Cell, Vol. 84, 735–744, March 8, 1996, Copyright ©1996 by Cell Press
The Making of a Compound Leaf:
Genetic Manipulation of Leaf Architecture
in Tomato
Dana Hareven,* Tamar Gutfinger,* Ania Parnis,* a simple versus compound leaf have not been investigated.
Compound leaves have been considered to rep-Yuval Eshed,† and Eliezer Lifschitz*
*Department of Biology resent the reiteration of the program that makes simple
leaves. To study the molecular-genetic basis for thisTechnion–Israel Institute of Technology
Haifa 32000 dichotomy between simple and compound leaves, we
have exploited the maize homeobox-containing Knot-Israel
†Department of Genetics and Field Crops ted-1 (Kn1)gene inconjunction witha rangeof mutations
that alter the tomato compound leaf.Faculty of Agriculture
Hebrew University The diverse forms of plant organs are shaped by developmental
events in their respective meristems (Sus-Rehovot 70700
Israel sex, 1989), and the identification of Kn1 as a meristematic
homeobox gene has permitted molecular-genetic
analyses of leaf morphogenesis. Dominant mutations in
the Kn1 locus of maize result in distinct alterations inSummary
cells along the vasculature of the blade. Formations of
pocketed outgrowths (knots) on lateral veins, overallThe most distinctive morphogenetic feature of leaves
growth retardation, wider leaves, distorted patterns ofis their being either simple or compound. To study the
lateral veins, disappearance of the ligule appendage,basis for this dichotomy, we have exploited the maize
and ectopic formation of sheath tissues on blades char-homeobox-containing Knotted-1 (Kn1) gene in conacterize
the KNOTTED syndrome (Freeling and Hake,junction with mutations that alter the tomato com-
1985; Sinha and Hake, 1994). The Kn1 gene was clonedpound leaf. We show that misexpression of Kn1 conby
transposon tagging and shown to represent a con-fers different phenotypes on simple and compound
served class of plant genes coding for homeodomain-leaves. Up to 2000 leaflets, organized in compound
containing plant proteins (Hake et al., 1989; Volbrechtreiterated units, are formed in tomato leaves expresset
al., 1991; Hake, 1992; Kerstetter et al., 1994; Becrafting Kn1. In contrast, Kn1 induces leaf malformations
and Freeling, 1994). It was convincingly shown that thebut fails to elicitleaf ramification in plantswith inherent
dominant nature of Kn1 mutations is a consequence ofsimple leaves such as Arabidopsis or in tomato mutant
its ectopic expression in the lateral veins of the leafplants with simple leaves. Moreover, the tomato Kn1
blade (Smith et al., 1992; Jackson et al., 1994). The Kn1ortholog, unlike that of Arabidopsis, is expressed in
gene is normally expressed in vegetative and inflores-the leaf primordia. Presumably, the two alternative leaf
cence apical meristems, but not in leaf primordia, norforms are conditioned by different developmental proin
developing leaves or in floral organs. When the Kn1grams in the primary appendage that is common to
gene was misexpressed in dicot plant species with sim-all types of leaves.
ple leaves, like tobacco and Arabidopsis, severe morphogenetic
alterations were induced(Sinha et al.,1993b;Introduction
Lincoln et al., 1994). The leaves of these species turn
lobed and rumpled, as well as shorter and wider and,All types of leaves, regardless of their eventual architecin
extreme cases, ectopic shoots appear on the mainture, arise as dorsiventral appendages from the flanks
veins. Growth retardation and loss of apical dominanceof the shoot apical meristem. The dorsiventral nature of
have also been observed. Significantly, and despite theleaf primordia contrasts with axillary branches or stems
excessive meristematic activities causing all the growththat arise as radial primordia (Cutter, 1971; Kaplan,
malformations, leaves of Arabidopsis and tobacco plant1973). In angiosperms, vegetative leaves come mainly
overexpressing Kn1, like those of maize Kn1 overpro-in two basic arrangements: simple and compound. The
ducers, remain simple.single blade of the simple leaf, as well as the indepenWe
hypothesize that simple and compound leavesdent blades (leaflets) of the compound leaf, can be sesare
the result of different patterns of meristematic activi-sile or can be carried on a petiole, and their margins can
ties. We also propose that simple leaves are morphoge-be entire, lobed, parted, dentate, or palmate. Leaflets of
netically rigid, while compound leaves are developmen-a compound leaf are distinguished from the leaves, as
tally more flexible, thus permitting the phenotypiconly the latter form axillary buds (see Smith and Hake,
manifestation of a wide scope of gene mutations. To-1992).
mato is well suited for the analysis of simple versusVariations in leaf shape have beenanalyzed by genetic
compound in isolation from other parameters of leafor anatomical means in maize, pea, cotton, and a wide
shape, since a range of dominant and recessive muta-variety of other species (Marks, 1987; Dolan and Poetions,
which change leaves from supercompound tothig, 1991; Sinha et al., 1993a; Steeves and Sussex,
simple, are available (Stevens and Rick, 1986; Figure1989), and growth parameters of blade units in simple
3) and since all appendages of its compound leaf areleaves have been studied by clonal analysis (Poethig,
identical.1987; Poethig and Sussex, 1985; also Freeling, 1992;
We suspected that the morphogenetic versatility ofSmith and Hake 1992). However, at present, no coherent
developmental-geneticframework for leaf morphogene- the tomato compound leaf, in conjunction with the demonstrated
meristematic function of the Kn1 gene familysis has been derived, and the pathways that determine
Cell
736
Figure 1. Ultracompound Leaves in Kn1-Expressing Tomato Plants
(A) Wild-type compound leaf: the prototype unit. The terminal leaflet (TL) emerges first and the pairs of lateral leaflets (LT) appear in basipetal
order. Note that all leaflets are anatomically similar, each is petiolated (P), and with serrate margins. Folioles (F) appear occasionally between
leaflets along the rachis (R) or on either side of the petioles.
(B) A supercompound leaf of transgenic tomato plants expressing the Kn1 gene. Terminal and lateral leaflets of the prototype unit now acquire
the compoundness of the wild-type leaf.
