Parallel evolution of TCP and B-class genes in Commelinaceae flower bilateral symmetry.

Abstract

Flower bilateral symmetry (zygomorphy) has evolved multiple times independently across angiosperms and is correlated with increased pollinator specialization and speciation rates. Functional and expression analyses in distantly related core eudicots and monocots implicate independent recruitment of class II TCP genes in the evolution of flower bilateral symmetry. Furthermore, available evidence suggests that monocot flower bilateral symmetry might also have evolved through changes in B-class homeotic MADS-box gene function.
In order to test the non-exclusive hypotheses that changes in TCP and B-class gene developmental function underlie flower symmetry evolution in the monocot family Commelinaceae, we compared expression patterns of teosinte branched1 (TB1)-like, DEFICIENS (DEF)-like, and GLOBOSA (GLO)-like genes in morphologically distinct bilaterally symmetrical flowers of Commelina communis and Commelina dianthifolia, and radially symmetrical flowers of Tradescantia pallida.
Expression data demonstrate that TB1-like genes are asymmetrically expressed in tepals of bilaterally symmetrical Commelina, but not radially symmetrical Tradescantia, flowers. Furthermore, DEF-like genes are expressed in showy inner tepals, staminodes and stamens of all three species, but not in the distinct outer tepal-like ventral inner tepals of C. communis.
Together with other studies, these data suggest parallel recruitment of TB1-like genes in the independent evolution of flower bilateral symmetry at early stages of Commelina flower development, and the later stage homeotic transformation of C. communis inner tepals into outer tepals through the loss of DEF-like gene expression.

Full text

Page 1

RESEARCHOpen Access
Parallel evolution of TCP and B-class genes in
Commelinaceae flower bilateral symmetry
Jill C Preston*and Lena C Hileman
Abstract
Background: Flower bilateral symmetry (zygomorphy) has evolved multiple times independently across
angiosperms and is correlated with increased pollinator specialization and speciation rates. Functional and
expression analyses in distantly related core eudicots and monocots implicate independent recruitment of class II
TCP genes in the evolution of flower bilateral symmetry. Furthermore, available evidence suggests that monocot
flower bilateral symmetry might also have evolved through changes in B-class homeotic MADS-box gene function.
Methods: In order to test the non-exclusive hypotheses that changes in TCP and B-class gene developmental
function underlie flower symmetry evolution in the monocot family Commelinaceae, we compared expression
patterns of teosinte branched1 (TB1)-like, DEFICIENS (DEF)-like, and GLOBOSA (GLO)-like genes in morphologically
distinct bilaterally symmetrical flowers of Commelina communis and Commelina dianthifolia, and radially
symmetrical flowers of Tradescantia pallida.
Results: Expression data demonstrate that TB1-like genes are asymmetrically expressed in tepals of bilaterally
symmetrical Commelina, but not radially symmetrical Tradescantia, flowers. Furthermore, DEF-like genes are
expressed in showy inner tepals, staminodes and stamens of all three species, but not in the distinct outer tepal-
like ventral inner tepals of C. communis.
Conclusions: Together with other studies, these data suggest parallel recruitment of TB1-like genes in the
independent evolution of flower bilateral symmetry at early stages of Commelina flower development, and the
later stage homeotic transformation of C. communis inner tepals into outer tepals through the loss of DEF-like gene
expression.
Keywords: B class genes, Commelinaceae, CYCLOIDEA, homeotic change, monocots, tepals, teosinte branched1
Background
Evolutionary transitions between flower radial symmetry
(polysymmetry and actinomorphy) and bilateral symmetry
(monosymmetry and zygomorphy) have occurred multiple
times independently across angiosperms, and are asso-
ciated with increased pollinator specialization and specia-
tion rates [1-6]. Indeed, some of the largest angiosperm
families have species with predominantly bilaterally sym-
metrical flowers, including the legumes (Leguminosae,
rosids, core eudicots), daisies (Asteraceae, asterids, core
eudicots) and orchids (Orchidaceae, monocots) [7]. Recent
functional studies in distantly related core eudicots -
including Antirrhinum majus (Plantaginaceae), Iberis
amara (Brassicaceae), Pisum sativum and Lotus japonicus
(Leguminosae) and Gerbera hybrida (Asteraceae) - have
demonstrated a role for class II TEOSINTE BRANCHED1
(TB1)/CYCLOIDEA (CYC)/PROLIFERATING CELL
NUCLEAR ANTIGEN GENE-CONTROLLING ELE-
MENT BINDING FACTOR (PCF) (TCP) transcription
factors in establishing flower symmetry by specifying iden-
tity to the dorsal (adaxial) region of the flower [8-13].
Since bilateral symmetry has evolved independently in
these lineages, genetic data suggest parallel recruitment of
class II TCP genes in the evolution of a convergent trait
[11,14-16]. Whether homologous TCP genes have been
similarly utilized in monocot flower bilateral symmetry
remains largely untested [but see [17,18]].
The genetic basis of flower bilateral symmetry is best
understood in the model species A. majus, and involves
* Correspondence: jcpxt8@ku.edu
Department of Ecology and Evolutionary Biology, University of Kansas,1200
Sunnyside Avenue, Lawrence, KS 66045, USA
Preston and Hileman EvoDevo 2012, 3:6
http://www.evodevojournal.com/content/3/1/6
© 2012 Preston and Hileman; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative
Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and
reproduction in any medium, provided the original work is properly cited.

