Research

Duplication and expression of CYC2-like genes in the origin and

maintenance of corolla zygomorphy in Lamiales
Jinshun Zhong1,2 and Elizabeth A. Kellogg1,3
1
Department of Biology, The University of Missouri-St Louis, One University Blvd, St Louis, MO 63121, USA; 2Department of Plant Biology, The University of Vermont, Carrigan Drive,
Burlington, VT 05405, USA; 3Donald Danforth Plant Science Center, North Warson Road, St Louis, MO 63132, USA

Summary
Author for correspondence:
Duplication, retention, and expression of CYCLOIDEA2 (CYC2)-like genes are thought to
Jinshun Zhong affect evolution of corolla symmetry. However, exactly what and how changes in CYC2-like
Tel: +1 802 656 7670
genes correlate with the origin of corolla zygomorphy are poorly understood.
Email: Jinshun.Zhong@uvm.edu

We inferred and calibrated a densely sampled phylogeny of CYC2-like genes across the
Received: 4 June 2014 Lamiales and examined their expression in early diverging (EDL) and higher core clades (HCL).
Accepted: 29 August 2014

CYC2-like genes duplicated extensively in Lamiales, at least six times in core Lamiales (CL)
around the Cretaceous–Paleogene (K–Pg) boundary, and seven more in EDL relatively more
New Phytologist (2015) 205: 852–868 recently. Nested duplications and losses of CYC2-like paralogs are pervasive but may not cor-
doi: 10.1111/nph.13104 relate with transitions in corolla symmetry. We found evidence for dN/dS (x) variation follow-
ing gene duplications. CYC2-like paralogs in HCL show differential expression with higher
Key words: corolla symmetry, CYCLOIDEA2
expression in adaxial petals.
(CYC2)-like gene, gene duplication and
Asymmetric expression but not recurrent duplication of CYC2-like genes correlates with the
retention, Lamiales, positive selection. origin of corolla zygomorphy. Changes in both cis-regulatory and coding domains of CYC2-
like genes are probably crucial for the evolution of corolla zygomorphy. Multiple selection
regimes appear likely to play important roles in gene retention. The parallel duplications of
CYC2-like genes are after the initial diversification of bumble bees and Euglossine bees.

Almeida, 2002; Corley et al., 2005). TCP transcription factors

Introduction
Corolla zygomorphy has evolved many times during the diversifi-
cation of angiosperms and is thought to affect the interaction
between pollinators and plants, that is, pollination efficiency
(Donoghue et al., 1998; Endress, 1999, 2001; Ree & Donoghue,
1999; Preston & Hileman, 2009). Bees, Euglossine bees and
bumble bees, in particular, are attracted to zygomorphic corollas
through either learning or innate preference (Møller, 1995;
Neal et al., 1998; Rodr
ıguez et al., 2004; Westerkamp &
Claßen-Bockhoff, 2007). Correlational studies of corolla zygo-
morphy and species richness showed that corolla zygomorphy is
associated with several of the most species-rich angiosperm lin-
eages (e.g. core Lamiales (CL), Fabaceae, Orchidaceae), suggest-
ing that corolla zygomorphy can greatly accelerate speciation
rates (Sargent, 2004). This hypothesis is further supported by
experimental evidence showing that plants with zygomorphic
corollas have higher fitness than those with actinomorphic ones
in natural conditions (Erysimum mediohispanicum, Brassicaceae)
(G
omez et al., 2006).
In the model species Antirrhinum majus (snapdragon, Planta-
ginaceae, Lamiales), the development of corolla zygomorphy is
controlled by a genetic network including two TCP genes,
CYCLOIDEA (CYC) and DICHOTOMA (DICH) (Luo et al.,
1996, 1999), and two MYB transcription factors (Galego &
are named for the three founding members of the gene family,
CYC, TEOSINTE BRANCHED1 (TB1) and Proliferation Cell
Factor (PCF). Within the TCP family, CYC and TB1 (CYC/
TB1 or ECE clade) proteins are characterized by a unique ECE
(glutamate–cysteine–glutamate) domain (Howarth & Don-
oghue, 2006) in addition to two highly conserved domains, TCP
and R. ECE clade genes have duplicated extensively and indepen-
dently in eudicots and monocots (Howarth & Donoghue, 2006;
Damerval et al., 2007; Mondrag
on-Palomino & Theißen, 2009;
Bartlett & Specht, 2011; Preston & Hileman, 2012). Phyloge-
netic analyses have identified three types of ECE clade genes
(CYC1, CYC2, and CYC3) that have resulted from two duplica-
tions at the base of core eudicots (Howarth & Donoghue, 2006).
However, only the ECE-CYC2 clade (e.g. CYC/DICH) in core
eudicot species has been shown to be involved in corolla zygo-
morphy (Luo et al., 1996, 1999; Feng et al., 2006; Busch &
Zachgo, 2007; Broholm et al., 2008; Wang et al., 2008).
In snapdragon, flowers of cyc/dich double mutants are actino-
morphic, while flowers of cyc mutants are semipeloric (weakly
zygomorphic or actinomorphic) and dich mutants are zygomor-
phic but with less internal asymmetry of the adaxial petals (Luo
et al., 1996, 1999). The effects of CYC and DICH are partially
redundant but not identical. Both genes are expressed exclusively
in the adaxial regions of the flowers (Luo et al., 1996, 1999).

852 New Phytologist (2015) 205: 852–868 Ó 2014 The Authors

www.newphytologist.com New Phytologist Ó 2014 New Phytologist Trust
New
Phytologist Research 853

CYC/DICH regulate expression of the MYB transcription factor Origanum

RADIALIS (RAD), which antagonizes another MYB protein DI- Mentha
VARICATA (DIV) that is thus restricted to the abaxial part of Lamium
Tectona
the flower and helps to establish adaxial-abaxial asymmetry Congea
(Galego & Almeida, 2002; Corley et al., 2005). Lamiaceae
Callicarpa
A CYC2-mediated pathway has apparently been recruited Phrymaceae
Mimulus
independently for the development of corolla zygomorphy in Ruellia dp dp
some eudicot lineages, such as Brassicaceae, Fabaceae, and Astera- Hygrophila
lp lp
HCL Acanthaceae
ceae (Busch & Zachgo, 2007; Broholm et al., 2008; Wang et al., Schaueria
2008). Extensive investigation of the phylogeny and expression Avicennia
bp
Verbena CL
of CYC2-like genes in different lineages has shown that multiple
Scrophulariaceae Buddleja
copies of CYC2-like genes tend to be maintained in lineages with Scrophularia
zygomorphic corollas in both rosids and asterids (Citerne et al., Mohavea
2000, 2003; Hileman & Baum, 2003; Howarth & Donoghue, CL Plantaginaceae Antirrhinum
2005; Zhang et al., 2010). In addition, studies in Arabidopsis Gratiola
(Brassicaceae, rosids) and Primulina heterotricha (Gesneriaceae, Veronica
asterids) indicate that persistent expression of CYC2-like genes in Gesneriaceae Opithandra
petals in later developmental stages is important for the develop- Primulina
Bournea
ment and/or maintenance of corolla zygomorphy, as in CL Calceolariaceae
Calceolaria
(Cubas et al., 2001; Busch & Zachgo, 2007; Yang et al., 2012). EDL Tetrachondraceae
Polypremum
It remains unclear whether the duplication of CYC2-like genes Syringa
occurred early in Lamiales and is directly correlated with the ori- Oleaceae Ligustrum
EDL
gin of corolla zygomorphy, how important retention of CYC2- Jasminum
Carlemanniaceae
like paralogs is for the maintenance of corolla zygomorphy, what Silvianthus
EDL Plocospermataceae EDL
evolutionary forces (selection regime) would most likely have Plocosperma

helped to retain CYC2-like paralogs, and how conservation and
diversification of CYC2-like gene expression pattern following
duplication might have led to evolutionary transitions of corolla
symmetry. Most previous comparative studies have compared
zygomorphic corollas and derived actinomorphy within CL or
corolla zygomorphy and actinomorphy between distantly related
taxa (e.g. snapdragon vs Arabidopsis; cf. Zhang et al., 2010;
Howarth et al., 2011). Surprisingly few studies have investigated
the duplication, retention, and expression of CYC2-like genes in
lineages with actinomorphic corollas closely related to those with
zygomorphic corollas (e.g. Zhang et al., 2010; Howarth et al.,
2011). In addition, the selection regime following duplication of
CYC2-like genes is poorly studied and known. Previous studies
showed that retention of CYC2-like genes may be driven by
either relaxed purifying selection, as in Plantaginaceae-
Antirrhineae (Hileman & Baum, 2003), or by positive selection
on one paralog, as in Zingiberales (Bartlett & Specht, 2011).
The angiosperm order Lamiales, the focus of this study,
includes c. 12% of eudicot diversity, including c. 24 000 species
in 24 families; a number of relationships between these families
are well resolved (Sch€aferhoff et al., 2010; cf. McDade et al.,
2012). The early diverging lineages (EDL), including Plocosper-
mataceae, Oleaceae + Carlemanniaceae, and Tetrachondraceae
(Fig. 1), contain c. 3% of the species diversity of Lamiales; these
families are successively sister to CL (c. 97%). Within CL, Ges-
neriaceae + Calceolariaceae and Plantaginaceae are successive sis-
ters to the remaining taxa, a clade that is sometimes called higher
core Lamiales (HCL; Fig. 1; cf. Sch€aferhoff et al., 2010). Lami-
ales are a rich system for the study of floral evolution; zygomor-
phic corollas originated at least twice independently (Fig. 1, CL
and Carlemanniaceae) early in the clade, and derived
Fig. 1 Evolution of corolla symmetry and summary of duplication of floral
symmetry gene CYC2-like genes in Lamiales. The phylogeny of Lamiales
was modified from Sch€aferhoff et al. (2010), based on three plastid
markers. Corolla zygomorphy (black) has originated at least twice
independently in core Lamiales (CL) and in Silvianthus (Carlemanniaceae),
and ‘reversals’ to actinomorphy (red) multiple times independently in CL
(Kadereit, 2004). Vertical double lines indicate duplication events
discovered previously in Gesneriaceae and Plantaginaceae (Citerne et al.,
2000; Hileman & Baum, 2003; Hileman et al., 2003; Zhou et al., 2008;
Preston et al., 2009; Yang et al., 2012). Species in italics were used for
phylogenetic analyses (subset); species in Plantaginaceae and
Gesneriaceae were from previous studies (Citerne et al., 2000; Hileman
et al., 2003; Zhou et al., 2008; Preston et al., 2009; Yang et al., 2012).
Simplified floral diagrams (actinomorphic vs zygomorphic) represent the
most common floral form in corresponding clades, that is, actinomorphy in
early diverging Lamiales (EDL) and zygomorphy in CL. Black petals in the
diagram indicate expression of CYC2-like genes; white petals indicate lack
of expression. The expression pattern of CYCLOIDEA (CYC) and
DICHOTOMA (DICH) in the adaxial petals (dp) in snapdragon (Luo et al.,
1996, 1999) is hypothesized as the ancestral expression pattern in CL.
Lack of expression in petals is hypothesized as the expression pattern in
EDL. Underlined species with a picture of a flower are focal species for the
study of gene expression. lp, lateral petals; bp, abaxial petal; HCL, higher
core Lamiales.
actinomorphy evolved separately multiple times within lineages
in CL (e.g. Donoghue et al., 1998; Kadereit, 2004).
Here we found pervasive parallel and nested duplication events
of CYC2-like genes in Lamiales; gene duplication is not corre-
lated with the floral shift from actinomorphy to zygomorphy
early in Lamiales. Rather, asymmetrical expression of CYC2-like
genes along the adaxial/abaxial floral axis evolved in major lin-
eages within CL and is directly correlated with the origin of
corolla zygomorphy in CL. Furthermore, we estimated that par-
allel major duplications of CYC2-like genes mostly occurred in