(C) Transgenic tomato plant expressing Kn1 under the control of the 35S promoter. Note the bushy appearance of the plant, the consequence
of lost apical dominance.
in both mono- and dicotyledonous plants, would provide theloss of apical dominance,resulting in dwarfed,bushy
plants (Figure 1C).new opportunities to study leaf arrangement. In this reRamified
primary, secondary, and tertiary lateral leaf-port, we show that the ubiquitous misexpression of the
lets excised from the leaf shown in Figure 1B are illus-maize Kn1 gene in the tomato leaf confers dramatic
trated in Figure 2A. In such leaves,each lateral extensionadditional orders of subdivisions on the already comacquires
the complexity of the primary compound de-pound leaf. Such a ramification is completely prevented
sign, as shown in Figure 1A. Independent transformantsin the simple leaves of the tomato mutant Lanceolate
yielddifferent types of ultracompound leaves,producing(La). Our observations suggest that a compound leaf is
leaflets with altered vein-to-lamina ratio. The type ofnot a trivial reiteration of a simple leaf and that the
leaflets shown in Figures 1B and 2B are the most promi-making of either leaf type depends onmutually exclusive
nent. Reiteration of such units (the term phyllomere isgrowth patterns.
suggested to describe a given prototypic leaf architecture)
results in an increase in the number of leaflets fromResults
9 to 700–2000. However, note that although the overall
number of leaflets per compound leaf is greatly in-Misexpression of Kn1 Makes the Tomato Leaf
creased, the dimensions of the leaf remain mostly unStrikingly
More Compound
changed. The excessive proliferation of lateral leaf ap-In plants of the wild-type progenitor line, the prototype
pendages that is associated with alterations of the
compound leaf is composed of a midvein (rachis), a
lamina-to-vein ratio in a given leaflet is also associated
terminal (distal) leaflet, and three to four pairs of lateral,
with altered relative size of the terminal leaflets in each
petiolated, and dentate leaflets. Occasionally, several
compound unit, or with the modified spacing of such
additional folioles emerge on the midvein between leafunits
along the midveins. Leaves representing extreme
lets or on the petioles of the lateral leaflets (Figure 1A). alterations of these parameters, which were formed by
The effect of the Kn1 gene on morphogenesis of com- independent transformants, are shown in Figure 2C.
pound leaves was observed by generating transgenic Several additional major anatomical features charactomato
plants. Generated in the determinate TRK9/8 terize Kn1-expressing primordia and leaf blades in
and VF36 lines were 42 kanamycin-resistant primary 35S::Kn1 plants.Nearly equal adaxial (toward thecenter)
transformants (T1 plants) expressingKn1 under thecon- and abaxial cell growth confers, from the outset, an
trol of the potent and ubiquitous cauliflower mosaic virus erect shape on the wild-type leaf primordium. First and
35S promoter (Benfey and Chua, 1990). All butfive trans- secondprimordia of the lateral leaflets appearin a basipformants
exhibited extreme alterations in the degree of etal order when the primary, peg-like, primordium of the
structural ramification of the leaf. In the 37 independent wild-type leaf reaches about 300–400 ␮m and 800 ␮m
M series T1 primary transformants displaying altered in length, respectively (Dengler, 1984; stars inFigure 2D).
morphogenesis, mature leaves are subdivided to the In 35S::Kn1 plants, multiple lateral leaflet buds develop
fourth, fifth, or sixth order, forming a supercompound prematurely, in a much more distal position on the leaf
leaf (Figure 1B). The appearance of supercompound primordium (stars in Figures 2E and 2F). New meristematic
centers in a secondary leaflet primordium showleaves is always associated with growth retardation and
The Making of a Compound Leaf
737
Figure 2. Growth Parameters in Kn1-Expressing Tomato Plants
(A) Degree of subdivision of a single supercompound leaf. Primary (P), secondary (S), and tertiary (T) lateral leaflets excised from the leaf
shown in Figure 1B illustrate the ramification of the supercompound leaf.
(B) A supercompound leaf with altered lamina-to-vein ratio. This type of leaf and the one shown in Figure 1B are the most prominent among
tomato plants expressing the 35S::Kn1 transgene.
(C) Extreme variation in blade and leaflet morphology in tertiary branches of supercompound leaves of four independent primary transformants.
(D–E) Scanning electron micrographs (SEMs) of wild-type and 35S::Kn1 transgenic apices. (D) Wild-type apex with leaf primordia. Two older
leaf primordia (L) and a newly emerging one (marked with an arrow) are seen. Leaflet primordia are marked with stars. Note the erect stage
of primordia and the relative sites of emerging lateral leaflets.
(E–F) Apices of 35S::Kn1 plants with premature appearance of distal lateral appendages (stars) and inward curving of the primordia.
(G) Fiddle-head shape of supercompound leaves (L) before expansion.
a similar pattern of leaflet proliferation. Leaf primordia later as part of Figure 4) revealed high levels of expression
in leaves and flowersof all affected plants, irrespec-in the shoot apex, as well as young leaflets of the emerging
leaf of Kn1-expressing plants, presumably as a con- tive of the severeness of the leaflet phenotype or the
degree of subdivision (Figure 4A, lanes 1–4 and 9–12).sequence of unequal abaxial and adaxial growth, tend
to curl inward toward the center and display a “fiddlehead”
shape, reminiscent of fern leaves. This shape is
retained until just prior to final leaf expansion (Fig- Kn1 Induces Leaf Ramification in the Compound
Leaves of Petroselinum, but Not in the Simpleure 2G).
In both tomato and tobacco leaves expressing the Leaves of La Plants
Phenotypic consequences of misexpression of Kn1 in35S::Kn1 transgene, the prominence of the midvein is
reduced to a condensed palmate-like design. The ter- the simple leaves of maize (Sinha and Hake, 1994) or in
transgenic rice, tobacco, and Arabidopsis plants aretiary vein system is more diffused, and areoles (the
smallest lamina fields confined by veins) are 2- to 3- restricted to local distortions of the blade (Matsuoka et
al., 1993; Sinha et al., 1993; Lincoln et al., 1994). Infold larger. Every vein thus serves more cells (data not
shown). contrast with the dramatic effect on tomato leaves, the
misexpression of Kn1 in these other species does notUnlike leaves,morphology of the inflorescences, flowers,
and floral organs of tomato are not visibly affected change the basic simple design of the leaves. The difference
could be attributed to unknown species-specificby the misexpression of the Kn1 gene. Two thirds of
the primary transformants are fertile, and the 35S::Kn1 factors, or to inherent variations in the programs that
condition simple and compound leaves. To address thisphenotype is transmitted to T2 plants in the expected
Mendelian proportions for a single locus–dominant mu- question directly in a single plant species, misexpression
of Kn1 was examined in several mutants of tomato.tation. The early degeneration of flowers in the nonfertile
one third of the transgenicplants is attributed to second- Petroselinum (Pts) and La represent the two extreme
variations of the compound leaf. Pts leaves have elabo-ary effects such as overall growth retardation.