Page 2

the action of four asymmetrically expressed transcription
factors [8,9,19-21]. Dorsal identity is specified by the
class II TCP genes CYCLOIDEA (CYC) and DICHOT-
OMA (DICH), and the MYB gene RADIALIS (RAD),
whereas ventral (abaxial) identity is conferred by the
MYB gene DICHOTOMA (DICH). CYC and DICH are
derived from a recent duplication event at the base of the
Antirrhineae tribe [22], and have distinct but overlapping
functions, as inferred from their mutant phenotypes
[8,9]. Whereas wild type A. majus plants have five petals,
four stamens and a dorsal staminode, cyc single mutants
often have extra dorsal petals that are reduced in size,
and a fully developed dorsal stamen. By contrast, dich
single mutants only lack the internal asymmetry of wild
type dorsal petals. Together with the fully ventralized,
and, hence, radially symmetrical flowers of cyc:dich dou-
ble mutants, these data demonstrate a role for CYC in
dorsal stamen abortion, petal growth and organ number
determination, and a role for DICH in shaping dorsal
petal growth [8,9].
A similar range of dorsal identity functions has been
assigned to CYC-like genes of other core eudicots, includ-
ing Linaria vulgaris, P. sativum, I. amara, Lotus japonicus,
and G. hybrida [10-13,20,23]. Furthermore, although
monocots do not have a strict CYC gene ortholog due to
two separate duplication events at the base of core eudi-
cots [24], the observation that the class II TCP gene
RETARDED PALEA1 (REP1) in Oryza sativa (rice, Poa-
ceae) is expressed only in the dorsally positioned palea,
suggests further recruitment of TCP genes in the evolution
of monocot flower bilateral symmetry [17]. Indeed, it was
recently found that bilaterally symmetrical flowers of Cos-
tus spicatus (Costaceae; Zingiberales) and Heliconia stricta
(Heliconiaceae) have asymmetric TCP gene (CsTB1a and
HsTBL2b, respectively) expression at early to late stages of
flower development [18]. However, in contrast to the core
eudicots, CsTB1a and HsTBL2b expression is restricted to
the ventral, rather than the dorsolateral, side of the flower
[18].
In addition to a hypothesized role for TCP genes, the
floral homeotic B-class genes DEFICIENS (DEF) and GLO-
BOSA (GLO) have been implicated in the multiple inde-
pendent derivations of flower bilateral symmetry in
monocots [25-29]. Evidence supporting this comes largely
from the observation that DEF-like gene paralogs are dif-
ferentially expressed in the morphologically distinct outer
tepals, dorsolateral inner tepals and ventral inner tepal (lip
or labellum) of Phalaenopsis equestris (Orchidaceae)
[25,26]. In both core eudicots and monocots, DEF- and
GLO-like proteins function as obligate heterodimers to
confer identity to the second and third whorls, although
there are exceptions [30,31]. In the second whorl of Arabi-
dopsis thaliana (Brassicaceae) and A. majus APETALA3
(AP3)/DEF and PISTILLATA (PI)/GLO form tetramers
with SEPALLATA (SEP) and APETALA1 (AP1)/SQUA-
MOSA (SQUA) MADS-box proteins, resulting in the acti-
vation of downstream genes that confer petal identity
[32-36]. Thus, since DEF/GLO-, and to a lesser extent
SEP- and SQUA-like, gene function is largely conserved
across angiosperms [37-45], shifts between radial and
bilateral perianth symmetry could be explained by changes
in expression of these floral homeotic genes, resulting in
the loss, gain or modification of second whorl identity.
The monocot order Commelinales contains species with
both radially and bilaterally symmetrical flowers, is sister
to the Zingiberales and comprises five families: Commeli-
naceae (for example, Commelina and Tradescantia),
Hanguanaceae (Hanguana only), Philydraceae (for exam-
ple, Philydrella and Helmholtzia), Haemodoraceae, and
Pontederiaceae [46,47] (Figure 1). Phylogenetic analyses
suggest that showy, insect-pollinated and bilaterally sym-
metrical flowers are plesiomorphic to the Commelinales/
Zingiberales clade [47]. However, since Hanguana and
Cartonema (Commelinaceae) are both progressively sister
to the two major Commelinaceae clades, and have radially
symmetrical flowers, it has been suggested that radial
flower symmetry is ancestral to Commelinaceae [29,48].
Based on the prevalence of radial flower symmetry in its
sister families - including Joinvilleaceae and Flagellariaceae
- an ancestral state of bilateral flower symmetry has simi-
larly been invoked for Poaceae [48-50]. Thus, flower bilat-
eral symmetry probably evolved independently in the
Commelinaceae (Figure 1), Philydraceae, Zingiberales and
Poaceae, as well as in other monocots outside the comme-
linid clade (for example, Orchidaceae) [29]. A critical
question in evolutionary developmental biology is whether
similar changes at the level of gene networks, genes and/
or gene regions underlie these convergent shifts in floral
form.
Flower bilateral symmetry in the genus Commelina
involves differential organ development in all of the whorls
(Figures 1 and 2) [51-55]. In C. communis and C. dianthi-
folia, each hermaphroditic flower comprises two whorls of
three tepals, two whorls of three stamens/staminodes and
a whorl of three fused carpels (Figure 2A, B) [54]. In the
first whorl of both species, all three outer tepals are mem-
branous, but the lateroventral two are distinct from the
dorsal one in both size and shape (Figure 2G). In the sec-
ond whorl of C. dianthifolia all inner tepals are large and
showy, but vary slightly in overall size and shape along the
dorsoventral axis (Figure 2E, H). By contrast, in the second
whorl of C. communis, only the dorsolateral inner tepals
are large and showy; the ventral inner tepal is small and
membranous, similar to outer tepals (Figure 2D, G) [54].
In the stamen whorls of both species, the dorsal (outer
whorl) or dorsolateral (inner whorl) organs are underdeve-
loped staminodes that function to attract pollinators, but
produce few sterile pollen grains (Figure 2D, G, H). By
Preston and Hileman EvoDevo 2012, 3:6
http://www.evodevojournal.com/content/3/1/6
Page 2 of 14