Ó 2014 The Authors New Phytologist (2015) 205: 852–868

New Phytologist Ó 2014 New Phytologist Trust www.newphytologist.com

854 Research
geological periods subsequent to the initial diversification of
major bee lineages. We also found evidence that multiple evolu-
tionary forces have probably facilitated the retention of CYC2-
like duplicates.
Materials and Methods
Taxon sampling
Ninety-four species from 15 of the 24 families of Lamiales,
mostly EDL and HCL, were sampled for this study (Supporting
Information, Table S1). Additional sequences of CYC2-like genes
of some Calceolariaceae, Gesneriaceae, and Plantaginaceae were
obtained from GenBank. Species for the study of gene expression
were chosen on the basis of previous phylogenetic findings
(Sch€aferhoff et al., 2010), including taxa from the EDL like Olea-
ceae and two major subclades from HCL (Fig. 1), and also repre-
senting CYC2-like paralogs derived from separate duplication
events (this study).
DNA and RNA extraction
Total genomic DNA was extracted from individual plant silica
gel dried or fresh leaves with a DNeasy Plant Mini Kit following
the manufacturer’s protocol (Qiagen). Plant material for RNA
extraction was collected in RNAlater solution (Life Technologies,
Grand Island, NY, USA) and preserved at 20°C; total RNA
isolation was done using TRI Reagent (Life Technologies).
Gene isolation
CYC2-like genes were amplified from genomic DNA for most
species with various sets of primers (Table S2) using GoTaq Flexi
DNA polymerase kits (Promega) with the annealing temperature
Tm-5°C. All amplified PCR products were cleaned with a QIA-
quick Gel Extraction Kit (Qiagen) and subcloned to pGEM-T
Vector (Promega). To obtain all possible paralogs, at least three
sets of degenerate primers for the TCP to ending box regions
(CYCF2/CYCP2R, CYCF1/LCYC1R, and CYC118F/
CYCP2R) were used for amplification, and additional degenerate
primer pairs (Table S2) were used for species in which only a sin-
gle type of CYC2-like gene was found. In some cases, several
shorter fragments were amplified between the TCP and the R
regions and between the R and the ending box regions using
degenerate primers anchoring the relatively conserved TCP, R,
and ending box regions. For Oleaceae species, CYC2-like genes
of several species (Chionanthus, Abeliophyllum, Jasminum
angustifolium, Ligustrum, and Syringa) were amplified using
paired primers CYCF2 and CYCP2R (or LCYCR), CYCF2 and
562R, and 562F and CYCP2R (or LCYCR), and then more spe-
cific and efficient internal primers (CYC126F and CYC693R)
were designed. Likewise, CYC2-like genes in Polypremum
procumbens (Tetrachondraceae) were assembled from four over-
lapping fragments amplified from RNA using SuperScript III
One-Step RT-PCR System with PlatinumTaq (Life Technolo-
gies). Only one CYC2-like gene was recovered using all the
New Phytologist (2015) 205: 852–868
www.newphytologist.com
New
Phytologist
amplification strategies described for some Lamioideae species
(e.g. Lamium and Stachys). In addition, a premature stop codon
was found for all sampled Lamioideae species (Lamium, Stachys,
and Pogostemon); thus cDNA sequences were also amplified to be
compared with genomic CYC2-like gene sequences to confirm the
presence of an intron and the possibility of multiple alternatively
spliced transcripts. At least 12 colonies were screened by PCR for
each independent amplification using separate primer sets, and
plasmids were extracted using the 5 PRIME* FastPlasmid Mini
Kit (Fisher Scientific, Pittsburgh, PA, USA). Clones were then
sequenced in both directions with the BigDye Terminator v3.1
Cycle Sequencing Kit (Life Technologies) and analyzed on an
ABI3730 DNA sequencer (Life Technologies) at the University of
Missouri, St Louis, USA, and/or the Nucleic Acid Facility at
Pennsylvania State University, USA. To confirm the copy number
of CYC2-like genes in these species, we conducted Southern blots
for two Oleaceae species and two core Lamioideae species follow-
ing previous protocols (Malcomber & Kellogg, 2004).
Motif-based sequence analysis
The unaligned amino acid matrix of 31 CYC2-like sequences was
submitted to the motif-based sequence analysis website (MEME;
Bailey & Elkan, 1994, http://meme.nbcr.net/meme/) to search
for additional conserved motifs of CYC2-like proteins across the
order Lamiales. All sequences of CYC2-like proteins include the
TCP domain to ending box region, except those of the species
Polypremum in which the TCP domain was c. 100 bp shorter as a
result of primer problems.
Phylogenetic analysis
Sequencing trace files were trimmed of vector and assembled
from both sequencing directions using Geneious Pro 6.0.5 (Bio-
Matters, Auckland, New Zealand). Preliminary phylogenetic
analyses were conducted with the neighbor-joining algorithm in
MEGA 5.10 (Tamura et al., 2011) for all clones. Clones of the
same accession that formed a monophyletic clade and shared at
least 99.5% identical bases were considered as representing the
same genomic sequence, and the clone with the shortest branch
length was chosen for subsequent analyses. Much lower sequence
divergence among clones of individual Oleaceae species made it
impossible to distinguish paralogous and allelic variation, so at
least two clones were included for phylogenetic analyses for com-
parison even if they shared over 99.5% identity. Reduced
sequences were translated, aligned with MAFFT Version 7
(Katoh & Standley, 2013) in Seaview (Gouy et al., 2010), and
then converted back to nucleotides for later manual refinement
with MEGA 5.10 (Tamura et al., 2011). Eighty-eight models of
molecular evolution were assessed using jModeltest (Darriba
et al., 2012) to find the best-fit model for phylogenetic inference.
To test whether our new CYC2-like sequences fall into the
broad ECE-CYC2 clade, we created an ECE-CYC dataset that
included CYC2-like sequences of representatives from major
clades of core eudicots (The Angiosperm Phylogeny Group,
2009) from Phytozome (http://www.phytozome.net/).
Ó 2014 The Authors
New Phytologist Ó 2014 New Phytologist Trust
New
Phytologist
A second dataset included only ECE-CYC2 orthologs in
Lamiales. Helianthus annuus (Asteraceae, campanulids), Petunia
9 hybrida and Solanum lycopersicum (both Solanaceae, lamiids)
were included as outgroups, and only the conserved sequences of
the TCP domain to the R domain from these species could be
aligned with CYC2-like sequences from Lamiales for phyloge-
netic analyses. The aligned matrix consisted of 298 sequences
and 1344 sites; ambiguous regions were removed for phyloge-
netic inference. The GTR + I + G model (loge = 55 756.4228)
was chosen based on Akaike information criterion.
Maximum likelihood analyses were conducted using PhyML
(Guindon et al., 2010) (http://www.atgc-montpellier.fr/phyml/)
and RAxML in CIPRES (Ludwig et al., 2002) with 100 boot-
strap replicates. The matrices were partitioned by codon position,
and Bayesian inference was performed using MrBayes 3.2.1
(Ronquist et al., 2012) in CIPRES, with the nucleotide substitu-
tion model GTR + I + G, 10 000 000 generations and sampling
every 1000 generations. The first 25% of the trees (2500) were
discarded as burn-in. A majority-rule consensus of the remaining
trees was produced to assess Bayesian posterior probabilities
(PPs). Bayesian analysis with enforced monophyly of HCL was
also conducted using MrBayes 3.2.1 (Ronquist et al., 2012) in
CIPRES with other settings the same as in unconstrained analy-
ses. Tracer v1.5 (Rambaut & Drummond, 2009) was used to
summarize parameter estimates for a Bayes factor comparison
(Kass & Raftery, 1995).
A third dataset was constructed removing sequences that had
> 50% missing data in the TCP domain, and some sequences
from densely sampled clades (Bignoniaceae, Lamiaceae and
Acanthaceae). This dataset included 195 Lamiales sequences and
was analyzed with RAxML analysis with 100 bootstrap replicates.
A fourth reduced dataset included only 154 Lamiales sequences
without any missing data in the TCP domain; this data set was
analyzed using RAxML with 100 bootstrap replicates, and was
later used for tests of selection.
Tests for selection
The ratio of nonsynonymous to synonymous nucleotide substitu-
tions (x = dN/dS) was used to measure selection on protein-
coding sequences. If x > 1.0, the sequences of interest are under
positive selection, whereas if x < 1.0 they are under purifying
selection, and if x = 1.0 they are under neutral selection (Yang,
2007 and references therein).
To detect shifts in selection following gene duplication, we
used branch-based models of selection in which x varies among
different branches in the phylogeny (Yang, 1998; Yang & Niel-
sen, 1998). We tested hypotheses of x variation across the phy-
logeny with CODEML in PAML 4.7 (Yang, 2007). We also
conducted branch-site random effects likelihood (REL) imple-
mented in the HyPHY package to infer episodic positive selec-
tion along all branches (Kosakovsky Pond & Frost, 2005b;
Kosakovsky Pond et al., 2005, 2011).
To identify sites under selection, we employed site-specific
models and branch-site models using CODEML in PAML 4.7
(Yang, 2007) and three different likelihood-based methods:
Ó 2014 The Authors
New Phytologist Ó 2014 New Phytologist Trust
Research 855
single likelihood ancestor counting (SLAC), fixed effects likeli-
hood (FEL) and a fast, unconstrained Bayesian approximation
(FUBAR) using the Datamonkey web server (Kosakovsky Pond
& Frost, 2005a).
Molecular dating
To estimate and compare the relative divergence times of major
independent duplications, we applied a relaxed-clock Bayesian
Markov chain Monte Carlo method as implemented in BEAST
v1.7.4 (Drummond et al., 2012) with a Yule tree prior,
GTR + I + G substitution model parameters, and both an uncor-
related relaxed lognormal clock (ucld) and an uncorrelated
relaxed exponential clock (uced). Fossil calibration points from
Oleaceae and Bignoniaceae were applied. The prior age of the
Fraxinus clade that includes Olea, Osmanthus, Syringa, Philyrea,
Noronhia and Chionanthus (Oleaceae) was set using a lognormal
distribution with mean = 1.5, SD = 0.5 and offset = 37 million yr
ago (Mya; i.e. estimated age for Fraxinus fossils; Manchester,
1999), thus setting a hard lower bound to the age of the group of
37 Mya. Likewise, the age of Catalpa/Oroxylum from Bignonia-
ceae was set as 28.4 Mya based on fossils of Catalpa
(Bignoniaceae) (Manchester, 1999). Two additional calibration
points were used for the nodes of HCL-CYC2A Verbenaceae-Big-
noniaceae and HCL-CYC2B Verbenaceae-Bignoniaceae whose
prior ages were set using an exponential distribution with
mean = 1.0 and offset = 49.4 Mya (Nie et al., 2006). The
MCMC chain was run for 10 000 000, 20 000 000, 50 000 000
and 100 000 000 generations to make sure that effective sample
sizes (ESS) were > 200, with tree and parameter values being
saved every 1000th generation. The marginal likelihoods of two
different clock models, ucld and uced, were compared using a Ba-
yes factor test for best fit (Drummond et al., 2006).
TreeAnnotator was used to summarize the information and
calculate the maximum clade credibility phylogenetic tree with
the removal of an appropriate burn-in (the first 25% of the
samples) after visual inspection in Tracer v1.5 (Rambaut &
Drummond, 2009).
Expression of CYC2-like genes in EDL and HCL
RNA was isolated from young inflorescences, and adaxial, lateral
and abaxial petals of floral buds (petals for EDL) that were 1–3 d
before anthesis, and young leaves for each focal species. Paralog-
specific primers (Table S2) were designed for the amplification of
CYC2-like paralogs. The RT-PCR used the SuperScript III One-
Step RT-PCR System with PlatinumTaq (Life Technologies)
that started with cDNA synthesis for 30 min at 60°C, and 30
cycles of regular PCR amplification. The housekeeping gene actin
was used as reference. Amplified products were subsequently sub-
cloned to pGEM-T Vector (Promega) and sequenced to verify
the specificity of primers. For quantitative PCR (qPCR), c. 1 lg
of total RNA was reverse-transcribed using the iScript cDNA
synthesis kit (Bio-Rad, Hercules, CA, USA). qPCR was done on
the MyIQ single-color real-time detection system (Bio-Rad)
using iQ SYBR Green Supermix (Bio-Rad). The data presented
New Phytologist (2015) 205: 852–868
www.newphytologist.com
 New
856 Research Phytologist

are the summary of three biological replicates and three qPCR
technical replicates. Data were normalized against EF1alpha
expression.
Results
CYC2-like sequence properties
One hundred and ninety-nine CYC2-like sequences from 94
Lamiales species (EDL, Calceolariaceae and HCL) were newly
generated in this study. All CYC2-like sequences for analyses
include the TCP (partial) to ending box regions ranging from
366 bps (Jasminum spp.) to 844 bps (Mendoncia spp). Most
CYC2-like sequences contain no intron, but sequences from
Lamiaceae-Lamioideae have a c. 90 bp intron located at the 30
terminus of the R-domain. Sequence identity for characterized
genes (Asteraceae, Plantaginaceae and Gesneriaceae) is included
in Table 1 for comparison. If clones from individual plants
formed a monophyletic group and had sequence identity
> 99.5%, they were inferred to represent a single gene, assuming
that ‘polymorphisms’ represent PCR errors or allelic variation
(e.g. core Lamioideae (Stachys and Lamium), most Oleaceae spe-
cies (e.g. Olea, Ligustrum and Syringa)) (Fig. S1, Supporting
Information). By contrast, if clones from a single individual
formed a clade but had < 95% identity, they were inferred to be
paralogs derived from gene duplication (e.g. Polypremum, 77.7%
identity). In general, paralogous sequences from a single individ-
ual fell into separate clades in the gene tree and had sequence