Analysis of Kn1 mRNA in transgenic plants (shown rate leaflets with three to four pairs of secondary leaflets
Cell
738
lobed, and frequently rumpled (Figure 3F); growth of
La/ϩ; 35S::Kn1 plants is severely retarded and apical
dominance is lost. La/ϩ; 35S::Kn1 plants are fertile and,
among progeny of a La/ϩ; 35S::Kn1 (M1) plant, La/
ϩ::ϩ/ϩ; La/ϩ; 35S::Kn1; and ϩ/ϩ; 35S::Kn1 phenotypes
segregate in the expected proportions. Thus, the failure
of La; 35S::Kn1 leaves to subdivide is not attributed to
failure in the transcriptional expression of the 35S:Kn1
transgene. Presumably, the meristematic deficiency associated
with the Lanceolate allele cannot be rescued
by Kn1 ectopic activity and leaf ramification remains
arrested in these plants.
In contrast with La/ϩ plants, Pts/Pts primary transformants
expressing the 35S::Kn1 are not readily disFigure
4. Blot Analysis of Kn1 and TKn1 Transcripts in Total RNA tinguishable from wild-type plants possessing the
Samples
35S::Kn1 construct. Furthermore, one half of the F1
(A) Expression of Kn1 in floral buds of independent primary (T1) progeny of Pts/Pts plants pollinated by transgenic plant
transformants driven by the 35S and dUTPase promoters: lanes 1–4
M1 also form supercompound leaves, with multiple diand
9–12; 35S::Kn1 transgenic plants; lanes 5–8; PdUTPase::Kn1
minutive leaflets typical of the parent plant. Leaves oftransgenic plants.
approximately one quarter (6 out of 26) of F2 plantsThe letters a, b and c, which appear above lanes 1–12, denote
strong, medium, and weak phenotypic expression, respectively. exhibit extreme growth retardation and leaflets with very
(B) Expression pattern of the tomato TKn1 gene in different plant narrow blades, in addition to being supercompound.
organs: 3 cm long leaves of the 93-137 wild-type line (lane 1); 5 cm Thus, the dominant Pts allele neither antagonizes nor
long leaves of 93-137 plants (lane 2); 3–5 mm long leaves of 93-137
enhances the ramification effect of Kn1.
plants (lane 3); 5 mm long intact apices of 93-137 plants (lane 4);
The Lanceolate leaf is simple and also differs from10 mm long stem sections of 93-137 plants (lane 5); 5 mm long stem
wild type in that its margins are entire (compare Figuresections of 93-137 plants (lane 6). Apices include the actual apex,
up to five-leaf primordia, the longer of which is 5 mm, and two to 3E with 1A). To determine which of these two features
four floral primordia of the stage shown in Figure 6. Stem sections prevents the ramification effect of Kn1, the 35S::Kn1
are barren stems 5 mm long just below the shoot apex. transgene was introduced into potato-leaf (c) mutant
Mature flowers two to three days before anthesis (lane 7). Floral
plants. Leaves of c/c plants are compound, but their
organs of mature flowers: sepals (lane 8); petals (lane 9); stamens
margins are entire rather than dentate, and they bear(lane 10); carpels (lane 11); The amount of carpel RNA loaded is
onlytwo pairs(rather than three tofour) of leaflets (Figureonly one half of that loaded for the other organs; Anantha floral
meristems (lane 12). Flowers are good representatives of the level 3C). As shown in Figure 3G, such leaves clearly respond
of Kn1 transcripts because they exhibit no phenotypic variation to misexpression of Kn1 by increased subdivision, as
between and within transgenic plants. do regular compound leaves, but ramification is restricted
to the terminal portion of the midveins, leaving
relatively long naked petioles with no or only a slender
(Figure 3A), each of which resembles the wild-type phyl- lamina.
lomere (compare with Figure 1A). Such leaves are said The pattern of the ramified leaves in wild-type, Pts,
to be divided to the third order. A simple and entire leaf, and potato-leaf plants suggested that Kn1 will elicit the
composed of one petiolated blade similar to that of multiplication of preexisting compound patterns, but is
wild-type tobacco or Arabidopsis, is formed in plants unable to increase the complexity of a given phyllomere
heterozygote to the dominant La gene (La/ϩ) (Figure or to rescue the basic compound prototype in mutant
3E). Homozygote La/La seedlings are practically lethal, plants. To determine the relation between alterations of
as no apical shoot meristems are produced (Mathan the basic prototype and additional ramifications further,
and Jenkins, 1962; Stettler, 1964; Caruso, 1968). Three and to examine the possibility to manipulate leaf archimutants
in which compoundness is intermediate be- tecture in a predictable manner, the 35S::Kn1 transgene
tween wild type and La illustrate the genetic control of was introduced via regular crosses into trifoliate mutant
the development of the basic prototype. In entire (e/e) plants. A trifoliate (tf/tf) leaf of plants expressing one
homozygote plants, a pair of reduced and sessile lateral dose of the 35S::Kn1 transgene is shown in Figure 3H.
leaflets is fused to the terminal one to generate a pseu- In such leaves, every appendage is converted into a
dosimple leaf (Figure 3B). The recessive potato-leaf ternate design itself so that three triplets rather than one,
gene (c/c, sometimes referred to as solanifolia) permits and nine leaflets rather than three, are formed (compare
the formation of only two, rather than three to four pairs with Figure 3D). The wild-type prototype, though, was
of lateral leaflets (Figure 3C). Leavesof trifoliate homozy- not restored.
gote plants have long petioles and bear only one pair
of lateral leaflets (Figure 3D).
The 35S::Kn1 transgene was introduced into La/ϩ The Tomato Kn1 Gene: Sequence,
Analysis, Genetic Mapping,mutant plants by transformation and by crossing with
the transgenic plant M1 shown in Figure 1C. Results and Developmental Expression
To explore the possibilities that either Pts or La mayfrom both experiments were identical. With the exception
of leaf ramification, La/ϩ plants expressing represent mutations in the Kn1 ortholog of tomato and
that the developmental expression of Kn1 genes in spe-35S::Kn1 display all facets of Kn1 misexpression: leaves
are much smaller, sometimes relatively wider, slightly cies with simple and compound leaves are different, we
The Making of a Compound Leaf
739
Figure 3. Phenotypic Expression of Kn1 in Leaf Arrangement Mutants of Tomato
(A–E) Genes affecting leaf compoundness in tomato. (A) Petroselinum. Chromosome VI. Note the additional order of subdivision in comparison
with the wild-type prototype shown in Figure 1A. The terminal and lateral leaflets acquire the architecture of the wild-type leaf phyllomere.
