechnol.7:181–86

84. Ow DW, Medberry SL. 1995. Genomemanipulation through site-specific recom-bination.Crit. Rev. Plant Sci.14:239–61

85. Paques F, Haber JE. 1999. Multiplepathways of recombination induced bydouble-strand breaks inSaccharomycescerevisiae. Microb. Mol. Biol. Rev.63:349–404

86. Parinov S, Sundaresan V. 2000. Functio-nal genomics inArabidopsis: large-scaleinsertional mutagenesis complements the

23

genome sequencing project.Curr. Opin.Biotechnol.11:157–61

87. Paszkowski J, Baur M, Bogucki A, Potry-kus I. 1988. Gene targeting in plants.EMBO J.7:4021–26

88. Paszkowski J, Shillito RD, Saul M, Man-dak V, Hohn T, et al. 1984. Direct genetransfer to plants.EMBO J.3:2717–22

89. Puchta H, Hohn B. 1996. From centiMor-gans to base pairs: homologous recombi-nation in plants.Trends Plant Sci.1:340–48

90. Quatrano R, Bashiardes S, Cove D, Cum-ing A, Knight C, et al. 1999. Leeds Wash-ington University Moss EST Project.http://www.ncbi.nlm.nih.gov

91. Regan S, Moffat BA. 1990. Cytochemicalanalysis of pollen development in wild-type Arabidopsis and a male-sterile mu-tant.Plant Cell2:877–89

92. Reski R. 1998. Development, genetics andmolecular biology of mosses.Bot. Acta111:1–15

93. Reski R. 1998. Physcomitrella and Ara-bidopsis: the David and Goliath of reversegenetics.Trends Plant Sci.3:209–10

94. Reski R. 1999. Molecular genetics of Phy-scomitrella.Planta208:301–9

95. Reski R, Faust M, Wang X-H, Wehe M,Abel WO. 1994. Genome analysis of amoss,Physcomitrella patens. Mol. Gen.Genet.244:352–59

96. Reski R, Reynolds S, Wehe M, Kleber-janke T, Kruse S. 1998. Moss (Physco-mitrella patens) expressed sequence tagsinclude several sequences which are novelfor plants.Bot. Acta111:143–49

97. Reutter K, Atzorn R, Hadeler B, Sch-mulling T, Reski R. 1998. Expression ofthe bacterial ipt gene in Physcomitrellarescues mutations in budding and plastiddivision.Planta206:196–203

98. Richmond T, Somerville S. 2000. Chasingthe dream: plant EST microarrays.Curr.Opin. Plant Biol.3:108–16

99. Rong YS, Golic KG. 2000. Gene target-ing by homologous recombination in Dro-sophila.Science288:2013–18

100. Rothstein R. 1991. Targeting, disruption,replacement and allele rescue: integrativeDNA transformation in yeast.MethodsEnzymol.194:281–301

101. Russell AJ, Knight MR, Cove DJ, KnightCD, Trewavas AJ, et al. 1996. The moss,Physcomitrella patens, transformed withapoaequorin cDNA responds to coldshock, mechanical perturbation and pHwith transient increases in cytoplasmiccalcium.Trans. Res.5:167–70

102. Sakakibara K, Nishiyama T, Kato M,Hasebe M. 2001. Isolation of homeodo-main-leucine zipper genes from the mossPhyscomitrella patensand the evolutionof homeodomain-leucine zipper genes inland plants.Mol. Biol. Evol.18:491–502

103. Sauer B. 1993. Manipulation of the trans-gene by site-specific recombination: useof cre recombinase.Methods Enzymol.225:890–900

104. Sauer B. 1994. Recycling selectable mar-kers in yeast.Biotechnology16:1086–88

105. Sauer B. 1998. Inducible gene targetingin mice using the Cre/lox system.Meth-ods14:381–92

106. Sawahel W, Onde S, Knight C, Cove D.1992. Transfer of foreign DNA intoPhy-scomitrella patensprotonemal tissue byusing the gene gun.Plant Mol. Biol. Rep.10:314–15

107. Schaefer DG. 1994.Molecular geneticapproaches to the biology of the mossPhyscomitrella patens. PhD Thesis. Univ.Lausanne, Switz. www.unil.ch/lpc/docs/DSThesis.htm

