Farnesylation is involved in meristem organization in Arabidopsis

Abstract. Although studies in plant and animal cellculture systems indicate farnesylation is required fornormal cell cycle progression, how this lipid modi®ca-tion of select proteins translates into whole-organismdevelopmental decisions involving cell proliferation ordi�erentiation is largely unknown. The era1 mutant ofthe higher plant Arabidopsis thaliana (L.) Heynh. o�ers aunique opportunity to understand the role farnesylationmay play in regulating various processes during thedevelopment of a multicellular organism. Loss offarnesylation a�ects many aspects of Arabidopsis growthand development. In particular, apical and axillarymeristem development is altered and these phenotypesare contingent on the growth conditions.

Key words: Arabidopsis (farnesylation, mutant) ±Farnesylation ± Lipid modi®cation ± Meristemdevelopment ± Mutant (Arabidopsis, farnesylation) ±Photoperiod

Introduction

Protein prenylation involves the attachment of a farnesyl(15-carbon) or geranylgeranyl (20-carbon) isoprenoid toa select group of molecules thereby facilitating proteininteraction with membrane lipids and/or other proteins(Zhang and Casey 1996). The farnesyltransferase(FTase) and geranylgeranyltransferases (GGTase I andII) that carry out these respective reactions have beenidenti®ed in animals, plants and lower eukaryotes and sothese mechanisms of lipid modi®cation of proteins havebeen evolutionarily conserved (Casey and Seabra 1996;

Rodriguez-Concepcion et al. 1999). Although bothfarnesylation and geranylgeranylation are biochemicallyrelated, the target proteins of these enzymes are usuallyvery di�erent. Recently, farnesylation has garneredmuch attention because in animal cells mitogenic RASneeds to be farnesylated to be fully functional (Caseyet al. 1989; Der and Cox 1991). The potential to alleviateoncogenic transformation by inhibiting FTase activityhas focused a great deal of research in the molecularcharacterization of FTases from both mammalian andfungal systems. Consequently, there is now a clear senseof the kinetics and the substrate speci®cities of thisenzyme. The farnesyl lipid is covalently attached to acysteine located in a carboxyl-terminal `CAAX box'motif, where the C is the cysteine, A is usually analiphatic amino acid and X can be any amino acid(Moores et al. 1991; Reiss et al. 1991). Apart from theRAS superfamily of small GTP-binding proteins, thereis a growing roster of proteins that are involved in signaltransduction and are farnesylated in fungi and animals(Davey et al. 1998; Inglese et al. 1992).

Although the biochemical characterization of farn-esylation is advanced, the full extent of this modi®cationas a regulatory mechanism in higher eukaryotes is stillunclear. Studies in yeast and cultured plant and animalcells using FTase inhibitors suggests that proteinsassociated with cell division and growth are commonlya�ected (Miquel et al. 1997). In plants, the peak activ-ities of FTases and the growth sensitivity of tobacco cellcultures to FTase inhibitors are coincident, and FTasepeak activities precede the onset of mitosis (Randallet al. 1993). Furthermore, FTase inhibitors are e�ectiveat blocking cell cycle progression before the onset of G1and S phase but not at G2. Together, these studiessuggest plant farnesylation is essential for normal cellcycle progression (Qian et al. 1996). The diversity offarnesylated proteins in fungi, plants and animals,however, has made it di�cult to determine how speci®cthis type of modi®cation is in regulating cell cycle events.Furthermore, extrapolations from cell culture to wholeorganisms, particularly with the use of inhibitors isalways problematic due to the inherent di�culties of

*Present address: The University of Hong Kong, Department ofBotany, Pokfulam Road, Hong Kong

Abbreviations: ABA = abscisic acid; CL = continuous light;FTase = farnesyltransferase; SD = short day

Correspondence to: P. McCourt;E-mail: [email protected]; Fax: +1-416-978-5878

Planta (2000) 211: 182±190

Farnesylation is involved in meristem organization in Arabidopsis

Dario Bonetta, Peter Bayliss, Susanna Sun*, Tammy Sage, Peter McCourt

Department of Botany, University of Toronto, Toronto, Ontario, Canada M5S 3B2

Received: 25 October 1999 /Accepted: 22 December 1999

application, uniform accessibility and continued main-tenance of inhibitor concentrations over the life of theorganism. For these reasons FTase inhibitor experimentsin animals have been limited to oncology studies andlittle detailed developmental analysis has been reported.Thus, for both plants and animals the need for in vivomodel systems for studying farnesylation-dependentprocesses is essential for developmental studies.

In the higher plant, Arabidopsis thaliana, loss-of-function mutations in the ERA1 gene result in plantsde®cient in protein farnesylation (Cutler et al. 1996).The ERA1 gene encodes the b-subunit of a proteinFTase. ERA1 was identi®ed in a genetic screen thatcauses the plant to become supersensitive to the planthormone abscisic acid (ABA). Seeds of era1 are inhib-ited at concentrations of exogenous ABA that are ®vetimes lower than concentrations that inhibit germinationof the wild type. In addition, ABA-stimulated closure ofstomata in era1 is about three times more sensitive thanin the wild type (Pei et al. 1998). The phenotypesobserved in era1 mutants are consistent with thephysiological roles of ABA in plants. We still, however,do not have a clear idea of how loss of farnesylation canconfer ABA supersensitivity. Although the simplestinterpretation is that farnesylation acts as a negativeregulator of ABA signal transduction, the lack of amolecular target for this pathway suggests interpreta-tions of era1 phenotypes must be done with caution(Bonetta and McCourt 1998).