(B) A pseudosimple leaf of plants homozygote for the entire (e) recessive gene. One or more pair(s) of leaflets is merged (or fused) with the
terminal leaflet, as suggested also by the altered orientation of veins at the distal half of the structure. The borderline between the terminal
leaflet and the fused laterals is indicated by an arrow. Depending on genetic background, less and more extreme leaf arrangements are
formed. Chromosome IV. (C) Potato-leaf. Chromosome VI. Only two pairs of lateral leaflets with entire margins are formed. Number of folioles
is also reduced. (D) trifoliate. Chromosome V. Only terminal leaflets and the most adjacent pairs of laterals are formed. (E) Simple and entire
leaf of Lanceolate heterozygote (La/ϩ) plants. Chromosome VII.
(3F–G) Phenotypic expression of Kn1 in La, potato-leaf, and trifoliate mutant plants. (F) Kn1 does not rescue the compoundness of the simple
La/ϩ leaf. Three modified leaves of La/ϩ; Kn1 are shown. Note the reduced size and altered lamina shape and compare with Tobacco::Kn1
leaves in Figure 7B and Sinha et al. (1993).
(G) Kn1-induced ramification of potato-leaf leaves. Left: young potato-leaf with entire margins and only one pair of major leaflets. Right: a
ramified c/c::Kn1 leaf (top) and excised subdivided lateral leaflet (bottom). (H) trifoliate::Kn1 leaf. Each of the three appendages of the ramified
leaf acquires the ternate arrangement and elongated petiole of the progenitor leaf. Compare with Figure 3D.
undertook the isolation of the tomato Kn1 (TKn1) gene. by various procedures, and subsequent isolation and
sequencing revealed that at least five genes belong toUsing the maize gene as a probe, a cDNA clone, (designated
TKn1), with extensive homology in the homeodo- the Kn1 family of tomato. They do not cross-hybridize
under stringent conditions, and only TKn1 exhibits ex-main and flanking sequences was isolatedfrom a tomato
shoot cDNA library(Figure 5). All features that character- tensive homology outside the homeodomain with the
maize and Arabidopsis Kn1genes. RestrictionFragmentize the Kn1-type homeodomain (for reviews, see Kerstetter
et al., 1994; Lincoln et al., 1994; Ma et al., 1994, Length Polymorphism. (RFLP) mapping using the N-terminal
half of the TKn1 gene unambiguously placed itare conserved in the TKn1 gene (see Figure 5).
Southern blot analysis (data not shown), screening on chromosome IV rather than chromosomes VI or VII,
Cell
740
Figure 5. The Tomato Kn1 (Tkn1) Gene
Bold and wavy underline, respectively, mark the conserved homeodomain
and ELK that characterize all Kn1 class 1 genes. Note the
extensive homology in the 100 residues long presumptive acidic
region immediately upstream of the ELK domain. Within this acidic
domain most hydrophobic positions are also conserved. The
N-terminal one third of the genes is the most variable, but reveals
common features as well. It is extremely histidine-rich in maize and
soybean, and less so in tomato. It is also dominated by hydrophilic
residues: asparagine repeats in Arabidopsis, excess serine and asparagine
in soybean, and a very high proportion of glycine in tomato.
Identical residues are shaded, and a more detailed analysis of homeodomains
of Kn1 genes is provided in Kerstetter et al. (1994) and
Figure 6. TKn1 Expression in Leaf and Floral Primordia of Tomato:Ma et al. (1994).
In Situ Hybridization of DIG-Labeled Antisense Probes
(A) Anantha floral meristems. (Longitudinal sections). Apical cells
where Pts and La, respectively, reside. Mapping withthe (AC) and provascular bundles (PV) are labeled.
homeodomain alonegave identical results. No dominant (B) Shoot apex. A longitudinal section of a floral bud (FL) is shown
to the left, and the next sympodial apex (AP) to the right. Stars onleaf mutants are known to be linked to chromosome
the right mark tangential section through lateral leaflets (LL) of theIV, but at least six recessive mutations that alter leaf
emerging compound leaf. In the floral bud, the floral meristem (FM)
development are located on this chromosome (Stevens
and the vascular bundles (VB) are heavily labeled. TKn1 RNA is
and Rick, 1986). The nearest gene to TKn1 is entire foundalso in the parenchyma cells of the cortex (CT).In the emerging
(Figure 3B), butTKn1 cDNA clones originated from entire sympodial apex (AP), the apical cells and provascular derivatives
plants were found to be identical at DNA sequence level are marked and the growing pointsand provascular strands of newly
emerged lateral leaflets are also labeled.to the wild-type gene.
(C) Cross-sections of leaf primordia. TKn1 transcripts are found inAs shown in Figure 4B, TKn1 transcripts were not
the lateral tips that will form the lamina (arrowheads) and in the
observed in samples isolated from 3 cm and 5 cm long
provascular tissue (PV).
leaves (lanes 1 and 2), but were found in 0.5 cm Complete cDNA and a 477 bp BamHI–HindIII fragment from the 5Ј
long leaves (lane 3). TKn1 mRNA is abundant in 0.5 cm end of the gene give identical distribution of signals.
long shoot apices that carry leaf and floral primordia (D) Expression of the dUTPase gene in leaf primordia and shoot
apex. AC, apical cells; CT, cortex; FM, floral meristem; FP, floral(lane 4), and more so in the upper part of the stems
primordia; GE, growing ends of leaves; IM, inflorescence meristem;when stripped of these appendages (lanes 5 and 6). Low
LT, lateral leaflets; PV, provascular strands; S, sepals; SA, shoot
levels of mRNA are found in mature flowers (lane 7), due
apex; VB, vascular bundles.
probably to the floral pedicles and carpels (lane 11),
and in arrested floral meristems of the anantha mutant
inflorescences (lane 12). TKn1 is also expressed at the of the apical meristem AC and in the provascular (PV)
strands. The sympodial shoot of tomato is composedwild-type level in shoot apices of La/ϩ plants (data not
shown). of reiterated units of three leaves and a terminal inflorescence.