Table 1 Summary of pairwise divergence of putative CYCLOIDEA2 (CYC2)-like paralogs derived from independent duplication events in Lamiales

Clade Speciesa Putative paralogsb Percentage identityc Taxonomic group

Non-Lamiales Helianthus annuus HaCYC2s 50.3–78.9 Non-Lamiales

Plantaginaceae Antirrhinum majus↑ CYC/DICH 63.40 Plantaginaceae
Plantaginaceae Mohavea confertiflora↑ CYCs/DICHs 60.0–93.2 Plantaginaceae
Plantaginaceae Veronica serpyllifolia↑ VCYC2A/VCYC2B 58.00 Plantaginaceae
Plantaginaceae Gratiola officinalis↑ GoCYC2s 61.2–91.3 Plantaginaceae
Gesneriaceae Bournea leiophylla* GCYC2A/GCYC2B 66.30 Gesneriaceae
Calceolariaceae Calceolaria rivularis↑ CCYC2A/CCYC2B 64.60 Calceolariaceae
EDL Fraxinus americana FaCYC2s 94.20 Oleaceae
EDL Jasminum nudiflorum* JnCYC2s 88.70 Oleaceae
EDL Osmanthus fragrans* OfCYC2s 96.90 Oleaceae
EDL Polypremum procumbens* PpCYC2A/PpCYC2B 77.70 Tetrachondraceae
HCL Acanthus mollis↑ AmCYC2A2/AmCYC2B 56.50 Acanthaceae
HCL Acanthus mollis↑ AmCYC2A1/AmCYC2B 57.50 Acanthaceae
HCL Acanthus mollis↑ AmCYC2A1/AmCYC2A2 87.00 Acanthaceae
HCL Avicennia germinans* CYC2A/CYC2B 65.00 Acanthaceae
HCL Hygrophila corymbosa↑ RCYC2A1/RCYC2A2 61.30 Acanthaceae
HCL Ruellia tweediana↑ RCYC2A1/RCYC2A2 60.70 Acanthaceae
HCL Schaueria calicotricha↑ CYC2A/CYC2B 53.00 Acanthaceae
HCL Thunbergia erecta↑ CYC2A/CYC2B 53.00 Acanthaceae
HCL Thunbergia mysorensis↑ TmCYC2A1/TmCYC2A2 84.80 Acanthaceae
HCL Mentha longifolia* MlCYC2B1/MlCYC2B2 92.30 Lamiaceae
HCL Buddleja davidii* SCYC2A/SCYC2B 57.00 Scrophulariaceae
HCL Scrophularia marilandica↑ SCYC2A1/SCYC2B2 53.40 Scrophulariaceae
HCL Scrophularia marilandica↑ SCYC2A1/SCYC2B1 54.00 Scrophulariaceae
HCL Scrophularia marilandica↑ SCYC2A1/SCYC2A2a 70.10 Scrophulariaceae
HCL Scrophularia marilandica↑ SCYC2A1/SCYC2A2b 71.80 Scrophulariaceae
HCL Scrophularia marilandica↑ SCYC2B1/SCYC2B2 90.30 Scrophulariaceae
HCL Scrophularia marilandica↑ SCYC2A2a/SCYC2A2b 95.30 Scrophulariaceae
HCL Scrophularia orientalis↑ SCYC2A1/SCYC2B 53.60 Scrophulariaceae
HCL Scrophularia orientalis↑ SCYC2A2/SCYC2B 58.30 Scrophulariaceae
HCL Scrophularia orientalis↑ SCYC2A1/SCYC2A2 71.80 Scrophulariaceae
HCL Scrophularia variegata↑ SCYC2A1/SCYC2A2 61.60 Scrophulariaceae
HCL Lantana velutina↑ LvCYC2B1/LvCYC2B2 97.50 Verbenaceae
HCL Lippia myriocephala↑ LmCYC2B1/LmCYC2B2 93.50 Verbenaceae
HCL Rehdera trinervis↑ CYC2A/CYC2B 54.10 Verbenaceae
HCL Verbena canadensis↑ VcCYC2B1/VcCYC2B2 96.80 Verbenaceae

EDL, early diverging lineages; HCL, higher core Lamiales.

a
Species with an asterisk (*) have actinomorphic corolla, while those with an upward arrow (↑) have zygomorphic corolla; the only exception is Fraxinus
americana, which is apetalous.
b
Sequences with < 95% identity are interpreted as paralogs. Sequences with 95–98% identity are provisionally considered as putative paralogs here, but
they might represent divergent alleles.
c
Percentage of identity = 100% 9 (overall aligned matched (identical) bases in pairwise alignment of CYC2-like paralogs)/(total length of pairwise align-
ment of CYC2-like paralogs); CYC2-like genes include sequences from the TCP to the ending box domains. CYC2-like paralogs of Antirrhinum, Bournea,
Gratiola, Mohavea and Helianthus were included for reference.

New Phytologist (2015) 205: 852–868 Ó 2014 The Authors

www.newphytologist.com New Phytologist Ó 2014 New Phytologist Trust
New
Phytologist Research 857

identities < 95% (e.g. most species within CL). However, in a
few cases, sequences with greater identity (Table 1; e.g.
Osmanthus, 96.9% identity; Verbena canadensis, 96.8% identity;
and Lantana velutina, 97.5% identity) were provisionally consid-
ered as paralogs from very recent gene duplications, although
they might represent divergent alleles.
Motif analyses found four new conserved motifs for CYC2-
like proteins in Lamiales (Figs 2 (motifs 3, 5, 7 and 8), S2)
besides the TCP, R, ending-box and ECE domains (Fig. 2,
motifs 1, 2, 4, 9). In addition, two lineage-specific motifs were
discovered for the Oleaceae proteins (Fig. 2, motifs 6 and 10).
Within the conserved TCP domain, two lineage-specific amino
acid replacements were observed in helix 2 (position 50, K-N, in
Oleaceae-CYC2s; and position 56, K-E in Oleaceae-CYC2s).
Phylogenetic analyses
Our phylogenetic analyses of ECE-CYC sequences showed that
all sequences amplified from this study are grouped with CYC/
DICH from Antirrhinum majus (Plantaginaceae) in the CYC2
(CYC2-like) clade with high Bayesian posterior probability and
moderate bootstrap support (PP, 0.96; pBT, 53%; rBT, 67%;
Fig. S3).
Phylogenetic results of a reduced set with only the CYC2-like
clade show that all sequences from Lamiales form a monophyletic
group with high support (PP, 1; pBT, 99%; rBT, 97%; Fig. 3a,
b). CYC2-like sequences of Tetrachondraceae plus Oleaceae form
a clade (PP, 1; pBT, 99%; rBT, 97%; Fig. 3a) that is sister to the
CL clade (PP, 1; pBT, 52%; rBT, 50%; Fig. 3a). However, we
believe that this sister relationship of Oleaceae and Tetrachondra-
ceae is caused by rooting with a distantly related taxon Helianthus
annuus (asterid II/campanulid clade). It is more likely that Olea-
ceae and Tetrachondraceae are successive sister groups to CL, as
shown in the phylogeny of Lamiales based on multiple chloro-
plast loci, and we assume this relationship in our later discussion
(Sch€aferhoff et al., 2010).
Our phylogenetic analyses of CYC2-like genes found at least
13 gene duplication events in Lamiales (Fig. 3a,b). Interestingly,

Combined
Name Motif location
P-value
CongeaTOM1_8_CYC2A 1.79e-156

PaulowniaTOM12_5_CYC2A 9.09e-169

MimulusRIN1_1_CYC2A 5.40e-148

ThunbergiaERE4_5_CYC2A 1.48e-114

StachytarphetaCAL10_1_CYC2A 9.60e-138

ScrophulariaMAR4_2_CYC2A1 1.14e-125

ScrophulariaMAR8_8_CYC2A2b 1.37e-148

ScrophulariaMAR9_3_CYC2A2a 8.48e-152

CongeaTOM1_4_CYC2B 6.78e-155

PaulowniaTOM12_1_CYC2B 4.45e-170

MimulusRIN1_4_CYC2B 3.13e-134
ThunbergiaERE4_7_CYC2B 9.84e-114
StachytarphetaCAL10_7_CYC2B2 1.06e-136
StachytarphetaCAL10_6_CYC2B2 1.75e-140
ScrophulariaMAR3_8_CYC2B1 6.91e-136
ScrophulariaMAR3_1_CYC2B2 2.60e-133
AntirrhinumMAJ_DICH 4.29e-123
AntirrhinumMAJ_CYC 2.71e-111
BourneaLEI_CYC2 1.35e-117
BourneaLEI_CYC1 1.16e-112
CalceolariaRIV2_5 6.42e-119

CalceolariaRIV2_8 6.42e-119

CalceolariaRIV3_4 9.30e-118

PolypremumPRE_11_5 6.82e-60

PolypremumPRE_11_3 3.19e-51

AbeliophyllumDIS8_4 7.94e-209

AbeliophyllumDIS8_2 4.14e-209

AbeliophyllumDIS2_2 1.15e-200

AbeliophyllumDIS2_1 2.26e-199

ChionanthusVIR12_7 8.69e-188

ChionanthusVIR12_8 1.29e-183

0 50 100 150 200 250

Motif 1 Motif 2 Motif 3 Motif 4 Motif 5 Motif 6 Motif 7 Motif 8 Motif 9 Motif 10
TCP domain R domain Ending-box ECE domain
Fig. 2 CYC2-like protein (ECE-CYC2 clade) evolution across the Lamiales. Italic names are sequences from published studies. Bold names are from early
diverging Lamiales. All sequences of CYC2-like proteins include the TCP (motif 1), ECE (motif 9), R (motif 2) (Cubas et al., 2001) and ending box (motif 4;
Zhou et al., 2008), except for those of Polypremum, which included only a small portion of the TCP domain. Most Lamiales species appear to share motif 1
(TCP domain), motif 2 (R domain), motif 3, motif 4 (ending box), motif 5 and motif 8. Motifs 6 and 10 are unique to sampled Oleaceae species, whereas
motif 9 (ECE domain) is specific to core Lamiales. The scale relates to amino acid residues.