Although mutants de®cient in geranylgeranylationexist in Drosophila, the era1 mutant is currently the onlyFTase loss-of-function mutant in higher eukaryotes,which a�ords the unique opportunity of assessing therole of farnesylation in processes during the develop-ment of a multicellular organism. Here we report thatfarnesylation in Arabidopsis is required for a number ofprocesses ranging from meristem organization to meio-sis. Furthermore, the role of farnesylation in theseprocesses is dependent upon the photoperiod in whichthe plants are grown.

Materials and methods

Plant material, growth conditions and mutant screen. Seeds ofArabidopsis thaliana (L.) Heynh. used in the present study were in aMeyerowitz Columbia background. Mutations used included theenhanced response to ABA alleles, era1-2 and era1-3. Both of thesealleles are complete deletions of the ERA1 gene. For seedlinganalysis, seeds were imbibed for 4 d at 4 °C on 0.8% agar platescontaining half-strength MS salts, and subsequently moved togrowth shelves at 20 °C with 150 lmol photons m)2 s)1 illumina-tion. Otherwise seeds were planted directly onto soil and grown at20 °C with either continuous illumination (CL) (200 lmol pho-tons m)2 s)1) or a short-day (SD) cycle of 10 h illumination.

Phenotypic characterization of wild-type and era1 plants. Plantgrowth was assessed by measuring both width and length expan-sion of an adult rosette leaf (leaf 8) every 2 d, for 20 plants of eachgenotype. Branch numbers were measured in 12±20 plants that hadreached senescence. Vegetative and reproductive tissues wereprepared for scanning and light microscopy as described by Sageet al. (1999). Serial sections (1.5 lm) were used to determinemorphometric characteristics of the median longitudinal section of

the apical meristem dome where the dome extended from the top ofthe meristem to the meristem base as de®ned by Laufs et al. (1998).Meristems of SD-grown wild-type and era1 in¯orescences weremeasured using Northern Exposure morphometric software (Em-pix Imaging Missisauga, Ont., Canada) Parameters measuredincluded: (i) dome height, width and area, (ii) dome cell number,and (iii) cell area = dome area/dome cell number. Values formedian longitudinal sections through in¯orescence meristems werecompared statistically using one-way analysis of variance (P £0.001) where each value represents the mean of 6±10 samples �standard error.

In-situ hybridization. In-situ hybridization with digoxigenin (DIG)-labeled RNA probes in wild-type plants was performed accordingto the Cold Spring Harbor Arabidopsis Molecular Genetics Coursemodi®cations. For the generation of DIG-labeled RNA probes, acDNA clone of ERA1 (Cutler et al. 1996; Accession No. U46574)was subcloned into the EcoRI and NotI sites of the pBC-SK+vector (Stratagene). The 3¢ terminal portion containing the poly-Aregion was removed by digesting with NruI and NotI, end-®llingand re-ligating to form pCBDN. Sense and antisense RNA probeswere generated by in-vitro-transcribing of BlpI- and HindIII-digested pCBDN, respectively. Probes were hydrolyzed to approx.100 nucleotides. Approximately 80±120 ng of probe was used permicroscope slide. Tissue was ®xed with either FAA (50% ethanol,5% acetic acid, 10% formalin) or 4% paraformaldehyde, dehy-drated in an ethanol-xylene series, embedded in Parap-last+ (Fisher Scienti®c), sectioned at 10 lm, and mounted onsilane-coated slides (Sigma). Prior to hybridization, slides weretreated with 26.4 mM acetic anhydride in 0.1 M triethanolamine(pH 8.0) for 10 min at room temperature, washed with 2´ SSC(1´ SSC = 0.15 M NaCl, 0.015 M Na3-citrate, pH 7) and dehy-drated through an ethanol series.

Analysis of RNA gel blots. Leaf number 5 was detached from wild-type and era1-2 plants at day 30 after planting. This time waspicked as the leaf had reached its fully expanded length and plantswere just beginning to ¯ower. The day-0 sample represents thispoint. RNA was extracted as described by Verwoerd et al. (1989).Equal amount (10 lg) of total RNA from days 0, 4, and 8 wereseparated by electrophoresis through a 1.1% formaldehyde gel andtransferred to hybond-N membrane (Amersham). Membranes werehybridized in 50% formamide, 5´ SSC, 0.1% SDS at 42 °Covernight and washed three times at 65 °C. The same blots werestripped and used for all three probes. The SAG12 (SenescenceActivated Gene) and SAG13 genes were a kind gift fromR. Amasino. The CAB gene (chlorophyll a/b-binding protein gene)was a gift from J. Coleman.