Page 3

contrast, the long lateroventral stamens in the outer whorl,
and the medium ventral stamen in the inner whorl are
showy and produce viable pollen (Figure 2D, G, H).
Finally, in the fifth whorl of both species, the dorsal carpel
is underdeveloped and sterile, while the two ventral car-
pels are fertile [54].
Based on detailed analyses of flower development, and
the known role of class II TCP genes in core eudicots,
Hardy et al. (2009) recently hypothesized a role for TB1-
like TCP genes in establishing bilateral symmetry in early
developmental stages of Commelina flower development
[54]. Furthermore, we propose a second, non-exclusive
hypothesis, that dorsal-specific expression of DEF/GLO-
like genes defines the dorsoventral axis of symmetry in
mid- to late-stage C. communis corollas. In this study, we
use gene expression and micromorphological data to test
three predictions of these hypotheses that: (1) TB1-like
genes are asymmetrically expressed in bilaterally symme-
trical C. communis and C. dianthifolia, but not radially
symmetrical Tradescantia (formerly Setcreasea) pallida,
Inner
Tepals Stamens
Helmholtzia glaberrima
Philydrella pygmaea
Commelina lukei
Commelina benghalensis
Commelina diffusa
Commelina communis
Commelina welwitschii
Commelina virginica
Commelina dianthifolia
Commelina africana
Murdannia simplex
Murdannia nudiflora
Tradescantia spathacea
Tradescantia zebrina
Tradescantia pallida
Tradescantia occidentalis
Tradescantia sillamontana
Spatholirion longifolium
Pollia secundiflora
Dischorisandra thrysiflora
Tripogandra serrulata
71
100
98
91
94
91
100
91
96
89
Figure 1 Evolution of inner tepal and stamen symmetry in Commelinaceae. Maximum likelihood (ML) trnL phylogeny reconstruction with
support values for 500 ML bootstrap replicates based on Burns et al. (2011), using Philydraceae species as outgroups [55]. Focal taxa are in bold.
Inner tepal and stamen morphology is depicted for each species as bilaterally symmetrical (gray) or radially symmetrical (black). The ancestral
state of Commelinaceae inner tepal and stamen symmetry is reconstructed as radial and bilateral, respectively. A shift to radial stamen symmetry
is inferred at the base of Tradescantia, whereas a late-developmental shift to weak bilateral symmetry is inferred for C. dianthifolia and C.
virginica.
Preston and Hileman EvoDevo 2012, 3:6
http://www.evodevojournal.com/content/3/1/6
Page 3 of 14