In Figure 6B, a longitudinal section of a floralSince Kn1-related genes are not expressed in leaf and
floral primordia of maize or Arabidopsis (Lincoln et al., bud is shown to the left, and a tangential section cutting
through a series of lateral leaflet primordia of a leaf1994; Kerstetter et al., 1994), in situ hybridization was
used to localize more precisely the TKn1 transcripts in on the right. Evidently, leaflet primordia (stars) and the
meristematic zone of the next sympodial apex arethese organs in tomato plants (Figure 6). In the floral
meristems of anantha mutant plants (Figure 6A), which stained. Staining is strong in the meristematic (FM) region
of the future inner three whorls of the floral bud,are arrested in the preorganogenesis stage (Helm, 1951;
Pri-Hadash et al., 1992), TKn1 is expressed in all layers but it is very weak in the emerging sepals (S). TKn1
The Making of a Compound Leaf
741
transcripts were detected in the newly emerged lateral
primordia (LT) inthe floral bud and their vascularbundles
(VB), and in the cortex parenchyma (CT) of the floral
pedicle. A more accurate picture of the localization of
TKn1 transcripts in leaf primordia is obtained from the
cross sections shown in Figure 6C, where provascular
strands and lateral growing tips (arrowheads) are labeled.
The internal growing tips give rise to the lamina,
upon primordium expansion. This pattern is practically
identical to that of the dUTPase gene in the very same
organs (Figure 6D; Pri-Hadash et al., 1992).
The Weak Meristem- and Vascular-Specific
dUTPase Promoter Is Sufficient to Elicit
the Kn1 Syndrome in Tomato
The ramification of leaves of transgenic 35S::Kn1 tomato
plants is apparent very early in the development of the
leaf primordium (see Figure 2). We have used the recently
isolated promoter of the tomato dUTPase gene
to examine the role of the meristematic and provascular
cells of the leaf primordium in determining the subdivi- Figure 7. The Vascular-Specific dUTPase Promoter Is Sufficient to
sion of the compound leaves. Similar to the TKn1 gene, Elicit Leaf Ramification.
the dUTPase gene functions predominantly in apical (A) A compound leaf of tomato transgenic plant expressing the
PdUTPase::Kn1 transgene. Leaves of such transgenic plants aremeristems of vegetative and floral organs, as well as in
distinguished from 35S::Kn1 plants by the longer petioles of theirprovascular cells with meristematic potential (Figure 6;
lateral leaflets and by the prominence of the terminal leaflet in each
Pri-Hadash et al., 1992). It is down-regulated in the pacompound
unit.
renchyma derivatives and other differentiated tissues, (B) Wild-type (left) and highly lobed leaves of transgenic tobacco
and its expression inmature leaves (5 cm long) or mature plant expressing the PdUTPase::Kn1 transgene. This phenotype is
flowers is negligible. A 380 bp proximal sequence of the similar to that reported for 35S::Kn tobacco plants (Sinha et al.,
1993).putative 5Ј regulatory region of the dUTPase gene was
shown to drive the expression of the ␤-glucuronidase
reporter gene in the above tissue domains (O. Cohen,
unpublished data) and was used in the experiments re- To find out whether the effect of the PdUTPase::Kn1
transgene is species specificor, likethat of the35S::Kn1,ported here.
We generated 31 kanamycin-resistant primary trans- depends on the developmental status of the leaf, we
have introduced it into tobacco plants as well. Indeed,formants expressing Kn1 under the control of the dUTPase
promoter (designated B series), and leaf ramifica- as shown in Figure 7B, we obtained transformed kanamycin-resistant
plants, with modified leaves (7 out oftion, as illustrated in Figure 7A, was observed in 21
plants. An important feature of this series of transgenic 12), similar to those reported in 35S-driven expression
in transgenic tobacco (Sinha et al., 1993).plants is that the extent of leaf ramification varied from
one transformant to another;this was not evident among The overall 50-fold difference in the expression of the
35S::Kn1 and dUTPase::Kn1 transgenes is not reflectedthe M series plants. Leaves of transgenic plant B1, for
example, are subdivided only once more and, in this in the severity of the phenotypes in tobacco or in the
degree of subdivision in tomato. It is the expression inrespect, precisely mimic the arrangement of leaves in
plants bearing the dominant Pts gene shown in Figure particular cells at particular times that matters most.
3A. One additional order of subdivision is exhibited by
plant B100 and more extreme ramifications by other B Discussion
series plants. These results, along with those obtained
with the trifoliate::Kn1 plants, illustrate our ability to ma- A General Scheme for Leaf Morphogenesis
Fortuitously, leaves of all species in which misexpres-nipulate at will the architecture of the compound leaf.
The expression of the PdUTPase:Kn1 transgene in sion of Kn1 was previously examined (i.e., maize, tobacco,
rice, and Arabidopsis) were simple. We havetransformed tomato plants is compared with that of the
35S::Kn1 transgene in Figure 4A (lanes 5–8). Kn1 tran- shown here that misexpression of Kn1 affects compound
leaves of tomato in a very different way than itscripts are hardly detected in the B1 plant with the very
weak phenotypic response, and they are also rare, in affects simple leaves and that this basic observation
reflects inherent fundamental differences in the devel-comparison with the 35S::Kn1 transformants, in other B
series plants that manifest full leaf ramification. Misex- opment of simple and compound leaves. This hypothesis
implies that simple and compound leaves are deter-pression of Kn1 in the dUTPase territories is sufficient,
therefore, to induce the full potential of leaf ramification. mined by two different developmental programs and
that the gene systems that condition them are con-It does not imply, by any means, that expression of Kn1
in mesophyll or epidermal cells will not result in altered served among species with simple and compound
leaves, respectively.morphogenesis.
Cell
742
An attempt to develop an experimental framework for extreme, allowing only one pair of lateral leaflet meristems.