Results

era1 mutants show subtle morphological phenotypes inconstant light. In CL conditions, era1 plants lookedrelatively similar to wild-type plants except for a numberof subtle phenotypes (Fig. 1A,B). Unlike the wild type,the margins of era1 leaf blades did not curve downward,giving the leaves a broadened appearance. Indeed, whenmeasured the average widths of mature leaves wereslightly increased in era1 (Table 1). Growth of era1plants was slow compared to the wild type and boltingwas delayed (Table 1). Also, era1 plants did not senesceat the same rate as wild-type plants (data not shown). Tofurther explore the reduced senescence, a number ofgene markers that show preferential transcript accumu-lation (SAG12, SAG13) or transcript reduction (CAB)during this developmental stage were assayed (Quirinoet al. 1999). As observed in Fig. 2, the two SAG genes

D. Bonetta et al.: Farnesylation is involved in meristem organization in Arabidopsis 183

that normally show mRNA accumulation under senes-cence conditions in the wild type did not increase at thesame developmental stage in era1. In contrast, CABRNA levels that decrease during senescence in wild-typeleaves were not a�ected in era1 mutants at thisdevelopmental stage. Together, it appears the era1mutation retards the senescence program in Arabidopsis.

In wild-type plants, axillary meristems typicallybecome evident in either the axils of cauline leaves onthe primary shoot or in the axils of rosette leaves. Theseaxillary meristems form additional ¯owering axes orbranches; most of which arise from the axils of caulineleaves and to a lesser extent from the axils of rosetteleaves. For simplicity, branches originating from rosette-leaf axils will be referred to as rosette branches and thoseoriginating from cauline-leaf axils, as cauline branches.In CL, wild-type and era1 plants produced on averagethe same number of cauline branches (Table 1). How-ever, compared with the wild type, there was a slightreduction in rosette branches in era1 mutants grown inCL (Table 1). Finally, era1 ¯owers showed a stage-speci®c extension of carpels above sepals, which resultedin a protruding carpel in young, ¯owers (Fig. 1).

era1 mutants have reduced branching in SDs. In an e�ortto better assess the e�ects of ERA1 loss-of-function wedetermined the growth characteristics of era1 under anumber of conditions. A number of era1 phenotypes,which were not obvious in CL, were unmasked whenplants were grown under SD conditions. Under theseconditions the appearance of era1 rosette leaves wassimilar to that of wild-type rosette leaves; however, thepetioles were shortened (Table 1). The onset of ¯oweringwas slightly but not signi®cantly delayed in era1 versuswild-type (Table 1). To determine if growth rates weredecreased in era1, the expansion of the eighth rosette leafin wild-type and era1 plants was determined. Thesemeasurements indicated that complete leaf expansion inwild-type plants requires approximately 30 d and thesame leaf in era1 lags behind by about 10 d (Table 1).This lag is due to a decrease in initial rates of era1 leafexpansion, since subsequent rates of expansion were notobviously di�erent (data not shown).

In comparison to CL conditions, wild-type plantsgrown under SDs produced fewer branches from rosetteaxils and cauline-axil branch initiation increased(Fig. 1). In contrast, era1 cauline branches remainedconstant (Table 1). In addition, while there was almostalways a branch produced wherever a cauline leafemerged along a wild-type stem, many of the caulineleaves along an era1 stem did not produce a branch(Fig. 1). These observations suggest that there is a

Fig. 1A±G. Whole-plant morphology of wild-type and era1 Arabid-opsis thaliana plants grown in CL or SDs. Wild-type plants grown inCL conditions (A) show a reduction in cauline branches compared tothose grown under SL (C). Although era1mutant plants grown underCL (B) are similar to the wild-type, by contrast SD-grown era1 plants.(D) show a dramatic reduction in branching and aborted siliquedevelopment along the stem (E). The in¯orescence development ofwild-type (F) and era1 (G) plants grown in CL also di�ers in thatmutant ¯owers are open and often show protruding carpels (arrow)before fertilization

Table 1. Phenotypic measurements of wild-type and era1 A. thaliana plants under di�erent photoperiods. The leaf measurements were forleaf number 8. Values are means �SD, with the number of plants measured given in parentheses

Continuous light Short days

Wild-type era1 Wild-type era1

Blade length 37.2 � 5.0 (20) 38.3 � 3.3 (20) 41.3 � 4.6 (20) 38.7 � 2.6 (20)Petiole length 11.2 � 1.5 (20) 9.5 � 1.4 (20) 16.4 � 2.3 (20) 13.2 � 1.3 (20)*Blade width 12.4 � 2.3 (20) 17.5 � 1.4 (20)* 13.5 � 1.4 (20) 15.6 � 1.8 (20)Complete leaf expansion (d) 24.8 � 1.5 (20) 33.1 � 1.8 (20) 33.8 � 2.3 (20) 42.2 � 2.5 (20)Time to bolting (d) 20.0 � 0.5 (20) 24.6 � 1.7 (20) 45.4 � 1.3 (20) 51.8 � 3.7 (20)Rosette leaves 10.4 � 1.0 (20) 12.5 � 2.6 (20) 45.0 � 4.1 (20) 49.0 � 4.9 (20)Cauline branches 2.0 � 0.8 (12) 2.5 � 1.5 (18) 7.0 � 1.0 (22) 1.8 � 1.3 (29)*Rosette branches 2.9 � 0.9 (12) 1.2 � 1.3 (18) 1.7 � 1.1 (17) 0.5 � 0.8 (19)

* Signi®cant di�erence between mutant and wild-type plants (Student's t-test, P = 0.05)

184 D. Bonetta et al.: Farnesylation is involved in meristem organization in Arabidopsis

general shift under SD conditions to a greater contri-bution of cauline branches to overall branching patternsin the wild-type and that this aspect is a�ected in era1.

