The flowering hormone florigen functions as ageneral systemic regulator of growth and terminationAkiva Shalita, Alexander Rozmana, Alexander Goldshmidtb, John P. Alvarezb, John L. Bowmanc, Yuval Eshedb,and Eliezer Lifschitza,1

aDepartment of Biology, Technion-Israel Institute of Technology, Haifa 32000, Israel; bDepartment of Plant Sciences, Weizmann Institute of Science, Rehovot76100 Israel; and cSchool of Biological Sciences, Monash University, Melbourne, Victoria 3800, Australia

Edited by Elliot M. Meyerowitz, California Institute of Technology, Pasadena, CA, and approved March 9, 2009 (received for review October 27, 2008)

The florigen paradigm implies a universal flowering-inducing hor-mone that is common to all flowering plants. Recent work identifiedFT orthologues as originators of florigen and their polypeptides as thelikely systemic agent. However, the developmental processes tar-geted by florigen remained unknown. Here we identify local balancesbetween SINGLE FLOWER TRUSS (SFT), the tomato precursor offlorigen, and SELF-PRUNING (SP), a potent SFT-dependent SFT inhib-itor as prime targets of mobile florigen. The graft-transmissibleimpacts of florigen on organ-specific traits in perennial tomato showthat in addition to import by shoot apical meristems, florigen isimported by organs in which SFT is already expressed. By modulatinglocal SFT/SP balances, florigen confers differential flowering re-sponses of primary and secondary apical meristems, regulates thereiterative growth and termination cycles typical of perennial plants,accelerates leaf maturation, and influences the complexity of com-pound leaves, the growth of stems and the formation of abscissionzones. Florigen is thus established as a plant protein functioning as ageneral growth hormone. Developmental interactions and a phylo-genetic analysis suggest that the SFT/SP regulatory hierarchy is arecent evolutionary innovation unique to flowering plants.

growth hormone � SFT/SP ratio � perennial � compound leaf �abscission zone

The florigen paradigm was conceived from the study ofphotoperiod-sensitive plants but implies, in its general form,

a universal graft-transmissible flowering signal that althoughactivated in leaves by species-specific stimuli is common to allplants (1–3). Unequivocal evidence for the critical tenets of grafttransmissibility and universality of the systemic mechanism wasobtained in tomato. SFT, the FT homologue encoding florigen(4, 5), triggers graft-transmissible signals that complement lateflowering in sft plants and substitutes for light dose stimuli inday-neutral tomato and tobacco, for short days in MarylandMammoth tobacco and for long days in Arabidopsis (6). On thebasis of the absence of SFT mRNA beyond the graft joints, wesuggested that f lorigenic signals are generated by cell-autonomous SFT transcripts. This implicated the protein as alikely systemic agent (7), which was supported by strong circum-stantial evidence (8–14). However, the developmental mecha-nisms targeted by florigen to transform vegetative meristemsinto reproductive organs remain unknown, and their study, byand large, is indifferent to florigen being a protein or RNA.

A clue as to the target of florigen was inferred from theobservation that overexpression of SFT induces, in addition toprecocious flowering, an overall growth retardation (6). Thisseemingly trivial phenomenon associated with flowering in manyplants might be the consequence of stress upon flowering, butbecause growth retardation and precocious flowering were trig-gered by a single gene, we hypothesized that they represent 2 facetsof the same mechanism. In other words, boosting flowering is just1 of the pleiotropic functions of florigen (6). To identify thedevelopmental targets of florigen system-wide, we dissected itsoverall growth effects by using grafting in conjunction with mutantsthat sensitize organ-specific responses to florigen.

The tomato plant presents unique opportunities to study multipleaspects of florigen. Its shoots consist of developmental moduleswith homology to monopodial annuals but also feature regularvegetative/reproductive oscillations typical of woody sympodialperennials. Unlike other systems, tomato is photoperiod insensitive,thereby eliminating the influence of day length on the functionalanalysis of florigen. Finally, the ease of grafting and a fortunatebattery of gene mutations can be used to monitor the effects offlorigen on diverse aspects of growth. Pivotal to the system are SFT,encoding florigen (6), and SP, a homologue of TFL1 that promotesgrowth and represses flowering (15–19). Being members of thesame gene family, these 2 CETS (CEN, TFL, SP) genes encodesignaling factors with multiple options for protein–protein interac-tions (20). For example, the interaction with FD proved essential forthe floral-inducing function of FT (21, 22). The interaction of the2 genes with the same proteins (20) and the contrasting floweringmodes of primary and sympodial apices of the self-pruning plantsindicated to us that SP is a major component of the floweringresponse mechanism (7). Here we analyze the broad developmentalconsequences of changes in the SFT/SP balances as modified, in acontext-dependent manner, by the mobile, graft-transmissible formof SFT, florigen. Florigen is thus established as a plant proteinshown to function as a general growth hormone.

ResultsContext-Specific Termination of Vegetative Growth in Shoot ApicalMeristems by Florigen. WT tomato plants terminate with a primaryinflorescence after forming 8 to 12 leaves, and subsequent sympo-dial units (SUs) consist of 3 leaves and a terminal inflorescence(Fig. 1A). Isogenic sp plants also terminate after 8 to 12 leaves, butsubsequent SUs form progressively fewer leaves until the shoot isterminated by 2 consecutive inflorescences (SI Text, Fig. S1, andTable S1). It was inferred therefore that the flowering programs forthe primary and sympodial shoots in tomato might be different (7).Here we describe the role of SFT and its mobile form, florigen, inthe 2 flowering programs. We show that both local SFT and florigenimpact differential termination and flowering in the primary andsympodial apices.