In the less extreme potato leaf, however, wherethe genetic dissection of leaf arrangement requires the
formulation of a likely developmental scenario, a judi- two pairs of leaflets are formed, the blades are entire
rather than dentate, suggesting that the two genes affectcious classification of the many genes involved, and
their isolation and characterization. Formally, the forma- meristematic functions in different ways. The lesions in
both cases, we surmise, favor early lamina expansiontion of the prototype compound unit of tomato entails
the establishment, in a regular basipetal pattern, of pairs and, consequently, also limit to a certain extent the ability
to subdivide in response to Kn1.of lateral meristems along the peg-like structure of the
primary dorsiventral primordium, and the concomitant The Pts dominant mutation, most likely an overproducer
allele, shifts the balance in the opposite direction;inhibition of lamina expansion. Additional orders of subdivision
require the reiterated formation of lateral meri- lamina growth is delayed for one additional cycle with
no associated alteration in the complexity of the phyl-stems on the secondary primordia before lamina expansion
ensues. We speculate that the interplay between lomer. By the same token, the dominant effect of Kn1
results in extra ramification with no increase in the num-apical growth, formation of lateral meristems, and activation
of the diffused leaf meristems that condition the ber of lateral meristems in each unit, and Kn1 fails to
rescuethe compound prototype of potato-leaf, trifoliate,expansion of the lamina determine whether a leaf will
be compound or simple. As long as the primordium apex and probably entire as well. Thus, Kn1 can only ramify
a preexisting plan as it allows the proliferation of activecontinues to “grow,” emanating signals that induce the
formation of lateral meristems, and a coupled system cell type during a given developmental window. In so
doing, Kn1 dismisses the fine-tuning conferred by thedelays the activation of the diffused meristem, the leaf
will be compound. Maintaining this balance among the wild-type Pts system, and it is not unrealistic to suggest
that the endogenous TKn1 gene regulates the activitydevelopmental programs allows for more subdivisions
to ensue. Presumably, an additional, species-specific of Pts in leaf primordia.
La is epistatic to all known leaf shape mutations (E.developmental plan dictates the developmental window
in which “active” cell types, i.e., apical, provascular, and Lifschitz, unpublished data), and in addition, arrest meristematic
activity in all aerial meristems. La/La embryoslateral lamina, are allowed to proliferate.
The inherent inability of simple leaves to respond produce root meristems, but fail to develop shoot apical
meristems (Mathan and Jenkins, 1962; Caruso, 1968;properly to Kn1 overexpression is reflected by ectopic
formation of sheath tissues and “knots” in blade territor- Dengler, 1984); the shoot apical meristems are “normal”
in La/ϩ plants. La/ϩ fruits are the size of small cherryies in maize (Sinha and Hake, 1994) and in irregular
expansion of the lamina and the ectopic formation of tomatoes, and meristematic activity in leaves is differentially
reduced, preventing the formation of lateral meri-shoots on tobacco and Arabidopsis leaves (Sinha et al.,
1993; Lincoln et al., 1994). The developmental require- stems and dentate margins, but not of the appearance
of the primary leaf primordium itself. From the dosements for compoundness cannot be satisfied under the
restrictions imposed by the “simple” program, and the effect studies of Stettler (1965) in tetraploid plants, we
infer that La is an antimorph or a neomorph, rather thanKn1 gene product that enhances meristematic activity
in leaves affects only secondary growth parameters, a haploinsufficient or an overproducer. La/ϩ leaves are
morphogenetically simple and thus similar to Arabi-resulting in a variety of malformations. The reiterated
ramification of the compound prototype unit of the to- dopsis and tobacco leaves. They remain simple following
misexpression of Kn1, albeit displaying the othermato leaf is, according to this view, due to developmental
programs that distinguish compound from sim- aspects of the 35S::Kn1 phenotype, just like leaves of
Arabidopsis and tobacco. Pts and Kn1 fail to rescue theple leaves.
compound prototype or to induce ramification of La/ϩ
leaves. The response of simple leaves of tomato (i.e.,
Genetic Evidence for the Developmental La), Arabidopsis, and tobacco to Kn1 is thus species
Program Controlling Compoundness independent, which suggests that theyare developmenAll
aerial meristems have a common onthogeneticorigin. tally, and not merely morphogenetically, in the same
Consequently, mutations in basic functions of meri- state.
stems, as amply illustrated by Kn1, are expected to have We suggest expanding the analogy between La leaves
multiple pleiotropic effects. The actual most prominent and simple leaves to sepals as well. Sepals of tomato,
manifestation of such mutations will depend on local tobacco, and Arabidopsis are also simple and entire,
developmental programs. In accord with the proposed and similar to simple leaves, sepals do not respond to
developmental scenario, we suggest that the recessive Kn1 misexpression by any elaboration of their form, but
potato-leaf and trifoliate mutations, as well as the domi- their venation pattern is distorted and the areole size
nant Pts and Kn1 alleles, modify preferentially the bal- is increased. In contrast with sepals, the potential for
ance between lamina expansion and lateral leaflet ramification is silent but otherwise intact in tomato juvemeristems.
The La dominant mutation subdues, prefer- nile leaves. Juvenile leaves of tomato normallyhave only
entially, meristematic activities without which thepoten- one to three leaflets, but they ramify to the same extent
tial for compoundness, and thus ramification, cannot be proportionally, as adult leaves do upon ectopic expresmaterialized.
sion of Kn1.
potato-leaf and trifoliate reduce the complexity of the The phenotypic manifestation of Kn1 misexpression
compound phyllomer but are not involved directly in can thus be used to verify the inherent nature of leaf
architecture. It is possible, for example, that in somethe plan that permits compoundness. trifoliate is more
The Making of a Compound Leaf
743
plant species, compound leaves are actually modified (Hareven et al., 1994). While this by itself does not constitute
evidence that expression of TKn1 in the leaf primor-simple leaves (i.e., Kaplan, 1983). In this context, it will
be interesting to see the fate of tendrils and leaflets in dium is required for the formation of the compound
structure or for its ramification, the induction of addi-pea, orto find out whether the basicternate arrangement
in clover leaves will be converted to multiples of three tional subdivisions by the PdUTPase-drivenKn1 expression
in the same domains where TKn1 is expressedfollowing misexpression of the Kn1 gene.
supports such a premise.
While the overall scenario suggested here for the forThe
Possible Role of Apical and Provascular Cells mation of compound leaves is open to question, there
of the Primary Dorsiventral Primordium is no doubt that leaf morphogenesis is presently open
in the Formation and Reiteration to molecular and genetic dissection. Since no other orof
the Compound Prototype gan of living creatures exploits analogous sets of genes
The difference between tomato and the species with to acquire such a developmental flexibility and to genersimple
leaves is that the Kn1-induced meristematic ac- ate such a range of diverse forms both between and
tivity in tomato is exploited very early during primordia within species, understanding the underlying mechadevelopment
for the ramification of the compound phyl- nisms will be of fundamental importance.
lomer. The knots in maize, and the malformations in the
leaves of transgenic 35S::Kn1 tobacco and Arabidopsis Experimental Procedures
plants, are not visibly detected until relatively late in leaf
Tomato Linesdevelopment (Sinha et al., 1993; Lincoln et al., 1994).