In most higher plants, removal of the apical meristemhas the e�ect of inducing growth of axillary meristemsby relieving the suppression that is exerted upon them bythe dominant apical meristem. To determine whether thereduction in branch formation in era1 was due toextreme apical dominance, era1 plants with only oneprimary in¯orescence were decapitated at the base of thestem shortly after bolting. Of the 10 plants tested, only 1plant initiated a branch from a rosette-leaf axil after15 d from the time of decapitation. The basis for thealtered branching pattern was determined by longitudi-nally sectioning era1 and wild-type rosette or caulineaxils. None of era1 plants examined showed any obviousmeristematic tissue present in positions where meristemsare found in wild-type axils (Fig. 3). Therefore, thereduction in rosette branching seen in era1 grown underSDs is due to an inability of the mutant to di�erentiatethese structures rather than an imposed inhibition ofalready developed meristems.

era1 mutants a�ect axillary meristem development inSDs. In in¯orescences, most of the ¯owers of SD-grownera1 plants tended to cluster at the in¯orescence apexrather than distribute along the length of stem (Fig. 1).In addition, the primary in¯orescence in era1 plantsbecame fasciated about 15% of the time and many¯owers showed abnormal organ number. While wild-type ¯owers almost invariably consisted of four sepals,four petals, six stamens and two carpels, typically inera1, at least one ¯oral whorl was a�ected (Fig. 4).Although a mutant whorl usually contained either one

extra or one less organ, the average number of organsper whorl was not signi®cantly di�erent from the wild-type average (Fig. 4). Consequently the total number oforgans per ¯ower was not signi®cantly di�erent from thewild-type mean of 16.

The clustered ¯oral apex, fasciated stems and abnor-mal number of ¯oral organs, raised the possibility thatera1 apical meristem organization is abnormal. Arabid-opsis shoot apical meristems consist of a central zonewhich is surrounded by a peripheral zone and a rib zonewhich is located beneath the central zone. During thegrowth of the apical meristem undi�erentiated centralzone cells divide, displacing older cells towards theperipheral zones where they di�erentiate into organprimordia. While meristems are developmentally vege-tative, the peripheral zones give rise to the leaves. Whenthe vegetative meristem developmentally converts to areproductive meristem this zone produces ¯oral meris-tems that form ¯owers. To determine when possiblechanges to era1 meristems might be occurring, meris-tems of both mutant and wild-type genotypes grownunder SDs were sampled every 5 d up to 60 d afterplanting. Figure 3 shows representative meristems ofdevelopmentally similar era1 and wild-type vegetativemeristems. As shown in this ®gure and for all othervegetative meristems samples (data not shown), noobvious di�erences could be detected between mutantand wild type. Upon transition to a reproductive apicalmeristem, however, mutant samples did show increasesin meristem size over wild-type samples. Median

Fig. 2. RNA gel blot analysis of senescence-regulated gene expressionin wild-type and era1mutant leaves of A. thaliana. RNA samples wereonly isolated from the ®fth leaf to emerge after germination. Time 0represents a leaf taken from a plant that has begun to ¯ower (30 dafter germination under CL). Times 4 and 8 are subsequent days aftertime 0. The total RNA loaded on the gel is shown by ethidiumbromide staining below the blots. The same blots were stripped andused for all three hybridizations

Fig. 3A±F. Meristem development in wild-type and era1 A. thalianain SDs. A, B Longitudinal sections of typical rosette-leaf axils of wild-type (A) and era1 (B) plants. C, D Scanning electron microscopy ofwild-type (C) and era1 (D) mutant vegetative meristems of similardevelopmental stages grown under SDs. E, F Median longitudinalsections of wild-type (E) and era1 (F) in¯orescence meristems.Bars = 80 lm

D. Bonetta et al.: Farnesylation is involved in meristem organization in Arabidopsis 185

longitudinal sections through in¯orescence meristems ofSD-grown wild-type and era1 were compared and theapical dome of era1 meristems appeared to be largerthan that of the wild type (Fig. 3). This increase in apicalmeristem size in era1 was re¯ected in quantitativemeasurements of dome width and height and averagecell area in the meristematic region (Table 2). Further-more, the number of cells in the epidermal layer andbelow this layer down to the level where the nextprimordia was initiated was approximately three timeshigher in era1 meristems (Table 2).

era1 a�ects the development of reproductive apicalmeristems in SDs. Although aberrant ¯owers wereproduced at some frequency at the in¯orescence apex,¯oral structures below the apex appeared to be moreseverely a�ected by SD conditions in era1 mutants(Fig. 5). These defects which consisted of individualshort ®laments ¯anked by what appeared to be stipules

and subtended by a mound of cells were visible shortlyafter ¯oral in¯orescences began to elongate (Fig. 5).This organization suggests that the short ®lament is avestigial axillary meristem and the mound of cells belowit a presumptive cauline leaf, indeed the components of acauline branch. These vestigial structures remainedarrested in this form and did not di�erentiate into anymature structure. The defects in di�erentiation ofreproductive structures were represented in less extremecases by ¯owers which consisted of only sepals or onlypedicels (Fig. 5). Elongation of the primary in¯ores-cence revealed that successful production of complete¯owers along the stem occurred in a stop-and-gofashion. Groups of mature ¯owers were separated byintervening lengths of stem where no ¯owers weresuccessfully produced (Fig. 5). The phenotypic severityof era1 showed an acropetal decrease and most complete¯owers formed at the apex when plants were nearingsenescence. Mature era1 ¯owers appeared to be quali-tatively larger than wild-type ¯owers.