Unlike in sp, termination of the primary shoot in sft is delayed by5 to 6 leaves resulting in a terminating vegetative inflorescenceshoot that arrests the sympodial branching, thereby replacing thenormal sympodial shoot system (SI Text and Fig. S1). Surprisingly,despite the opposite effects of sp and sft on WT shoot architecture,sft single mutants and sft sp double mutants are indistinguishable(23). Furthermore, overexpression of SP delays primary termina-tion, increases the number of leaves per SU, and promotes leafy

Author contributions: A.S., Y.E., and E.L. designed research; A.S., A.R., A.G., J.P.A., Y.E., andE.L. performed research; J.L.B., Y.E., and E.L. analyzed data; and Y.E. and E.L. wrote thepaper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

1To whom correspondence should be addressed. E-mail: lifs@techunix.technion.ac.il.

This article contains supporting information online at www.pnas.org/cgi/content/full/0810810106/DCSupplemental.

8392–8397 � PNAS � May 19, 2009 � vol. 106 � no. 20 www.pnas.org�cgi�doi�10.1073�pnas.0810810106

Dow

nloa

ded

by g

uest

on

Mar

ch 2

2, 2

020

inflorescences (17), but these effects are essentially masked in sftplants (Fig. S2). Thus, the terminating effect of sp is largely relevantonly in the presence of a functional SFT.

In cultivars such as Money Maker (MM) and VFNT (resistant toverticillium wilt, fusarium wilt, nematodes, and tobacco mosaicvirus), primary shoots of WT plants expressing the constitutive35S:SFT transgene terminate after only 3 leaves (6) (Table S1). Incontrast, their sympodial program, although initially delayed (Fig.1B), maintains a typical robust regularity: 3 leaves in MM 35S:SFTand surprisingly, only 2 leaves in VFNT 35S:SFT (Fig. S2). Thus,although primary shoots were extremely sensitive to overexpressionof SFT, the regularity with which growth and termination alternatein SUs did not change, although the size of SUs could be shortened.As shown in Fig. 1C, 35S:SFT donors induced a 2-leaf sympodialmode in VFNT receptors, indicating that in regulating the size ofSUs, SFT and its mobile form are interchangeable.

The differential response of primary and sympodial apices to35S:SFT (Fig. 1 and Fig. S2) contrasted with their opposite sensi-tivity to the inactivation of SP. Whereas florigen, or a 200-foldexcess of SFT (Fig. S3), failed to disrupt the regularity of sympodialcycles, this regularity readily collapses in sp plants. To analyze thecontrasting response of primary and sympodial apices to overex-pression of florigen and to sp inactivation, we bred sp plantsoverexpressing SFT. Primary termination in sp 35S:SFT plants alsooccurred after 3 to 4 leaves, but the terminating inflorescencemeristem produced no or at most 2 flowers, and sympodial branch-ing was completely suppressed. Although distal axillary shoots werereadily released from apical dominance in these plants, they pro-duced only 1 or 2 leaves before terminating with a single flower ora blind apex (Fig. 1D). SFT is therefore a potent terminator ofprimary, sympodial, and inflorescence apices, but its role in sym-podial and flower meristems is checked primarily by SP. Although1 dose of SP in sp/� heterozygotes is sufficient to maintain the 3-leaf

cycles in WT plants, 2 doses are required to maintain it under excessof SFT, as suggested by the fluctuation in leaf numbers (between1 and 3) per SUs in sp/�35S:SFT plants (Table S1). Significantly,all features of sp 35S:SFT were also induced in sp receptors byflorigenic SFT signals emanating from a grafted 35S:SFT donor(Fig. 1E).

To examine the effect of other flowering genes on the differentialresponse of the primary and sympodial apices to SFT and florigen,we bred additional flowering mutant lines that also expressed the35S:SFT transgene. falsiflora ( fals, the tomato LFY), macrocalyx(mc, an AP1-like), blind (bl), and other mutant lines (Table S2)expressing the 35S:SFT transgene terminated, like 35S:SFT, after 3to 4 leaves (Fig. 1F and Fig. S3). However, the phenotypes of theirmutant inflorescences and their proper sympodial patterns weremaintained. The role of SP in checking SFT is therefore unique,such that SFT confers termination but not identity, and this functiondoes not require the activity of FALS or MC.

To facilitate the analysis of perception and transmission offlorigen by the different genotypes, uniflora (uf), a late-flowering,light-sensitive tomato mutant used previously to examine theuniversal functions of SFT (6), was combined with sft to generatea tester line that does not flower under any growth condition unlessgrafted with a florigen donor (Fig. 1 G and H, SI Text, and TableS3). Grafting experiments showed that all mutant lines expressing35S:SFT as graft donors induced efficient flowering in uf sftreceptors and that as receptors, all these mutants responded tomobile florigen (Fig. S4 and Table S3). Thus, the sensitivities of themutant lines to endogenous SFT and a mobile florigen are com-parable. Together these results indicate that the tested genes are notrequired for the mobility or perception of florigen and reveal thatthe differential response of primary and sympodial meristems toflorigen is maintained under a wide range of genetic backgrounds.Evidence pertinent to the 2 flowering programs in Arabidopsis andother plants is discussed in Fig. S2.