The following mutant lines were providedby Prof. C. M. Rick (Univer-To understand the formation of compound leaves in
sity of California, Davis): anantha (an) LA536; entire (e) LA2922;
tomato, we will have to identify the target cells for the
Lanceolate (La) LA335; potato-leaf (c) LA3211; self-pruning (sp)
Kn1-induced ramification. LA154; Petroselinum (Pts) LA2532; trifoliate2 (tf2) LA579. VF36
The ectopic expression of the dominant mutation Kn1 seeds were the gift of S. McCormick, PGEC, Albany. Indeterminate
in maize is restricted to vein tissues (Smith et al., 1992), line 93–137 (Sp/Sp) was provided by D. Zamir (Rehovot).
but the 35S-driven expression in tobacco, Arabidopsis,
Cytological Proceduresand tomato is ubiquitous. We do not, therefore, know
In situ hybridization using digoxygenin-labeled RNA probes werewhat the phenotypic consequence of Kn1 misexpresaccording
to J. H. Doonan and E. Coen (personal communication)
sion outside the vasculature will be, or, expression in
and Jackson (1991), with the following minor modifications: buffer
which tissue is responsible for the variety of pleiotropic 1 was 100 mM maleic acid 150 mM NaCl (pH 7.5), as recommended
effects seen in the different species. by Boehringer. Dilution of antidigoxygenin-AP of Boehringer was
1:1000 and incubation time with the antibody was 1.5 h.Expression of Kn1 driven by the meristem and provasObservations
of venation patterns were conducted on clearedcular-specific PdUTPasepromoter was shown to induce
leaves following 5 min boiling in 85% ethanol and 3 h in lactic acidthe full potential of leaf ramification in tomato (Figure 7)
at 70ЊC. Preparation of tissues for SEM were performed according to
and the loss of apical dominance as well. While it is
a protocolprovidedby R. Meeks-Wagner (personal communication).
possible to relate the formation of laterals to differential
localization of hypothetical KNOTTED signals along the Nucleic Acids Procedures
The maize Kn1 cDNA clone (Volbrecht et al., 1991) in the pBIN19provascular strands of the dorsiventral leaf primordia,
vector (Bevan, 1984) was provided by S. Hake. The Kn1 gene drivenit is impossible to exclude a role for the primordium
by the 380 bp PdUTPase promoter (O. Coen, unpublished data) wasapical cells. Apical cells were not expected to respond
also cloned into the same binary vector, and transformations of both
to misexpression of Kn1, since the Kn1 gene was shown
were performed via Agrobacterium tumefasciens strain LBA4404 in
to express normally in apices of shoots and inflores- the RK9/8 line (Pnueli et al., 1994b) according to Horsch et al. (1985)
cences, and its overexpression there had no obvious or in VF36 according to McCormick (1991).
The screen for the TKn1 gene was conducted with the maize Kn1-morphogenetic consequences in tobacco or Arabilabeled
cDNA under relaxed hybridization conditions: 15% for-dopsis (Smith et al., 1992; Jackson et al., 1994; Sinha
mamide, 5ϫ SSC, 5ϫ Denhardt’s solution and 1% SDS in 42ЊC.et al., 1993; Lincoln et al., 1994).
Filters were washed with 3ϫ SSC, 1% SDS in 37ЊC and exposed
That apical cells of the dorsiventral leaf primordium
for 48 hr. The shoot cDNA library representing1.2 ϫ 106
independent
are required for the initiation of the compound leaf is inserts was constructed in the ␭ZAPII vector (Stratagene) from the
supported by the expression pattern of TKn1 in the plant second and third sympodial shoots of the indeterminate line 93–137.
The 5 mm long apices contain the apex, three to four leaf primordia,meristems (Figure 6). Expression of all Kn1-subclass
and one to three floralprimordia in whichonly sepals emerged. Othergenes in Arabidopsis and maize is clearly excluded from
nucleic acid procedures followed published protocols (Ausubel etleaf and floral primordia (Lincoln et al., 1994; Kerstetter
al., 1988).
et al., 1994). The observation that tomato and Arabidopsis
display different expression patterns of homolo-
Acknowledgments
gous regulatory genes in their respective meristems is
not surprising, however, and conceptually similar to We thank R. Meeks-Wagner, S. McCormick, T. Zachs, B. Horowitz,
and G. Eitan for their valuable discussions and for a critical readingwhat was described for the homeotic genes in the aniof
the manuscript. We are particularly grateful to S. Hake for provid-mal kingdom (see Carroll, 1995). The two dicot plant
ing the maize Kn1 gene and for her many helpful comments and tospecies must employ homologous regulatory genes in
D. Zamir for providing all the expertise that was essential for the
order to execute contrasting meristematic programs:
mapping experiments. M. Egea Cortines helped us during the initial
monopodial shoots, indeterminate inflorescences, and characterization of the TKn1 gene, and O. Coen kindly provided us
simple leaves in Arabidopsis; sympodial shoots, deter- with the characterized 380 bp PdUTPase promoter. This work was
supported by research grants to E. L. from the United States-Israelminate inflorescences, and compound leaves in tomato
Cell
744
Binational Agricultural Research and Development Fund, the Ger- in the vegetative meristem and dramatically alters leaf morphology
when overexpressed in transgenic plants. Plant Cell 6, 1859–1876.man-Israel Biotechnology Projects, and the Israel Academy of Science.
This project is conducted under the auspices of the Israeli Ma, H., McMullen, M.D., and Finer, J.J. (1994). Identification of a
Plant Genome Center. homeobox-containing gene with enhanced expression during soybean
(Glycine max L.) somatic embryo development. Plant Mol. Biol.
Received January 2, 1996; revised January 15, 1996. 24, 465–473.
Marks, G.A. (1987). A suit of mutants that modify pattern formation
References in pea leaves. Plant Mol. Biol. Reporter 5, 311–335.
Mathan, D.S., and Jenkins, J.A. (1962). A morphogenetic study ofAusubel, F.M., Brent, R., Kingston, R.E., Moor, D.D., Seidman, J.G.,
leaf shape of lanceolate, a leaf-shape mutant in the tomato. Am. J.Smith, J.A., and Struhl, K., eds. (1988). Current Protocols in MolecuBot.
49, 504–514.lar Biology (New York: Wiley).