Fig. 4A,B. Organ number frequencies (A) and total number of organs(B) in wild-type and era1 ¯owers of A. thaliana. (A) The frequency oforgan numbers for 50 mature ¯owers of each genotype is shown. Eachbar represents a particular whorl and the pattern represents theproportion of organs found in that whorl. The number indicated inparentheses beside the whorl is the average number of organs (�SD)found in that whorl. B The bar pattern represents the proportion oforgans found in the ¯ower. The numbers indicated in parentheses arethe average number of organs (�SD) per ¯ower

Table 2. Morphometric characteristics of the median longitudinal section through an in¯orescence meristem of wild-type and era1A. thaliana. Values are means �SD

Dome width (lm) Dome height (lm) # cells in L1 # cells below L1 Dome area (lm2) Cell area (lm2)

Wild type 82.9 � 4.1 23.3 � 1.4 20.3 � 0.7 38.3 � 5.1 1336.9 � 148.6 22.9 � 1.7era1 149.8 � 6.7 39.9 � 3.0 31.4 � 1.3 111.3 � 11.4 4300.9 � 648.5 37.5 � 2.5

Fig. 5A±G. Characteristics of morphology of reproductive structuresin era1 in SDs. In SDs, unde®ned structures are produced in era1 (A±C) consisting of a short ®lament (f) subtended by a small mound ofcells (m) and ¯anked by stipules (s) (B). C A higher magni®cation ofthese structures. The short ®laments are not always present andtrichomes (t) are occasionally visible coming o� the cell mounds or®laments (B, C). Along the length of era1 stems, regions where ¯owersare successfully formed are separated by regions where no di�eren-tiated structures are formed (arrow head inD). Flowers formed in era1plants are either complete (E), produce only sepals (F) or only pedicels(G) Bars = 100 lm

186 D. Bonetta et al.: Farnesylation is involved in meristem organization in Arabidopsis

era1 mutants show defective meiosis. When crossing era1for genetic studies it was noted that era1 pollen clumpedand plants were semi-sterile. Although most aspects ofpollen development in era1 plants proceeded normally,there was an apparent variability in the progression ofmeiosis between individual cells and often an uncou-pling of aspects of pollen development from meioticevents. Wild-type microsporogenesis is characterized bythe development of enlarged microsporocytes (micros-pore and pollen mother cells) from sporangenous tissue.While microsporocytes are in the early-late stages ofprophase, callose is deposited around each microsporo-cyte (Fig. 6A). Meiosis is synchronous and cytokinesisis simultaneous followed by additional deposition ofcallose around each microspore (Fig. 6C). Microsporo-genesis results in a tetrahedral arrangement of microsp-ores that become separated from each other followingdegradation of callose walls (Fig. 6E). Exine formationis followed by intine formation and two mitoticdivisions give rise to a tricellular pollen grain (Fig. 6G).Microsporogenesis in era1 is similar to that of wild-typeup to late prophase (Fig. 6B). However, meiosis isasynchronous, although cytokinesis appears to besimultaneous during formation of microspores(Fig. 6D). Tetrads of microspores may be arranged ina tetrahedral con®guration but also in a more lineararray (Fig. 6F). Many microspore mother cells continueto undergo meiosis and subsequently degenerate whileother microsporocytes continue to undergo meiosisduring callose degradation (Fig. 6F,H). In manymicrospores, exine and intine formation and twomitotic divisions are completed resulting in the forma-tion of tricellular pollen grains. Wall deposition appearsto occur around other microsporocytes and subsequentdegeneration of the cytoplasm results prior the mitoticdivision (Fig. 6I). Partial wall formation is alsoobserved around enlarged microspores or persistentmicrosporocytes (Fig. 6J).

ERA1 expression patterns are correlated with SD-grownera1 phenotypes. High-stringency Southern blot analysisusing the ERA1 gene indicates there is only one farnesyltransferase gene in the Arabidopsis genome (Cutler et al.1996). Hence, in-situ hybridization using the ERA1 geneon wild-type plants may be informative with respect tothe mutant phenotypes observed. Since era1 phenotypesimplied a greater dependence for ERA1 function in SDconditions, in-situ hybridization of ERA1 mRNA wascarried out on tissue of SD-grown wild-type plants. Abasal level of ERA1 transcript was present in all tissues.Higher levels were present in regions of active celldivision but this could be due to the cytoplamic densityof cells that are active in cell proliferation (Fig. 7).Nevertheless, expression patterns of ERA1 RNA corre-lated well with many of the era1 phenotypes seen inplants grown in SD. For example, in both vegetative andin¯orescence meristems higher transcript levels werepresent in the peripheral region of the meristem and inprimordia (Fig. 7). In ¯oral tissues, RNA was mostlylocalized to developing tissues such as ovule primordiaand microsporocytes (Fig. 7). In embryos, a high

amount of signal which began to decrease in matureembryos was observed from at least the globular stage to