A B

B1

B2

C D

D1

D2

E

H

GF

Fig. 1. Florigen and SP-dependent termination of apical meristems. (A) A scheme of a WT main shoot of the tomato plant composed of the primary shoot (leaves1–10) and reiterated SUs (SL1–3). LS, lateral shoot; PL, primary leaf; SL, sympodial leaf. (B) Precocious primary termination and modes of first sympodial branchingin MM 35S:SFT transgenic plants. (B1) Normal resumption of sympodial branching, with the first inflorescence (arrow) positioned correctly between 2 leaves (SIText). (B2) A prospective sympodial bud was permanently suppressed, and growth resumed from a more distal axillary bud. All subsequent SUs consisted of 3leaves in both modes. (C) A robust 2-leaf sympodial cycling (numbered) in a receptor VFNT plant stimulated by a grafted 35S:SFT donor (arrow). (D and E) Florigenregulates 2 flowering programs in an SP-dependent manner. (D1) sp 35S:SFT plants terminate after 3 leaves, as do 35S:SFT plants, but the terminal inflorescenceconsists of just 1 or 2 flowers, and subsequent sympodial branching is completely arrested. (D2) Distal lateral shoots of sp 35S:SFT plants (arrow) form 1 or 2 leaveswith ‘‘blind’’ apices as shown here, or with terminal flowers as in D1 (arrow). (E) Systemic induction of flowering and termination in a uf sp receptor shoot (boxesmark grafting joint). The induced shoot is terminated after 3 leaves with 2 consecutive flowers (arrow). (F) Florigen is epistatic to late-flowering inflorescenceidentity genes. mc bl sp plants terminate after more than 20 leaves with 1 abnormal flower with enlarged sepals, and complete arrest of laterals. mc bl sp 35S:SFTplants terminate prematurely with a similar flower (arrow) and serve as ideal donors of florigen (Fig. S4). (G) The main shoot (few laterals removed) of anever-vegetative, 75-day-old, 2-m-tall uf sft double-mutant plant. (H) Flowering on a 35S:SFT//uf sft graft. The flowering shoot (circle) arose as a lateral from theaxil of a receptor leaf below the graft joint.

Shalit et al. PNAS � May 19, 2009 � vol. 106 � no. 20 � 8393

PLA

NT

BIO

LOG

Y

Dow

nloa

ded

by g

uest

on

Mar

ch 2

2, 2

020

Florigen Regulates Stem and Leaf Meristems in an SP-DependentManner. Because of their developmental versatility, the compoundleaves of tomato provide a highly sensitized forum to illustrate theeffects of florigen as a general growth hormone (24) (Fig. 2B). SPand sp leaves of isogenic lines were indistinguishable except for aslightly reduced serration in the sp leaf margins. However, sft leavesdisplayed a distinct morphology and spacing of leaflets and carriedadditional folioles, suggesting a role for SFT in regulating midribmeristems (Fig. 2B). Leaves of 35S:SFT plants, although maintain-ing a compound architecture, usually lacked 1 pair of leaflets and

were almost devoid of folioles. Strikingly, leaves of sp 35S:SFTplants were reduced to only 1 pair of lateral leaflets, and some weresimple with entire margins (Fig. 2B). The arrest of lateral leafletmeristems suggests an early function of SP during leaf primarymorphogenesis. In contrast, overexpression of SP induced largerblades with undulating surfaces, indicative of unregulated growth(Fig. 2B), an effect that was not displayed by sft 35S:SP plants (Fig.S2). As shown in Fig. 2C, all of these features were also induced bymobile SFT, suggesting that florigen is imported by leaves, whereSFT is already expressed (see below), and arrest leaflet formationin an SP-dependent manner. Thus, in sft and sp 35S:SFT (i.e., underlow or high SFT/SP ratios, respectively), SFT, SP, and florigenicSFT function as leaf meristem factors.

The response of stem meristems to different doses of SFT andSP is demonstrated in Fig. 2 A. Although normal sympodialcycling in 35S:SFT plants resumed after primary termination,normal radial expansion of sympodial internodes was perma-nently suppressed, indicating a context-specific function of flo-rigen. In general, a high SFT/SP ratio promoted growth restric-tion at the shoot apical meristems (SAMs) and lateral leafletmeristems, leading to a faster floral transition and reduced leafcomplexity respectively. But higher SFT/SP ratio as in sp35S:SFT plants resulted in the complete suppression of bothvegetative and inflorescence meristems (Fig. 1D).

Florigen Mediates Its Own Distribution by Regulating Sink–SourceRelations. The reiterated phase transitions along the tomatoshoot and its evergreen nature and day length insensitivityrequire that the distribution of florigen be regulated by adynamic balance on a daily basis. To characterize the balancebetween the dynamic needs for endogenous local SFT and itsmobile form in the shoot system, we determined the minimalsource required to induce a flowering response in the tomatobush. A single mature 35S:SFT donor leaf induced flowering insft uf receptor shoots for 2 months and in apices 2 m above thegraft points (Fig. 2D). Therefore, every mature leaf is capable ofexporting florigen to all parts of the tomato bush.