Ó 2014 The Authors New Phytologist (2015) 205: 852–868

New Phytologist Ó 2014 New Phytologist Trust www.newphytologist.com
 New
858 Research Phytologist

(a) Higher core Lamiales (HCL, Fig. 3b)

1/68/76
HCL HCL - CYC2A
Antirrhineae 1/72/77
HCL - CYC2B
CL Gratiola 1/93/95 Columnea_byrsina
0.83/67/73 Alloplectus_panamensis
Veronica+Digitalis Paradrymonia_decurrens
Gesneriaceae Sinningia spp.
CL Spaherorrhiza_sarmentiana
Calceolariaceae Gesneria_ventricosa
Goyazia_rupicola
Polypremum 0.96/66/57 Niphaea_oblonga
1/89/93 Rhytidophyllum_tomentosum
Jasminum 0.83/66/88 0.99/88/91 Lenbrassia_australis
nudiflorum 1/99/100 Negria_rhabdothamnoides
Syringa 0.78/69/88 Coronanthera_clarkeana GCYC2B1
EDL 1/87/87 1/98/100 Fieldia_australis
Mitraria_coccinea
1/99/99 Asteranthera_ovata
1/75/69 Fieldia_australis GCYC2
1/58/50 Mitraria_coccinea
1/99/100 Asteranthera_ovata GCYC2B2
1/98/98 Depanthus_glaber
Negria_rhabdothamnoides
Coronanthera_clarkeana
Titanotrichum_oldhamii
0.94/88/95 Napeanthus_reitzii
1/82/76 Bournea_leiophylla
0.73/-/52 Opithandra_dinghushanensis
1/99/100 Cyrtandra_apiculata
Core Lamiales 0.98/-/77
Conandron_ramondioides
Ramonda_myconi
(CL) Haberlea_ferdinandi_coburgii
1/52/50 Opithandra_dinghushanensis
Oreocharis_benthamii
0.66/74/73 1/90/93 Bournea_leiophylla
1/98/99 Opithandra_dinghushanensis
0.72/60/90 Loxostigma_sp GCYC2A2
0.88/63/58 Didymocarpus_citrinus
Gesneriaceae 0.87/85/90
Primulina_tabacum
Didymocarpus_citrinus
1/88/90 Loxostigma_sp GCYC2A1
Cyrtandra_apiculata
Opithandra_dinghushanensis
GCYC1
1/89/90
0.99/66/80 Saintpaulia_spp.
1/98/100 Streptocarpus_primulifolius
0.93/58/90 Saintpaulia_spp.
Streptocarpus_primulifolius
1/99/100
1/99/100 Jankaea_heldreichii
Ramonda_myconi
Haberlea_ferdinandi_coburgii
Calceolaria_engleriana
Calceolaria_rivularis
Calceolaria_arachnoidea
CCYC2A
Calceolaria_rivularius
Calceolaria_arachnoidea CCYC2B
Calceolariaceae Mohavea_confertiflora
0.79/ - / - Sairocarpus_nuttallianus
Misopates_orontium
Antirrhinum_majus
1/77/94
Asarina_procumbens
Mohavea_confertiflora DICH
1/91/94 1/82/94 Linaria_spp.
Chaenorhinum_villosum
Lophospermum_sp
1/99/100 Collinsia_heterophylla
1/89/85 Tonella_floribunda
1/99/100 Keckiella_cordifolia
1/99/99 Chelone_glabra
Penstemon_hartwegii
Lamiales 0.67/–/– 1/93/98
0.73/-/-
Mohavea_confertiflora
Sairocarpus_nuttallianus
1/97/99 1/85/91
Mohavea_confertiflora
1/85/94 Misopates_orontium
1/92/97 Antirrhinum_majus
Chaenorhinum_villosum
CYC
0.96/62/77
1/89/95 Linaria_spp.
1/94/96 Cymbalaria_muralis
Asarina_procumbens
Lophospermum_sp
1/83/86
1/88/92 0.79/64/67 Digitalis_purpurea
Plantago_lanceolata
Plantago_major VCYC2A
Veronica_serpyllifolia
Plantaginaceae Gratiola_officinalis
GoCYC2s
Scoparia_sp
Digitalis_purpurea
1/100/98 Veronica_serpyllifolia VCYC2B
Noronhia_candicans NcCYC2s
Olea_europaea
1/94/98 Osmanthus_fragrans OfCYC2s
Phillyrea_angustifolium
1/88/92 Syringa_pekinensis
0.92/63/56 Chionanthus_virginicus
0.82/53/67
0.8/57/63 Fraxinus_americana FaCYC2s
Ligustrum_lucidum
0.99/86/88
Comoranthus_minor
Oleaceae 1/92/90
JtCYC2A
Jasminum_spp. JtCYC2B
1/99/100
Jasminum_nudiflorum JnCYC2A
1/97/99

1/99/97
JnCYC2B
Abeliophyllum_distichum
Early diverging Forsythia_suspensa
Polypremum_procumbens PpCYC2A
Polypremum_procumbens PpCYC2B
Lamiales (EDL) Tetrachondraceae

Fig. 3 Phylogram from Bayesian inference of CYC2-like genes of Lamiales. (a) Basal portion of Lamiales CYC2-like gene phylogeny showing relationships
among early diverging Lamiales (EDL, Oleaceae and Tetrachondraceae), Calceolariaceae, Gesneriaceae, and Plantaginaceae. (b) Detailed phylogeny of
CYC2-like genes within higher core Lamiales (HCL). Heavy branches have support values of 100/100/100 (Bayesian posterior probability/PhyML
bootstrap/RAxML bootstrap); lower values are labeled above the branches. Simplified floral diagrams are drawn to show the corolla zygomorphy in core
Lamiales (CL) and actinomorphy in EDL. Vertical bars indicate putative duplication events of CYC2-like genes. Open vertical bars are nested duplications;
closed vertical bars point to major duplication events in major taxonomic groups (Calceolariaceae, Gesneriaceae, Plantaginaceae, and HCL), and gray
triangles are still additional duplications for which paralogs are not shown here. Species in red in EDL, Plantaginaceae, and HCL have completely
actinomorphic flowers, whereas the species Fraxinus americana in purple lacks petals. The inset in the upper left corner shows duplication events that are
plotted on a species phylogeny, showing two of a putative seven duplication events in EDL (Polypremum and Jasminum nudiflorum), and six duplication
events in CL (HCL, Antirrhineae, Gratiola, Veronica + Digitalis, Gesneriaceae and Calceolariaceae).

New Phytologist (2015) 205: 852–868 Ó 2014 The Authors

www.newphytologist.com New Phytologist Ó 2014 New Phytologist Trust
New
Phytologist Research 859

(b) Phrymaceae Amphitecna_spp.

1/94/93 Crescentia_spp.
Orobanchaceae Tabebuia_impetiginosa
Markhamia_lutea
Lamiaceae 0.84/–/–
1/96/97
Martinella_obovata
Phryganocydia_corymbosa
0.98/56/58
Bignonia_capreolata
Acanthaceae 1/82/88 1/94/96
Macfadyena_tweediana
Melloa_quadrivalvis
Verbenaceae 1/82/87
Anemopaegma_orbiculatum
Paragonia_pyramidata
1/97/98 1/99/98
Tanaecium_crucigerum
Scrophulariaceae 0.97/66/83 Jacaranda_cuspidifolia
Catalpa_spp.
Oroxylum_indicum
Campsis_radicans
Orobanchaceae
1/99/100 Gerardia_grandiflora
Digomphia_densicoma Agalinis_tenuifolia
0.84/89/90
1/89/92 Scutellaria_lateriflora
0.96/56/65 Scutellaria_incana
Scutellaria_incana
Lamiaceae0.8/–/– 1/92/99 1/99/99
Teucrium_chamaedrys
Perovskia_atriplicifolia
1/96/96
Ocimum_basilicum Origanum_vulgare
Lamium_maculatum
Holmskioldia_sanguinea Stachys_byzantina
0.79/–/–
0.94/67/68 Pogostemon_cablin
1/99/100 Pogostemon_heyneanus
0.81/61/60
LCYC2A1
1/100/99 Pogostemon_heyneanus
Premna_fulva
LCYC2A2
0.81/–/– 0.91/–/–
0.92/–/–
Gmelina_arborea
0.92/57/68
Congea_tomentosa
Tectona_grandis
Callicarpa_cathayana
Paulownia_tomentosa

HCL - CYC2A
Acanthaceae Thunbergia_mysorensis
Thunbergia_erecta
1/80/87 Mendoncia_spp.
0.99/94/94 Eranthemum_pulchellum
Hygrophila_corymbosa
Ruellia_tweediana RCYC2A1
Schaueria_calicotricha
Avicennia_germinans
1/73/83 Crossandra_infundibuliformis
1/98/98 Acanthus_mollis
Hygrophila_corymbosa
Verbenaceae Ruellia_tweediana RCYC2A2
1/99/100 Bouchea_fluminensis
0.99/90/97 Stachytarpheta_calderonii
0.74/–/– Citharexylum_schottii
1/68/76 Rehdera_trinervis
0.7/–/– 0.98/–/–
Phryma_leptostachya
Mimulus_ringens
Phrymaceae
Sesamum_triphyllum
Thomandersia_laurifolia
Schlegelia_parviflora 1/89/92
0.73/52/55 Scrophularia_marilandica SmCYC2A2b
SmCYC2A2a
Scrophulariaceae 0.89/65/69 Scrophularia_orientalis
Verbascum_thapsus
Scrophularia_variegata SCYC2A2
0.99/76/79 Verbascum_chaixii
/-
0.75/–/– Scrophularia_orientalis
0.95/65/60 1/98/99
Scrophularia_marilandica SmCYC2A1
1/95/94 Verbascum_chaixii Scrophularia_variegata SCYC2A1
Buddleja_davidii
1/97/98 Lantana_velutina
0.83/70/59
Lippia_myriocephala
0.94/73/81 Lippia_nodiflora
1/90/98 Acantholippia_trifida
1/54/76

Verbenaceae
Verbena_canadensis
1/100/99 Glandularia_canadensis
Junellia_seriphioides
1/92/98 Citharexylum_schottii
Rehdera_trinervis
1/95/98 Citharexylum_schottii
0.92/89/95 Bouchea_fluminensis
1/84/88
Stachytarpheta_calderonii
Acanthaceae 1/100/99 Mendoncia_spp.

1/89/94 Thunbergia_erecta
1/91/96 Crossandra_infundibuliformis
Acanthus_mollis
Schaueria_calicotricha
Avicennia_germinans
Digomphia_densicoma
Sesamum_triphyllum HCL - CYC2B
1/99/95
Mentha_longifolia
1/92/96 Origanum_vulgare
0.82/–/– Perovskia_atriplicifolia
Lamiaceae 0.83/–/–
0.98/55
Ocimum_basilicum
Teucrium_chamaedrys
1/52/50
Premna_fulva
1/77/85 Gmelina_arborea
1/66/50
Scutellaria_incana
1/83/84 Scutellaria_lateriflora
1/86/89 1/99/95 Holmskioldia_sanguinea
Congea_tomentosa
Tectona_grandis
Callicarpa_cathayana Orobanchaceae
1/89/92
1/66/60 Agalinis_tenuifolia
Gerardia_grandiflora
Phrymaceae
0.94/86/88 Mimulus_ringens
1/72/77 Paulownia tomentosa Phryma_leptostachya
Schlegelia_parviflora
0.92/61/– Thomandersia_laurifolia
Scrophulariaceae 0.9/74/70 Scrophularia_orientalis
Scrophularia_marilandica
SCYC2B1
Scrophularia_marilandica SCYC2B2
Verbascum_thapsus
Buddleja_davidii
1/89/90
GCYC2 + CCYC2
1/88/92
CYC/DICH + VCYC2 + GoCYC2 Fig. 3a
1/99/97
Oleaceae_CYC2 + Tetrachondraceae_CYC2
0.2

at least seven duplication events are found in EDL whose corollas duplications likely occurred in CL (one each in Calceolariaceae,
are actinomorphic (Fig. 3a, one event in Tetrachondraceae Gesneriaceae, and HCL, and three in Plantaginaceae; Fig. 3a,b).
(Polypremum) and at least seven duplications in Oleaceae The duplication before the diversification of the HCL clade
(Noronhia, Fraxinus, Osmanthus, Abeliophyllum, and two (Fig. 3b, HCL-CYC2A and HCL-CYC2B) is not well supported
Jasminum species)). Meanwhile, at least six separate gene (< 50%). We tested this hypothesis with phylogenetic constraint