Fig. 6A±J. Pollen development in wild-type and era1 A. thaliana. AWild-type anther locule containing microsporocytes (M ) in earlyprophase.Arrow indicates callose.BAn era1 anther locule at equivalentstage where the microsporocytes are in late prophase. Arrow denoteshighly condensed chromosomes. C Wild-type microsporocytes under-going synchronous anaphase of meiosis.DMicrosporocytes of era1 indi�erent stages of meiosis with some in prophase ( p), some undergoingeither meiosis I or II (m) and other where meiosis is completed (c) andcytokinesis is occurring. E Wild-type tetrads (t) formed aftercytokinesis. F Arrangement of era1 microspores after cytokinesis iseither tetrahedral (thin arrow) or linear. Note the presence of amicrosporocyte that failed to undergo meiosis (thick arrow).GMaturetricellular pollen grains (pg) in a wild-type locule. H Degeneratedmicrospores (arrow) in era1 after cytokinesis. I Occurrence ofcontinued meiosis (m) in microsporocytes after callose degradationin era1. J In mature pollen grains (pg) of era1, pollen grains where thecytoplasm has degenerated (f ) and enlarged microspores or persistentmicrosporocytes (arrow) are present. Bars = 25 lm

D. Bonetta et al.: Farnesylation is involved in meristem organization in Arabidopsis 187

the torpedo stage (Fig. 7). Expression of the ERA1 peaorthologue has previously been analyzed using a pro-moter b-glucuronidase (GUS) reporter system (Zhouet al. 1997) and high GUS activity was found in theshoot apex, ¯ower receptacles, regions of axillary budformation and in vascular tissue. Similarly, FTaseactivity was detectable in all tissues tested with thehighest activity in the apical bud, stem and developingfruits. The previous studies on pea and tomato, and theexpression data presented here for ERA1 of Arabidopsisindicate that there is some level of regulation of FTaseabundance in various tissues.

Discussion

Previous studies on era1 mutants have linked farnesyla-tion with two classically de®ned ABA responses inplants, seed dormancy and water relations (Cutler et al.1996; Pei et al. 1998). However, given that farnesylationis a ubiquitous function and that there are potentiallynumerous cellular target proteins, the fact that onlysubtle phenotypes are observed in these mutants underCL growth conditions was intriguing. Possibly, the lackof phenotypes re¯ects functional redundancy for farn-esylated processes in plants. For example, the cross-speci®city of FTases with other prenyltransferases, likegeranylgeranyltransferase I (Trueblood et al. 1993;Yokoyama et al. 1997), means that some compensationfor loss of ERA1 function is likely to exist in era1 nulls.In addition, it is also possible that some degree ofredundancy for ERA1 function is present, as we are notcertain that ERA1 is the only b-FTase gene in Arabid-opsis. In this study, however, we describe a novelcollection of era1 mutant phenotypes that are contingenton the length of the photoperiod. This suggests that thecontributions of any redundant functions are not enoughto compensate for loss of the ERA1 FTase activity underSD conditions. The ERA1 FTase is, therefore, necessaryfor a number of developmental processes in plants andthis requirement is environmentally sensitive.

Under SD conditions two major farnesylation-de-pendent phenotypes are observed in mature era1 mutantplants: the inability to initiate or maintain organs on thesides of the main stem and an enlargement of the apicalmeristem. Two models can be used to explain thesephenotypes. Possibly, aspects of axillary and apicalmeristem development are regulated by independentsignaling pathways that both require di�erent farnesy-lated intermediates to function correctly. A secondpossibility is a single process is a�ected in era1 mutants,but because axillary meristem initiation/maintenanceand regulation of apical meristem size may not bedevelopmentally equivalent, the defect is manifesteddi�erently in the two era1 mutant meristem types.Mutations in a number of genes in Arabidopsis havebeen identi®ed that preferentially a�ect one meristemtype over another, suggesting these tissues are notdevelopmental equivalent. In revoluta mutants, forexample, axillary meristems have a greater tendency toabort development while the primary apical meristem isless a�ected (Talbert et al. 1995). Similarly, pin andpinoid mutations cause a reduction in the capacity togenerate mature ¯owers, ¯oral organs and caulineleaves; however, primordia are initiated from the apicalmeristem (Okada and Shimura 1994; Bennett et al.1995). By contrast, mutations in a number of genes havebeen identi®ed in Arabidopsis that increase meristem sizebut do not appear to a�ect axillary meristem initiationor maintenance (Barton 1998). Although these geneticexperiments argue for discrete signaling steps for apicalversus axillary meristem development, it is also possiblethat these phenotypes represent the sensitivity of di�er-ent meristematic tissue to genetic perturbation. Forexample, regulators such as LEAFY and APETELA2,