The developmental expression of SFT and SP at the whole-plantlevel was studied by comparing each gene’s age-dependent expres-sion gradients within sequentially growing leaves along primaryshoots. Initially a series of leaves was collected from WT VFNTseedlings, having 10 leaves larger than 1 cm and a primordialinflorescence. Leaves of all ages were collected 4 h after dawn,corresponding to the diurnal SFT peak (Fig. S3). As shown in Fig.2E, SFT and SP display opposing age-dependent expression gra-dients in which SFT RNA was relatively high in expanded matureleaves and SP RNA relatively high in the immature leaves. Forcomparison, profiles of other relevant genes were also included(Fig. 2E). Note that the age-dependent gradients of SFT and SP areindependent of plant age, and similar series of leaves taken frompostflowering plants will display the same gradients.

As shown in Fig. 2F, the expression patterns of SP were notaltered in sft plants and vice versa, suggesting that these patterns arenot interdependent. We next explored the possibility that overex-pression of one of the genes will affect the endogenous expressionof the other (Fig. 2G). Here, the expression of 35S:SFT or 35S:SPresults in high (200-fold and more) but equal expression in allleaves; the endogenous differential expression of SP and SFT,respectively, in old and young leaves was maintained. We thereforeinferred that the functional antagonism between SFT and SP doesnot primarily involve mutual transcriptional feedback loops.

Age-dependent gradients were also evident in leaflets along theproximal–distal axis of immature leaves up to 15 cm long. But whenleaves reached approximately three quarters of their final size(approximately 25 cm), all their major leaflets expressed SFTequally (Fig. 2H).

Removal of mature leaves delays flowering in tomato (7), and wespeculated that by the time intraleaf expression gradients have

A B

C D

IE F

G

H

Fig. 2. Florigen regulates termination of sympodial and leaf meristems in anSP-dependent manner. (A) Florigen regulates radial expansion of the stems.Note the effect of sft and of over-expression of SP on stem girth. (B and C) SFTand florigen regulate leaflet meristems in an SP-dependent manner. (B)Compound leaves with different SFT/SP ratios. WT leaves have a terminalleaflet, 3 to 5 pairs of independently formed lateral leaflets, and late inter-calary folioles (24). sft leaves feature elongated rachises, an increased numberof folioles, and extended leaflet petioles. 35S:SFT leaves are proportionallysmaller, with some reduction in leaflet number. The seventh leaf is shown. Thethird to fifth sp 35:SFT leaves, respectively, lost the majority of leaflets andfolioles, and their blade margins were smooth. 35S:SP blades display in-terveinal bulging, indicative of ectopic cell proliferation. (C) Long-range,SP-dependent regulation of leaf morphology by florigen. uf sp leaf of acontrol homograft, and a receptor uf sp shoot grafted with a 35S:SFT donor.(D–I) The florigen balance in the tomato perennial bush. (D) Systemic induc-tion of flowering by a single leaf. Left: a uf sft receptor shoot of a single leafgraft (see Fig. S2) formed its first flowers (circle) after 19–21 days and 7 to 8leaves. Secondary and tertiary branches continued to generate flowers for �2months. By that time, the receptor had developed 30 mature leaves (Right)and �100 leaves altogether. (E) Opposite age-dependent expression gradi-ents of SFT and SP in tomato leaves. RNA was extracted from consecutiveleaves of seedlings with 10 leaves (numbered) and primordial inflorescences.Leaf no. 10 is 0.5–1.0 cm long. Expression profiles of other relevant genes arealso shown, including SP2G and SP5G, 2 CETS genes with no known role inflowering. (F and G) SFT and SP are not involved in an intergenic regulatoryloop. (F) Expression gradients of SFT or SP are maintained in mutant plants ofthe opposite genotype. (G) Expression profiles of the endogenous SFT and SPgenes are not altered in 35S:SP and 35S:SFT plants, respectively. (H) Intraleafgradients of SFT or SP. Top: expression gradients of SFT and SP along the rachisof 5-cm-long primary leaves. TerL, a terminal leaflet; PI-PIII, leaflet pairs.Bottom: Intraleaf expression gradients of SFT persist in �15-cm-long leavesbut level off in 25-cm long leaves. (I) Inactivation of SP sensitizes the responseof shoot and leaf meristems to mobile florigen. sp 35S:SFT donor shoot with4 leaves and all potential growing points removed, grafted onto a uf sft spstock. Note the termination by 2 consecutive inflorescences and the reducedcomplexity of the top compound leaves of the receptor shoot.

8394 � www.pnas.org�cgi�doi�10.1073�pnas.0810810106 Shalit et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 2

2, 2

020

leveled off (Fig. 2H), leaves become better exporters of florigen.Because SP promotes growth, its inactivation, or an elevatedSFT/SP ratio brought about by imported florigen, are likely toaccelerate leaf maturation, converting leaf status from sink tosource and leading to early release of florigen. This conjecture issupported by experiments comparing 35S:SFT and sp 35S:SFTdonors. Donors of both genotypes were either regular-growingshoots with leaves, axillary buds and growing apices, or 2-leaf stemsections with apices and axillary buds removed. Regular sp 35S:SFTdonors were significantly more effective inducers of both uf sft anduf sft sp receptors, and, as expected, sp receptors were moreresponsive (compare Fig. 2 D and I). However, when sp and SPdonors having mature leaves only were compared, they were equallyeffective (Table S3). Therefore, the superiority of sp donors can beattributed to their early maturation: 35S:SFT donors continue togenerate sink organs in the form of new branches, but branching issuppressed in sp 35S:SFT donors (Fig. 1D). Florigen can thereforebe seen as a hormone that by regulating leaf maturation in anSP-dependent manner, adjusts its own distribution at the whole-plant level.