Ó 2014 The Authors New Phytologist (2015) 205: 852–868

New Phytologist Ó 2014 New Phytologist Trust www.newphytologist.com
 New
860 Research Phytologist

analysis using MrBayes by constraining the HCL-CYC2A and
HCL-CYC2B clades as sisters. The Bayes factor comparison
shows that the constrained tree (logeLc = 55 163.680) is not sig-
nificantly more strongly supported by the data (GTR + I + G)
than an unconstrained analysis (logeLu = 55 163.209; Fig. S3).
However, HCL-CYC2A and HCL-CYC2B are supported as sister
clades with 73 and 66% RAxML bootstrap support using two
reduced datasets that include only Lamiales sequences and fewer
missing data (Figs S4,S5).
In addition to six parallel gene duplications in CL, nested
additional duplications (Fig. 3b, gray bars and triangles) are
common within GCYC2A, GCYC2B, HCL-CYC2A and
HCL-CYC2B; in some cases these additional duplications fol-
lowed gene losses of one paralog, thus restoring copy number
back to two. Thunbergia mysorensis, Ruellia and Hygrophila
(Acanthaceae), and Pogostemon (Lamiaceae) lack CYC2B paralogs
(Fig. 3b), but maintain two copies of CYC2A genes, while
Lantana, Lippia, Verbena, Glandularia (Verbenaceae), and
Mentha (Lamiaceae) have no CYC2A paralogs but two CYC2B
paralogs (Fig. 3b).
Species with derived actinomorphic flowers in HCL (e.g.
Buddleja, Callicarpa, Tectona, and Avicennia; Fig. 3b, coded in
red) retain both HCL-CYC2A and HCL-CYC2B paralogs. Con-
versely, species with conspicuously zygomorphic corollas may
have only one type of CYC2-like gene as in Lamium and Stachys
(Lamiaceae) (with only HCL-CYC2A; Fig. S1).
on the HCL-CYC2A branch that immediately follows the gene
duplication event (Fig. 3b, HCL-CYC2A; Table 2, H1,
xHCLCYC2A = 1.842 18, xo = 0.205 44, P = 0.0095). We also
found evidence for x variation along the CCYC2-like paralog lin-
eages following duplications, in that both CCYC2A and
CCYC2B (Calceolariaceae) appear to be under purifying selection
(Table 2, H7, xCCYC2A = 0.109 05, xo = 0.207 31, P = 0.0387).
However, we did not find significant evidence for changes in x
along branches following CYC2-like gene duplications in Ges-
neriaceae or Plantaginaceae (Table 2).
HyPHY found six local branches that are under episodic
positive selection in Plantaginaceae (Fig. S6, EPS1–EPS3, dotted
lines) and the HCL clade (Fig. S6, EPS4–EPS6, dotted lines)
following more recent nested duplications of CYC2-like paralogs.
The TCP- and R- domains were found to be under strong
purifying selection by all models using HyPHY. One site in the
C-terminus of the genes (codon 197, E/D/Q/R/G/H/T) (Table
S3, Fig. S2) was found to be under positive selection using all
methods using HyPHY. Codon 192 (S/V/I/L/G/P/A) may be
under positive selection using site-specific models implemented
in PAML, but is not statistically significant (Table 3, M8). The
branch-site model found two sites (Table 3; codon 147, G/E/D
(Fig. 2, Motif 2) and codon 251, C/S/T (Fig. 2, Motif 7)) that
might be under positive selection in the HCL-CYC2A foreground
branch, though again this result is not significant (Table 3,
branch-site model A; Fig. S2).

Tests for selection Molecular dating

Branch-based models using PAML found varied rates of nonsyn- Results from 100 000 000 generations were used for discussion,
onymous vs synonymous nucleotide substitution (x = dN/dS) as all ESS from these runs are > 200. Bayes factor analysis found
following duplication in HCL, which indicates positive selection that the uced model (logeL = 57 059.1816) was a better fit for
Table 2 Parameter estimates under models of variable dN/dS ratios (x) along lineages following duplications of CYCLOIDEA2 (CYC2)-like genes in
Lamiales

Modela Estimates of parameters LogeL P-valueb

Hypothesis of one ratio H0: x x = 0.206 25 33 831.046

Hypothesis of two ratios H1: xHCLCYC2A, xo xHCLCYC2A = 1.842 18, xo = 0.205 44 33 827.686 0.0095**
H2: xHCLCYC2B, xo xHCLCYC2B = 0.513 55, xo = 0.205 99 33 830.620 0.3557
H3: xCYC, xo xCYC = 0.221 31, xo = 0.206 14 33 831.022 0.8239
H4: xDICH, xo xDICH = 0.4196, xo = 0.2059 33 830.563 0.3253
H5: xGCYC2, xo xGCYC2 = 0.198 05, xo = 0.206 31 33 831.040 0.9068
H6: xGCYC1, xo xGCYC1 = 0.238 09, xo = 0.205 98 33 830.927 0.625
H7: xCCYC2A, xo xCCYC2A = 0.109 05, xo = 0.207 31 33 828.910 0.0387*
H8: xCCYC2B, xo xCCYC2B = 0.169 34, xo = 0.206 55 33 830.871 0.5532
H9: xVCYC2A, xo xVCYC2A = 0.342 11, xo = 0.205 81 33 830.624 0.358
H10: xVCYC2B, xo xVCYC2B = 0.210 52, xo = 0.206 22 33 831.045 0.9531
H11: xGoCYC2A, xo xGoCYC2A = 0.302 82, xo = 0.205 53 33 830.271 0.2128
H12: xGoCYC2B, xo xGoCYC2B = 0.219 97, xo = 0.206 15 33 831.031 0.8598
Hypothesis of three ratios H13: xHCLCYC2A, xCCYC2A, xo xo = 0.2065, xHCLCYC2A = 1.845 31, 33 825.561 0.0096**
xCCYC2A = 0.108 83
a
The topology and branch-specific x values are shown in Fig. S6. xo, x ratios of ancestral lineages in each hypothesis; xi (i = one CYC2-like duplicate), prespeci-
fied x following gene duplications that might exhibit elevated rates of nonsynonymous vs synonymous nucleotide substitution. H1–H12 were tested
against the null hypothesis H0. H13 was against H7, to test whether we could still detect varied x values following HCL-CYC2 duplication given the x vari-
ation after CCYC2 duplication in Calceolariaceae.
b
The likelihood ratio test (LRT) was used to test the hypotheses of x variation following six CYC2-like gene duplication events in the lineages Calceolariaceae,
Gesneriaceae, Plantaginaceae-Antirrhineae, Plantaginaceae-Veronica, Plantaginaceae-Gratiola, and higher core Lamiales (HCL) across the core Lamiales
(CL; v2df = 1, a = 0.05 = 3.841; *, 5% significance level; **, 1% significance level). Values of x ratios > 1 and significant P-values are both highlighted in bold.

New Phytologist (2015) 205: 852–868 Ó 2014 The Authors

www.newphytologist.com New Phytologist Ó 2014 New Phytologist Trust
New
Phytologist Research 861

Table 3 Parameter estimates under models of site-specific and branch-site models of CYCLOIDEA2 (CYC2)-like genes in Lamiales

Modelsa #pb LogeL Estimates of parametersc Positively selected sitesd

Site-specific models
M0 (one ratio) 1 33 831.046 42 x = 0.206 25 None
M1a (nearly neutral) 2 33 027.434 93 p0 = 0.633 51, (p1 = 0.366 49); Not allowed
x0 = 0.130 76 (x1 = 1)
M2a (positive selection) 4 33 027.434 93 p0 = 0.633 51, p1 = 0.154 58, None
(p2 = 0.211 92);
x0 = 0.130 76, (x1 = 1), x2 = 1
M7 (beta) 2 32 446.597 69 P = 0.540 27, q = 1.348 10; Not allowed
M8 (beta and x) 4 324 46.599 58 p0 = 0.999 99, (p1 = 0.000 01), 192S (P = 0.594)
P = 0.540 29, q = 1.347 97,
x = 2.381 03
Branch-site models
Model A (foreground 4 33 027.007 88 p0 = 0.585 91, p1 = 0.337 39, Site for foreground lineage
x2 > 1) (p2a + p2b = 0.0767) (HCL-CYC2A): 147E (P = 0.515),
x0 = 0.130 69, x2 = 2.438 79 251S (P = 0.550)
Model A1 (foreground 3 33 027.070 60 p0 = 0.541 41, p1 = 0.311 88, Not allowed
x2 = 1) (p2a + p2b = 0.1467)
x0 = 0.130 69
a
Likelihood ratio tests compare M1a against M2a (df = 2), M7 against M8 (df = 2), and model A against model A1 (df = 1).
b
#p is the number of free parameters in the x distribution. Parameters in parentheses are not free and should not be counted but are presented for clarity
here.
c
x ratios > 1 indicate positive selection and are highlighted in bold.
d
Sites potentially under positive selection were identified using Congea tomentosa HCL-CYC2B as the reference. Results are from Bayes empirical Bayes
(BEB) analyses. P is the posterior probability for site class.

the data than the ucld model (logeL = 57 082.5751; Fig. S6).
Multiple parallel duplication events in major lineages (e.g. HCL,
Calceolariaceae, Gesneriaceae, Antirrhineae, Ruellieae) in CL
appear to have occurred around the K–Pg boundary (mean esti-
mates of the most recent common ancestor, 40.1–71.3 Mya)
(Fig. 4; e.g. HCL, Gesneriaceae, Calceolariaceae, Ruellieae, An-
tirrhineae). The paralogs that arose around these geological peri-
ods exhibit much higher sequence divergence (< 70% identity;
Table 1). By contrast, all seven duplications in EDL appear to
have occurred more recently (mean estimates < 23 Mya) (Figs 4,
S9) and the sequence divergence is much lower (Table 1; > 77%
identity in Polypremum CYC2-like genes).
Expression of CYC2-like genes in EDL and HCL
The results of gene expression for six focal species are shown in
Fig. 5(a–k). In three EDL species, J. angustifolium (Fig. 5b),
Ligustrum lucidum, and Syringa vulgaris (Fig. 5a; Oleaceae),
CYC2-like transcripts are barely detectable in petals (Fig. 5f,k).
By contrast, in our HCL representative species (Mimulus ringens
(Fig. 5c, Phrymaceae), Ruellia tweediana (Fig. 5e, Acanthaceae)
and Schaueria calicotricha (Fig. 5d, Acanthaceae)), one CYC2-like
paralog (HCL-CYC2A, RCYC2A1) is detected only in adaxial
and lateral petals, while the other paralog (HCL-CYC2B,
RCYC2A2) is widely transcribed in all petals (Fig. 5k). Our
qPCR further showed that both HCL-CYC2A and HCL-CYC2B
genes are differentially expressed along the floral adaxial/abaxial
axis, with much higher expression in the adaxial side (Fig. 5g–j).
The similar expression pattern in two Acanthaceae species
(Fig. 5k, HCL-CYC2A and HCL-CYC2B, RCYC2A1 and
RCYC2A2) is independently derived (Fig. 6). However, we were
unable to test this pattern further with qPCR because not enough
RNA/replicates of R. tweediana were available.
Discussion
CYC2-like gene duplication events are not correlated with
a single origin of corolla zygomorphy early in Lamiales
Gene duplication has long been postulated as an important pro-
cess in the generation of evolutionary novelty (e.g. Ohno, 1970;
Force et al., 1999). Phylogenetic investigations in distantly
related lineages of eudicots with zygomorphic corollas have
shown that CYC2-like genes have experienced multiple indepen-
dent duplications, hinting that duplication of CYC2-like genes is
required for the origin of corolla zygomorphy (Citerne et al.,
2000, 2003; Hileman & Baum, 2003; Zhang et al., 2010;
T€ahtiharju et al., 2012).
Our results indicate that duplications of CYC2-like genes are
not correlated with the origin of corolla zygomorphy, and that
CYC2-like genes have recently duplicated independently in spe-
cies of EDL whose corollas are actinomorphic (Figs 3a,4).
Indeed, duplication of CYC2-like genes may be neither necessary
nor sufficient for the evolution of zygomorphic corollas.
Zygomorphic corollas may occur in species that have only one
CYC2-like gene; this may be through cis-regulatory changes in
Brassicaceae (Busch et al., 2012) or via alternatively spliced prod-
ucts as shown in Lamioideae (Fig. S8) that might function redun-
dantly. Rather, it appears that more ancient parallel duplications
of CYC2-like genes around the K–Pg boundary (Fig. 4; e.g.

Ó 2014 The Authors New Phytologist (2015) 205: 852–868

New Phytologist Ó 2014 New Phytologist Trust www.newphytologist.com
 New
862 Research Phytologist

Fig. 4 Chronogram of Lamiales: maximum clade credibility tree from the BEAST analysis. Posterior estimates of divergence times were inferred using an
uncorrelated relaxed exponential clock (uced) model; minimum age constraints were provided by a Fraxinus (Oleaceae) fossil date of 37 million yr ago
(Mya; calibration C1), and a Caltapa (Bignoniaceae) fossil date of 28.4 Mya (calibration C2). Two additional calibration points were derived from fossil
information for the nodes Verbenaceae/Bignoniaceae (C3 and C4). Some nodes are collapsed and relabeled at the tips for clarity, thick branches have
support values of 100 Bayesian posterior probability; lower values are labeled above the branches. Black vertical bars point to major duplication events in
major taxonomic groups (Calceolariaceae, Gesneriaceae, Plantaginaceae, and higher core Lamiales (HCL)), white bars are nested duplications, and gray
triangles are additional duplications for which paralogs are collapsed. The violet bars represent the 95% highest posterior density (HPD) interval for the
divergence time estimates. The geological timescale is drawn at the bottom for reference. Posterior probabilities > 0.5 are labeled above the nodes. Six
major duplications in core Lamiales (CL) are labeled. Mean estimates for the most recent common ancestor of parallel duplications of CYC2-like genes in
major lineages in core Lamiales (Gesneriaceae, Calceolariaceae, Antirrhineae, HCL and Ruellieae) ranged from 40.1 to 71.3 Mya, around the Cretaceous–
Paleogene (K–Pg) boundary. Conversely, the mean estimates for the most recent common ancestor of duplications in EDL appear to be much more recent
(< 23 Mya, e.g. Polypremum paralogs 22.1 Mya).