Page 4

Figure 2 Flower morphological diversity in representative Commelinaceae species. (A-C) Floral diagrams of mature C. communis (A), C.
dianthifolia (B) and T. pallida (C) flowers. Stamens (yellow) and gynoecia (black) of both Commelina species and inner tepals (blue or pink) of C.
communis are strongly bilaterally symmetrical, outer tepals (green) of both Commelina species and inner tepals of C. dianthifolia are weakly
bilaterally symmetry, and all organ whorls of T. pallida are nearly radially symmetry. (D) Front view of a bilaterally symmetrical Commelina
communis flower showing the two large dorsal inner tepals (blue), reduced ventral inner tepal, reduced dorsal staminodes (upper yellow organs),
and pollen-producing ventral stamens (lower yellow organs). (E) Side view of a C. dianthifolia flower showing the two large dorsal inner tepals
(upper blue) and the even larger ventral inner tepal (lower blue). (F) Front view of a radially symmetrical Tradescantia pallida flower showing the
three equal pink inner tepals, and six equal stamens. (G) Individual organs of post-anthesis C. communis flowers. (H) Individual organs of post-
anthesis C. dianthifolia flowers; outer tepals are not shown. (I) Individual organs of post-anthesis T. pallida flowers. (J) Scanning electron
micrograph (SEM) of a mid-stage C. communis flower; the ventral region develops faster than the dorsal region (this stage), and then the ventral
inner tepal arrests development (later stage). (K) SEM of a late stage C. dianthifolia flower. The ventral region develops faster than the dorsal
region from early to late stages. (L) SEM of an early stage T. pallida flower showing that radial symmetry is established early in development. ds,
dorsal outer tepal; ventral outer tepal; dp, dorsal inner tepal; vp, ventral inner tepal; oda, outer dorsal androecium; ida, inner dorsal androecium;
iva, inner ventral androecium; ova, outer ventral androecium; da; dorsal androecia; va, ventral androecia; gy, gynoecium.
Preston and Hileman EvoDevo 2012, 3:6
http://www.evodevojournal.com/content/3/1/6
Page 4 of 14