Restrictions on the long-range function of FT are imposed bythe size of the protein (9, 10, 12, 13). In tomato, 35S:SFT withC-terminal translational fusions of GFP, RFP, and GR inducedprecocious primary flowering and, in an sp background, a typicalarrest of lateral leaflets. It was, however, impossible to visualizefluorescent signals in the respective transgenic plants. In agree-ment with results in Arabidopsis (13), all 3 fusion constructsfailed to generate graft-transmissible flowering signals whengrafted onto uf sft sp receptors (Fig. 3A). Confirming previousresults (9), f lowering was delayed in uf tomato or Arabidopsisplants expressing 35S:amiR-FT/SFT (Fig. S3), generating in bothspecies effects that were similar to conventional loss of function.As shown in Table S3, it was possible to exploit the advantagesof the tomato system and to restore flowering when uf 35S:amiR-FT/SFT receptors were grafted with 35S:SFT donors.

Additional restrictions on the long-range effects of florigen wereimposed by the cells and tissues in which SFT is expressed (Fig. 3A–E and Fig. S5). In contrast to the 35S and the phloem-specificSUC2 promoters, expression of SFT, driven by 3 early leaf-specificpromoters (BLS, FIL, and 650), although inducing precocious

flowering, failed to generate graft-transmissible florigenic signals(see Fig. S5 for further discussion).

Florigen Links the Transition to Flowering with Leaf Architecture. Tounderstand the genetic basis for phenotypic responses resultingfrom changing SFT/SP ratios, we examined the effects of dif-ferent SFT/SP balances in mutant backgrounds with altered leafmorphology. Leaves of trifoliate (tf ) plants form only 1 pair oflateral leaflets (24), but we observed that leaves of tf sp plantsgradually lose their lateral leaflets, resembling the sequentialreduction of leaflets in the compound leaves of the floweringrose shoots (Fig. 4A). A dysfunctional TF therefore sets highersensitivity thresholds for changing SFT/SP ratios.

The effect of sft on the trifoliate leaf is analogous to its mildeffect on WT leaves (compare Fig. 4B with Fig. 2B), but if SFTis inactivated in tf sp background the effect of sp is completelysuppressed: tf sft sp and tf sft leaves were indistinguishable fromeach other (Fig. 4B). Significantly, sft and sft sp leaves were alsoindistinguishable (Fig. S2), suggesting that the allelic status of SPis irrelevant in sft leaves as in sft shoots (Table S1).

However, when SFT was overexpressed in tf (i.e., tf 35S:SFT),a functional SP, as expected of the sensitized tf background, wasno longer sufficient to support the formation of lateral leaflets,and almost all leaves were simple (Fig. 4D and Fig. S4). As shownin Fig. 4E, tf did not affect the expression profiles of SFT or SP.

To determine whether mobile florigen also inhibits leafletmeristems in the sensitized tf leaves, tf 35S:SFT donor shootswere grafted onto tf and tf sft sp receptors. Both receptors beartrifoliolate leaves (Fig. 4B) and have a similar SFT/SP balance,but with functional and dysfunctional SFT and SP genes, respec-tively. As shown in Fig. 4 C and D, f lorigen induced simple leavesin both tf and tf sft sp receptors. In addition, shoots of the tfreceptors shifted to 2-leaf SUs, as in tf 35S:SFT (Fig. 4C). In tfsft sp receptors, f lorigen complemented the sft gene and inducedtf sp-like shoots. Thus, trifoliate leaves monitor both SFT/SP andsft/sp as balanced 1:1 ratios, and both genotypes are similarlymodified by imported florigen, which is the ultimate manifes-tation of the florigen-dependent SFT/SP regulatory hierarchy.

The tomato leaf is initiated as a terminal leaflet and anelongated rachis. Independent pairs of lateral leaflets, eachcapable of duplicating the compound pattern, are then formed(24). In WT and tf, SFT and florigen regulate, in an SP-dependent context (Fig. 2), the arrest of lateral leaflets in agradient opposite to their formation. The same is observed in tfsp or tf leaves importing florigen or overexpressing SFT (Fig. 4).In all these cases the terminal leaflets and their elongated rachisare unaffected, suggesting distinct regulatory mechanisms forthe initiation of classes of leaflets. We surmise that florigen,mediated by the SFT/SP ratio, regulates the leaflet initiationgradient in conjunction with auxin (see SI Text and below).