New Phytologist (2015) 205: 852–868 Ó 2014 The Authors

www.newphytologist.com New Phytologist Ó 2014 New Phytologist Trust
New
Phytologist Research 863

(a) (b) (c) (d) (e)

dp dp
dp
lp lp
lp
petal lp lp lp
petal
bp bp bp

(f) 0.0016 SvCYC2 gene expression (g) 0.010 MrCYC2A gene expression (h) 0.016 MrCYC2B gene expression
Relative expression level

Relative expression level

Relative expression level

0.0014 0.014
0.008
0.0012 0.012
0.0010 0.006 0.010
0.0008 0.008
0.0006 0.004 0.006
0.0004 0.004
0.002
0.0002 0.002
0.0000 0.000 0.000
infl petal Leaf dp lp bp Leaf dp lp bp Leaf

(i) 0.020 ScCYC2A gene expression (j) 0.00035 ScCYC2B gene expression
Relative expression level
Relative expression level

0.00030

0.015 0.00025

0.00020
0.010
0.00015

0.00010
0.005
0.00005

0.000 0.00000
dp lp bp Leaf dp lp bp Leaf

(k)

Fig. 5 Gene expression of CYC2-like paralogs. (a–e) Flowers of focal species for the study of gene expression: (a) Syringa vulgaris and Ligustrum lucidum
(not shown) (Oleaceae, early diverging Lamiales (EDL)) have an actinomorphic corolla with four petals; (b) Jasminum angustifolium (Oleaceae, EDL) has
an actinomorphic corolla with five or six petals; (c–e) Mimulus ringens (c), Schaueria calicotricha (d) and Ruellia tweediana (e) are sampled from higher
core Lamiales (HCL) and all have zygomorphic corollas. (f–j) Gene expression detected using quantitative PCR (qPCR) and flowers of focal species (the bars
show means + SD of three biological replicates): (f) Syringa vulgaris; (g, h) Mimulus ringens; (i, j) Schaueria calicotricha. (k) expression pattern detected by
reverse transcription polymerase chain reaction (RT-PCR). infl, inflorescence; dp, adaxial petals; lp, lateral petals; bp, abaxial petals.

HCL, Gesneriaceae, Calceolariaceae, Ruellieae, Antirrhineae), recently (Figs 4, S9), and the sequence divergence is much lower
and considerable changes in coding (Fig. 2; Table 1) and cis-regu- (Table 1). Our protein motif analyses suggest the structure and
latory regions (Yang et al., 2012) are critical for the origin and probably the function of CYC2-like proteins in Oleaceae and
formation of the complex zygomorphic corollas as in Lamiales. Tetrachondraceae are quite different from those of the proteins
The more ancient paralogs that arose around the K–Pg boundary in CL (Fig. 2). Acquisition of TCP recognition sites in the regu-
in CL exhibit greater sequence divergence (Table 1), and also latory regions of CYC2-like genes is critical for persistent expres-
considerable changes in the overall structure of the conserved sion of CYC2-like genes in the adaxial and lateral floral parts
motifs and lineage-specific amino acid replacements (Fig. 2). By during later developmental stages in Gesneriaceae (Yang et al.,
contrast, the duplications in EDL probably occurred much more 2012), but the importance of this feedback loop in other lineages

Ó 2014 The Authors New Phytologist (2015) 205: 852–868

New Phytologist Ó 2014 New Phytologist Trust www.newphytologist.com
 New
864 Research Phytologist

RuelliaTWECYC2A1
HygrophilaCORCYC2A1 Pervasive duplication and losses of CYC2-like genes and
RuelliaTWECYC2A2
HygrophilaCORCYC2A2
shifts in corolla symmetry in the HCL clade
SchaueriaCALCYC2A
MimulusRINCYC2A Retention of CYC2-like paralogs is not necessary for the mainte-
LamioideaeCYC2A
ScrophulariaMARCYC2A1
nance of corolla zygomorphy. Species having all paralogs of
HCL ScrophulariaORICYC2A1 CYC2-like genes may have actinomorphic corollas (e.g. Buddleja,
ScrophulariaMARCYC2A2
ScrophulariaORICYC2A2 Callicarpa, Tectona and Avicennia; Fig. 3b), while species having
BuddlejaDAVCYC2A only one type of CYC2-like gene may have either strongly zygo-
MimulusRINCYC2B
SchaueriaCALCYC2B morphic (e.g. HCL-CYC2A, Stachys, Lamium, Ruellia and
ScrophulariaORICYC2B Hygrophila) or weakly actinomorphic corollas (e.g. HCL-CYC2B,
ScrophulariaMARCYC2B
BuddlejaDAVCYC2B Mentha, Lantana, Lippia, and Verbena; Fig. 3b). Nested duplica-
MohaveaDICH1
MohaveaDICH2 tion of CYC2-like paralogs is pervasive (Fig. 3a,b). These
AntirrhinumDICH subsequent additional duplications sometimes restore the num-
MohaveaCYC1
dp dp MohaveaCYC2 ber of CYC2-like paralogs from a gene loss to two copies (e.g.
lp lp AntirrhinumCYC two HCL-CYC2As in Scrophularia variegata, Ruellia and
GratiolaCYC1
GratiolaCYC2 Hygrophila, and two HCL-CYC2Bs in Mentha and Lippia;
bp CL GratiolaCYC3
VeronicaCYC2 Fig. 3b).
VeronicaCYC1 Stachys and Lamium (Lamoideae) have only one CYC2-like
OpithandraCYC1C
PrimulinaCYC1C gene (CYC2A; Fig. S1) but strongly zygomorphic corollas. Inter-
OpithandraCYC1D estingly, the one CYC2A gene has acquired an intron; this could
PrimulinaCYC1D
BourneaCYC1 potentially allow alternative splicing to produce two types of
OpithandraCYC2A
PrimulinaCYC2A CYC2-like transcripts to help maintain the conspicuously zygo-
OpithandraCYC2B morphic corolla in this lineage. Multiple possible splice sites that
PrimulinaCYC2B
BourneaCYC2 allow alternative splicing were predicted using an online program
CCYC2A (http://wangcomputing.com/assp/index.html), and there might
CCYC2B
EDL PpCYC2A be only two amino acid differences between alternatively spliced
PpCYC2B
JnCYC2A transcripts (Fig. S8). Indeed, introns of different length in two
EDL JnCYC2B LCYC2As of the species Pogostemon heyneanus are found at similar
JaCYC2
SvCYC2 positions (30 - of the R domain) (Lamioideae) and exhibit only
LlCYC2 seven nucleotide differences (Fig. S8). Although we did not detect
Fig. 6 Summary of extensive independent and nested duplications of any alternatively spliced transcripts of CYC2A using RT-PCR,
CYC2-like genes in both early diverging Lamiales (EDL) and core Lamiales alternative splicing cannot be ruled out. Certain transcripts could
(CL) lineages, and conservation and diversification of CYC2-like gene
exhibit low expression levels, our primers might have preferen-
expression in Lamiales. Vertical double lines indicate duplications of CYC2-
like genes. The oblique double line indicates the acquisition of an intron of tially amplified a highly expressed isoform or we might have failed
the paralogs (Citerne et al., 2000; Hileman & Baum, 2003; Zhou et al., to sequence the clones with alternatively spliced transcripts.
2008; Preston et al., 2009; Yang et al., 2012). Expression of color-coded The evolutionary history of CYC2-like genes in HCL appears
paralog names detected using quantitative PCR (qPCR), in situ quite complex, with the lineage-specific acquisition of introns,
hybridization and/or reverse transcription polymerase chain reaction (RT-
and extensive parallel and nested duplication events of certain
PCR). Red-coded paralogs represent the expression in adaxial/lateral
petals, green-coded paralogs exhibit broad expression in all petals, paralogs, the latter sometimes probably following gene losses.
whereas gray-coded paralogs represent lack of expression in petals (Luo However, how these changes in gene number and structure relate
et al., 1996, 1999; Hileman et al., 2003; Zhou et al., 2008; Preston et al., to shifts in corolla symmetry remains unclear (but see the later
2009; Yang et al., 2012). No expression data were available for the discussion on the expression of RCYC2As of a nested duplica-
paralogs in black. Simplified floral diagrams (actinomorphy vs
tion). Further finer-scale analyses and functional characterization
zygomorphy) were drawn to represent the asymmetrical expression
pattern of CYC2-like genes along the adaxial/abaxial floral axis. CYC2-like of CYC2-like genes in HCL may help us to better understand this
genes were highly expressed in the adaxial petals (dp, black) with much complex evolutionary history of CYC2-like genes and their possi-
lower or barely detectable expression in lateral (lp) and/or abaxial petals ble roles in the evolution of corolla symmetry.
(bp, gray). Asymmetrical expression of CYC2-like genes along adaxial/
abaxial floral axis in CL appears to be correlated with the corolla transition
from actinomorphy in EDL to zygomorphy in CL. Changes in CYC2-like Selection test and retention of CYC2-like duplicates
paralog expression are correlated with a transition in corolla symmetry
and/or floral modification (e.g. Mohavea, Bournea, and Veronica; Multiple selection schemes may have facilitated preservation of
Hileman et al., 2003; Zhou et al., 2008; Preston et al., 2009). CYC2-like duplicates in CL. Our analyses found significant evi-
dence of episodic changes but not a long-term shift in selection
is unclear. Therefore, parallel, more ancient duplications of pressure on the HCL-CYC2A branch that immediately follows
CYC2-like genes and both cis- and trans- changes in CYC2-like the gene duplication event (Table 2, H1 vs H0) and varied values
genes are likely to play important roles in the evolution of com- of dN/dS (x) on CCYC2A and CCYC2B branches following the
plex corolla zygomorphy in Lamiales. gene duplication event in Calceolariaceae (Table 2, H7 vs H0).