108. Schaefer DG. 2001. Gene targeting inPhyscomitrella patens. Curr. Opin. PlantBiol. 4:143–50

109. Schaefer DG, Bisztray G, Zr¨yd J-P. 1994.Genetic transformation of the mossPhys-comitrella patens. In Plant Protoplastsand Genetic Engineering V, ed. YPS Ba-jaj, pp. 349–64. Berlin/Heidelberg/NewYork: Springer-Verlag

110. Schaefer DG, Zr¨yd J-P. 1997. Efficientgene targeting in the mossPhyscomitrellapatens. Plant J.11:1195–206

24

111. Schaefer DG, Zr¨yd J-P. 2001. The mossPhyscomitrella patensnow and then.Plant Physiol.127:1430–38

112. Schaefer DG, Zr¨yd J-P, Knight CD, CoveDJ. 1991. Stable transformation of themossPhyscomitrella patens. Mol. Gen.Genet.226:418–24

113. Schumaker KS, Dietrich MA. 1997. Pro-grammed changes in form during moss de-velopment.Plant Cell9:1099–107

114. Schumaker KS, Dietrich MA. 1998.Hormone-induced signaling during mossdevelopment.Annu. Rev. Plant Physiol.Plant Mol. Biol.49:501–23

115. Sedivy JM, Dutriaux A. 1999. Gene tar-geting and somatic cell genetics—a re-birth or a coming of age?Trends Genet.15:88–90

116. Seifert HS, Chen EY, So M, Heffron F.1986. Shuttle mutagenesis: a method oftransposon mutagenesis forSaccharomy-ces cerevisiae. Proc. Natl. Acad. Sci. USA83:735–39

117. Springer PS. 2000. Gene traps: tools forplant development and genomics.PlantCell 12:1007–20

118. Stinchcomb DT, Shaw JE, Carr SH, HirshD. 1985. Extrachromosomal DNA trans-formation of Caenorhabditis elegans.Mol. Cell. Biol.5:3484–96

119. Strepp R, Scholz S, Kruse S, Speth V,Reski R. 1998. Plant nuclear gene knock-out reveals a role in plastid division for thehomolog of the bacterial cell division pro-tein FtsZ, an ancestral tubulin.Proc. Natl.Acad. Sci. USA95:4368–73

120. Struhl K. 1983. The new yeast genetics.Nature305:391–97

121. Theissen G, M¨unster T, Henschel K. 2001.Why don’t mosses flower?New Phytol.150:1–8

122. Thiellement H, Bahrman N, Damerval C,Plomion C, Rossignol M, et al. 1999. Pro-teomics for genetic and physiological stu-dies in plants.Electrophoresis20:2013–26

123. Thomas KR, Capecchi MR. 1987. Site-directed mutagenesis by gene targeting in

mouse embryo derived stem cells.Cell51:503–12

124. van Nocker S, Sadis S, Rubin DM, Glick-man M, Fu H, et al. 1996. The multi-ubiquitin-chain-binding protein mcb1 is acomponent of the 26S proteasome isSac-charomyces cerevisiaeand plays a non-essential, substrate-specific role in proteinturnover.Mol. Cell. Biol.16:6020–28

125. Vasquez KM, Marburger K, Intody Z,Wilson JH. 2001. Manipulating the mam-malian genome by homologous recombi-nation. Proc. Natl. Acad. Sci. USA98:8403–10

126. Vergunst AC, Hooykaas PJJ. 1999. Re-combination in the plant genome and itsapplication in biotechnology.Crit. Rev.Plant Sci.18:1–31

127. Voges D, Zwickl P, Baumeister W. 1999.The 26S proteasome: a molecular machin-ery designed for controlled proteolysis.Annu. Rev. Biochem.68:1015–68

128. Walbot V. 1992. Strategies for mutagene-sis and gene cloning using transposon tag-ging and T-DNA insertional mutagenesis.Annu. Rev. Plant Physiol. Plant Mol. Biol.43:49–82

129. Weigel D, Ahn JH, Blazquez MA, Bore-vitz JO, Christensen SK, et al. 2000. Acti-vation tagging in Arabidopsis.Plant Phys-iol. 122:1003–13

130. Zeidler M, Gatz C, Hartmann E, HughesJ. 1996. Tetracycline-regulated reportergene expression in the mossPhyscomit-rella patens. Plant Mol. Biol.30:199–205

131. Zeidler M, Hartmann E, Hughes J. 1999.Transgene expression in the mossCer-atodon purpureus. J. Plant Physiol.154:641–50

132. Zuo J, Niu Q-W, Chua N-H. 2000. An es-trogen receptor-based transactivator XVEmediates highly inducible gene expres-sion in transgenic plants.Plant J.24:265–73