Florigen Links the Vegetative/Reproductive Balance in the Inflores-cence with the Generation of Abscission Zones. Abscission zones(AZs) mark the sites where plant organs, particularly fruits andleaves (Fig. 5A), are eventually separated from the main body ofthe plant. The formation of AZs is regulated by day length,auxin, and ethylene, and the deciduous habit is considered animportant innovation of angiosperms (25, 26). In tomato, for-mation of floral pedicel AZs is preceded by site-specific celldivisions (27), and AZs are completely missing from jointless1(j1) and j2 pedicels (Table S2).

mc, sft, and bl condition partial vegetative inflorescencessimilar to those seen upon overexpression of SP (17) (Table S2).In the course of studying the interactions between mc, bl, andflorigen, we noticed that AZs in floral pedicels of mc, bl, and sftwere incomplete, irregular, or mislocated. We found that similarto j1, f loral pedicels of sft mc, sft bl, or mc bl double mutantscompletely lacked AZs (Fig. 5 B and C), indicating that mc and

A B

D E

C

Fig. 3. Restrictions on the movement of florigen in the perennial tomato bush.(A) Florigenic potential of tagged-SFT proteins and of SFT driven by specificpromoters via either direct (:) or transactivation (��). (B and C) Expression ofpSUC2 is confined to the companion cells (B) of the phloem strands (C) of matureveins. (D and E) Expression by pBLS is limited to young blades (D) and is excludedfrom mature veins (E). For the detailed analyses of these patterns, see Fig. S5.

Shalit et al. PNAS � May 19, 2009 � vol. 106 � no. 20 � 8395

PLA

NT

BIO

LOG

Y

Dow

nloa

ded

by g

uest

on

Mar

ch 2

2, 2

020

bl act redundantly with sft in specifying these tissues. Graftingexperiments showed that mobile florigenic signals, which com-plement sft, penetrate floral pedicels to rescue AZs in sft mc andsft bl pedicels (Fig. 5 D and F). Although SFT, SP, MC, BL, andJ1 are all expressed in pedicels and floral buds (Fig. 5E), the

corresponding proteins did not interact in any combination inyeast 2-hybrid tests (data not shown).

AZs serve as a classical example of a morphogenetic traitregulated by auxin (25, 26). Auxin is also involved in regulatingradial expansion of stems, formation of leaflet meristems, andvarious forms of apical dominance (28, 29), all shown here toinvolve florigen. One possible scheme by which florigen and auxininteract to regulate the sympodial cycles and leaf architecture isdiscussed briefly in the SI Text. It is expected that the interactionsbetween the 2 hormones in regulating the growth balance in shootsystems will attract much attention in coming years.

DiscussionFlorigen as a Growth Hormone and the Evolution of the SFT/SPRegulatory Hierarchy. The need for plant organs to respondeffectively to changing environmental signals dictates quantita-tive regulatory schemes with inherent potentials for reversibility.Such tenets are satisfied by florigen functioning as a generalgrowth hormone. SFT and SP modulate a variety of signalingpathways but are not directly involved in the fate of cells ororgans. Rather, they regulate the balance of diverse growthprocesses (20) and thus facilitate the potential for plasticity ingrowth, not much different from the proposed role of theNotch/Wnt system that pleiotropically modulates multiple sig-naling processes but is not essential for any of them (30).

FT is an integrator of all flowering pathways (4), but as shownhere (Figs. 1, 2, and 4), SFT is imported by organs in which it isalready produced, with the exception of the SAM. It was recentlysuggested that genes of the autonomous pathway carry pleiotropicvegetative functions (31). Mutations in SOC1 and FUL, 2 majorflowering genes in Arabidopsis, affect vegetative functions (32).Likewise, GA promotes flowering (33) and modifies growth. Givenour finding that florigen is not only a flowering-specific agent,perhaps there are no designated flowering genes at all. One reasoncould be that ‘‘flowering’’ is an arbitrary external description (34)that does not match the internal description of the plant. By thesame token, the mechanism requiring movement of florigen fromleaves to apices (21, 22) may not be specific to flowering. Rather itis one aspect of a more general mechanism in which the florigen

E

A

B

C D

Fig. 4. Leaf architecture in tf plants is determined by the florigen-dependent,SFT/SP regulatory hierarchy. (A) Left: A flowering tf plant. All leaves have 3leaflets. Middle: A tf sp flowering shoot with a gradual reduction in leaf com-plexity. Right: Stepwise elimination of leaflets in compound leaves of the gardenrose toward flowering. (B) Inactivation of TF sensitizes the growth response ofleaves to changing SFT/SP ratios (ratios are listed below the corresponding im-ages). SFT/SP ratio (tf, far left) and sft/sp ratio (tf sft sp, far right) result in similartrifoliolate leaves. High ratios, as in overexpression of SFT or inactivation of SP(leaves 3 and 4 from left, respectively) induced simple LANCEOLATE-like leaves(24, 36). Note that tf sft leaves (second from left) have low SFT/SP ratio but areindistinguishable from tf sft sp leaves (right-most), and both have extendedrachisesandadditional leafletscharacteristicof sft leaves (Fig.3). (CandD)Mobileflorigen modulates the SFT/SP balance to generate 2 leaf SUs and to arrest leafletmeristems. (C)Florigendonor inducedslenderstems, simple leaves,andareducednumber of leaves per SU in a tf receptor (Inset). (D) A systemic reconstitution ofhigh SFT/SP balance in tf leaves. Left: The contribution of florigen by a WT donoris insufficient to reduce the complexity of tf receptor leaves. Middle: Systemicreconstitution of a high SFT/SP ratio in tf sft leaves induces simple leaves. Right:Mobile florigen elevates the level of SFT in tf sft sp leaves, thereby inducing‘‘reversion’’ to a simple architecture. (E) tf does not affect expression gradients offlowering genes.