New Phytologist (2015) 205: 852–868 Ó 2014 The Authors

www.newphytologist.com New Phytologist Ó 2014 New Phytologist Trust
New
Phytologist
But our analyses found no significant evidence for different x
along branches following duplications in Gesneriaceae or Planta-
ginaceae (Table 2). Indeed, a previous study in snapdragon and
its relatives (Antirrhineae) on the CYC/DICH duplication found
no positive selection on CYC/DICH, but showed less constraint
in the DICH lineage than in the CYC lineage by purifying selec-
tion (Hileman & Baum, 2003). Furthermore, a REL model
found that three branches following two separate duplications of
CYC2-like genes in Plantaginaceae (Fig. S6; EPS1 in GoCYC2As,
EPS2–EPS3 in CYC) and an additional three branches in HCL
(Fig. S6; EPS4–EPS6) appear under episodic positive selection.
All site-specific models found strong evidence for purifying
selection in the TCP domains, as in a previous study (Hileman &
Baum, 2003), which suggests strong constraint in the functional
domain that is important in (homo- and/or hetero-) dimerization
and DNA recognition/binding of TCP proteins. Several sites
(Table 3; Fig. S2) could be under positive selection, although sta-
tistical support is low. The signal of positive selection could be
obscured by nested duplications producing paralogs that subse-
quently diverged. For example, position 147 in most HCL-CYC2A
is glycine and is under positive selection. By contrast, the amino
acid at this position of Ruellieae CYC2As may be either G
(RCYC2A2) or an acidic amino acid D/E (RCYC2A1; Fig. S2).
Furthermore, changes may occur at sites under positive selection in
the species with derived corolla actinomorphy, which might also
mask such sites in the selection test. For instance, at position 147
in HCL-CYC2A, species of Buddleja and Tectona with actinomor-
phic corollas both have an acidic amino acid D/E (Fig. S2). The
roles of the conserved domains within which we found sites under
positive selection (Motifs, 3, 5, 7) are poorly known; one possible
role could be interactions with other transcription factors.
Conservation and diversification of CYC2-like gene expres-
sion
Although the shift from corolla actinomorphy to zygomorphy is
not correlated with increased numbers of copies of CYC2-like
genes in CL, it does correlate with distinct expression patterns of
CYC2-like paralogs between EDL and CL (Luo et al., 1996,
1999; Hileman et al., 2003; Yang et al., 2012; Fig. 5). CYC2-like
genes in three sampled Oleaceae species are transcribed in both
vegetative and reproductive tissues, but barely in petals (Fig. 5;
J. angustifolium, L. lucidum and S. vulgaris). Lack of expression of
CYC2-like genes in later developmental stages in Oleaceae species
may represent the ancestral pattern of CYC2-like genes in Lami-
ales, and could thus account for floral actinomorphy in the Olea-
ceae. Our qPCR data show high expression of CYC2-like
paralogs (CYC2A and CYC2B) in the adaxial corolla lobes in
sampled HCL species. This asymmetric expression of CYC2-like
genes in adaxial petals also occurs in P. heterotricha (PhCYC1C,
PhCYC1D, Gesneriaceae) and snapdragon (CYC and DICH)
despite their different duplication origins, hinting at their con-
served roles in determining the polarity along the adaxial/abaxial
floral axis. This conserved asymmetrical expression pattern of
CYC2-like paralogs may thus have played a critical role in the
development of corolla zygomorphy in CL (Figs 5,6).
Ó 2014 The Authors
New Phytologist Ó 2014 New Phytologist Trust
Research 865
Divergent expression of paralogs after a duplication event is
found in all duplication events in lineages in CL, and the expres-
sion of paralogs varies to some extent with their separate duplica-
tion origin (Fig. 6; Luo et al., 1996, 1999; Gao et al., 2008;
Preston & Hileman, 2009). P. heterotricha (Gesneriaceae) has
two major types and four copies of ECE-CYC2 genes (PhCYC1C,
PhCYC1D, PhCYC2A and PhCYC2B), none of which have tran-
scripts in abaxial petals. PhCYC2A and PhCYC2B paralogs are
not expressed in petals at all (Gao et al., 2008), while PhCYC1C
and PhCYC1D are expressed asymmetrically, with PhCYC1D
restricted to adaxial petals and PhCYC1C in both adaxial and
lateral petals (Gao et al., 2008). In Antirrhinum majus (Plantagin-
aceae), both copies of CYC2-like genes (CYC/DICH) are
expressed only in adaxial petals, but not in abaxial petals (Luo
et al., 1996, 1999). Likewise, GoCYC and VmCYC paralogs each
have distinct expression patterns, consistent with their derivation
from separate duplications (Fig. 6; Preston et al., 2009). The
expanded expression of certain CYC2-like paralogs in Gesneria-
ceae and Plantaginaceae may not correlate with changes in corolla
symmetry, but rather with changes in other aspects in floral mor-
phology (e.g. internal symmetry of adaxial petals, development of
stamens; Hileman et al., 2003; Zhou et al., 2008; Preston &
Hileman, 2009; Song et al., 2009). Likewise, CYC2-like genes
(CYC2B) are expressed in the abaxial corolla in our sample of
HCL species, a pattern uncorrelated with the transition in floral
symmetry or other aspects in floral morphology. Unlike relatively
even levels of expression of VmCYC2 in Veronica, the expanded
expression of HCL-CYC2B in abaxial petals is much lower com-
pared with the levels in adaxial corollas shown by qPCR (Fig. 5h,
j). Our data and previous results thus indicate diversification of
CYC2-like gene expression following extensive duplication events
(e.g. Luo et al., 1996, 1999; Zhou et al., 2008; Preston &
Hileman, 2009; Song et al., 2009).
The origins of corolla zygomorphy in Lamiales and the
initial diversification of core major bee clades
Character optimization points to a single major floral transition
from corolla actinomorphy to zygomorphy at the base of CL,
although weakly zygomorphic corollas are also occasionally
recorded in EDL (Fig. 1, Carlemanniaceae) and the sister-taxon
Solanales (e.g. Petunia 9 hybrid, Solanaceae). In addition, some
weakly zygomorphic corollas are quite incidental and even unsta-
ble within individual plants (e.g. J. angustifolium). However,
what appears to be a single origin of corolla zygomorphy in CL
may actually be underlain by at least six independent duplications
of CYC2-like genes at different time periods (40.1–71.3 Mya;
Figs 3a,b,4). The evolutionary transition from corolla actinomor-
phy to zygomorphy early in Lamiales might thus represent
multiple cryptic parallel origins of corolla zygomorphy. Alterna-
tively, zygomorphic corollas might have originated before the
extensive duplications of CYC2-like genes; that is, the single copy
of CYC2-like genes in EDL could have provided the initial
molecular signal for corolla zygomorphy as found in some Brass-
icaceae species (e.g. Cubas et al., 2001; Busch & Zachgo, 2007).
Extensive duplications of CYC2-like genes followed by
New Phytologist (2015) 205: 852–868
www.newphytologist.com