133. Zuo JM, Niu Q-W, Møller SG, Chua N-H. 2001. Chemical-induced, site-specificDNA excision in transgenic plants.Nat.Biotechnol.19:157–61

25

16 Apr 2002 15:1 AR AR156-20-COLOR.tex AR156-20-COLOR.SGM LaTeX2e(2002/01/18)P1: GDL

(See legend on next page)

26

16 Apr 2002 15:1 AR AR156-20-COLOR.tex AR156-20-COLOR.SGM LaTeX2e(2002/01/18)P1: GDL

Figure 1 (See figure on previous page) The life cycle of the mossPhyscomitrellapatens(a–c). The three developmental stages of its life cycle: (a) the juvenile gameto-phyte or protonema displaying characteristic filamentous growth (bar= 1 cm); (b) theadult gametophyte composed of a colony of gametophores developing by caulinarygrowth (5 mm); (c) the parasitic diploid sporophyte that develops at the apex of a game-tophore to form a spore capsule where spores are differentiated (1.5 mm). (d–h) Closerview of the different stages of protonema development. (d) A germinating protoplast72 h after isolation displays a highly polarized structure; it has already divided once,and the daughter cell has resumed apical growth (25µm). (e) Typical filamentousgrowth pattern of a branched chloronemal filament with its tip cell undergoing a devel-opmental transition to form a caulonema (100µm). ( f–g) Two cell types with distinctultrastructural and biological characteristics form the protonema: The photosyntheticchloronema (f ) is filled with chloroplast and has cell walls perpendicular to the axis ofthe filament. The adventitious caulonema (g) contains less chloroplasts and forms cellwalls that are oblique with respect to the axis of the filament (40µm). (h) Transitionfrom filamentous growth to caulinary growth can be pinpointed to a single cell, thecaulonemal side-branch initial that differentiates into a primitive meristem, the bud(50µm). Caulonemal side-branch initials can also form chloronemal or caulonemalcells, and the relative ratio of these different developmental fates can be modulated bychemical and environmental factors and accounts for the general morphology of theprotonema. (i–k) The gametophore displays a body plan that is simpler but analogous tohigher plants. (i ) A fully differentiated gametophore is composed of a nonvascularizedphotosynthetic stem carrying simple leaves formed by a single layer of photosyntheticcells and displaying phyllotaxis along the stem axis and of filamentous pigmentedrhizoids that radiate from the base of the stem (2 mm). Reproductive organs differen-tiate in the apical part of the gametophore: the male antheridia (j ), borne in the leafaxils, and the female archegonia (k), borne terminally (100µm). Ciliated antherozoidsformed in the antheridia swim and fuse with the egg cell located in the archegonia togive rise to a diploid cell that further differentiates into the sporophyte (c).

27

16 Apr 2002 15:1 AR AR156-20-COLOR.tex AR156-20-COLOR.SGM LaTeX2e(2002/01/18)P1: GDL

(See legend on next page)

28

16 Apr 2002 15:1 AR AR156-20-COLOR.tex AR156-20-COLOR.SGM LaTeX2e(2002/01/18)P1: GDL

Figure 2 (See figure on previous page) (A) Insertion vectors contain one homol-ogous targeting fragment (F box) of your favorite sequence (YFS) cloned next to amarker encoding a resistance gene (R box). Targeted integration occurs by a single HRevent (dashed lines) and results in the insertion of one or several copies of the vectorflanked by two repeats of the targeting sequence (F ). Insertion vectors are suitablefor generating insertional gene disruption (AI ) but not for generating point mutations.(B) In replacement vectors, the resistance gene (R box) flanked by 2loxPsites in directorientation (yellow triangle) is inserted between two homologous targeting fragments(M∗ andS boxes, where M∗ carries a point mutation) of your most favorite sequence(YMFS). Targeted integration occurring by two HR events within M and F leads to thereplacement either of the endogenous chromosomal sequence by one (BII ) or severalrepeats of the replacement cassette (BIII ), or of the entire plasmid (not shown). Whenthe selectable marker flanked by twoloxP sites is cloned in an intron, subsequentexpression of the Cre recombinase eliminates the selectable marker and the plasmidrepeats to generate accurate targeted point mutations (BIV) (M→M∗). Yet, replace-ment vectors can also integrate by a single HR event within either the M or F sequenceleading to an insertional disruption, as described with insertion vectors (V for HR in M).

29