A B C D

E F

Fig. 5. Florigen links the formation of AZs with inflorescence genes. (A)Normal AZ of mature tomato fruit. (B) mc bl floral pedicels produce no AZs. (C)SFT is required for the formation of AZs in an mc background. (D) Graft-transmissible signals donated by a 35S:SFT scion restore AZs in mc sft floralpedicels. (E) Genes involved in generating floral AZs are expressed in WT floralpedicels. YP, young pedicels before AZs can be observed; MP, mature pedicelswith developed AZs; YF, young flowers with pedicels removed. (F) Summary oflong-range complementation tests of AZs by florigen.

8396 � www.pnas.org�cgi�doi�10.1073�pnas.0810810106 Shalit et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 2

2, 2

020

hormone is exploited to change local SFT/SP balances. If correct,florigen may regulate flowering in species in which SFT orthologuesare normally expressed in apices, just as it regulates leaflet meris-tems in the compound leaf.

In shoots and leaves, meristems react to florigen in an SP-dependent manner. Here we offer an evolutionary scenario for thegeneration of this mechanism in flowering plants. Consider thefollowing. (i) Overexpression phenotypes define SFT as a growthretardant (6). It terminates primary SAMs, sympodial SAMs, andinflorescence meristems and arrests stem and leaflet meristems(Figs. 1, 2, and 4). (ii) The arrest of meristematic activities bySFT/florigen is alleviated by even the low WT level of SP. SP istherefore an amazingly potent inhibitor of SFT. (iii) SP is function-ally relevant only in the presence of a functional SFT, and con-versely, its inactivation is irrelevant without a functional SFT. (iv)Expression of FT orthologues is correlated with photoperiod budsetting in slow-growing conifers (35). With their highly conductivevasculature, fast-growing meristems, and need for a rapid responseto environmental signals, the ability of flowering plants to exploitnew habitats likely required high levels of FT/florigen, but theselevels were also detrimental. We speculate that SP evolved specif-ically to alleviate the detrimental effects of SFT. This is supported,as shown in Fig. 6, by the absence of genes of the SP/TFL1/CENclade from all known nonflowering plant genomes.

MethodsPlants were grown and grafted as specified previously (6). Monogenic linesand WT cultivars were obtained from the Rick Center at University of Califor-nia, Davis. Combinations mentioned in the text and in Tables S1–S3 werespecifically bred for this work. Cloning, RNA isolation, and PCR were per-

formed as previously described (6). A list of primers, fragment sizes, andnumber of cycles for each gene is given in Table S4. Confocal imaging andmicroscopy were performed as described previously (39).

ACKNOWLEDGMENTS. We thank Sheila McCormick, Zach Lippman, andBenny Horowitz for their critical reading of the manuscript. This work wassupported by grants from the International Science Foundation (ISF) andthe Human Frontier Science Program (to E.L.) and from the ISF and MINERVA(to Y.E.).

1. Chailakhyan MK (1936) [About the mechanism of the photoperiodic response (inRussian)]. Dokl Akad Nauk SSSR 1:85–89.

2. Zeevaart JAD (1976) Physiology of flower formation. Ann Rev Plant Phys 27:321–348.3. Lang A, Chilakhian MK, Frolova LA (1977) Promotion and inhibition of flower forma-

tion in a day-neutral plant in graft with a short-day plant and a long-day plant. ProcNatl Acad Sci USA 74:2412–2416.

4. Kobayashi Y, Weigel D (2007) Move on up, it’s time for change—mobile signalscontrolling photoperiod-dependent flowering. Genes Dev 21:2371–2384.

5. Turck F, Fornara F, Coupland G (2008) Regulation and identity of florigen: FLOWERINGLOCUS T moves center stage. Ann Rev Plant Biol 59:573–594.

6. Lifschitz E, et al. (2006) The tomato FT ortholog triggers systemic signals that regulategrowth and flowering and substitute for diverse environmental stimuli. Proc Natl AcadSci USA 103:6398–6403.

7. Lifschitz E, Eshed Y (2006) Universal florigenic signals triggered by FT homologuesregulate growth and flowering cycles in perennial day-neutral tomato. J Exp Bot57:3405–3414.

8. Tamaki S, Matsuo S, Wong HL, Yokoi S, Shimamoto K (2007) Hd3a protein is a mobileflowering signal in rice. Science 316:1033–1036.

9. Mathieu J, Warthmann N, Kuttner F, Schmid M (2007) Export of FT protein fromphloem companion cells is sufficient for floral induction in Arabidopsis. Curr Biol17:1055–1060.

10. Jaeger KE, Wigge PA (2007) FT protein acts as a long-range signal in Arabidopsis. CurrBiol 17:1050–1054.

11. Lin MK, et al. (2007) FLOWERING LOCUS T protein may act as the long-distanceflorigenic signal in the cucurbits. Plant Cell 19:1488–1506.

12. Corbesier L, et al. (2007) FT protein movement contributes to long-distance signalingin floral induction of Arabidopsis. Science 316:1030–1033.

13. Notaguchi M, et al. (2008) Long-distance, graft-transmissible action of ArabidopsisFLOWERING LOCUS T protein to promote flowering. Plant Cell Physiol 49:1645–1658.

14. Yang Y, Klejnot J, Xuhong Y, Xuanming LX, Lin C (2007) Florigen (II): It is a mobileprotein. J Int Plant Biol 49:1665–1669.