866 Research
considerable divergence in sequences and expression patterns of
the paralogs in CL might thus have greatly reinforced the initial
event. Indeed, the CYC2-like genes are expressed asymmetrically
in the adaxial region in Arabidopsis at very early developmental
stages. It cannot be ruled out that CYC2-like genes may be
expressed asymmetrically in early developmental stages in EDL
species (Table 1; Figs 3a,b,6).
The parallel duplications of CYC2-like genes occurred after
the initial diversification of bumble bees and Euglossine bees.
Corolla zygomorphy has long been thought to have evolved
repeatedly from actinomorphy under strong selection by special-
ized pollinators. Zygomorphic corollas conceal and protect sta-
mens within the corolla and thus allow more precise pollination
and/or reduce pollen wastage (Neal et al., 1998; Westerkamp &
Claßen-Bockhoff, 2007). Zygomorphy is specifically associated
with bee pollination, particularly Euglossine bees and bumble
bees, either through learning or innate preference (Møller, 1995;
Neal et al., 1998; Rodr
ıguez et al., 2004; Westerkamp &
Claßen-Bockhoff, 2007). Fossil-calibrated molecular dating
analyses showed that most bee lineages diversified around the
K–Pg boundary (Cardinal & Danforth, 2013). Comparison of
our mean age estimates of major parallel duplications of CYC2-
like genes in CL (40.1–71.3 Mya) and previous estimates for the
diversification of bees indicate that the origin, stabilization, and/
or reinforcement of corolla zygomorphy may follow the initial
diversification of major core bee lineages, especially those that
include Euglossine bees and bumble bees (65–75 Mya; Cardinal
& Danforth, 2013).
Acknowledgements
We thank Cynthia Hong-Wa, George Yatskievych, the Missouri
Botanical Garden, the Shaw Nature Reserve, and the Plant
Phylogenetics and Conservation Biology Laboratory in
Xishuangbanna Tropical Botanical Garden for help in sampling,
and Wang lab and Geliang Wang at UMSL for assistance with
the qPCR. Thanks are also due to Peter Stevens, Jill Preston, and
three anonymous reviewers for their constructive comments on
the manuscript. This work was supported by a Doctoral Disserta-
tion Improvement Grant from the National Science Foundation
(grant number DEB 1210540 to E.A.K. and J.Z.), the Jane Har-
ris Scholarship in Tropical Botany from Whitney R. Harris
World Ecology Center at the University of Missouri, St Louis, a
Graduate Student Research Award from the Botanical Society of
America, and an Alumni Fellowship from the Missouri Botanical
Garden.
References
Bailey TL, Elkan C 1994. Fitting a mixture model by expectation maximization
to discover motifs in biopolymers. Proceedings of the Second International
Conference on Intelligent Systems for Molecular Biology. Menlo Park, CA, USA:
AAAI Press, 28–36.
Bartlett ME, Specht CD. 2011. Changes in expression pattern of the teosinte
branched1-like genes in the Zingiberales provide a mechanism for
evolutionary shifts in symmetry across the order. American Journal of Botany
98: 227–243.
New Phytologist (2015) 205: 852–868
www.newphytologist.com
New
Phytologist
Broholm SK, T€a htiharju S, Laitinen RAE, Albert VA, Teeri TH, Elomaa P.
2008. A TCP domain transcription factor controls flower type specification
along the radial axis of the Gerbera (Asteraceae) inflorescence. Proceedings of the
National Academy of Sciences, USA 105: 9117–9122.
Busch A, Horn S, M€ uhlhausen A, Mummenhoff K, Zachgo S. 2012. Corolla
monosymmetry: evolution of a morphological novelty in the Brassicaceae
family. Molecular Biology and Evolution 29: 1241–1254.
Busch A, Zachgo S. 2007. Control of corolla monosymmetry in the Brassicaceae
Iberis amara. Proceedings of the National Academy of Sciences, USA 104: 16714–
16719.
Cardinal S, Danforth BN. 2013. Bees diversified in the age of eudicots.
Proceedings. Biological sciences/The Royal Society 280: 20122686.
Citerne HL, Luo D, Pennington RT, Coen E, Cronk QCB. 2003. A
phylogenomic investigation of CYCLOIDEA-like TCP genes in the
Leguminosae. Plant Physiology 131: 1042–1053.
Citerne HL, M€oller M, Cronk QCB. 2000. Diversity of cycloidea-like genes in
Gesneriaceae in relation to floral symmetry. Annals of Botany 86: 167–176.
Corley SB, Carpenter R, Copsey L, Coen E. 2005. Floral asymmetry involves an
interplay between TCP and MYB transcription factors in Antirrhinum.
Proceedings of the National Academy of Sciences, USA 102: 5068–5073.
Cubas P, Coen E, Zapater JMM. 2001. Ancient asymmetries in the evolution of
flowers. Current Biology 11: 1050–1052.
Damerval C, Le Guilloux M, Jager M, Charon C. 2007. Diversity and evolution
of CYCLOIDEA-like TCP genes in relation to flower development in
Papaveraceae. Plant Physiology 143: 759–772.
Darriba D, Taboada GL, Doallo R, Posada D. 2012. jModelTest 2: more
models, new heuristics and parallel computing. Nature Methods 9: 772.
Donoghue MJ, Ree RH, Baum DA. 1998. Phylogeny and the evolution of flower
symmetry in the Asteridae. Trends in Plant Science 3: 311–317.
Drummond AJ, Ho SYW, Phillips MJ, Rambaut A. 2006. Relaxed
phylogenetics and dating with confidence. PLoS Biology 4: e88.
Drummond AJ, Suchard MA, Xie D, Rambaut A. 2012. Bayesian phylogenetics
with BEAUti and the BEAST 1.7. Molecular Biology and Evolution 29: 1969–
1973.
Endress PK. 1999. Symmetry in flowers: diversity and evolution. International
Journal of Plant Sciences 160: S3–S23.
Endress PK. 2001. Evolution of floral symmetry. Current Opinion in Plant
Biology 4: 86–91.
Feng X, Zhao Z, Tian Z, Xu S, Luo Y, Cai Z, Wang Y, Yang J, Wang Z, Weng L
et al. 2006. Control of petal shape and floral zygomorphy in Lotus japonicus.
Proceedings of the National Academy of Sciences, USA 103: 4970–4975.
Force A, Lynch M, Pickett FB, Amores A, Yan Y, Postlethwait J. 1999.
Preservation of duplicate genes by complementary, degenerative mutations.
Genetics 151: 1531–1545.
Galego L, Almeida J. 2002. Role of DIVARICATA in the control of dorsoventral
asymmetry in Antirrhinum flowers. Genes & Development 16: 880–891.
Gao Q, Tao J-H, Yan D, Wang Y-Z, Li Z-Y. 2008. Expression differentiation of
CYC2-like floral symmetry genes correlated with their protein sequence
divergence in Chirita heterotricha (Gesneriaceae). Development Genes and
Evolution 218: 341–351.
G
omez JM, Perfectti F, Camacho JPM. 2006. Natural selection on Erysimum
mediohispanicum flower shape: insights into the evolution of zygomorphy. The
American Naturalist 168: 531–545.
Gouy M, Guindon S, Gascuel O. 2010. SeaView version 4: a multiplatform
graphical user interface for sequence alignment and phylogenetic tree building.
Molecular Biology and Evolution 27: 221–224.
Guindon S, Dufayard J-F, Lefort V, Anisimova M, Hordijk W, Gascuel O.
2010. New algorithms and methods to estimate maximum-likelihood
phylogenies: assessing the performance of PhyML 3.0. Systematic Biology 59:
307–321.
Hileman LC, Baum DA. 2003. Why do paralogs persist? Molecular evolution of
CYCLOIDEA and related floral symmetry genes in Antirrhineae
(Veronicaceae). Molecular Biology and Evolution 20: 591–600.
Hileman LC, Kramer EM, Baum DA. 2003. Differential regulation of symmetry
genes and the evolution of floral morphologies. Proceedings of the National
Academy of Sciences, USA 100: 12814–12819.
Ó 2014 The Authors
New Phytologist Ó 2014 New Phytologist Trust
New
Phytologist
Howarth DG, Donoghue MJ. 2005. Duplications in CYC-like genes from
Dipsacales correlate with floral form. International Journal of Plant Sciences 166:
357–370.
Howarth DG, Donoghue MJ. 2006. Phylogenetic analysis of the ‘ECE’ (CYC/
TB1) clade reveals duplications predating the core eudicots. Proceedings of the
National Academy of Sciences, USA 103: 9101–9106.
Howarth DG, Martins T, Chimney E, Donoghue MJ. 2011. Diversification of
CYCLOIDEA expression in the evolution of bilateral flower symmetry in
Caprifoliaceae and Lonicera (Dipsacales). Annals of Botany 107: 1521–1532.
Kadereit JW. 2004. Lamiales: introduction and conspectus. In: Kadereit JW, ed.
VII. Flowering plants Dicotyledons Lamiales (except Acanthaceae including
Avicenniaceae). Berlin, Germany: Springer Verlag, 1–4.
Kass RE, Raftery AE. 1995. Bayes factors. Journal of the American Statistical
Association 90: 773–795.
Katoh K, Standley DM. 2013. MAFFT multiple sequence alignment software
version 7: improvements in performance and usability. Molecular Biology and
Evolution 30: 772–780.
Kosakovsky Pond SL, Frost SDW. 2005a. Not so different after all: a
comparison of methods for detecting amino acid sites under selection.
Molecular Biology and Evolution 22: 1208–1222.
Kosakovsky Pond SL, Frost SDW. 2005b. Datamonkey: rapid detection of
selective pressure on individual sites of codon alignments. Bioinformatics 21:
2531–2533.
Kosakovsky Pond SL, Frost SDW, Muse SV. 2005. HyPHY: hypothesis testing
using phylogenies. Bioinformatics 21: 676–679.
Kosakovsky Pond SL, Murrell B, Fourment M, Frost SDW, Delport W,
Scheffler K. 2011. A random effects branch-site model for detecting episodic
diversifying selection. Molecular Biology and Evolution 28: 3033–3043.
Kosugi S, Ohashi Y. 2002. DNA binding and dimerization specificity and
potential targets for the TCP protein family. Plant Journal 30: 337–348.
Ludwig T, Lindermeier M, Stamatakis A, Rackl G. 2002. Tool environments in
CORBA-based medical high-performance computing. Future Generation
Computer Systems 18: 841–847.
Luo D, Carpenter R, Copsey L, Vincent C, Clark J, Coen E. 1999. Control of
organ asymmetry in flowers of Antirrhinum. Cell 99: 367–376.
Luo D, Carpenter R, Vincent C, Copsey L, Coen E. 1996. Origin of floral
asymmetry in Antirrhinum. Nature 383: 794–799.
Malcomber ST, Kellogg EA. 2004. Heterogeneous expression patterns and
separate roles of the SEPALLATA gene LEAFY HULL STERILE1 in grasses.
Plant Cell 16: 1692–1706.
Manchester SR. 1999. Biogeographical relationships of North American tertiary
floras. Annals of the Missouri Botanical Garden 86: 472–522.
McDade LA, Daniel TF, Kiel CA, Borg AJ. 2012. Phylogenetic placement,
delimitation, and relationships among genera of the enigmatic Nelsonioideae
(Lamiales: Acanthaceae). Taxon 61: 637–651.
Møller AP. 1995. Bumblebee preference for symmetrical flowers. Proceedings of
the National Academy of Sciences, USA 92: 2288–2292.
Mondrag
on-Palomino M, Theißen G. 2009. Why are orchid flowers so diverse?
Reduction of evolutionary constraints by paralogues of class B floral homeotic
genes. Annals of Botany 104: 583–594.
Neal PR, Dafni A, Giurfa M. 1998. Floral symmetry and its role in
plant-pollinator systems: terminology, distribution, and hypotheses. Annual
Review of Ecology and Systematics 29: 345–373.
Nie Z-L, Sun H, Beardsley PM, Olmstead RG, Wen J. 2006. Evolution of
biogeographic disjunction between eastern Asia and eastern North America in
Phryma (Phrymaceae). American Journal of Botany 93: 1343–1356.
Ohno S. 1970. Evolution by gene duplication. New York, NY, USA: Springer
Verlag.
Preston JC, Hileman LC. 2009. Developmental genetics of floral symmetry
evolution. Trends in Plant Science 14: 147–154.
Preston JC, Hileman LC. 2012. Parallel evolution of TCP and B-class genes in
Commelinaceae flower bilateral symmetry. EvoDevo 3: 6.
Preston JC, Kost MA, Hileman LC. 2009. Conservation and diversification of
the symmetry developmental program among close relatives of snapdragon
with divergent floral morphologies. New Phytologist 182: 751–762.
Rambaut A, Drummond AJ. 2009. Tracer v1.5. [WWW document] URL http://
beast.bio.ed.ac.uk/tracer [accessed 17 February 2013].
Ó 2014 The Authors
New Phytologist Ó 2014 New Phytologist Trust
Research 867
Ree RH, Donoghue MJ. 1999. Inferring rates of change in flower symmetry in
asterid angiosperms. Systematic Biology 48: 633–641.
Rodr
ıguez I, Gumbert A, de Ibarra NH, Kunze J, Giurfa M. 2004. Symmetry is
in the eye of the ‘beeholder’: innate preference for bilateral symmetry in
flower-na€ıve bumblebees. Naturwissenschaften 91: 374–377.
Ronquist F, Teslenko M, van der Mark P, Ayres DL, Darling A, H€ohna S,
Larget B, Liu L, Suchard MA, Huelsenbeck JP. 2012. MrBayes 3.2: efficient
Bayesian phylogenetic inference and model choice across a large model space.
Systematic Biology 61: 539–542.
Sargent RD. 2004. Floral symmetry affects speciation rates in angiosperms.
Proceedings of the Royal Society of London. Series B: Biological Sciences 271:
603–608.
Sch€aferhoff B, Fleischmann A, Fischer E, Albach DC, Borsch T, Heubl G,
M€ uller KF. 2010. Towards resolving Lamiales relationships: insights
from rapidly evolving chloroplast sequences. BMC Evolutionary Biology 10:
352.
Song CF, Lin QB, Liang RH, Wang YZ. 2009. Expressions of ECE-CYC2 clade
genes relating to abortion of both dorsal and ventral stamens in Opithandra
(Gesneriaceae). BMC Evolutionary Biology 9: 244.
T€a htiharju S, Rijpkema AS, Vetterli A, Albert VA, Teeri TH, Elomaa P. 2012.
Evolution and diversification of the CYC/TB1 gene family in Asteraceae –
a comparative study in Gerbera (Mutisieae) and sunflower (Heliantheae).
Molecular Biology and Evolution 29: 1155–1166.
Tamura K, Peterson D, Peterson N, Stecher G, Nei M, Kumar S. 2011.
MEGA5: molecular evolutionary genetics analysis using maximum likelihood,
evolutionary distance, and maximum parsimony methods. Molecular Biology
and Evolution 28: 2731–2739.
The Angiosperm Phylogeny Group. 2009. An update of the Angiosperm
Phylogeny Group classification for the orders and families of flowering plants:
APG III. Botanical Journal of the Linnean Society 161: 105–121.
Wang Z, Luo Y, Li X, Wang L, Xu S, Yang J, Weng L, Sato S, Tabata S,
Ambrose M. 2008. Genetic control of floral zygomorphy in pea (Pisum sativum
L.). Proceedings of the National Academy of Sciences, USA 105: 10414–10419.
Westerkamp C, Claßen-Bockhoff R. 2007. Bilabiate flowers: the ultimate
response to bees? Annals of Botany 100: 361–374.
Yang X, Pang H-B, Liu B-L, Qiu Z-J, Gao Q, Wei L, Dong Y, Wang Y-Z.
2012. Evolution of double positive autoregulatory feedback loops in
CYCLOIDEA2 clade genes is associated with the origin of floral zygomorphy.
Plant Cell 24: 1834–1847.
Yang Z. 1998. Likelihood ratio tests for detecting positive selection and
application to primate lysozyme evolution. Molecular Biology and Evolution 15:
568–573.
Yang Z. 2007. PAML 4: phylogenetic analysis by maximum likelihood. Molecular
Biology and Evolution 24: 1586–1591.
Yang Z, Nielsen R. 1998. Synonymous and nonsynonymous rate variation
in nuclear genes of mammals. Journal of Molecular Evolution 46:
409–418.
Zhang W, Kramer EM, Davis CC. 2010. Floral symmetry genes and the origin
and maintenance of zygomorphy in a plant-pollinator mutualism. Proceedings
of the National Academy of Sciences, USA 107: 6388–6393.
Zhou XR, Wang YZ, Smith JF, Chen R. 2008. Altered expression patterns of
TCP and MYB genes relating to the floral developmental transition from initial
zygomorphy to actinomorphy in Bournea (Gesneriaceae). New Phytologist 178:
532–543.
Supporting Information
Additional supporting information may be found in the online
version of this article.
Fig. S1 Southern blots in species with a single CYC2-like gene,
verifying the lack of paralogs.
Fig. S2 MEME analyses of CYC2-like proteins in Lamiales.
New Phytologist (2015) 205: 852–868
www.newphytologist.com
 New
868 Research Phytologist

Fig. S3 Phylogram from Bayesian inference of ECE-CYC Fig. S9 Chronogram of Lamiales showing only early diverging
sequences across the eudicots. Lamiales.

Fig. S4 Bayesian phylogeny of constrained dataset enforcing
monophyly of HCL.
Fig. S5 Phylogeny of CYC2-like genes (only 195 Lamiales
sequences included) rooted with Oleaceae using RAxML with
100 bootstrap replicates.
Table S1 Sampled taxa and GenBank accession number
Table S2 Primers used in this study
Table S3 Summary of selection test results using HYPHY in the
Datamonkey web server

Fig. S6 Phylogeny of CYC2-like genes (only 154 Lamiales
sequences included) rooted with Oleaceae using RAxML with
100 bootstrap replicates.
Fig. S7 Results from BEAST dating analyses using uncorrelated
relaxed log-normal clock (ucld).
Please note: Wiley Blackwell are not responsible for the content
or functionality of any supporting information supplied by the
authors. Any queries (other than missing material) should be
directed to the New Phytologist Central Office.

Fig. S8 Structure of CYC2-like genes in representatives of the

subfamily Lamioideae showing the position and sequences of
(putative) introns.

New Phytologist is an electronic (online-only) journal owned by the New Phytologist Trust, a not-for-profit organization dedicated
to the promotion of plant science, facilitating projects from symposia to free access for our Tansley reviews.

Regular papers, Letters, Research reviews, Rapid reports and both Modelling/Theory and Methods papers are encouraged.
We are committed to rapid processing, from online submission through to publication ‘as ready’ via Early View – our average time
to decision is <25 days. There are no page or colour charges and a PDF version will be provided for each article.

The journal is available online at Wiley Online Library. Visit www.newphytologist.com to search the articles and register for table
of contents email alerts.

If you have any questions, do get in touch with Central Office (np-centraloffice@lancaster.ac.uk) or, if it is more convenient,
our USA Office (np-usaoffice@lancaster.ac.uk)

For submission instructions, subscription and all the latest information visit www.newphytologist.com

New Phytologist (2015) 205: 852–868 Ó 2014 The Authors

www.newphytologist.com New Phytologist Ó 2014 New Phytologist Trust