Page 5

corollas consistent with a role in differential organ
growth, (2) DEF and GLO orthologs are co-expressed in
staminodes, stamens and the dorsolateral inner tepals,
but not in outer tepals or the outer tepal-like ventral
inner tepal of C. communis, and (3) cellular morphology
of the C. communis ventral inner tepal is more similar to
outer tepals than dorsolateral inner tepals, suggesting
asymmetric loss of inner tepal identity in the second
whorl.
Methods
Plant material
Inflorescence material was harvested from wild-collected,
flowering individuals of C. communis and T. pallida in
Missouri and Kansas (USA) during the summer months,
and from C. dianthifolia plants grown under standard
greenhouse conditions at the University of Kansas. For
scanning electron microscopy (SEM) and in situ hybridi-
zation, whole inflorescences were fixed overnight in for-
malin acetic alcohol (FAA: 47.5% (v/v) ethanol, 5% (v/v)
acetic acid, 3.7% (v/v) formaldehyde), and gradually taken
through an alcohol series to 100% ethanol. For quantita-
tive reverse transcriptase polymerase chain reaction
(qRT-PCR), flower organs from 10 to 16 flowers were
individually dissected twice from 3 to 5 mm long buds,
pooled according to identity/position, and snap-frozen in
liquid nitrogen. Total RNA was extracted using TriRea-
gent (Applied BioSystems, Carlsbad, CA, USA), DNA
contamination was removed using TURBO DNase
(Applied BioSystems), and 1 μg of RNA was reverse tran-
scribed using the iScript Select cDNA synthesis kit (Bio-
Rad, Hercules, CA, USA). Genomic DNA was extracted
from leaf material using the DNeasy Plant Mini Kit (Qia-
gen, Valencia, CA, USA).
SEM
Ethanol-dehydrated inflorescences were dissected to reveal
internal floral organs as necessary. Tissues were critical
point dried in a Tousimis critical point drier, mounted on
stubs, sputter-coated with gold and viewed with a D. Leo
field emission scanning electron microscope (VTT, Espoo,
Finland). For micromorphological analyses of inner and
outer tepals, both adaxial and abaxial surfaces were ana-
lyzed at regions proximal, medial and distal to the floral
axis.
Gene isolation and phylogenetic analysis
In order to isolate and sequence all TB1-like orthologs
from genomic DNA of C. communis, C. dianthifolia and
T. pallida, we used the degenerate forward primers,
CYCF1, CYCF2, CYC73aaF, CYC73bF, which were pre-
viously designed to amplify CYC/TB1 genes from a wide
range of angiosperm taxa, in combination with the reverse
primers CYCR1 and CYCR2 [56,57]. Amplicons were
cloned into the pGEM-T vector (Bio-Rad), and 5 to 10
colonies per successful PCR reaction were sequenced. To
determine if the Commelinaceae sequences corresponded
to CYC/REP/TB1-like genes, and to determine orthology
and copy number, amino acid sequences spanning the
TCP to R domain from C. communis, C. dianthifolia and
T. pallida were aligned by MAAFT [58,59] and then by
eye in MacClade [60] with previously designated CINCI-
NATTA (CIN)-like, CYC-like, REP-like and TB1-like TCP
genes [24,61]. Nucleotide alignments were subjected to
maximum likelihood (ML) analyses in GARLI 0.951 and
Bayesian analyses in MrBayes 3.1.2 following model opti-
mization in MrModelTest [62-64]. ML analyses were run
using 10 random addition sequences with 500 bootstrap
replicates. Bayesian analyses were run twice for 10 million
generations, sampling every 1,000 generations, with 25%
of trees discarded as burn-in. Newly generated TB1-like
sequences were submitted to Genbank with accession
numbers [Genbank: JQ622131-JQ622135 and JQ622142].
DEF and GLO orthologs were isolated and sequenced
from cDNA of C. communis, C. dianthifolia and T. pallida
using previously described degenerate primers [65].
Amplicons were cloned and sequenced as previously
described, and aligned nucleotide sequences were sub-
jected to ML and Bayesian analyses as described for the
TB1-like genes. Newly generated DEF- and GLO-like
genes were submitted to Genbank with accession numbers
[Genbank: JQ622136-JQ622141].
qRT-PCR
To compare patterns of TB1-like and B-class gene expres-
sion in mid-stage floral organs of T. pallida, C. dianthifolia
and C. communis, qRT-PCR analyses using DyNAzyme II
Hot Start DNA Polymerase (GE Healthcare, Pittsburgh,
PA, USA) and SYBR Green I (Invitrogen, Carlsbad, CA,
USA) were conducted on a DNA Engine Opticon 2 real-
time PCR machine (MJ Research. Waltham, MA, USA) as
previously described [66]. Two primers pairs per target
gene were designed in Primer 3, and the most efficient pri-
mer pair was selected for expression analysis (Additional
file 1) [67]. Where possible, primer pairs were designed to
span introns to rule out DNA contamination. b-ACTIN
(ACT) and EUKARYOTIC TRANSLATION ELONGATION
FACTOR 1 ALPHA 1 (EF1alpha) showed little transcrip-
tional variation across different tissues of C. communis and
C. dianthifolia, and T. pallida, respectively, and were,
therefore, selected as reference genes. A negative cDNA
control containing RNA from dorsolateral inner tepals and
lacking the reverse transcriptase enzyme was initially used
as a negative control template for all qRT-PCR analyses.
Furthermore, when no transcription was detected in floral
material, genomic DNA was used as a positive control. For
each gene the mean and standard deviation was deter-
mined for three to four technical replicates. Results were
Preston and Hileman EvoDevo 2012, 3:6
http://www.evodevojournal.com/content/3/1/6
Page 5 of 14