Fig. 7A±M. In-situ hybridization of wild-type tissue from SD-grownA. thaliana plants using ERA1 digoxigenin-RNA probe. Positivesignals are red-brown to purple-blue. Sections in A, C±G, I, and K±Mwere hybridized to anti-sense RNA probe. A Longitudinal sectionthrough an in¯orescence. B Comparable section to A probed withsense probe. C Longitudinal section through developing ¯owers.D Longitudinal section through a vegetative meristem. E Sectionthrough an anther locule containing microsporocytes. F Sectionshowing hybridization to microspore tetrads and tapetum in an antherlocule. G Section through a gynoecium where ovule primordia areevident. H Sense control of a comparable section to G. I Sectionthrough a silique showing hybridization to a globular stage embryo.J Similar section of an embryo probed with sense RNA. K, LHybridization in torpedo state (K) and late torpedo stage (L)embryos. M Positive signal decreases in mature embryos.Bars = 50 lm

188 D. Bonetta et al.: Farnesylation is involved in meristem organization in Arabidopsis

are expressed at di�erent stages of plant development inArabidopsis, yet loss-of-function mutations in thesegenes preferentially a�ect only ¯oral initiation and ¯oralorgan identity (Jofuku et al. 1994; Blazquez et al. 1997).It is possible that a lack of phenotypes in tissues wheregene products are produced may simple re¯ect geneticredundancy. Resolution of the role of farnesylatedproteins in meristem organization will require theidenti®cation of farnesylated targets and characteriza-tion of their functions in both axillary and apicalmeristems. Genetic analysis has identi®ed a number ofgene products that appear to be involved in controllingmeristem size. The clavata mutations, for example, havede®ned a putative receptor kinase (CLV1) and it'spotential ligand (CLV3) (Clark et al. 1997; Fletcheret al. 1999). One interpretation of these genes is theyencode negative regulators of cell proliferation. Alter-natively, loss of functions necessary to promote thetransition of slowly dividing central zone cells intoactively dividing peripheral zone cells, could result in anoverall increase in undi�erentiated central zone cells. Atthis time we cannot conclude whether the increased¯oral meristem size seen in era1 is caused by increasedcell division or aberrant di�erentiation.

Loss of ERA1 function appears to be more detri-mental to axillary meristem development than the shootapical meristem. Mutant axillary meristems appear to bedefective in either cell division and/or di�erentiation.Mutant side shoot structures can be initiated, but thesedo not produce any di�erentiated tissues. In an instruc-tive scenario, farnesylation could be required for di�er-entiation of cells into axillary meristematic cells.Alternatively, meristem di�erentiation may occur nor-mally in the mutant but farnesylation is required topermit further cell proliferation of an axillary meristem-atic cell. This permissive scenario is similar to thefunction of many growth factors in animals, which arerequired cell division but not di�erentiation. The mor-phology of era1 abnormal side shoots is reminiscent of¯oral phenotypes reported for ufo mutants (Wilkinsonand Haughn 1995). In these mutants, initial commitmentto form side shoots is initiated; however, this commit-ment is either not maintained or is insu�ciently strong.Interestingly, the UFO gene encodes a protein with anF-box motif that is thought to be involved in cell cycleregulation in both Arabidopsis and Antirrhinum (Ingramet al. 1997). The production of vestigial structures andthe reduction of axillary meristem formation in era1mutants may be due to a decrease in the mitoticcompetence of cells comprising these structures. Thatboth cauline leaves and rosette-leaf axils with no axillarymeristems can be produced in era1 mutants may justre¯ect di�erent degrees of severity of this phenomenon.

In conclusion, our results suggest that aside fromregulating ABA-dependent plant processes such as seeddormancy and stomatal aperture size, ERA1 functions ina diverse number of cellular events ranging from meiosisand mitosis to di�erentiation. Some degree of regulationof farnesylation activity must exist and based on what isknown of protein farnesylation this modi®cation can actas a switch to a�ect the function of multiple target

signaling molecules. The ability to address how farnesy-lated signals are integrated within the context of a wholeorganism should now be possible.

We thank Nocha van Theilan for leaf measurements and SaraSarkar for reading the manuscript. This research was funded bygrants to P.M. and T.L.S. from the Natural Sciences andEngineering Research Council.

References

Barton MK (1998) Cell type speci®cation and self renewal in thevegetative shoot apical meristem. Curr Opin Plant Biol 1: 37±42

Bennett SRM, Alvarez J, Bossinger G, Smyth DR (1995)Morphogenesis in pinoid mutants of Arabidopsis thaliana. PlantJ 8: 505±520

Blazquez MA, Soowal LN, Lee I, Weigel D (1997) LEAFYexpression and ¯ower initiation in Arabidopsis. Development19: 3835±3844

Bonetta D, McCourt P (1998) Genetic analysis of ABA signaltransduction pathways. Trends Plant Sci 3: 231±235

Casey PJ, Seabra MC (1996) Protein prenyltransferases. J BiolChem 271: 5289±5292

Casey PJ, Solski PA, Der CJ, Buss JE (1989) p21ras is modi®ed bya farnesyl isoprenoid. Proc Natl Acad Sci USA 86: 8323±8327

Clark SE, Williams RW, Meyerowitz EM (1997) The CLAVATA1gene encodes a putative receptor kinase that controls shoot and¯oral meristem size in Arabidopsis. Cell 89: 575±585

Cutler S, Ghassemian M, Bonetta D, Cooney S, McCourt P (1996)A protein farnesyl transferase involved in abscisic acid signaltransduction in Arabidopsis. Science 273: 1239±1241

Davey J, Davis K, Hughes M, Ladds G, Powner D (1998) Theprocessing of yeast pheromones. Semin Cell Dev Biol 9: 19±30