15. Bradley D, et al. (1996) Control of inflorescence architecture in Antirrhinum. Nature379:791–797.

16. Bradley D, Ratcliffe O, Vincent C, Carpenter R, Coen E (1997) Inflorescence commitmentand architecture in Arabidopsis. Science 275:80–83.

17. Pnueli L, et al. (1998) The SELF-PRUNING gene of tomato regulates vegetative toreproductive switching of sympodial meristems and is the ortholog of CEN and TFL1.Development 125:1979–1989.

18. Kobayashi Y, Kaya H, Goto K, Iwabuchi M, Araki T (1999) A pair of related genes withantagonistic roles in mediating flowering signals. Science 286:1960–1962.

19. Kardailsky I, et al. (1999) Activation tagging of the floral inducer FT. Science 286:1962–1965.

20. Pnueli L, et al. (2001) Tomato SP-interacting proteins define a conserved signalingsystem that regulates shoot architecture and flowering. Plant Cell 13:2687–2702.

21. Wigge P, et al. (2005) Integration of spatial and temporal information during floralinduction in Arabidopsis. Science 309:1056–1059.

22. Abe M, et al. (2005) FD, a bZIP protein mediating signals from the floral pathwayintegrator FT at the shoot apex. Science 309:1052–1056.

23. Molinero-Rosales N, Latorre A, Jamilena M, Lozano R (2004) SINGLE FLOWER TRUSSregulates the transition and maintenance of flowering in tomato. Planta 218:427–434.

24. Hareven D, Gutfinger T, Parnis A, Eshed Y, Lifschitz E (1996) The making of a compoundleaf: Genetic manipulation of leaf architecture in tomato. Cell 84:735–744.

25. Roberts JA, Elliot KA, Gonzallezz-Carranza ZH (2002) Abscission, dehiscence, and othercell separation processes. Ann Rev Plant Biol 35:131–158.

26. Taiz L, Zeiger E, eds (2006) Plant Physiolology (Sinaur Associates, Sunderland, MA).27. Jensen TE, Valdovinos JG (1967) Fine structure of abscission zones. I. Abscission zones

and pedicels of Tobacco and Tomato flowers at anthesis. Planta 77:298–318.28. Avasarala S, Yang J, Caruso JL (1996) Production of phenocopies of the lanceolate

mutant in tomato using polar auxin transport inhibitors. J Exp Bot 47:709–712.29. Ongarno V, Leyser O (2008) Hormonal control of shoot branching. J Exp Bot 59:67–74.30. Hayward P, Kalmar T, Martinez-Arias A (2008) Wnt/Notch signaling and information

processing during development. Development 135:411–424.31. Marquardt S, Boss PK, Hadfield J, Dean C (2006) Additional targets of the Arabidopsis

autonomous pathway members, FCA and FY. J Exp Bot 57:3379–3386.32. Melzer S, et al. (2008) Flowering-time genes modulate meristem determinacy and

growth form in Arabidopsis thaliana. Nat Genet 40:1489–1492.33. Hisamatsu T, King RW (2008) The nature of floral signals in Arabidopsis. II. Roles for

FLOWERING LOCUS T (FT) and gibberellin. J Exp Bot 59:3821–3829.34. Brenner S (1975) Cell patterning. CIBA Found Symp 29:119–130.35. Gyllenstrand N, Clapham D, Kallman T, Lagercrantz U (2007) A Norway spruce FLOW-

ERING LOCUS T homolog is implicated in control of growth rhythm in conifers. PlantPhysiol 144:248–257.

36. Ori N, et al. (2007) Regulation of LANCEOLATE by miR319 is required for compound-leaf development in tomato. Nat Genet 39:787–791.

37. Hanzawa Y, Money T, Bradley D (2005) A single amino acid converts a repressor to anactivator of flowering. Proc Natl Acad Sci USA 102:7748–7753.

38. Ahn JH, et al. (2006) A divergent external loop confers antagonistic activity on floralregulators FT and TFL1. EMBO J 25:605–614.

39. Goldshmidt A, Alvarez JP, Bowman JL, Eshed Y (2008) Signals derived from YABBY geneactivities in organ primordial regulate growth and partitioning of Arabidopsis shootapical meristems. Plant Cell 20:1217–1230.

Fig. 6. Genes of the SP/TFL1/CEN clade are not found in nonflowering plants.CETS genes from EST databases and sequenced genomes of land plants (moss,red; lycophytes, orange; gymnosperms, blue; angiosperms, green) were sub-jected to Bayesian phylogenetic analyses (for the complete gene phylogenyand a detailed evolutionary analysis, see Fig. S6). The resulting tree was rootedwith moss sequences. Four distinct clades were identified: MFT, a putativegrowth suppressor, is present in all lineages; SFT/FT, the universal precursor offlorigen, is present in all flowering plants and highly related to the gymno-sperms FT-like; the SP/TFL1/CEN clade is unique to flowering plants. W, Y, andH are amino acids characterizing the ligand-binding pocket of the 3 majorclades (37, 38).

Shalit et al. PNAS � May 19, 2009 � vol. 106 � no. 20 � 8397

PLA

NT

BIO

LOG

Y

Dow

nloa

ded

by g

uest

on

Mar

ch 2

2, 2

020