Matsuoka, M., Ichikawa, H., Saito, A., Tada, Y., Fujimura, T., andBecraft, P.W., and Freeling, M. (1994). Genetic analysis of Rough
Kano-Murakami, Y. (1993). Expression of a rice homeobox genesheath1 developmental mutants of maize. Genetics 136, 295–311.
causes altered morphology of transgenic plants. Plant Cell 5, 1039–
Benfey, P.N., and Chua, N.-H. (1990). The cauliflower mosaic virus
1048.
35S promoter: combinatorial regulation of transcription in plants.
McCormick,S. (1991). Transformation of tomato with AgrobacteriumScience 250, 959–966.
tumefaciens. Plant Tissue Culture Manual B6, 1–9.
Bevan, M. (1984). Binary Agrobacterium vectors for plant transforPnueli,
L., Hareven, D., Broday, L., Hurwitz, C., and Lifschitz, E.
mation. Nucl. Acids Res. 12, 8711–8721.
(1994). The TM5 MADS box gene mediates organ differentiation in
Carroll, S.B. (1995).Homeotic genes and theevolution of anthropods the three inner whorls of tomato flowers. Plant Cell 6, 175–186.
and chordates. Nature 376, 479–485.
Poethig, R.S. (1987). Clonal analysis of cell lineage patterns in plant
Caruso, J.L. (1968). Morphogenetic aspects of a leafless mutant in development. Am. J. Bot. 74, 581–594.
tomato. I. General patterns of development. Am. J. Bot. 55, 1169–
Poethig, S., and Sussex, I.M. (1985). The cellular parameters of leaf
1176.
development in tobacco: a clonal analysis. Planta 165, 170–184.
Cutter, E.G. (1971). The leaf. In Plant Anatomy: Experiment and
Pri-Hadash, A., Hareven, D., and Lifschitz, E. (1992). A meristemInterpretation:
Organs (London: Edward Arnold Publishers, Limited),
related gene from tomato encodes a dUTPase: Analysis of exprespp.
117-196.
sion in vegetative and floral meristems. Plant Cell 4, 149–159.
Dengler, N.G. (1984). Comparison of leaf development in normal
Sinha, N., and Hake, S. (1994). The Knotted1 leaf blade is a mosaic
(ϩ/ϩ), entire (e/e), and lanceolate (La/ϩ) plants of tomato, Lycoperof
blade, sheath and auricle identities. Dev. Genet. 15, 401–414.
sicon esculentum “Ailsa Craig”. Bot. Gaz. 145, 66–77.
Sinha, N., Hake, S., and Freeling, M. (1993a). Genetic and molecular
Dolan, L., and Poethig, R.S. (1991). Genetic analysis of leaf developanalysis
of leaf development. Curr. Topics Dev. Biol. 28, 47–80.
ment in cotton. Development Suppl. 113, 39–46.
Sinha, N., Williams, R., and Hake, S. (1993b). Overexpression of the
Freeling, M. (1992). A conceptual framework for maize leaf developmaize
homeobox gene, Knotted1, causes a switch from determinate
ment. Dev. Biol. 53, 44–58.
to indeterminate cell fates. Genes Dev. 7, 787–795.
Freeling, M., and Hake, S. (1985). Developmental genetics of muSmith,
L.G., and Hake, S. (1992). The initiation and determination of
tants that specify Knotted leaves in maize. Genetics 111, 617–634.
leaves. Plant Cell 4, 1017–1027.
Hake, S. (1992). Unraveling the knots in plant development. Trends Smith, L.G., Greene, B., Veit, B., and Hake, S. (1992). A dominant
Genet. 8, 109–114. mutation in the maizehomeobox gene, Knotted-1, causes its ectopic
Hake, S., Volbrecht, E., and Freeling, M. (1989). Cloning Knotted, expression in leaf cells with altered fates. Development 116, 21–30.
the dominant morphological mutantin maize using Ds2 asa transpo- Steeves, T.A., and Sussex, I.M. (1989). Patterns in Plant Developson
tag. EMBO J. 8, 15–22. ment (Cambridge, England: Cambridge University Press).
Hareven, D., Gutfinger, T., Pnueli, L., Bauch, L., Cohen, O., and Stettler, R.F. (1964). Dosageeffects of the lanceolate gene in tomato.
Lifschitz, E. (1994). The floral system of tomato. Euphytica 79, Am. J. Bot. 51, 253–264.
235–243.
Stevens, A.M., and Rick, C.M. (1986). Genetics and breeding. In The
Helm, J. (1951). Vergleichende Betrachtungen u¨ ber die Entwicklung Tomato Crop, J.G. Atherton and J. Rudick, eds. New York: Chapman
der infloreszenz bei Lycopersicon esculentum Mill und bei einer and Hall).
Ro¨ ntgemutante. Zuchter 21, 89–95.
Sussex, I.M. (1989). Developmental programming of the shoot meriHorsch,
R.B., Fry, J.E., Hoffmann, N.L., Eichholtz, D., Rogers, S.G., stem. Cell 56, 225–229.
and Fraley, R.T. (1985). A simple and general method for transfering
Volbrecht, E., Veit, B., Sinha, N., and Hake, S. (1991). The developgenes
into plants. Science 227, 1229–1231.
mental gene Knotted-1 is a member of a maize homeobox gene
Jackson, D. (1991). In situ hybridization in plants. In Molecular Plant family. Nature 350, 241–243.
Pathology: A Practical Approach, D.J. Bowles, S.J. Gurr, and M.
McPherson, eds. (Oxford: IRL Press). GenBank Accession Number
Jackson, D., Veit, B., and Hake, S. (1994). Expression of maize KnotThe
accession number for the sequence reported in this paper isted1-related homeobox genes in the shoot apical meristem predicts
U32247.patterns of morphogenesis in the vegetative shoot. Development
120, 405–413.
Kaplan, D. (1973). Comparative developmental analysis of the heteroblastic
leaf series of axillary shoots of Acorus calamus L. (Araceae).
Cellule 69, 251–290.
Kaplan, D. (1983). The development of palm leaves. Sci. Am. 249,
98–105.
Kerstetter, R., Volbrecht, E., Lowe, B., Veit, B., Yamaguchi, J., and
Hake, S. (1994). Sequence analysis and expression patterns divide
the maize Knotted1-like homeobox genes into two classes. Plant
Cell 6, 1877–1887.
Lincoln, C., Long, J., Yamaguchi, J., Serlkawa, K., and Hake, S.
(1994). A Knotted-like homeobox gene in Arabidopsis is expressed