Der CJ, Cox AD (1991) Isoprenoid modi®cation and plasmamembrane association: critical factors for ras oncogenicity.Cancer Cells 3: 331±340

Fletcher JC, Brand U, Running MP, Simon R, Meyerowitz EM(1999) Signaling of cell fate decisions by CLAVATA3 inArabidopsis shoot meristems. Science 283: 1911±1914

Inglese J, Glickman JF, Lorenz W, Caron MG, Lefkowitz RJ(1992) Isoprenylation of a protein kinase. Requirement offarnesylation/alpha-carboxyl methylation for full enzymaticactivity of rhodopsin kinase. J Biol Chem 267: 1422±1425

Ingram GC, Doyle S, Carpenter R, Schultz EA, Simon R, Coen ES(1997) Dual role for ®mbriata in regulating ¯oral homeoticgenes and cell division in Antirrhinum. EMBO J 21: 6521±6534

Jofuku KD, den Boer BG, Van Montagu M, Okamuro JK (1994)Control of Arabidopsis ¯ower and seed development by thehomeotic gene APETALA2. Plant Cell 9: 1211±1225

Laufs P, Grandjean O, Jonak C, Kieu K, Traas J (1998) Cellularparameters of the shoot apical meristem in Arabidopsis. PlantCell 10: 1375±1390

Miquel K, Pradines A, Sun J, Qian Y, Hamilton AD, Sebti SM,Favre G (1997) GGTI-298 induces G0-G1 block and apoptosiswhereas FTI-277 causes G2-M enrichment in A549 cells.Cancer Res 57: 1846±1850

Moores SL, Schaber MD, Mosser SD, Rands E, O'Hara MB,Garsky VM, Marshall MS, Pompliano DL, Gibbs JB (1991)Sequence dependence of protein isoprenylation. J Biol Chem266: 14603±14610

Okada K, Shimura Y (1994) The PIN-FORMED gene. In:Bowman J (eds), Arabidopsis, an atlas of morphology anddevelopment. Springer New York, pp 180±183

Pei ZM, Ghassemian M, Kwak CM, McCourt P, Schroeder JI(1998) Role of farnesyltransferase in ABA regulation of guardcell anion channels and plant water loss. Science 282: 287±290

Qian D, Zhou D, Ju R, Cramer CL, Yang Z (1996) Proteinfarnesyltransferase in plants: molecular characterization andinvolvement in cell cycle control. Plant Cell 8: 2381±2394

D. Bonetta et al.: Farnesylation is involved in meristem organization in Arabidopsis 189

Quirino BF, Normanly J, Amasino RM (1999) Diverse range ofgene activity during Arabidopsis thaliana leaf senescenceincludes pathogen-independent induction of defense-relatedgenes. Plant Mol Biol 40: 267±278

Randall SK, Marshall MS, Crowell DN (1993) Protein isopreny-lation in suspension-cultured tobacco cells. Plant Cell 5: 433±442

Reiss Y, Stradley SJ, Gierasch LM, Brown MS, Goldstein JL(1991) Sequence requirement for peptide recognition by ratbrain p21ras protein farnesyltransferase. Proc Natl Acad SciUSA 88: 732±736

Rodriguez-Concepcion M, Yalovsky S, Gruissem W (1999) Proteinprenylation in plants: old friends and new targets. Plant MolBiol 39: 865±870

Sage TL, Strumas F, Cole WW, Barrett SCH (1999) Di�erentialovule development following self- and cross-pollination: thebasis of self-sterility in Narcissus triandrus (Amaryllidaceae).Am J Bot 86: 855±870

Talbert PB, Adler HT, Parks DW, Comai L (1995) The REVO-LUTA gene is necessary for apical meristem development andfor limiting cell divisions in the leaves and stems of Arabidopsisthaliana. Development 121: 2723±2735

Trueblood CE, Ohya Y, Rine J (1993) Genetic evidence of in vivocross-speci®city of the CaaX-box protein prenyltransferasesfarnesyltransferase and geranylgeranyltransferase-I in Sacchar-omyces cerevisiae. Mol Cell Biol 13: 4260±4275

Wilkinson MD, Haughn GW (1995) UNUSUAL FLORALORGANS controls meristem identity and organ primordia fatein Arabidopsis. Plant Cell 7: 1485±1499

Verwoerd TC, Dekker BM, Hoekema A (1989) A small-scaleprocedure for the rapid isolation of plant RNAs. Nucleic AcidsRes 17: 2362

Yokoyama K, Zimmerman K, Scholten J, Gelb MH (1997)Di�erential prenyl pyrophosphate binding to mammalianprotein geranylgeranyltransferase-I and protein farnesyltrans-ferase and its consequence on the speci®city of proteinprenylation. J Biol Chem 272: 3944±3952

Zhang FL, Casey PJ (1996) Protein prenylation: molecularmechanisms and functional consequences. Annu Rev Biochem65: 241±269

Zhou D, Qian D, Cramer CL, Yang Z (1997) Developmental andenvironmental regulation of tissue- and cell-speci®c expressionfor a pea protein farnesyltransferase gene in transgenic plants.Plant J 4: 921±930

190 D. Bonetta et al.: Farnesylation is involved in meristem organization in Arabidopsis